Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Servo-controlled fixed bias
Please,
As I understand it, a common way of doing this is: Detect cathode current with low-value resistor and op-amp. Clip the output of the op-amp to make it's AC component symmetrical. Smooth the result with a filter having a time period of several seconds. Amplify and/or invert this as necessary and use the output voltage to bias the grid. It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian |
#2
|
|||
|
|||
"Ian Iveson" wrote in message ... Please, As I understand it, a common way of doing this is: Detect cathode current with low-value resistor and op-amp. Clip the output of the op-amp to make it's AC component symmetrical. Smooth the result with a filter having a time period of several seconds. Amplify and/or invert this as necessary and use the output voltage to bias the grid. It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. Well, yes and no. The aim is to balance the current through two OP valves across a PP transformer. Not just at idle but at max rated output. Proteus says that all valves are created equal and this is not the case. Even the best 'matched pairs' exhibit slight differences, hence the common practice of seperate bias pots for each valve. The application of the same bias voltage does not automatically result in the same current. I have a 'matched quad' of 6L6GCs. which I can balance perfectly, measuring current, with a DVM across a 10 ohm cathode sense resistor. At perfect current balance conditions, none of the four has the same bias voltage. I have a spread of 1.3V between them. Unfortunately, simulation assumes uniformity and no allowance is made for, manufacturing tolerances. A chap, in Krapograd is nailing up a KT88 as we speak.. Sensing current of one valve, converting it to a voltage and applying it to the others is not the way to go... It's the difference between a PP pair, you need to sense. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. You cannot use one valve as a reference. It could go bad on you and **** the whole lot up. A reference point must be fixed. Two 1% resistors. as a PD, across a voltage and earth. regards jim Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian |
#3
|
|||
|
|||
Ian Iveson wrote:
Please, As I understand it, a common way of doing this is: Detect cathode current with low-value resistor and op-amp. Clip the output of the op-amp to make it's AC component symmetrical. Smooth the result with a filter having a time period of several seconds. Amplify and/or invert this as necessary and use the output voltage to bias the grid. That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book. Kevin O'Connor in his book Principles of Power gives another example. Cheers, John Stewart It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian |
#4
|
|||
|
|||
"John Stewart" wrote in message ... Ian Iveson wrote: Please, As I understand it, a common way of doing this is: Detect cathode current with low-value resistor and op-amp. Clip the output of the op-amp to make it's AC component symmetrical. Smooth the result with a filter having a time period of several seconds. Amplify and/or invert this as necessary and use the output voltage to bias the grid. That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book Sure, it gives the advantages of fixed bias... no FB from a normal CB resistor, no wasted HT, but does nothing to balance the two valves in a PP pair which you cannot do with two simple preset pots. The circuit may provide an accurate reference voltage for the other valve but does not sense it's current. If we assume OP pairs are always perfectly matched, that's OK. They rarely are, however. MJ occasionally makes some terrifying assumptions. If you want to go that way, surely linking the two cathodes via the OP amp to sense the DC difference and ramp the bias on the two valves up or down to achieve DC balance is the answer ? Or perhaps I don't know what I'm talking about, again.. jim Kevin O'Connor in his book Principles of Power gives another example. Cheers, John Stewart It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian |
#5
|
|||
|
|||
"jim" wrote It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. Well, yes and no. The aim is to balance the current through two OP valves across a PP transformer. Not just at idle but at max rated output. Yes, and maintaining the same current through them all acheives that, and more. Proteus says that all valves are created equal and this is not the case. Holy ****! Bugger. Time to get my valve tester out the cellar. Thought I only had to measure one valve. Even the best 'matched pairs' exhibit slight differences, hence the common practice of seperate bias pots for each valve. Right, that's what I use at present. But every now and then they need resetting. Hence the servo wonderings. The application of the same bias voltage does not automatically result in the same current. I have a 'matched quad' of 6L6GCs. which I can balance perfectly, measuring current, with a DVM across a 10 ohm cathode sense resistor. Yup, that's what I do. Common practice. At perfect current balance conditions, none of the four has the same bias voltage. I have a spread of 1.3V between them. That's pretty good, as a proportion of bias voltage, probably. Retailers don't match at your operating point, perhaps? Mine were matched at first, then after a couple of months I had to shuffle them all about to rematch across pairs. Unfortunately, simulation assumes uniformity and no allowance is made for, manufacturing tolerances. Simulation can simulate anything. Spice has limitations, but this is not one of them. It is even possible to do a monte-carlo analysis and get an output distribution showing the range of possible outputs as component parameters wander. I don't bother because there is nothing I can do about it anyway. I do sometimes check sensitivity to mismatching simply by setting the bias pots a little different for each, in simulation. Yes, I even simulate the pots, and can watch the effect of moving the sliders in real time. This is the 21st century, right? Can't quite listen in real time though. Need heaps more processing power for that, and I suspect Spice is batch-processing...can't do input and output at the same time. So I can't make real-time effects pedals, unfortunately. A chap, in Krapograd is nailing up a KT88 as we speak.. Sensing current of one valve, converting it to a voltage and applying it to the others is not the way to go... It's the difference between a PP pair, you need to sense. Yes, as above. If they are both the same set value, then there is no difference. If they are merely the same, they could both run away together. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. You cannot use one valve as a reference. It could go bad on you and **** the whole lot up. A reference point must be fixed. Two 1% resistors. as a PD, across a voltage and earth. Hmm, this is one of the key issues I am trying to get my head around. Check pp293-4 of Morgan Jones. I have more or less paraphrased his description, including the admittedly cryptic bit about the reference. Also, if you look at Norman Koren's Totally Perfect Amplifier, or whatever: You can see that he uses the same kind of description, and uses current mirrors for the purpose, but it gets too complicated for me to follow in one dense diagram. As I first thought, the procedure for setting is that the current through one master valve would be measured whilst fine-setting the voltage reference for that valve. The other valves would be referenced then to the ensuing current, rather than to the voltage reference. As you say, it is the current you wish to replicate, rather than the voltage. Then I wondered how, as the chap fine-setting that voltage reference, you would know which valve to measure the current through? There would be no apparent difference between the master and the others. Then I got hung up on the definition of reference, and whether they can be nested. That's when I thought I should ask... Aside from the confusion, there is a real issue here. As valves age, is it best to change the voltage to maintain current, or the current to maintain voltage, or somewhere in between? This is one of the differences in behaviour between cathode and fixed bias. cheers, Ian |
#6
|
|||
|
|||
Ian Iveson wrote: Please, As I understand it, a common way of doing this is: Detect cathode current with low-value resistor and op-amp. Clip the output of the op-amp to make it's AC component symmetrical. Smooth the result with a filter having a time period of several seconds. For class A circuits, there is little need with such a system when its easy to use cathode bias, and indeed a CCS instead of the Rk, bypassed with a big cap at the cathode. One can use a CCS, and use fixed bias, so only say 15v is across the Ck, and the applied Eg is at say -30, so the total bias is -45v. But away from class A, such a scheme allows Ek drift, and underbiasing. Amplify and/or invert this as necessary and use the output voltage to bias the grid. It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. In class AB amps, the cathode current increases, and any attempt to alter the grid bias with a rise in cathode DC as you suggest results in the grid being sent more negative at the power out increases, and this overbiases the amp into cut off, and has the equivalent effect of having a rising cathode DV when a sine wave is used to test an AB amp with cathode bias up to full power. The whole idea about fixed bias is to fix the electrode voltages, but *allow* the variation in tube DC, which comes in the form of pulses of AC on the +ve half of the cycles. Since tubes are unmatched, using one tube to guide the bias conditions of the others is unreliable. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian I have tried a differential amp using a pair of pnp transitors to make a diff amp, to keep the DC in each side of the PP circuit equal, but it was far more trouble than it was worth, and I got LF oscillations with NFB applied. It was far simpler in a two tube stage to have a bjt diff pair and a couple of leds to monitor the dc balance, and have an adjust pot for the owner to twiddle so that the leds stayed extinguished when dc balance was present. Loud music signals make the DC balance drift considerably, and any attempt to try to counter the drift by altering the grid bias fast enough was unsuccessful. The only simple way I know to have automatic bias adjustment and also to stop the drift in bias, and the upward drift in cathode DV with increasing signals is to use the simple AC bypass circuit I posted recently at ABSE, which uses a couple of power transistors to shunt the +ve signal current peaks, rather than allowing then to charge up the cathode C and cause underbiasing, and a pile more crossover distortions. By dynamically shunting excess cathode current, the electrode voltages are held constant, and circuit works as well as a fixed bias circuit. In class A, the AC shunting circuit virtually does not do anything, since reliance is made on the non linear threshold turn on voltage character of a humble transistor. To sum up, I think servo controlled fixed bias don't work, although the idea has fascinated many even since before they wrote the RDH4, in which there is at least a sample using 807. But afaik, there is no other way to keep the cathode voltage stable whilst the grid bias voltage is fixed, using tubes, ie, if one was to try to do with tubes what I achieved with a pair of garden variety bjts, the circuit would be very complex, expensive, and not really worth all the effort. Am I wrong? Patrick Turner. |
#7
|
|||
|
|||
"Patrick Turner" wrote in message ... Ian Iveson wrote: Please, As I understand it, a common way of doing this is: Detect cathode current with low-value resistor and op-amp. Clip the output of the op-amp to make it's AC component symmetrical. Smooth the result with a filter having a time period of several seconds. For class A circuits, there is little need with such a system when its easy to use cathode bias, and indeed a CCS instead of the Rk, bypassed with a big cap at the cathode. One can use a CCS, and use fixed bias, so only say 15v is across the Ck, and the applied Eg is at say -30, so the total bias is -45v. But away from class A, such a scheme allows Ek drift, and underbiasing. Amplify and/or invert this as necessary and use the output voltage to bias the grid. It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. In class AB amps, the cathode current increases, and any attempt to alter the grid bias with a rise in cathode DC as you suggest results in the grid being sent more negative at the power out increases, and this overbiases the amp into cut off, and has the equivalent effect of having a rising cathode DV when a sine wave is used to test an AB amp with cathode bias up to full power. The whole idea about fixed bias is to fix the electrode voltages, but *allow* the variation in tube DC, which comes in the form of pulses of AC on the +ve half of the cycles. Since tubes are unmatched, using one tube to guide the bias conditions of the others is unreliable. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian I have tried a differential amp using a pair of pnp transitors to make a diff amp, to keep the DC in each side of the PP circuit equal, but it was far more trouble than it was worth, and I got LF oscillations with NFB applied. It was far simpler in a two tube stage to have a bjt diff pair and a couple of leds to monitor the dc balance, and have an adjust pot for the owner to twiddle so that the leds stayed extinguished when dc balance was present. Loud music signals make the DC balance drift considerably, and any attempt to try to counter the drift by altering the grid bias fast enough was unsuccessful. The only simple way I know to have automatic bias adjustment and also to stop the drift in bias, and the upward drift in cathode DV with increasing signals is to use the simple AC bypass circuit I posted recently at ABSE, which uses a couple of power transistors to shunt the +ve signal current peaks, rather than allowing then to charge up the cathode C and cause underbiasing, and a pile more crossover distortions. By dynamically shunting excess cathode current, the electrode voltages are held constant, and circuit works as well as a fixed bias circuit. In class A, the AC shunting circuit virtually does not do anything, since reliance is made on the non linear threshold turn on voltage character of a humble transistor. To sum up, I think servo controlled fixed bias don't work, although the idea has fascinated many even since before they wrote the RDH4, in which there is at least a sample using 807. But afaik, there is no other way to keep the cathode voltage stable whilst the grid bias voltage is fixed, using tubes, ie, if one was to try to do with tubes what I achieved with a pair of garden variety bjts, the circuit would be very complex, expensive, and not really worth all the effort. Am I wrong? Patrick Turner. Two diodes, six resistors, no caps, and my PP, FB, pairs balance and talk to each other, continuously.... jim |
#8
|
|||
|
|||
But afaik, there is no other way to keep the cathode voltage stable whilst the grid bias voltage is fixed, using tubes, ie, if one was to try to do with tubes what I achieved with a pair of garden variety bjts, the circuit would be very complex, expensive, and not really worth all the effort. Am I wrong? Patrick Turner. Two diodes, six resistors, no caps, and my PP, FB, pairs balance and talk to each other, continuously.... jim Ian will probably scold me for hi-jacking the thread, or accuse me of being a colonial bully boy, but I am a how-will-we-do-it man, and perhaps there is some way which has escaped my notice, and it remains to be rediscovered here, or reinvented. All the more modern servo bias control methods I have seen in Wireless World including a couple using both opamps and others using discrete transistors look awfully complex. They probably measure very nicely, but my concerns about changing electrode voltages remain uppermost in my mind. Have you thought of posting your simple method at ABSE? Patrick Turner. |
#9
|
|||
|
|||
"Patrick Turner" wrote in message ... But afaik, there is no other way to keep the cathode voltage stable whilst the grid bias voltage is fixed, using tubes, ie, if one was to try to do with tubes what I achieved with a pair of garden variety bjts, the circuit would be very complex, expensive, and not really worth all the effort. Am I wrong? Patrick Turner. Two diodes, six resistors, no caps, and my PP, FB, pairs balance and talk to each other, continuously.... jim Ian will probably scold me for hi-jacking the thread, or accuse me of being a colonial bully boy, but I am a how-will-we-do-it man, **** off !! You'll have me crying on your shoulder, in a minute Just accept that you are a sheep shagging, incontinent, illeterate, unmarried, underdeveloped, colonial, Ozzie bully boy who knowns nothing, trying to thrust your filthy commercial Turnip Audio down our throats. No tears.. just live with it.... No, I don't want one of your ****in' OPTs !! Dave does mine and he's British !!! .... British Steel.... Huddersfield... My GOSS is personally, hand orientated by a chap called Norman. and perhaps there is some way which has escaped my notice, and it remains to be rediscovered here, or reinvented. Nothing is new. It's all in RDH4 or sitting under your nose, laughing at you. We're all quite small, nobody should be looking for a fight and nothing is very much bigger than a can of Fosters.. All the more modern servo bias control methods I have seen in Wireless World including a couple using both opamps and others using discrete transistors look awfully complex. They probably measure very nicely, but my concerns about changing electrode voltages remain uppermost in my mind. Have you thought of posting your simple method at ABSE? No, the clue is in the 0.6v drop across a diode, times two and a 4:1 low resistance divider across the cathodes. It may be a happy coincidence, not a proof. It may just relate to the big BTs I have, on hand. Fleming discovered penicillin. He could have just got a dirty plate/dish/bowl/anode.. depending on where you live.. Somebody will tell me I am a **** artist and I have a bathroom to plaster tomorrow. I am simple enough to be clever but not clever enough to be certain. If I have anything to contribute to this NG it is .... 50W..... That's 20 resistors and 4 caps... You got more than that .... you got it wrong jim Patrick Turner. |
#10
|
|||
|
|||
Have you thought of posting your simple method at ABSE? No, the clue is in the 0.6v drop across a diode, times two and a 4:1 low resistance divider across the cathodes. It may be a happy coincidence, not a proof. It may just relate to the big BTs I have, on hand. Fleming discovered penicillin. He could have just got a dirty plate/dish/bowl/anode.. depending on where you live.. Somebody will tell me I am a **** artist and I have a bathroom to plaster tomorrow. I am simple enough to be clever but not clever enough to be certain. If I have anything to contribute to this NG it is .... 50W..... That's 20 resistors and 4 caps... You got more than that .... you got it wrong jim Patrick Turner. Without a schematic of wht the hell you are talking about, I am lost. Patrick Turner. |
#11
|
|||
|
|||
"John Stewart" wrote
That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book. Kevin O'Connor in his book Principles of Power gives another example Thanks John. I have Jones but no O'Conner. I forgot to include the link to Koren's work. It is buried now in a very impressive photography site. http://www.normankoren.com/Audio/TENA.html Last time I looked I don't think he had actually built the amp. Now he has. Worth a look. cheers, Ian |
#12
|
|||
|
|||
"Patrick Turner" wrote
Am I wrong? Of course. Functionally, the approaches are identical. Perhaps you didn't see the significance of clipping the signal from the current sensing resistor, before integrating? That will allow the bias to respond, slowly, to 2H up to a point, but beyond that ignore it. Just like "yours". But, sensibly, control via the grid requires hardly any current, so the servo circuit dissipates little power and long time periods can be achieved with small components. Neither is it in the signal path, like "yours" is. No kind of servo bias is of much interest to amateur amp builders, or to anyone who takes enough pride in their amps to check the fixed bias from time to time. Unless they like building control systems for fun. Anyone using cathode bias presumably likes the way it sounds. Making it sound like fixed bias seems daft. Just use fixed bias if that's what you want it to sound like. For the convenience expected in a commercial amp, servo bias should allow all valves to be automatically set equal with one simple adjustment. Much more convenient for those who like to experiment with different valves. Check your copy of Morgan Jones, BTW, I may have explained it badly. cheers, Ian For class A circuits, there is little need with such a system when its easy to use cathode bias, and indeed a CCS instead of the Rk, bypassed with a big cap at the cathode. One can use a CCS, and use fixed bias, so only say 15v is across the Ck, and the applied Eg is at say -30, so the total bias is -45v. But away from class A, such a scheme allows Ek drift, and underbiasing. Amplify and/or invert this as necessary and use the output voltage to bias the grid. It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. In class AB amps, the cathode current increases, and any attempt to alter the grid bias with a rise in cathode DC as you suggest results in the grid being sent more negative at the power out increases, and this overbiases the amp into cut off, and has the equivalent effect of having a rising cathode DV when a sine wave is used to test an AB amp with cathode bias up to full power. The whole idea about fixed bias is to fix the electrode voltages, but *allow* the variation in tube DC, which comes in the form of pulses of AC on the +ve half of the cycles. Since tubes are unmatched, using one tube to guide the bias conditions of the others is unreliable. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian I have tried a differential amp using a pair of pnp transitors to make a diff amp, to keep the DC in each side of the PP circuit equal, but it was far more trouble than it was worth, and I got LF oscillations with NFB applied. It was far simpler in a two tube stage to have a bjt diff pair and a couple of leds to monitor the dc balance, and have an adjust pot for the owner to twiddle so that the leds stayed extinguished when dc balance was present. Loud music signals make the DC balance drift considerably, and any attempt to try to counter the drift by altering the grid bias fast enough was unsuccessful. The only simple way I know to have automatic bias adjustment and also to stop the drift in bias, and the upward drift in cathode DV with increasing signals is to use the simple AC bypass circuit I posted recently at ABSE, which uses a couple of power transistors to shunt the +ve signal current peaks, rather than allowing then to charge up the cathode C and cause underbiasing, and a pile more crossover distortions. By dynamically shunting excess cathode current, the electrode voltages are held constant, and circuit works as well as a fixed bias circuit. In class A, the AC shunting circuit virtually does not do anything, since reliance is made on the non linear threshold turn on voltage character of a humble transistor. To sum up, I think servo controlled fixed bias don't work, although the idea has fascinated many even since before they wrote the RDH4, in which there is at least a sample using 807. But afaik, there is no other way to keep the cathode voltage stable whilst the grid bias voltage is fixed, using tubes, ie, if one was to try to do with tubes what I achieved with a pair of garden variety bjts, the circuit would be very complex, expensive, and not really worth all the effort. Patrick Turner. |
#13
|
|||
|
|||
Ian Iveson wrote:
"John Stewart" wrote That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book. Kevin O'Connor in his book Principles of Power gives another example Thanks John. I have Jones but no O'Conner. I forgot to include the link to Koren's work. It is buried now in a very impressive photography site. http://www.normankoren.com/Audio/TENA.html Last time I looked I don't think he had actually built the amp. Now he has. Worth a look. cheers, Ian Very Impressive, but unaffordable (& too complicated) for most. But give Norman an A+ for his attention to detail. And those Plitron Transformers are probably 63% of the reason the amp works so well. Plitron is built here locally not far from where I live. I met some of the guys there since they are using some of the equipment that we rep'ed. Lot's of stuff from Voltech (UK). But the Voltech development lab is now on Florida. Some of there products are built in Germany & then private branded with the Voltech logo. Cheers, JLS |
#14
|
|||
|
|||
Ian Iveson wrote: "John Stewart" wrote That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book. Kevin O'Connor in his book Principles of Power gives another example Thanks John. I have Jones but no O'Conner. I forgot to include the link to Koren's work. It is buried now in a very impressive photography site. http://www.normankoren.com/Audio/TENA.html Last time I looked I don't think he had actually built the amp. Now he has. Worth a look. From a look at the very-hard-to-understand schematic for the amp which he seems to have built, he uses the DC cathode current of the OP tubes to adjust the applied grid voltage to the CF driver to his class AB2 triode output stage, so that the cathode current will stay constant. Unfortunately, when tested with a sine wave to full power, this results in a displacement of the correct grid biasing point which should remain fixed regardless of the total DC current in the output tubes. IMHO, they guy is making a big mistake, and he doesn't understand why fixed bias means fixed bias. Am I wrong? RDH4 says fiixed bias is used precisely to get the low thd benefits of keeping electrode voltage fixed during operation, not variable according to the fluctuating DC demands of a class AB output stage. So, Mr Koren's amp operates no better than a fully cathode biased amp. In the latest version of my 300 watt mono amps with 12 x 6550, I have each output tube biased seperately with a 500 ohm Rk, byapassed with Panasonic 1,000 uF, so there is 6,000 uF per half to bypass the cathodes on each 1/2 of the OP stage. I use an "acoustical" output stage, with 20% of CFB, B+ = +526v, Eg2 = +400v, and Ek is at +23v, and a fixed bias of -17v drived from the rectified 12.6v heater supply is applied to all output tubes via separate RC grid couplings. The 500ohm Rk is sufficient to give good regulation to the Ik, and only 1 watt is dissipated in each Rk. The po is mainly class A into 8 ohms, and little Ek variation occurs, but with 3 ohms, without having AC dynamic shunting of the rectifying currents flowing into the 1,000 uF caps, the Ek would rise from 23 v at idle to 45v, with a sine wave test, and the thd would increase tenfold, even with the total of 14 dB of NFB which exists with a 3 ohm load. Fixed bias amps just don't have this problem with sine wave tests to full power. Mr Koren doesn't explain what does happen to the applied grid voltage with a sine wave test to full power. With the AC dynamic shunting, the Ek rises only a couple of volts with clipping into 3 ohms, which is at around 320 watts. There is a small sag in the PS, so max instantaneous peak power is about 350w. The undulation of cathode current one gets with normal cathode biased amps must contribute to some thd, beacuse the unbalanced DC from each 1/2 of the PP circuit means the core becomes DC biased one way then the other, ie, it lurches around, and affects the sound badly. This is mainly an effect at high levels of operation, where clipping in the music is just reached. With AC shunting I use, the cathode caps are protected from getting gobs of pulses of signal which would otherwise shift the Ek, and the bias. Its to be remembered that the cathode cap of 6,000 uF offers an impedance of only 0.53 ohms at 50 Hz, and any AC shunting to reduce the charge up effect of Ck has no effect on the voltage conditions in the amp, and makes it more like a true fixed bias amp, but in my case, without a dozen somewhat interactive bias pot adjustments to worry over. Just my 2c worth. Patrick Turner. cheers, Ian |
#15
|
|||
|
|||
Ian Iveson wrote: "Patrick Turner" wrote Am I wrong? Of course. Functionally, the approaches are identical. Perhaps you didn't see the significance of clipping the signal from the current sensing resistor, before integrating? That will allow the bias to respond, slowly, to 2H up to a point, but beyond that ignore it. Just like "yours". But, sensibly, control via the grid requires hardly any current, so the servo circuit dissipates little power and long time periods can be achieved with small components. Neither is it in the signal path, like "yours" is. I think the bias would still be made to drift. Once class AB operation is underway, clipping the voltage produced by the small Rk means the bias is adjusted to whatever the diode clipper forward voltage is, and all control of the bias is lost. But at least it is there for idle conditions, and for a major part of the amps po, the "sweet zone" class A portion, where lurches of the bias are not wanted. Where the bias isn't altered when class AB is entered, then I have no problems with the method, except to say that there really should be some bias manager circuit at each cathode of every output tube, which is possible but complex, if done with opamps and RC components, and a mosfet. I'd think that a couple of bjts would be easier, there is no need to have an opamp, with its huge DC open loop gain. A gain of 1,000 would be fine. In my circuit for 12 output tubes, there are only 2 power transistor "AB current shunters" since each cathode has a diode going off to a common point which has a 15 ohm R to the collector of a bjt, similar to my posted ciruit at ABSE. No kind of servo bias is of much interest to amateur amp builders, or to anyone who takes enough pride in their amps to check the fixed bias from time to time. Unless they like building control systems for fun. Well, why try to build bias servo circuits? DIYers are able to offer perfection to themselves. The commercial operators don't always manage to offer easy to work on, reliable components, relaible rugged bias adjustments, and usually SFA active protection against excessive cathode currents, or any indication of the bias condition. A couple of Wireless World schematics allowed the *balance* of dc in each half of the opt to be held constant by the monitoring of the Ik in each 1/2 of the PP amp. The fuctuation of Ik was then applied to the grids of the tubes to counter the swing in Ik. The common mode rise in both Ik did not affect the diff amp doing the balancing. A very simple two transistor LTP using say a pair of MJE350, pnp, can be set up to achieve this, but there are delays in the RC filtering, but at least it works well at low power. Once clipping occurs, wild corrections take place, and possible LF instabilities. Anyone using cathode bias presumably likes the way it sounds. Making it sound like fixed bias seems daft. Just use fixed bias if that's what you want it to sound like. A properly cathode biased AB amp sound identical to a fixed bias amp. I have tried both, using the same operating conditions, tubes, driver, opt, etc, and since most of the sound if not all is produced by class A, there is zero detectable difference. But for folks who want grunt, then something can be done. For the convenience expected in a commercial amp, servo bias should allow all valves to be automatically set equal with one simple adjustment. Much more convenient for those who like to experiment with different valves. But if one tube out of say 4 changes its condition, and its bias is set from a signal from another tube, you have no servo bias for the rogue tube. The idea that I would have the biasing of 6 of the output tubes based on the cathode current of just 1 of 6 of them seems plain balmy. One may as well just have the same fixed grid bias applied to all tubes, which wasn't uncommon, and all biased conservatively to allow for some to drift up and down say 20% since tubes don't saty matched after you buy a matched set. I like separate cathode bias, just plug in, and forget. KT88, 6550, KT90, KT66, 6L6GC, all can be plugged into the same amp. EL34 is the odd man out. Check your copy of Morgan Jones, BTW, I may have explained it badly. I will have a look at that, to find out for myself what he says. Patrick Turner. cheers, Ian For class A circuits, there is little need with such a system when its easy to use cathode bias, and indeed a CCS instead of the Rk, bypassed with a big cap at the cathode. One can use a CCS, and use fixed bias, so only say 15v is across the Ck, and the applied Eg is at say -30, so the total bias is -45v. But away from class A, such a scheme allows Ek drift, and underbiasing. Amplify and/or invert this as necessary and use the output voltage to bias the grid. It seems to makes sense to use the bias current of one master valve as a reference for the others, in multi-valve output stages. In class AB amps, the cathode current increases, and any attempt to alter the grid bias with a rise in cathode DC as you suggest results in the grid being sent more negative at the power out increases, and this overbiases the amp into cut off, and has the equivalent effect of having a rising cathode DV when a sine wave is used to test an AB amp with cathode bias up to full power. The whole idea about fixed bias is to fix the electrode voltages, but *allow* the variation in tube DC, which comes in the form of pulses of AC on the +ve half of the cycles. Since tubes are unmatched, using one tube to guide the bias conditions of the others is unreliable. The parts of the circuit I am particularly interested in are the clipping and the details of how one valve is used as a reference for the rest. The purpose of the clipping, BTW, is to stop the servo from responding to asymmetrical current peaks as the valve itself approaches clipping. Otherwise the bias would drift in response to extended AB operation. Anyone with experience, circuits, thoughts? A better description of how it works? I would appreciate a discussion. cheers, Ian I have tried a differential amp using a pair of pnp transitors to make a diff amp, to keep the DC in each side of the PP circuit equal, but it was far more trouble than it was worth, and I got LF oscillations with NFB applied. It was far simpler in a two tube stage to have a bjt diff pair and a couple of leds to monitor the dc balance, and have an adjust pot for the owner to twiddle so that the leds stayed extinguished when dc balance was present. Loud music signals make the DC balance drift considerably, and any attempt to try to counter the drift by altering the grid bias fast enough was unsuccessful. The only simple way I know to have automatic bias adjustment and also to stop the drift in bias, and the upward drift in cathode DV with increasing signals is to use the simple AC bypass circuit I posted recently at ABSE, which uses a couple of power transistors to shunt the +ve signal current peaks, rather than allowing then to charge up the cathode C and cause underbiasing, and a pile more crossover distortions. By dynamically shunting excess cathode current, the electrode voltages are held constant, and circuit works as well as a fixed bias circuit. In class A, the AC shunting circuit virtually does not do anything, since reliance is made on the non linear threshold turn on voltage character of a humble transistor. To sum up, I think servo controlled fixed bias don't work, although the idea has fascinated many even since before they wrote the RDH4, in which there is at least a sample using 807. But afaik, there is no other way to keep the cathode voltage stable whilst the grid bias voltage is fixed, using tubes, ie, if one was to try to do with tubes what I achieved with a pair of garden variety bjts, the circuit would be very complex, expensive, and not really worth all the effort. Patrick Turner. |
#16
|
|||
|
|||
Ian Iveson wrote: That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book. I have MJ's 2nd edtion book, which don't seem to have anything on servo bias. Patrick Turner. |
#17
|
|||
|
|||
"Patrick Turner" wrote in message ... From a look at the very-hard-to-understand schematic for the amp which he seems to have built, he uses the DC cathode current of the OP tubes to adjust the applied grid voltage to the CF driver to his class AB2 triode output stage, so that the cathode current will stay constant. Unfortunately, when tested with a sine wave to full power, this results in a displacement of the correct grid biasing point which should remain fixed regardless of the total DC current in the output tubes. IMHO, they guy is making a big mistake, and he doesn't understand why fixed bias means fixed bias. Am I wrong? Hi Patrick, Actually, I believe you are mistaken. The author states in the text that the three servo amplifiers match the DC currents of TU10, TU11, and TU12, to the DC current of TU9, which is operating in pure fixed bias. Although the schematic is quite confusing at first glance, you can see the simple fixed-bias adjustment in the bottom center. The servo is defined by three time constants, the 330s from the cathode of TU9, the 33s in the feedback of the servo amp, and the 109s on the output of the servo amp. The real question about operating point shift under high power operation revolves around the relationship of these time constants, and actually here is where the mistake lies. I would think that the time constants for the reference voltage and the error voltage for each servo amp should be the same . . . hence, the value of CBS2, 4, and 6 should be 3.3uF. If this adjustment is made, any long-term change in operating point would result in a symmetrical change in both the error voltage and reference voltage on each servo amp, keeping the operating point of their respective tubes in virtual fixed-bias operation. While I don't have an opinion on the benefits of this scheme vis-a-vis your modified cathode-bias approach (I haven't tested or listened to either), I think that there is a degree of elegance to Mr. Koren's design (and yours too) . . . the main flaw in this one being that a runaway current in TU9 will cause all of them to run away, while this isn't the case in your approach . . . but the normal methods of protecting against a lost bias supply will probably suffice in keeping anything too drastic from happening. Regards, Kirk Patton |
#18
|
|||
|
|||
"Patrick Turner" wrote
Ian Iveson wrote: That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book. I have MJ's 2nd edtion book, which don't seem to have anything on servo bias. That quote is from John, Patrick. He has the 1st edition. In second edition, see diagram p294 and text somewhere close. cheers, Ian |
#19
|
|||
|
|||
Kirk Patton wrote: "Patrick Turner" wrote in message ... From a look at the very-hard-to-understand schematic for the amp which he seems to have built, he uses the DC cathode current of the OP tubes to adjust the applied grid voltage to the CF driver to his class AB2 triode output stage, so that the cathode current will stay constant. Unfortunately, when tested with a sine wave to full power, this results in a displacement of the correct grid biasing point which should remain fixed regardless of the total DC current in the output tubes. IMHO, they guy is making a big mistake, and he doesn't understand why fixed bias means fixed bias. Am I wrong? Hi Patrick, Actually, I believe you are mistaken. The author states in the text that the three servo amplifiers match the DC currents of TU10, TU11, and TU12, to the DC current of TU9, which is operating in pure fixed bias. Although the schematic is quite confusing at first glance, you can see the simple fixed-bias adjustment in the bottom center. Perhaps I need to read it more carefully. The servo is defined by three time constants, the 330s from the cathode of TU9, the 33s in the feedback of the servo amp, and the 109s on the output of the servo amp. The real question about operating point shift under high power operation revolves around the relationship of these time constants, and actually here is where the mistake lies. I would think that the time constants for the reference voltage and the error voltage for each servo amp should be the same . . . hence, the value of CBS2, 4, and 6 should be 3.3uF. If this adjustment is made, any long-term change in operating point would result in a symmetrical change in both the error voltage and reference voltage on each servo amp, keeping the operating point of their respective tubes in virtual fixed-bias operation. That critique *will* take some digestion. While I don't have an opinion on the benefits of this scheme vis-a-vis your modified cathode-bias approach (I haven't tested or listened to either), I think that there is a degree of elegance to Mr. Koren's design (and yours too) . . . the main flaw in this one being that a runaway current in TU9 will cause all of them to run away, while this isn't the case in your approach . . . but the normal methods of protecting against a lost bias supply will probably suffice in keeping anything too drastic from happening. Regards, Kirk Patton Mr Koren should have simply tested the amp with a sine wave up to clipping, and a bit extra, and plotted the alterations to the applied grid biasing of the output stage. The voltages should stay about the same, the DC through each output tube should be the same up to clipping. Beyond steady state clipping, servo bias circuits can go wild, as they vainly try to correct the asymetrical DV conditions around the output stage and coupling caps. Meanwhile I have posted two schematics of servo bias ideas and comments at ABSE, one for a single tube, and one for a PP a pair of tubes. Patrick Turner. |
#20
|
|||
|
|||
Ian Iveson wrote: "Patrick Turner" wrote Ian Iveson wrote: That is how Morgan Jones does it in his book Valve Amplifiers. Refer to page 227 in the first edition of the book. I have MJ's 2nd edtion book, which don't seem to have anything on servo bias. That quote is from John, Patrick. He has the 1st edition. In second edition, see diagram p294 and text somewhere close. cheers, Ian Thanks Ian, Unfortunately, Morgan didn't include "servo bias" in his index. When I went to my book, I had underlined the part on pg 293 about clipping some of the class B cycle, so I must have red it a couple of years back. The auto setting of one tube, done so as to ignore the class B upward drift in DC in an amp isn't really addressed in the somewhat complex array around two opamps on pg 294. Using one tube assess the current and apply a correction to the others would only make sense to me if the other tubes all had their DC compared to the one used to set the lot, and then of course different Eg1 would have to be applied to each tube to achieve DC equality at all times. I don't think Morgan Jones really addresses the problems of servo bias in tube amps, and from waht I can see there is nothing that I would want to implement in my amps. Some other easier way should be found to simply limit the positive going current swings in the current sensing Rk, so that the simple two transistor circuit for an SE tube, like I have posted at ABSE, does not get a high signal applied to the bases of the transistors, therby preventing an unecessary high swing of the Eg1 to a low -ve voltage, and causing cut off, and paralysis, after the transients causing the DC drift have long gone. The way I do my circuit means the excess catde current is simply allowed to pass via a transistor and low R, so the Ek stays put, and there are no time lags or delays in the circuit, ie, the servo circuit isn't still trying to correct the Ia to correct value, years after something makes it drift. I might add that my circuit with a transistor current shunter can allso incorporate a small 1:1 transformer working off the 10 ohm Rk I have, to switch on the bjt, so that any large rise in the Ik DC does not try to turn on the bjt. The tranny method is the most elegant, and I tried a couple of tiny 600 : 600 ohm telphone trannies, but it wasn't successful, because a 50 : 50 ohm is really required, ie larger core, fewer turns, and thicker wire. Patrick Turner. |
#21
|
|||
|
|||
"Patrick Turner" wrote in message ... The way I do my circuit means the excess catde current is simply allowed to pass via a transistor and low R, so the Ek stays put, and there are no time lags or delays in the circuit, ie, the servo circuit isn't still trying to correct the Ia to correct value, years after something makes it drift. ** Shame the component values you gave do not allow the transistor to actually do what you say. ......... Phil |
#22
|
|||
|
|||
Phil Allison wrote: "Patrick Turner" wrote in message ... The way I do my circuit means the excess catde current is simply allowed to pass via a transistor and low R, so the Ek stays put, and there are no time lags or delays in the circuit, ie, the servo circuit isn't still trying to correct the Ia to correct value, years after something makes it drift. ** Shame the component values you gave do not allow the transistor to actually do what you say. ........ Phil Shame you ain't built the circuit to find out how well it works. This our Phil the Dill, who should know better. Patrick Turner. |
#23
|
|||
|
|||
"Patrick Turner" wrote in message ... Phil Allison wrote: "Patrick Turner" wrote in message ... The way I do my circuit means the excess catde current is simply allowed to pass via a transistor and low R, so the Ek stays put, and there are no time lags or delays in the circuit, ie, the servo circuit isn't still trying to correct the Ia to correct value, years after something makes it drift. ** Shame the component values you gave do not allow the transistor to actually do what you say. ........ Phil Shame you ain't built the circuit to find out how well it works. ** It looked suspicious so I did. Bad idea to post wild guesses as like that - Turner. ............. Phil |
#24
|
|||
|
|||
Shame you ain't built the circuit to find out how well it works. ** It looked suspicious so I did. So you had doubts about what you said last week, ie, that it couldn't work, and now you have tried the idea, you must have been able to get it to work. Bad idea to post wild guesses as like that - Turner. ............ Phil But after posting weekfuls of lies about me, and since I refused to agree with all your BS, I fully expect you to try to discredit me over this issue. Patrick Turner. |
#25
|
|||
|
|||
"Patrick Turner" wrote in message ... Shame you ain't built the circuit to find out how well it works. ** It looked suspicious so I did. So you had doubts about what you said last week, ** I tried the cct BEFORE posting any comments about it. ie, that it couldn't work, ** They are YOUR words - not mine. I said it failed to meet a major claim as published. But after posting weekfuls of lies about me, ** Not one - not ever. The libel is all coming form your poisoned, crippled mind. ........... Phil |
#26
|
|||
|
|||
Lat week I posted a schematic at ABSE for AC shunt regulation
of cathode voltage in cathode bias class AB power amps. I explained how it worked, and like all things, slightly altering the resistor values will achive the optimum regulation of the cathode bias regardless of load or power levels, right up to clipping, even when using a sine wave signal. Phil then exoressed serious doubts that it could work, and carried on stupidly about me being a charatan, liar, public menace, and all the other BS he goes on with, to try to escape my insistance that he proove to all here that my circuit does not achive what it sets out to do. I then wrote later:- Shame you ain't built the circuit to find out how well it works. He recently replied:- ** It looked suspicious so I did. I then said:- So you had doubts about what you said last week, And Phil had deleted the rest of the sentence, all of which was:- So you had doubts about what you said last week, ie, that it couldn't work, and now you have tried the idea, you must have been able to get it to work. Now he says :- ** I tried the cct BEFORE posting any comments about it. Then he includes a snippet of mine, ie, that it couldn't work, And then he denies he didn't say the circuit won't work:- ** They are YOUR words - not mine. And he tries to water down his stance that a silly old "bricklayer" has dreamed up another BS idea with :- I said it failed to meet a major claim as published. And my comment was :- But after posting weekfuls of lies about me, To which he replies :- ** Not one - not ever. The libel is all coming form your poisoned, crippled mind. And we all know who did all the lying last week, now don't we. If Phil expects to gain my respect, instead of me telling him to **** off, he should at least be a little more thorough and detailed, without all the BS, ie claims and names, all ficticious, all the vicious work of a dumb brain, depite uni education, and 30 years in the electronics industry. When asked prior to today if he tried my circuit before condemning it, he never said a word, so I assume he hasn't ever tried it, and still wants to maintain his negativity to the idea. I have nothing against the doubters and the skeptics here. I don't wish to proove the circuit works with much more info that I have extensively posted already. But after having used the circuit in a two very different class AB cathode biased amps, one of 20 watts, and the other of 300 watts, I know what can be achieved by active regulation of the cathode bias, and ppl are welcone to try the idea. Some fiddling of the base resistors and collector resistor values are required, depending on the bjt chosen to do the signal current shunting around the cathode bypass cap. It is to be noted that Phil has not answered any of the details of the circuit description, and so so far all of what I can get out of him is sheer BS couched in insults, again. Patrick Turner. |
#27
|
|||
|
|||
"Patrick Turner" Lat week I posted a schematic at ABSE for AC shunt regulation of cathode voltage in cathode bias class AB power amps. I explained how it worked, ** Wrong - you dribbled on about how you wished it worked. Phil then exoressed serious doubts that it could work, ** Post the actual words Turner - your re-phrasing is a evil thing. and carried on stupidly about me being a charatan, liar, public menace, ** But you are all those things Turner - and a malicious turd as well. and all the other BS he goes on with, to try to escape my insistence that he proove to all here that my circuit does not achive what it sets out to do. ** Those making claims get to prove them. I then wrote later:- Shame you ain't built the circuit to find out how well it works. He recently replied:- ** It looked suspicious so I did. I then said:- So you had doubts about what you said last week, And Phil had deleted the rest of the sentence, all of which was:- So you had doubts about what you said last week, ie, that it couldn't work, and now you have tried the idea, you must have been able to get it to work. Now he says :- ** I tried the cct BEFORE posting any comments about it. Then he includes a snippet of mine, ie, that it couldn't work, And then he denies he didn't say the circuit won't work:- ** They are YOUR words - not mine. And he tries to water down his stance that a silly old "bricklayer" has dreamed up another BS idea with :- I said it failed to meet a major claim as published. And my comment was :- But after posting weekfuls of lies about me, To which he replies :- ** Not one - not ever. The libel is all coming form your poisoned, crippled mind. And we all know who did all the lying last week, now don't we. ** Yes we do - it was you Turner. You WILL NOT POST UNDER MY WORDS !!!!!! You insist on removing everything from context and re-phrasing it to suit yourself. THAT is LYING !!!!! If Phil expects to gain my respect, ** ROTFLMAO - I would prefer you drop dead ASAP. When asked prior to today if he tried my circuit before condemning it, he never said a word, ** Post the words, I did not see any such question directed to me. Some fiddling of the base resistors and collector resistor values are required, depending on the bjt chosen to do the signal current shunting around the cathode bypass cap. ** Which is the "base resistor" ? It is to be noted that Phil has not answered any of the details of the circuit description, ** There **was** no damn detailed description of the operation of the schematic that was posted. Time for you to do one Turner. ......... Phil |
#28
|
|||
|
|||
Phil replied to my post below
re a cathode bias reg schematic with a posting that I was a weasel, and that all I said was wrong, etc, etc, etc. Unfortunately, for the calm rational minds here, who search daily for something of worth to read and discuss, Phil omits to say anything technical or otherwise that refutes my stance. All he can say in response of my explanations is :- ** Wrong - you dribbled on about how you wished it worked. I doubt he would have lowered himself to ever actually building the circuit, let alone actually conduct some real tests on it, along with optimising the function to get the best from the application of the idea. Patrick Turner. Patrick Turner wrote: Lat week I posted a schematic at ABSE for AC shunt regulation of cathode voltage in cathode bias class AB power amps. I explained how it worked, and like all things, slightly altering the resistor values will achive the optimum regulation of the cathode bias regardless of load or power levels, right up to clipping, even when using a sine wave signal. Phil then exoressed serious doubts that it could work, and carried on stupidly about me being a charatan, liar, public menace, and all the other BS he goes on with, to try to escape my insistance that he proove to all here that my circuit does not achive what it sets out to do. I then wrote later:- Shame you ain't built the circuit to find out how well it works. He recently replied:- ** It looked suspicious so I did. I then said:- So you had doubts about what you said last week, And Phil had deleted the rest of the sentence, all of which was:- So you had doubts about what you said last week, ie, that it couldn't work, and now you have tried the idea, you must have been able to get it to work. Now he says :- ** I tried the cct BEFORE posting any comments about it. Then he includes a snippet of mine, ie, that it couldn't work, And then he denies he didn't say the circuit won't work:- ** They are YOUR words - not mine. And he tries to water down his stance that a silly old "bricklayer" has dreamed up another BS idea with :- I said it failed to meet a major claim as published. And my comment was :- But after posting weekfuls of lies about me, To which he replies :- ** Not one - not ever. The libel is all coming form your poisoned, crippled mind. And we all know who did all the lying last week, now don't we. If Phil expects to gain my respect, instead of me telling him to **** off, he should at least be a little more thorough and detailed, without all the BS, ie claims and names, all ficticious, all the vicious work of a dumb brain, depite uni education, and 30 years in the electronics industry. When asked prior to today if he tried my circuit before condemning it, he never said a word, so I assume he hasn't ever tried it, and still wants to maintain his negativity to the idea. I have nothing against the doubters and the skeptics here. I don't wish to proove the circuit works with much more info that I have extensively posted already. But after having used the circuit in a two very different class AB cathode biased amps, one of 20 watts, and the other of 300 watts, I know what can be achieved by active regulation of the cathode bias, and ppl are welcone to try the idea. Some fiddling of the base resistors and collector resistor values are required, depending on the bjt chosen to do the signal current shunting around the cathode bypass cap. It is to be noted that Phil has not answered any of the details of the circuit description, and so so far all of what I can get out of him is sheer BS couched in insults, again. Patrick Turner. |
#29
|
|||
|
|||
"Patrick Turner" Phil A correctly said: ** Wrong - you dribbled on about how you wished it worked. I doubt he would have lowered himself to ever actually building the circuit, let alone actually conduct some real tests on it, along with optimising the function to get the best from the application of the idea. ** Man with no case invents facts to suit his non existant case. Weasel, weasel, weasel....... .......... Phil |
#30
|
|||
|
|||
The REAL weasel here is of course Phil Allison, who has not a shred of evidence that my circuit does not work as described. So Phil, **** Off Patrick Turner. ** Man with no case invents facts to suit his non existant case. Weasel, weasel, weasel....... ......... Phil |
#31
|
|||
|
|||
"Patrick Turner" wrote in message ... The REAL weasel here is of course Phil Allison, who has not a shred of evidence that my circuit does not work as described. ** Post a description that INCLUDES a current flow analysis - ****head. I tried the thing as shown and it FAILED to regulate the cathode voltage on sine waves. Anyone can verify that themselves. Anyone using basic electronic arithmetic can soon see why. ......... Phil |
#32
|
|||
|
|||
Under the banner of the insulting header, "Turner is a weasel"
Phil expects me to provide a current flow chart/diagram to prove the cathode voltage regulation circuit for an AB tube amp works. He demands, in filthy language :- ** Post a description that INCLUDES a current flow analysis - ****head. I tried the thing as shown and it FAILED to regulate the cathode voltage on sine waves. Anyone can verify that themselves. Anyone using basic electronic arithmetic can soon see why. Ah, so if the arithmetic is wrong, it cannot work. Well, I guess its his word against mine. But still he persists in providing NOT A SHRED of evidence that it cannot work, and that it cannot regulate the cathode voltage I don't need arithmetic to proove anything. Let me explain yet again, for the benefit of those in doubt of my circuit viability, and lets leave the arithmetic right out of this because nobody would understand, except mathematicians at a university, who don't build tube amps. Let's consider a PP amp with 2 x KT88 in UL, with an 8k to 8 ohm OPT, and with cathode bias, tested with a sine wave. The power output will be pure class A to clipping with a load that is slightly above say 8k a-a, maybe 30 watts, and the cathode bias voltages will not drift upwards by more than 10% up to the onset of clipping. There is not need for bias regulation in such an amp, or any servo bias circuit. The wave form of the current flowing into the cathode caps is substantially a sine wave, with some 2H distortion current, and because the positive going cycles rise higher than the -ve current cycles, the DV across the Ck rises slightly. But as RL a-a is reduced below 8k a-a, the amp starts producing class AB power, and the current flow shows that the -ve part of the current cycle in each tube is cut off, and as RL is reduced to say 3k a-a, the amp is nearly working in class B, with a small class A content. The average of the positive peak currents are far higher than the idle current, and under thse tests, the cathode voltage will move from say +55v to say +70v, at clipping, which has the effect of underbiasing the tubes, and instead of getting say 1% of thd, even with 15 dB of NFB, you will get maybe 8% at clipping. But with music, the actual cathode voltage does not move much at below 5 watts into 3k a-a, but as the music starts to clip, the DC flow in the tubes begins to fluctuate up and down and the amp sustains considerable distortion, both by mis biasing, and DC offset swings in the OPT. Now if our amp had been made using fixed bias, none of the cathode swings in voltage and DC occur, because the cathode caps are not in series with the "rectifying" cathode signal currents. But cathoed bias is a winner for owners, since not bias adjustments are required, and there is nothing for an owner to get confused with, of make mistakes with, and I assure you owners, or their sons, twiddle bias pots all wrong sometimes, with smoky results. It came into my mind that none of the bias servo circuits ever really worked. I then thought that if the part of the AB cathode wave form which exceeds the value of twice the idle current was able to be dynamically shunted to 0V, or around the power circuit, that the cathode caps wouldn't change their voltage. Whilst in class A, with a sine wave input signal, the cathode current increases to about twice the idle current, and then reduces to almost no current, on each wave cycle, so no change to Ek occurs. But once the +ve current cycle exceeds the -ve current cycle, the tube is placing a rectifying current into its Ck, and the DV rises. Now in my circuit, the 1,000 uF Ck feeds a 10 ohm current sensing R and this is where you may view the AB signal Ik, whiuch is like a sine wave, but with most of the -ve bottoms cut off, just under clipping. So in the case of a 3k a-a load, the class B load is 750 ohms for each tube, and if the anode voltage swing was 210 vrms, or 300 peak volt, so we would have a peak cathode current of 400 mA. This produces a peak voltage of +4 volts at the 10 ohm resistor, unless we were able to somehow shunt all the current above a threshold of say +1v because the idle voltage value is +0.5v because of the 50 mA of idle tube current. To achieve dynamic re-routing of the excess AB signal I, the 1,000 uF Ck is used as a coupling cap to a base of a bjt with a series R to limit the peak base I and prevent the bjt from failure. The emitter is at 0V, and the collector has a current limiting R to the cathode, which resides at +55v. During the AB wave cycle, the transistor is turned on for that part of the cycle which exceeds 100 mA. Since the transistor only conducts when I does exceed 100 mA, and because it only takes 0.7 volts to turn on the bjt quite hard, the peak voltage at 10 ohms has a struggle rising above around 1.2peak v. So the 1,000 uF is not exposed to larger charge currents, than the dischange currents Whilst the "current shunter" works, there is a near square wave which appears across the 10 ohm R of around 1v peak, but the increase in measured thd is utterly negligible, in a load of 1,000 ohms or less, because the class AB transition causes far more thd. So the AB part of the amp cycle is reserved for the drum beats, and momentary chains of transients, and the sound is not going to degrade any more than it already has when hard AB action takes place. Meanwhile, at 3.3 watts, and still in class A, the amp is not affected by the current shunters because the Rk volotage is below the threshold at which the bjt is turned on, which is at about +0.4v. The non linearity of the bjt current turn on transfer curve compliments the non linearities with AB cathode currents. Its possible to have too much cathode current shunting, and have the Ck voltage drop as input is increased, which shows that the tube is forced into conducting excess DC, and its equivalent of biasing the amp into heavier currented class A as the po rises. There is nothing to be gained by allowing this to happen, except that a lot more heat is generated. The correct working of the current shunting will produce a slight rise in Ek at clipping with a sine wave, but its a small % compared to without the shunting. With music, the stablity of the Ek is far better, and the behaviour of the amp resembles a fixed bias amp. So in my 300 watt amps with 12 x 6550, there is no need for 12 fixed bias pots, with would make biasing a real PITA, so separate cathode biasing is fine. The cathodes of the 6 tubes either side of the PP circuit are joined to one point by diodes so that none are tied together, and only one power transistor, one small filter cap, and 3 resistors are required for each side of the 300 watt PP circuit. This all fits on a small 100 mm long heatsink, which does not rise in temp more than warm, even when a sine wave is making the current shunters work at maximum. Patrick Turner. |
#33
|
|||
|
|||
"Patrick Turner" I don't need arithmetic to proove anything. ** ROTFLMAO *so hard* I can hardly stand it !!!!!! Turner the Turd is an absolute SCREAM !!!!!!! His arithmetic is just as dodgy as his ****ing awful spelling !!!!! This ** Charade of the Charlatan ** could go on producing more and more hysterical nonsense before the Turd has a clue !!!! ............. Phil |
#34
|
|||
|
|||
Phil's post below in response to
the issue of a cathode bias regulator has ZERO technical information. Phil should just **** Off. Patrick Turner Phil Allison wrote: "Patrick Turner" I don't need arithmetic to proove anything. ** ROTFLMAO *so hard* I can hardly stand it !!!!!! Turner the Turd is an absolute SCREAM !!!!!!! His arithmetic is just as dodgy as his ****ing awful spelling !!!!! This ** Charade of the Charlatan ** could go on producing more and more hysterical nonsense before the Turd has a clue !!!! ............ Phil |
#35
|
|||
|
|||
"Patrick Turner" wrote in message
... Phil's post below in response to the issue of a cathode bias regulator has ZERO technical information. Wow, this thread looks hilarious from my side... http://webpages.charter.net/dawill/Images/PatT.jpg Quit talking to yourself all damned ready. Tim -- "That's for the courts to decide." - Homer Simpson Website @ http://webpages.charter.net/dawill/tmoranwms |
#36
|
|||
|
|||
On Mon, 08 Dec 2003 22:46:57 +1100, Patrick Turner wrote:
Under the banner of the insulting header, "Turner is a weasel" Phil expects me to provide a current flow chart/diagram to prove the cathode voltage regulation circuit for an AB tube amp works. He demands, in filthy language :- ** Post a description that INCLUDES a current flow analysis - ****head. I tried the thing as shown and it FAILED to regulate the cathode voltage on sine waves. Anyone can verify that themselves. Anyone using basic electronic arithmetic can soon see why. Ah, so if the arithmetic is wrong, it cannot work. Well, I guess its his word against mine. But still he persists in providing NOT A SHRED of evidence that it cannot work, and that it cannot regulate the cathode voltage I don't need arithmetic to proove anything. Let me explain yet again, for the benefit of those in doubt of my circuit viability, and lets leave the arithmetic right out of this because nobody would understand, except mathematicians at a university, who don't build tube amps. Let's consider a PP amp with 2 x KT88 in UL, with an 8k to 8 ohm OPT, and with cathode bias, tested with a sine wave. The power output will be pure class A to clipping with a load that is slightly above say 8k a-a, maybe 30 watts, and the cathode bias voltages will not drift upwards by more than 10% up to the onset of clipping. There is not need for bias regulation in such an amp, or any servo bias circuit. The wave form of the current flowing into the cathode caps is substantially a sine wave, with some 2H distortion current, and because the positive going cycles rise higher than the -ve current cycles, the DV across the Ck rises slightly. But as RL a-a is reduced below 8k a-a, the amp starts producing class AB power, and the current flow shows that the -ve part of the current cycle in each tube is cut off, and as RL is reduced to say 3k a-a, the amp is nearly working in class B, with a small class A content. The average of the positive peak currents are far higher than the idle current, and under thse tests, the cathode voltage will move from say +55v to say +70v, at clipping, which has the effect of underbiasing the tubes, and instead of getting say 1% of thd, even with 15 dB of NFB, you will get maybe 8% at clipping. But with music, the actual cathode voltage does not move much at below 5 watts into 3k a-a, but as the music starts to clip, the DC flow in the tubes begins to fluctuate up and down and the amp sustains considerable distortion, both by mis biasing, and DC offset swings in the OPT. Now if our amp had been made using fixed bias, none of the cathode swings in voltage and DC occur, because the cathode caps are not in series with the "rectifying" cathode signal currents. But cathoed bias is a winner for owners, since not bias adjustments are required, and there is nothing for an owner to get confused with, of make mistakes with, and I assure you owners, or their sons, twiddle bias pots all wrong sometimes, with smoky results. It came into my mind that none of the bias servo circuits ever really worked. I then thought that if the part of the AB cathode wave form which exceeds the value of twice the idle current was able to be dynamically shunted to 0V, or around the power circuit, that the cathode caps wouldn't change their voltage. Whilst in class A, with a sine wave input signal, the cathode current increases to about twice the idle current, and then reduces to almost no current, on each wave cycle, so no change to Ek occurs. But once the +ve current cycle exceeds the -ve current cycle, the tube is placing a rectifying current into its Ck, and the DV rises. Now in my circuit, the 1,000 uF Ck feeds a 10 ohm current sensing R and this is where you may view the AB signal Ik, whiuch is like a sine wave, but with most of the -ve bottoms cut off, just under clipping. So in the case of a 3k a-a load, the class B load is 750 ohms for each tube, and if the anode voltage swing was 210 vrms, or 300 peak volt, so we would have a peak cathode current of 400 mA. This produces a peak voltage of +4 volts at the 10 ohm resistor, unless we were able to somehow shunt all the current above a threshold of say +1v because the idle voltage value is +0.5v because of the 50 mA of idle tube current. To achieve dynamic re-routing of the excess AB signal I, the 1,000 uF Ck is used as a coupling cap to a base of a bjt with a series R to limit the peak base I and prevent the bjt from failure. The emitter is at 0V, and the collector has a current limiting R to the cathode, which resides at +55v. During the AB wave cycle, the transistor is turned on for that part of the cycle which exceeds 100 mA. Since the transistor only conducts when I does exceed 100 mA, and because it only takes 0.7 volts to turn on the bjt quite hard, the peak voltage at 10 ohms has a struggle rising above around 1.2peak v. So the 1,000 uF is not exposed to larger charge currents, than the dischange currents Whilst the "current shunter" works, there is a near square wave which appears across the 10 ohm R of around 1v peak, but the increase in measured thd is utterly negligible, in a load of 1,000 ohms or less, because the class AB transition causes far more thd. So the AB part of the amp cycle is reserved for the drum beats, and momentary chains of transients, and the sound is not going to degrade any more than it already has when hard AB action takes place. Meanwhile, at 3.3 watts, and still in class A, the amp is not affected by the current shunters because the Rk volotage is below the threshold at which the bjt is turned on, which is at about +0.4v. The non linearity of the bjt current turn on transfer curve compliments the non linearities with AB cathode currents. Its possible to have too much cathode current shunting, and have the Ck voltage drop as input is increased, which shows that the tube is forced into conducting excess DC, and its equivalent of biasing the amp into heavier currented class A as the po rises. There is nothing to be gained by allowing this to happen, except that a lot more heat is generated. The correct working of the current shunting will produce a slight rise in Ek at clipping with a sine wave, but its a small % compared to without the shunting. With music, the stablity of the Ek is far better, and the behaviour of the amp resembles a fixed bias amp. So in my 300 watt amps with 12 x 6550, there is no need for 12 fixed bias pots, with would make biasing a real PITA, so separate cathode biasing is fine. The cathodes of the 6 tubes either side of the PP circuit are joined to one point by diodes so that none are tied together, and only one power transistor, one small filter cap, and 3 resistors are required for each side of the 300 watt PP circuit. This all fits on a small 100 mm long heatsink, which does not rise in temp more than warm, even when a sine wave is making the current shunters work at maximum. Patrick Turner. Hey Patrick, I just started building another tube amp, this time with push pull EL84s. I was planning on cathode bias, maybe I'll try your circuit for fun!! Any tips about it? I guess I'll have around 300 plate volts, 1k screen resistors. No ultra linear stuff, I already have the transformer... If I can't find your diagram I'll bug you to re-post it!! |
#37
|
|||
|
|||
Tim Williams wrote: "Patrick Turner" wrote in message ... Phil's post below in response to the issue of a cathode bias regulator has ZERO technical information. Wow, this thread looks hilarious from my side... http://webpages.charter.net/dawill/Images/PatT.jpg Quit talking to yourself all damned ready. Tim I see zero reason why I would reply to anyone posting a header which is an insult. I just keep trying to explain, but Phil explains SFA. Patrick Turner. -- "That's for the courts to decide." - Homer Simpson Website @ http://webpages.charter.net/dawill/tmoranwms |
#38
|
|||
|
|||
"Patrick Turner" I see zero reason why I would reply to anyone posting a header which is an insult. I just keep trying to explain, but Phil explains SFA. ** My explanation is clear - you are just to stupid and stubborn for words. Try doing what a asked you to - ****head. ........ Phil |
#39
|
|||
|
|||
Hey Patrick, I just started building another tube amp, this time with push pull EL84s. I was planning on cathode bias, maybe I'll try your circuit for fun!! Any tips about it? I guess I'll have around 300 plate volts, 1k screen resistors. No ultra linear stuff, I already have the transformer... If I can't find your diagram I'll bug you to re-post it!! When you get your amp built, email me and I'll send you the schematic, and help you through. In the case of EL84, the amount of current shunting is tiny compared to what I have in a 300 watt amp, so the transistors can be simply siliconed onto mica onto the chassis, almost any TO220 power transistor will do, and perhaps the value of the Rk current sensor needs to be more than 10 ohms. A small EL84 is often used at levels where the music provokes the balance of DC in the OPT to wander all over the place. So the current shunters should work very well. I like fixed bias, and I like to use 3 watt wire wound bias pots, but they cost more than the solid state current shunters. Patrick Turner. |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
OT Political | Pro Audio | |||
DNC Schedule of Events | Pro Audio |