[Sliding Bias]

1st post on here, hope you can help me....

History:
I have built 10s of SS amps and a couple of valve preamps over the last 20 years or so, but the time has come for me to build a valve/tube jobbie. A friend of mine has a 20 year old Grant Lumley GL50a which has had a new (larger) mains translformer fitted and been re-biased to use KT88s (originally KT77s). The amp always sounded very average with KT77s but sounds superb with KT88s - better than a pair of EAR 509II Anniversarys (~£10K nowadays!!!).

I am very tempted to clone the GL50A, the circuit very similar to this (and the Williamson amp):
http://www.drtube.com/schematics/grant/g60ams.gif
The B+ is 425V, KT88 bias is 40mA and o/p transformer turns ratio is 21.5: 1 (3K7 to 8 Ohms) with 45% taps.

If I go ahead with this I will go for 2 X 6SN7 instead of 2 X ECC82 because general opinion is that they sound better and M Jones reckons the distortion is much lower.

However there are two other circuits which are appealing:
1) P turner's 5050
2) The GEC 30W:
http://www.jacmusic.com/KT88/SCHEMATIC3.gif

My speakers are 87dB/W so I guess I need ~ 30W + a bit of headroom.

The 5050 seems to only have a gain of about X11 (ignoring V1). Can this be changed to a more usual X 20 to 30?
Why is a voltage doubler used?

I confess to minimal knowledge about impedance matching....
3K7:8 Ohms seems like a very low turns ratio compared to the 5050 whcih uses 6K6:4Ohms (with a higher B+). Looking at transformer manufacturers pages they seem to recommend 4K5 or 5K. Please help me understand the tradeoffs and how to choose.

I am also very confused about how to select a transformer supplier, here in the UK there seems to be 3 main options:
Lundahl - mid priced but seem to be very small cores for the claimed power rating.
Hammond - middle of the road? (No drop through style!).
Sowter - expensive, never heard a word against them. Custom windings at no extra cost.

I am tempted to go with Sowter just to be sure of getting something that will sound "right". Any comments?

I guess the main question is: Which circuit should I choose and why?

Regards,
Dave

[[email protected]]

Sowter's transformers are of good quality. But you can probably get
ones from Japan or Taiwan (James Wu) reasonably apart from the
shipping, and there may be someone who brings them in with container
loads from Japan or Taiwan at lower cost.

There may be vendors elsewhere in the EU for these too.

Supposedly the company that bought Partridge will still wind
Partridge parts for UK customers. You could contact them. They refuse
to talk to anyone in North America supposedly due to liability
reasons. It's horse**** of course, but that's what I've heard.

Anyway, I would look at the Marantz 2 and 5 and the early VTL amps in
the Manley book for examples of how to do it simply. This Lumley amp
looks something like a VTL at that.

Is this where Enid Lumley got her start?

[Ian Iveson[_2_]]

On Jan 15, 11:05*am, Sliding Bias Sliding.Bias.
wrote:
1st post on here, hope you can help me....

History:
I have built 10s of SS amps and a couple of valve preamps over the last
20 years or so, but the time has come for me to build a valve/tube
jobbie. A friend of mine has a 20 year old Grant Lumley GL50a which has
had a new (larger) mains translformer fitted and been re-biased to use
KT88s (originally KT77s). The amp always sounded very average with KT77s
but sounds superb with KT88s - better than a pair of EAR 509II
Anniversarys (~£10K nowadays!!!).

I am very tempted to clone the GL50A, the circuit very similar to this
(and the Williamson amp):http://www.drtube.com/schematics/grant/g60ams.gif
The B+ is 425V, KT88 bias is 40mA and o/p transformer turns ratio is
21.5: 1 (3K7 to 8 Ohms) with 45% taps.

If I go ahead with this I will go for 2 X 6SN7 instead of 2 X ECC82
because general opinion is that they sound better and M Jones reckons
the distortion is much lower.

However there are two other circuits which are appealing:
1) P turner's 5050
2) The GEC 30W:http://www.jacmusic.com/KT88/SCHEMATIC3.gif

My speakers are 87dB/W so I guess I need ~ 30W + a bit of headroom.

The 5050 seems to only have a gain of about X11. Can this be changed to
a more usual X 20 to 30?

I confess to minimal knowledge about impedance matching....
3K7:8 Ohms seems like a very low turns ratio compared to the 5050 whcih
uses 6K6:4Ohms (with a higher B+). Looking at transformer manufacturers
pages they seem to recommend 4K5 or 5K. Please help me understand the
tradeoffs and how to choose.

I am also very confused about how to select a transformer supplier, here
in the UK there seems to be 3 main options:
Lundahl - mid priced but seem to be very small cores for the claimed
power rating.
Hammond - middle of the road? (No drop through style!).
Sowter - expensive, never heard a word against them. Custom windings at
no extra cost.

I am tempted to go with Sowter just to be sure of getting something that
will sound "right". Any comments?

I guess the main question is: Which circuit should I choose and why?

Regards,
Dave

--
Sliding Bias

Useable bandwidth depends on quality transformers. Sowter have a
strong reputation, lots of continuous history, and are happy to give
advice once you've decided what you want to build. They can also
supply your power transformer with whatever taps you need.

You should read something about valve output transformers, because the
basics are too long for here. Menno van der Veen's "Modern valve
amplifiers" contains a good mathematical treatment of the subject. The
issues you should be aware of a leakage inductance, interwinding
capacitance, primary inductance, winding resistance and saturation
current. The load seen by the valves comprises the primary winding
resistance in series with (leakage inductance in series with the sum
of speaker impedance and secondary winding resistance multiplied by
the impedance ratio). The primary inductance and the winding
capacitance are each connected in shunt with the section in brackets.
The impedance ratio is the square of the winding ratio. It's worth
sketching that circuit if I've managed to make it clear. You should
see that there is a high pass filter resulting from the primary
inductance, and a low pass filter arising from the capacitance and
leakage. Also, note that the allowable current flowing through the
primary inductance is limited by core saturation, and that the current
through an inductor depends on frequency and signal amplitude. Hence
an output transformer imposes an extra, amplitude dependent limit to
bass. That's why Sowter will want to know your target full-power bass
frequency limit, in addition to your required small-signal roll-off
point.

Considering the lowish sensitivity of your speakers and the relatively
high output of most modern sources, you may be better looking for more
recently conceived circuits than the Williamson. An ultra-linear
output stage would give more power than triode. SS current sources,
voltage regulators and other odds and ends are cheap and easy, and
modern designs can use these to good effect. OTOH, if you know you
like the GL50 and you feel confident that you can modify the feedback
compensation to suit whatever output transformers you use, that's
likely to be a good way to go.

It may be worth reviewing your conception of how much power you need
for those speakers. Personally, I would want considerably more than
30W. Triode connection will give you a marginally cleaner output for
much of the time, but that might not be worthwhile if your short-term
signal peaks drive the amp into clipping. I try for a 60:1 ratio of
clipping point to mean listening power.

Finally, output valves in triode or UL mode are generally loaded with
between 2 to 3 times the anode resistance. Linearity is improved as
load rises, generally, but output power is reduced. As with mode, if
you have no power to spare, then what you choose might depend on which
source of distortion you prefer.

Ian

[Sliding Bias]

Thanks for the replies.

I have access to a 2nd hand pair of Sowter transformers. I believe they are 6K6 to 8 and 4 Ohms. Since Patrick Turner seems able to get 50W out of this impedance with ~500V B+ and fixed bias, the 5050 o/p stage and 50mA Iq seems to be a decent starting point for me. Should I go for these or not try to save a few pennies and get some new ones @ 4.5K?

I suspect there are some good reasons why he chose the topology he did for the 5050 (i.e. SET, LTP with transistor current source in the tail, PP UL o/p with fixed bias). The alternative is SET, Concertina, PP driver, PP UL o/p with fixed bias (like the GL50A). Any comments?

Hopefully I should be OK wrt tweaking the stability margins. I've done this for many SS amps and the GL50 owner is valve savvy, should I need some extra help. Complexity is not too much of an issue (as long as I can fit the whole lot in a normal sized chassis).

[[email protected]]

On Jan 16, 10:52*am, Ian Iveson wrote:
On Jan 15, 11:05*am, Sliding Bias Sliding.Bias.









wrote:
1st post on here, hope you can help me....

History:
I have built 10s of SS amps and a couple of valve preamps over the last
20 years or so, but the time has come for me to build a valve/tube
jobbie. A friend of mine has a 20 year old Grant Lumley GL50a which has
had a new (larger) mains translformer fitted and been re-biased to use
KT88s (originally KT77s). The amp always sounded very average with KT77s
but sounds superb with KT88s - better than a pair of EAR 509II
Anniversarys (~£10K nowadays!!!).

I am very tempted to clone the GL50A, the circuit very similar to this
(and the Williamson amp):http://www.drtube.com/schematics/grant/g60ams.gif
The B+ is 425V, KT88 bias is 40mA and o/p transformer turns ratio is
21.5: 1 (3K7 to 8 Ohms) with 45% taps.

If I go ahead with this I will go for 2 X 6SN7 instead of 2 X ECC82
because general opinion is that they sound better and M Jones reckons
the distortion is much lower.

However there are two other circuits which are appealing:
1) P turner's 5050
2) The GEC 30W:http://www.jacmusic.com/KT88/SCHEMATIC3.gif

My speakers are 87dB/W so I guess I need ~ 30W + a bit of headroom.

The 5050 seems to only have a gain of about X11. Can this be changed to
a more usual X 20 to 30?

I confess to minimal knowledge about impedance matching....
3K7:8 Ohms seems like a very low turns ratio compared to the 5050 whcih
uses 6K6:4Ohms (with a higher B+). Looking at transformer manufacturers
pages they seem to recommend 4K5 or 5K. Please help me understand the
tradeoffs and how to choose.

I am also very confused about how to select a transformer supplier, here
in the UK there seems to be 3 main options:
Lundahl - mid priced but seem to be very small cores for the claimed
power rating.
Hammond - middle of the road? (No drop through style!).
Sowter - expensive, never heard a word against them. Custom windings at
no extra cost.

I am tempted to go with Sowter just to be sure of getting something that
will sound "right". Any comments?

I guess the main question is: Which circuit should I choose and why?

Regards,
Dave

--
Sliding Bias

Useable bandwidth depends on quality transformers. Sowter have a
strong reputation, lots of continuous history, and are happy to give
advice once you've decided what you want to build. They can also
supply your power transformer with whatever taps you need.

You should read something about valve output transformers, because the
basics are too long for here. Menno van der Veen's "Modern valve
amplifiers" contains a good mathematical treatment of the subject. The
issues you should be aware of a leakage inductance, interwinding
capacitance, primary inductance, winding resistance and saturation
current. The load seen by the valves comprises the primary winding
resistance in series with (leakage inductance in series with the sum
of speaker impedance and secondary winding resistance multiplied by
the impedance ratio). The primary inductance and the winding
capacitance are each connected in shunt with the section in brackets.
The impedance ratio is the square of the winding ratio. It's worth
sketching that circuit if I've managed to make it clear. You should
see that there is a high pass filter resulting from the primary
inductance, and a low pass filter arising from the capacitance and
leakage. Also, note that the allowable current flowing through the
primary inductance is limited by core saturation, and that the current
through an inductor depends on frequency and signal amplitude. Hence
an output transformer imposes an extra, amplitude dependent limit to
bass. That's why Sowter will want to know your target full-power bass
frequency limit, in addition to your required small-signal roll-off
point.

Considering the lowish sensitivity of your speakers and the relatively
high output of most modern sources, you may be better looking for more
recently conceived circuits than the Williamson. An ultra-linear
output stage would give more power than triode. SS current sources,
voltage regulators and other odds and ends are cheap and easy, and
modern designs can use these to good effect. OTOH, if you know you
like the GL50 and you feel confident that you can modify the feedback
compensation to suit whatever output transformers you use, that's
likely to be a good way to go.

It may be worth reviewing your conception of how much power you need
for those speakers. Personally, I would want considerably more than
30W. Triode connection will give you a marginally cleaner output for
much of the time, but that might not be worthwhile if your short-term
signal peaks drive the amp into clipping. I try for a 60:1 ratio of
clipping point to mean listening power.

Finally, output valves in triode or UL mode are generally loaded with
between 2 to 3 times the anode resistance. Linearity is improved as
load rises, generally, but output power is reduced. As with mode, if
you have no power to spare, then what you choose might depend on which
source of distortion you prefer.

Ian

The Willy is not too stable at the low end. The circuit he shows IS
pretty Wiliamsonish on second look: it has a driving stage after the
phase splitter, which is unnecessary on Class AB1 or Class A output
stages.

I would look at the VTL, which has a long tailed pair phase inverter.
Or the Marantzes or even-horrors-the Dyna Mk III.

[John L Stewart]

The Willy is not too stable at the low end. The circuit he shows IS
pretty Wiliamsonish on second look: it has a driving stage after the
phase splitter, which is unnecessary on Class AB1 or Class A output
stages.

I would look at the VTL, which has a long tailed pair phase inverter.
Or the Marantzes or even-horrors-the Dyna Mk III.[/quote]

Referring to the Grant G60AMS,the nervous LF of a typical Willy is partially fixed by a 15 db shelf in this cct. But we don't know which OPT is used. A successful shelf depends on the OPT characteristics. Otherwise very ordinary.

The lower RH corner shows a cct with no labels, perhaps a low or hi bias indicator? And what does the mysterious 3.9 V off the heater supply do?

Cheers, John

[Sliding Bias]

Originally Posted by John L Stewart

The Willy is not too stable at the low end. The circuit he shows IS
pretty Wiliamsonish on second look: it has a driving stage after the
phase splitter, which is unnecessary on Class AB1 or Class A output
stages.

I would look at the VTL, which has a long tailed pair phase inverter.
Or the Marantzes or even-horrors-the Dyna Mk III.

Referring to the Grant G60AMS,the nervous LF of a typical Willy is partially fixed by a 15 db shelf in this cct. But we don't know which OPT is used. A successful shelf depends on the OPT characteristics. Otherwise very ordinary.

The lower RH corner shows a cct with no labels, perhaps a low or hi bias indicator? And what does the mysterious 3.9 V off the heater supply do?

Cheers, John[/quote]

I'm not sure about the 3.9V. The G50 is a more basic and earler version of the G60 it seems, and it does not use the opamps etc.

The Marantz model 5 looks interesting - similar topology to the 5050 except for the dc levels/coupling. It is hard to figure out what's goiong on with the feedback though - there seems to be a resistor shown in the "com" o/p conection that could be related to the same symbol in series with the extra o/p winding.

There seems to be an updated schematic for the Dyna MkIII which also seems to use (more or less ) this topology:
http://www.diytube.com/unidriver/poseidon.pdf
Not sure why they don't use a resistor for the o/p tube screens, but otherwise this looks decent. I would use tag strip to build though, not pcbs.

I cannot find any schematics for VTL. Maybe I need to buy the Manley book, where can I find this book??

[John L Stewart]

Originally Posted by Sliding Bias

Referring to the Grant G60AMS,the nervous LF of a typical Willy is partially fixed by a 15 db shelf in this cct. But we don't know which OPT is used. A successful shelf depends on the OPT characteristics. Otherwise very ordinary.

The lower RH corner shows a cct with no labels, perhaps a low or hi bias indicator? And what does the mysterious 3.9 V off the heater supply do?

Cheers, John

I'm not sure about the 3.9V. The G50 is a more basic and earler version of the G60 it seems, and it does not use the opamps etc.

The Marantz model 5 looks interesting - similar topology to the 5050 except for the dc levels/coupling. It is hard to figure out what's goiong on with the feedback though - there seems to be a resistor shown in the "com" o/p conection that could be related to the same symbol in series with the extra o/p winding.

There seems to be an updated schematic for the Dyna MkIII which also seems to use (more or less ) this topology:
http://www.diytube.com/unidriver/poseidon.pdf
Not sure why they don't use a resistor for the o/p tube screens, but otherwise this looks decent. I would use tag strip to build though, not pcbs.

I cannot find any schematics for VTL. Maybe I need to buy the Manley book, where can I find this book??[/quote]

----

I think I got my copy of Vacuum Tube Logic by David Manley about 10 years ago from Antique Electronic Supply of Arizona-

http://www.tubesandmore.com/

Seems to me AudioXpress Magazine also had the book-

www.audioxpress.com

Try a used book site like Alibris-

www.alibris.com

I got some nice copies of Fred Terman's books at Alibris at reasonable cost.

Perhaps VTL it is out of print.

All the grids in any ordinary audio amp should have grid stopper resistors to prevent RF interference & parasitics. I've seen parasitic oscillation to 150 MHz while using hi G tubes. This can be avoided by using 1/2 Watt resistors of one to 10K on all the control grids, mounted as close as possible to the socket. The screens with 100R, again of 1/2 Watt.

With the new front end the Dyna is no longer a Dyna in my opinion. In its original form the Dyna got its stability by using one less stage than the other guys. That was possible because they used high G tubes such as the EL34, not available when earlier amps such as the Wiliamson were introduced. But the more recent driver should work well anyway.

My 2€ antway, Cheers John

[John L Stewart]

Originally Posted by Sliding Bias

Referring to the Grant G60AMS,the nervous LF of a typical Willy is partially fixed by a 15 db shelf in this cct. But we don't know which OPT is used. A successful shelf depends on the OPT characteristics. Otherwise very ordinary.

The lower RH corner shows a cct with no labels, perhaps a low or hi bias indicator? And what does the mysterious 3.9 V off the heater supply do?

Cheers, John

I'm not sure about the 3.9V. The G50 is a more basic and earler version of the G60 it seems, and it does not use the opamps etc.

The Marantz model 5 looks interesting - similar topology to the 5050 except for the dc levels/coupling. It is hard to figure out what's goiong on with the feedback though - there seems to be a resistor shown in the "com" o/p conection that could be related to the same symbol in series with the extra o/p winding.

There seems to be an updated schematic for the Dyna MkIII which also seems to use (more or less ) this topology:
http://www.diytube.com/unidriver/poseidon.pdf
Not sure why they don't use a resistor for the o/p tube screens, but otherwise this looks decent. I would use tag strip to build though, not pcbs.

I cannot find any schematics for VTL. Maybe I need to buy the Manley book, where can I find this book??[/quote]

Here is a link to quite a good paper on taming the Williamson. Many of the points covered would be useful in other amplifiers.

http://www.frihu.com/content/diy/rc-...erstaerker.pdf

Cheers, John

[Sliding Bias]

Thanks John. I had previously found that article in the "Williamson Resources" thread. Very helpfull!!!

On balance (no pun intended) I think I will go for the 5050 type topology i.e. using an LTP phase splitter. General opinion seems to be that this sounds better than the concertina phase splitter + drivers topology and it's also easier to keep stable.
I think I'll also go for the 2nd hand 6K6 o/p transformers as these should allow a decent power o/p with good distortion figures.

Thanks for everyone's help.

Dave

PS I was rather hoping Patrick Turner would be along to explain why he chose a voltage doubler and if the low gian of the power amp part of the 5050 can be tweaked. But heyho.

[Patrick Turner]

On Jan 22, 4:45*am, Sliding Bias Sliding.Bias.
wrote:
Thanks John. I had previously found that article in the "Williamson
Resources" thread. Very helpfull!!!

On balance (no pun intended) I think I will go for the 5050 type
topology i.e. using an LTP phase splitter. General opinion seems to be
that this sounds better than the concertina phase splitter + drivers
topology and it's also easier to keep stable.
I think I'll also go for the 2nd hand 6K6 o/p transformers as these
should allow a decent power o/p with good distortion figures.

Thanks for everyone's help.

Dave

PS *I was rather hoping Patrick Turner would be along to explain why he
chose a voltage doubler and if the low gian of the power amp part of the
5050 can be tweaked. But heyho.

Remarkably, as if by some obscene Quirk of Fait, Patrick Turner has
materialised right in front of you in your down town local cyber hide-
out.

The 5050 design is based on the historical development of PP amps.
Before Mullard's 520 with EF86, 2 x 1/2 12AX7 in LTP, and UL OP stage
43% taps, maybe not many ppl used the Mullard topology, and perhaps
most used the Williamson config. I say that knowing that by 1955,
nearly everyone insisted on at least 12dB GNFB even in an amp with
triode output stage.
Before GNFB, people could leave out a whole gain stage in a given amp.

Anyway, the GOLD STANDARD of hi-fi performance by 1955 was the PP
triode amplifier with its disgustingly low gain OP stage where tubes
with µ of not more than 6 were considered best. 300B gave A 4, 6L6
or 807, also very low gain, but who cared? As long as the sound was
free of noise, had wide bandwidth, low THD, low IMD, low Rout, and
enough PO for typical speakers of 1955 which often had sensitivity of
95dB.

People tried using beam tetrodes and pentodes with hideous results
unless you applied a lot of NFB. OPTs needed to be extremely good to
allow say 26dB GNFB around a pure pentode/tetrode output stage, and
then the copious mix of ghastly upper harmonic products remained
intact, but just reduced in level by about 20dB. With a triode OP
stage, the local NFB operating within the triode minimised the higher
H products by a huge amount, so you'd need only 12dB GNFB which was
easier to apply, and the H mix just wasn't as complex as the pentode/
tetrode. The high gain of pentode and tetrode is due to the screen
preventing internal electrostatic FB from operating. Strapping the
screen to anode conveys the anode voltage field effect to the electron
stream, and voila, magic, and a shirt and trouser load of grunge is
nicely cleaned away. In the 1950s, if not officially in the late 1950s
and 1940s, people were experimenting with any possible arrangement of
tubes to get a marketting edge or better sound, or to get some darn
thing, as ppl do - given 26 letters in the alphabet, ppl are still
adding words listed in dictionaries. That's our nature. Anyway, the UL
stage was invented which gave the world OP stages with many benefits.
THD / IMD for the same PO as triode connected pentore/tetrode tubes
was found to be as low or lower, and Ra-a went down from say 60k for a
pair of OP tubes to 6k0, and about equal to the reflected load, so DF
went drom say 0.1 to 1.0, so not much GNFB was needed to increase DF
to say 5, which many ppl find OK. Gain was about twice triode gain. PO
was near that of tetrode and pentode, so after 1955, UL became king,
because for the price of 30 minutes of factory labour the amps could
be made to work far better.
Peter Walker marketed his Acoustical connection in his Quad-II amps.
This acts to reduce gain in an OP stage to be about equal to triode,
and give the same Ra-a as triode, further reducing the need for high
amounts of GNFB. In fact, the KT66 act as though they were in an amp
with 10% UL taps, but which also have 10% local series voltage FB. The
equations fully describing what is happening are far to complex to
mention here, but IMHO, the Acoustical Connection with local CFB is
the best way to cajole wayward larikinesque pentodes and tetrodes into
cultivated clever dicks who can be relied upon to put music to our
ears without many artifacts. I prefer using 20% of CFB. Some miht go
further, and sure, have 20% CFB, then have a CT choke feeding the two
screens at slightly less Eg2 than Ea, while bypassing each screen to
the cathodes of the opposite tube. Some would otherwise just have UL
taps so Ea = EG2 where both are not very high, then have the CFB, and
OP stage gain does go LOWER than the triode - but - the sound is
magnificanto!

Even with only 12.5% CFB, and with UL taps, and with 4 x 6550, one can
get 0.7% THD at 50W in class A, without any GNFB. I couldn't find any
way of achieving this while keeping the OP stage gain about the same
as a triode OP stage so that it is easily driven by any one of several
driver stage topologies found at my website and others and all of
which work better than the old fashioned crap of 1955. One does not
have to build a McIntosh or EAR 509 to get good results.

Voltage doublers with sililcon diodes offer better regulation of B+
than any PSU scheme with tube diodes. Modern cap values of 470uF offer
a decently LOW IMPEDANCE anchor to run any tube amp from, SE or PP.
For examples of pure crud in a PSU, see Quad-II, and ST70, and much
bean counter compromised 1955 crapolgy. Garbage. And winding the HT
secs on a given PT for a doubler for a given B+ involves 1/4 of the
turns used for a CT full wave jobby. The doubler wire is then much
thicker and easier to wind, without wasting so much space on enamel
and interlayer insulation, so for the same allowed current per sq.mm
of Cu section, more I out is available; ie, the doubler tranny can be
more efficient. Efficiency is not an obvious sonic artifact. But more
of it in PTs helps the sound.

Always assume you can do better, and never ever let me see anyone
assuming junk from the past was always fabulous stuff; it just wasn't,
when compared to what is possible now. Utter Crud is still being
pumped out as it always has and always will be because that's how
markets work.

BTW, 6k6 OPTs might be a compromise for the 5050 unless you have 4, 8
16 ohm outlets. The 5050 has Ea = 500V and the OPT is better if if has
well over 6K6. 50W is on the high side of the power. For best
performance, connect 8 ohm speakers to a "4 ohm" labelled outlet, thus
giving say 13k2 RLa-a and the tubes then are working in mainly class A
and it sounds best. Maybe you get less than 50W, but so what, the
first 10 watts are what count the most; you want your amp to give
finesse, and not just be a noisy brute.
The 16 ohm labelled outlet is about as useful as tits on a bull. If
you connect 8 ohms, sure, max PO is high, but all other parameters are
craperized.

Patrick Turner.

Sliding Bias

[Sliding Bias]

Wow, thanks for such a detailed answer Patrick!!!

Looks like I should be OK with the 6K6 to 4 and 8 Ohm transformers (Sowters usually have separate secondary windings, but I need to check in this case).

The voltage doubler bit is VERY interesting. I thought you were going to say "I had one lying around"

It seems that I can do quite a bit better than my friend's Grant which in turn sounds better than another friend's EAR 509IIs IMO.

I still have one question about the 5050 though. Have I understood correctly that the gain is ~11 without the SET stage V1 i.e a bit less than 1 + R33/R10?
To increase the gain to ~22 I could reduce R10 to 470R. This would of course 1/2 the feedback. Is this OK?

Thanks and regards,
Dave

[Patrick Turner]

On Jan 23, 2:47*am, Sliding Bias Sliding.Bias.
wrote:
Wow, thanks for such a detailed answer Patrick!!!

Looks like I should be OK with the 6K6 to 4 and 8 Ohm transformers
(Sowters usually have separate secondary windings, but I need to check
in this case).

The voltage doubler bit is VERY interesting. I thought you were going to
say "I had one lying around"

It seems that I can do quite a bit better than my friend's Grant which
in turn sounds better than another friend's EAR 509IIs IMO.

I still have one question about the 5050 though. Have I understood
correctly that the gain is ~11 without the SET stage V1 i.e a bit less
than 1 + R33/R10?
To increase the gain to ~22 I could reduce R10 to 470R. This would of
course 1/2 the feedback. Is this OK?

I think you have got mixed up. R10 is 100r beneath V2 in its cathode
circuit. NFB is established at top R10 with feed by R33, 1k0, from Vo
terminal. The GNFB can be reduced roughly 6dB by reducing R10 to 47r.
But why?
What's so wrong with gains as the amp is now?

If you want twice the input sensitivity, then you could use 6DJ8 /
6922 at V2 instead of 6CG7 and this doubles V2 gain. This doubles the
open loop gain of the whole power amp includingV2,3,4,5,6. Closed loop
gain will hardly change, so amount of GNFB applied will rise from 17dB
to 23dB, and the amp might become unstable. There is no need for 23dB
FB, so R10 could then be reduced to 47r to reduce NFB with V2 = 6DJ8
back down to 17 dB. If the the stability was OK with V2 = 6CG7, it
should remain so if when 6DJ8 is used if the amount of NFB is made the
same. Compensation values for shelving open loop HF and LF and Zobel
network on output and compo cap across R33 are what I used, but any
OPT made by some other dude will have different values or else there
**will be** oscillations. So C7, R11, C6, R12, C12, R32, C11, and R33
itself all have to be carefully chosen after building the amp.

A number of ppl have built the 5050, and most have had success. I've
answered the inevitable questions when they email to say they wanna
use say 5814A insead of 6CG7, or use some different OPT, different OP
tubes, operating voltages etc. So they end up building a 4747, or a
5353, but not a 5050. That does not matter. What matters is that you
know what you are doing while you are doing it, and that you can
addapt the basic design to suit what you have to get the best
performance.

But sadly, for every 5 ppl who start to buy parts for a tube power amp
project in their spare time, maybe only 1 gets a good working amp
after 12 months. I hear about their attempts, but there must be many
who don't ever contact me and don't ever finish. There are far more
Gunner Dooz than Igotit Dunns. And along the way they learn SFA, and
try to make changes but can't analyse anything, because they never
bother learning 101 basics. For 5 sets of transformers I've sold to
ppl only one build an amp after 12 months. One started 8 years ago, he
has a lovely box of bits. Spends time playing PC games, getting
divorced, working, and General Living. When he retires in 35 years and
maybe finds time, I will not be here. Nobody will even remember me.
Maybe the Green Police will arrest him for being in possession of
Earth damaging goods, ie, vacuum tubes etc, when what he must
purchase, by law, is the latest 2047 twigital amp, that produces sound
from the smell of compost.

Patrick Turner.


Thanks and regards,
Dave

--
Sliding Bias