[Patrick Turner]

This morning I well and truly fried a signal generator beyond
fixability.

Solid state generators don't like +580Vdc poked down their outputs, but
I managed it while testing a Dynaco MK6.

There was a sickening sausage sizzle sound as a a bunch of parts
converted quickly from plastic encased silicon to soot.

So, I am re-building a 5 range wien bridge oscillator to go from 2Hz to
200kHz.

So far, with NE5534 and with +/-6V rails I can just get the five full
ranges but only 2.8Vrms output. Using higher rails above 9V up to 12V
means the op amp struggles above 100kHz and the oscilations break up
into a series of bursts rather than continuous nice oscillations.

So what's the best opamp for wide bandwidth?

Positive FB networks have caps ranging from 4.7uF to 470pF, and variable
R from 1k5 to 16k.

NFB network is about 200 ohms plus two series tiny lamps which when
warmed are 100 ohms, or just right.

With 11V rails I got 5.6V at below 50kHz OK with the NFB ohms totaling
only 133 ohms with one lamp which means the opamp must move into class
AB operation to do this because at idle the Idc = 11mA.

I'd like 5 Vrms at 200kHz or higher F if I can get it.

Any thoughts?


Patrick Turner.

[Arny Krueger]

"Patrick Turner" wrote in message

This morning I well and truly fried a signal generator
beyond fixability.

Solid state generators don't like +580Vdc poked down
their outputs, but I managed it while testing a Dynaco
MK6.

That is nature's way of telling you that you should have put clamping diodes
into the output circuit.

There was a sickening sausage sizzle sound as a a bunch
of parts converted quickly from plastic encased silicon
to soot.

I regret that I've made similar mistakes. Since I don't do much with tubes,
microphone preamps with phantom power was the source of my most recent
waterloo, a few years back.

So, I am re-building a 5 range wien bridge oscillator to
go from 2Hz to 200kHz.

Been there, done that but I admit I stopped at 100 KHz.

So far, with NE5534 and with +/-6V rails I can just get
the five full ranges but only 2.8Vrms output.

It's pretty well known that NE5532-4 chips are most happy up in the +/- 18
range, and will stay glued together up to +/- 22 volts.

Using higher rails above 9V up to 12V means the op amp
struggles above 100kHz

Been there, done that as well. More properly stated the oscillator
struggles. I'm not that sure that the op amp is the whole reason why. With
VCC= +/- 22 I was able to get about 11 vrms @ 100 KHz.

and the oscilations break up into
a series of bursts rather than continuous nice
oscillations.

Many possible answers. Obviously an op map with more GBW and slew rate would
give relevant evidence to the hypothesis (which may be true) that the 5534
was the problem. Counter-evidence is the fact that

http://www.cordellaudio.com/instrume...d_analyzer.pdf

shows a full THD test set (both oscillator and analyzer) that is alleged to
work up to 200 KHz, and is based on 5534s.


So what's the best opamp for wide bandwidth?

The OPA 604 and OPA 2604 come to mind in the "classics" department

The LM4562 is the latest-greatest from National.

Positive FB networks have caps ranging from 4.7uF to
470pF, and variable R from 1k5 to 16k.

NFB network is about 200 ohms plus two series tiny lamps
which when warmed are 100 ohms, or just right.

With 11V rails I got 5.6V at below 50kHz OK with the NFB
ohms totaling only 133 ohms with one lamp which means the
opamp must move into class AB operation to do this
because at idle the Idc = 11mA.

I'd like 5 Vrms at 200kHz or higher F if I can get it.

Most of my experiments were based on the Wein bridge components of Heath's
IG-18/5218. There are schematics on the web.

[tubegarden]

Hi RATs!

Krone-Hite made a nice vacuum tube oscillator with three power
supplies (two regulated) and 0.00001 CPS to 99,999 CPS range with
output up to 10 volts, IIRC.

Mine was an A-440, but there were others. It had a 5x10 array of
buttons to select the freq

Traded it for a bunch of SE OPTs.

He checked it out with his AP at home. I believe his comment was: "We
often dismiss the past achievements of engineers more clever than us."

Or something along those lines.

Worth keeping an eye out for.

Happy Ears!
Al

[Ian Bell[_2_]]

wrote:
I zapped the holy f**** out of a solid state HP oscillator at work
once, but it was fixable. It took out three transistors and the output
pot and scorched the board but it cleaned up well enough. Probably you
can fix the old one, or just rebuild its schematic in new parts and
this time put in some diodes.

With HP 200CD lunchboxes going for $25 why **** with anything else,
in a high voltage environment?





Probably because their distortion is too high?

Cheers

Ian

[bigwig]

On 5 Feb, 15:17, Patrick Turner wrote:
This morning I well and truly fried a signal generator beyond
fixability.

Solid state generators don't like +580Vdc poked down their outputs, but
I managed it while testing a Dynaco MK6.

There was a sickening sausage sizzle sound as a a bunch of parts
converted quickly from plastic encased silicon to soot.

So, I am re-building a 5 range wien bridge oscillator to go from 2Hz to
200kHz.

So far, with NE5534 and with +/-6V rails I can just get the five full
ranges but only 2.8Vrms output. Using higher rails above 9V up to 12V
means the op amp struggles above 100kHz and the oscilations break up
into a series of bursts rather than continuous nice oscillations.

So what's the best opamp for wide bandwidth?

Positive FB networks have caps ranging from 4.7uF to 470pF, and variable
R from 1k5 to 16k.

NFB network is about 200 ohms plus two series tiny lamps which when
warmed are 100 ohms, or just right.

With 11V rails I got 5.6V at below 50kHz OK with the NFB ohms totaling
only 133 ohms with one lamp which means the opamp must move into class
AB operation to do this because at idle the Idc = 11mA.

I'd like 5 Vrms at 200kHz or higher F if I can get it.

Any thoughts?

Patrick Turner.

This could be interesting Patrick,
I am also playing about and gathering parts for a nice Wein bridge
oscilator. It seems that the NE5532/4 is are hard act to beat. Check
out this page on Doug. Selfs site:
http://www.dself.dsl.pipex.com/ampins/webbop/opamp.htm
There are a lot of new opamps about now but I m not really up to speed
on them.A couple I have been looking at are the LME49710 and 49720 and
their cousin the 49600 as a buffer. They are expensive though and the
NE5532/4 seems to be the best bang for your buck as they say.
Matt.

[Patrick Turner]

wrote:

I zapped the holy f**** out of a solid state HP oscillator at work
once, but it was fixable. It took out three transistors and the output
pot and scorched the board but it cleaned up well enough. Probably you
can fix the old one, or just rebuild its schematic in new parts and
this time put in some diodes.

With HP 200CD lunchboxes going for $25 why **** with anything else,
in a high voltage environment?

My Topward function gene was cheap in 1994? and had already been
repaired twice by the previous owner. I blew all the output transistors
about 3 times and repaired it. Then I began using a tube cathode
follower buffer and that stopped me blowing it up for awhile. The level
pot wore out, so I replace that and so did the F adjust pot which
altered DC for the chip making the sine waves so I put an 11 position
switch to change the F for each decade between 2Hz and 2MHz. It became
handy for testing OPTs and all sorts of things. But I forgot I had it
directly connected across 1/2 a primary in a Dynaco MK6 when I turned on
the PS to see what supply voltages I'd get. Although useful, the gene
also hummed all its life due to a dodgy mains tranny, which ran way too
hot. Distortion was about .3%.

I've settled on the NE5534 for now, and have got protection clamping
diodes in place and a 5uF cap on the output, 2 switched outputs for 0.0V
to 0.32V and 0.32V to 2.5V with 800 ohm and 5k ohm log pots. At 2.5V
maximum output voltage with +/- 6V rails I get the full 220kHz of
bandwidth I wanted, and I've never known a power amp needing more than
2.5Vrms to bring the output to clipping.

THD at 1kHz measures 0.006%, the best I've ever had from an opamp
oscillator, and range accuracy calibration on the dial is within 2%.
There's very little bouncing or amplitude modulation. I'm happy.

For higher F maybe I will upgrade a tuned LC type of oscillator.

To make a square wave I can amplify the sine wave and clip it off.

One can buy these things, but they always have faults.

Patrick Turner.

[Ian Bell[_2_]]

wrote:
On Feb 5, 2:15 pm, Ian Bell wrote:
wrote:
I zapped the holy f**** out of a solid state HP oscillator at work
once, but it was fixable. It took out three transistors and the output
pot and scorched the board but it cleaned up well enough. Probably you
can fix the old one, or just rebuild its schematic in new parts and
this time put in some diodes.
With HP 200CD lunchboxes going for $25 why **** with anything else,
in a high voltage environment?
Probably because their distortion is too high?


Nope. It's quite low.


Well, if you call 0.5% low, then I guess you are right ;-) but that's
not what I call quite low.

Cheers

Ian

[Arny Krueger]

"Patrick Turner" wrote in message
...

I've settled on the NE5534 for now, and have got protection clamping
diodes in place and a 5uF cap on the output, 2 switched outputs for 0.0V
to 0.32V and 0.32V to 2.5V with 800 ohm and 5k ohm log pots. At 2.5V
maximum output voltage with +/- 6V rails I get the full 220kHz of
bandwidth I wanted

So Cordell was right - the 5534 was not an inhibiter to 200 KHz operation?

, and I've never known a power amp needing more than
2.5Vrms to bring the output to clipping.

Stabilization by what, lightbulb?

THD at 1kHz measures 0.006%, the best I've ever had from an opamp
oscillator, and range accuracy calibration on the dial is within 2%.
There's very little bouncing or amplitude modulation. I'm happy.

Yeah, getting rid of that amplitude modulation can be hairy. With lightbulbs
distortion below 50 Hz can be a problem because the output stabilization
circuit's time constant whether a lightbulb or more complex means. One
solution is to just lower the gain of the stabilizer, but that can lead to
AM on the HF ranges.

For higher F maybe I will upgrade a tuned LC type of oscillator.

At higher frequencies it seems to be pretty economical to make low
distortion LC circuits.

To make a square wave I can amplify the sine wave and clip it off.

Schmidt trigger is best - nice discrete one in the IG 18/5218 schematic. I
think it would work well with your 5v supply.

Don't forget the clamping diodes for it! ;-)

[Ian Bell[_2_]]

wrote:
With HP 200CD lunchboxes going for $25 why **** with anything else,
in a high voltage environment?
Probably because their distortion is too high?
Nope. It's quite low.
Well, if you call 0.5% low, then I guess you are right ;-) but that's
not what I call quite low.



Hewlett-Packard 200CD Wide-Range Oscillator modifications as posted to
the Ampex Mailing List in 2000

Usual disclaimers. Follow instructions at your own risk. For use by
qualified personnel only. To avoid injury or equipment damage, take
usual and customary precautions.

Selling or modification of this information prohibited but you may
distribute if quoted as-is, with no changes, edits, modifications,
etc.

Dave Dintenfass
Full-Track Productions, Seattle, Washington
March 2006

www.fulltrackproductions.com

-----------------------------

From: David Dintenfass
Sent: Wednesday, February 02, 2000 10:13 AM
To: Ampex List
Subject: [AMPEX] OT: HP 200CD low-distortion modification

About ten years ago I did the low-distortion modification on one of my
HP 200CDs. Since I can't find the specific service note, I opened up
the oscillator to see what I had done (at the time, I followed the
procedures listed in the official HP service note).

To reduce distortion, change R12 from 3000 ohms to 2700 ohms. With a
600 ohms load on the oscillator output, crank the front-panel
"amplitude" knob wide open (fully clockwise), then adjust R11 until
you get 7.5 Vrms across the output.

Keep in mind that HP 200CDs can be divided into two broad categories
-- those with octal tubes and point-to-point wiring and those with
miniature tubes and a mix of point-to-point wiring and a small pc
board. I think the change occurred around 1962 going by the changes
reflected in various service manuals I have for the thing.

On older units, R11 is not identified but you can adjust it from the
top of the unit if you mesh the tuning capacitor. On newer units, R11
is on the pc board and is marked "output volts adj" with silkscreen
lettering. You will have to turn the unit on its side to adjust R11 on
newer units.

If you have trouble finding R12 on older units, look for R11 (which is
a pot) and R12 is the 3K resistor connected to the wiper of R12. With
newer units, R12 is marked on the pc board.

Note that after you make the mod, the oscillator will take longer to
settle when you change frequency. This is particularly noticeable at
low frequencies. This is a normal consequence of making this
modification.

THD (at full output) on my particular unit went from:

0.068% at 1 kHz before the mod to 0.024% at 1 kHz after the mod

That's moving in the right direction and is almost as good as my
Ferrograph RTS2 but I still don't consider that good enough for serious
work.


Note that this is THD, not THD+noise as it was measured on a spectrum
analyzer, not a THD meter.


That's odd. If he measured it on a spectrum analyser he should have seen
individual harmonics not THD.

Cheers


Ian