I use a string of 47 to 68 volt zener diodes and a little RC filtering, feeding a hexfet. Doesn’t need the last word in regulation or super temperature accuracy. Just get rid of the hum, gross voltage variations, and low frequency feedback paths through the supply. And the design of the entire front end just becomes easier.
So I connected up the 100ufs across the two 47s. Still, going from 0 to full power I drop 55V; 455 to 400. Measuring the secondary winding, it's falling from 180 to 170 RMS. 10V RMS is ~14.2 peak, but since doubler, that delta would account for ~28 of my 55V drop? So, I then assume even if I theoretically make the cap values infinite, I'll still drop 28V due to load on the winding itself.
So perhaps it's the wrong tail to chase, if my thinking is sound. I'll order replacements and in the mean time move on with a regulator for the preamp stages.
So perhaps it's the wrong tail to chase, if my thinking is sound. I'll order replacements and in the mean time move on with a regulator for the preamp stages.
Thanks, I've seen this circuit in another amp I once owned. I have one 0A2, but doubt I have any 7 pin sockets. I see I could get a 2nd one for ~$8; their manual page says they can be used in series for 300V. Only considering for the glow factor; I assume they could be RC'd and followed by a hexfet as well.
Here's what I've built so far (schematic capture done in LTSpice; pots are depicted by close R pairs);
No tone stack - SFW? Seems I can get plenty of variation in that just from the guitar! One aspect I like is it's pretty much hum free, while using the steel chassis for ground as per OEM build. With the actual guitar speaker instead of some rando, it sounds better than I have the ability to play. I'm afraid to turn it up for long, as there are neighbors and the SPL is much higher from this than my ordinary garage studio musings.
Yes, that's 3 volumes - the last I can dial down the output to zero and just listen to the 1st three differential stages. The preamp sound is not nearly as good as when the power amp is workin' at it as well. I have my suspicions as to why that is and what could be done as an educated guess to mitigate that.
So why the LTP phase splitter as the input tube and carry differential signals through the amplification chain? The answer lies in the next step I'd like to take, which is to compress just one of the two signal paths. That's going to involve generating a DC envelope control signal from the input amplitude and some way to attenuate an AC signal in proportion to the envelope. Also attack and decay time constants, which I'm hoping will be adjustable to give the amp's sound a musically usable dynamic tonal character. Something you can ride on while playing.
I have some Vactrols on the way and am looking at 5670 tubes available at...less than what I paid for the whole amp and everything I've got into it so far, including speaker. The 5670 is variable mu with grid bias;
Not that I'd know what to do with it, other than look at how others have used this tube in compressor application. It would have to go in place of the 2nd 12AX7 and I'd be varying the gain of both to just one side, hopefully with a continuous control pot. That way I can go from straight compression to collapsing the signal of just one side. The Vactrol is probably more my speed to design with. Getting "linear decibel" compression ala what DBX worked out all those years ago may be tough - any help would be appreciated.
Mr Collins worked out a box for Ham Radio voice compression ages ago - it's fun to read the Reverb sellers gush over that - what did I see, $3k? - circuit; "you can put any musical signal through it - dont even have to use the compression function - and its sound better". Gee, I wonder why...
Thanks, for any input on these ideas. Example; I have no idea what the transformers primary / secondary impedance would be; 10K:10k?
No tone stack - SFW? Seems I can get plenty of variation in that just from the guitar! One aspect I like is it's pretty much hum free, while using the steel chassis for ground as per OEM build. With the actual guitar speaker instead of some rando, it sounds better than I have the ability to play. I'm afraid to turn it up for long, as there are neighbors and the SPL is much higher from this than my ordinary garage studio musings.
Yes, that's 3 volumes - the last I can dial down the output to zero and just listen to the 1st three differential stages. The preamp sound is not nearly as good as when the power amp is workin' at it as well. I have my suspicions as to why that is and what could be done as an educated guess to mitigate that.
So why the LTP phase splitter as the input tube and carry differential signals through the amplification chain? The answer lies in the next step I'd like to take, which is to compress just one of the two signal paths. That's going to involve generating a DC envelope control signal from the input amplitude and some way to attenuate an AC signal in proportion to the envelope. Also attack and decay time constants, which I'm hoping will be adjustable to give the amp's sound a musically usable dynamic tonal character. Something you can ride on while playing.
I have some Vactrols on the way and am looking at 5670 tubes available at...less than what I paid for the whole amp and everything I've got into it so far, including speaker. The 5670 is variable mu with grid bias;
Not that I'd know what to do with it, other than look at how others have used this tube in compressor application. It would have to go in place of the 2nd 12AX7 and I'd be varying the gain of both to just one side, hopefully with a continuous control pot. That way I can go from straight compression to collapsing the signal of just one side. The Vactrol is probably more my speed to design with. Getting "linear decibel" compression ala what DBX worked out all those years ago may be tough - any help would be appreciated.
Mr Collins worked out a box for Ham Radio voice compression ages ago - it's fun to read the Reverb sellers gush over that - what did I see, $3k? - circuit; "you can put any musical signal through it - dont even have to use the compression function - and its sound better". Gee, I wonder why...
Thanks, for any input on these ideas. Example; I have no idea what the transformers primary / secondary impedance would be; 10K:10k?
Last edited:
The compressor circuit that you posted looks like part of a Fairchild 570 or a Gates Sta-Level. Both used the magic 6386 tube. The 6386 is exactly like the common and dirt cheap 5670 EXCEPT for the control grid. The 5670 is a twin triode RF amplifier tube from the WWII era. It is not variable Mu and is quite linear for audio preamp use. The 6386 is the variable Mu "remote cutoff" variant found in these compressors. Used tubes seem to sell for $50 to $100. There is (or was) a new production version that sold for over $100 each but for some reason it was not well liked. I built the Voltage Controlled Amplifier cell from the Fairchild 570 and tinkered with tubes and transformers until I had a good working design with no expensive tubes.
I also experimented with Vactrols. The drawback here is the delayed release time. When the LED lights up, the LDR drops in resistance quickly from megohms to ohms. When the LED goes dark the resistance goes up into the tens of K ohms pretty fast but can take 100 mS or more to reach the megohm region. Some are faster than others and the good ones like you got are graded out by release time. This may or may not be an issue in your circuit. I got a big bag full of "surplus" Vactrols from ma surplus dealer cheap, but they are likely factory rejects.
My version of the VCA used in the Fairchiild 570 was started in post #25 of this thread, I can dig up the schematic for it if you are interested. It did use low buck Edcor transformers and I think that they were 10K to 10K, but I would need to dig up the paperwork to be sure:
https://www.diyaudio.com/community/threads/modular-synth-build.258400/page-2
I will be away for most of today, but I can find the old VCA stuff tonight or tomorrow if you are interested. It is a fully differential design like the Fairchild.
Anyone starting down the guitar amp design road should see the Dogstar design. It was a site I used to visit often many years ago. The site owner passed away nearly 20 years ago and many here have probably never seen it. Those 20 years have certainly flown by. Fred Nachbaur had a web site for his electronics adventures, many of which involved vacuum tubes, and another site for the music he wrote, performed, and recorded. They are still being hosted. The electronics site has a Vacuum Tube Projects page, that includes some guitar amps. The Dogzilla is a must see. It will probably hold the record for the most tubes in a guitar amp forever at 18!
http://www.dogstar.dantimax.dk/
I also experimented with Vactrols. The drawback here is the delayed release time. When the LED lights up, the LDR drops in resistance quickly from megohms to ohms. When the LED goes dark the resistance goes up into the tens of K ohms pretty fast but can take 100 mS or more to reach the megohm region. Some are faster than others and the good ones like you got are graded out by release time. This may or may not be an issue in your circuit. I got a big bag full of "surplus" Vactrols from ma surplus dealer cheap, but they are likely factory rejects.
My version of the VCA used in the Fairchiild 570 was started in post #25 of this thread, I can dig up the schematic for it if you are interested. It did use low buck Edcor transformers and I think that they were 10K to 10K, but I would need to dig up the paperwork to be sure:
https://www.diyaudio.com/community/threads/modular-synth-build.258400/page-2
I will be away for most of today, but I can find the old VCA stuff tonight or tomorrow if you are interested. It is a fully differential design like the Fairchild.
Anyone starting down the guitar amp design road should see the Dogstar design. It was a site I used to visit often many years ago. The site owner passed away nearly 20 years ago and many here have probably never seen it. Those 20 years have certainly flown by. Fred Nachbaur had a web site for his electronics adventures, many of which involved vacuum tubes, and another site for the music he wrote, performed, and recorded. They are still being hosted. The electronics site has a Vacuum Tube Projects page, that includes some guitar amps. The Dogzilla is a must see. It will probably hold the record for the most tubes in a guitar amp forever at 18!
http://www.dogstar.dantimax.dk/
I looked at the perf board VCA and found that the input transformer is an Edcor UMM-8 (20K CT to 20K CT) and the output transformer is a PCW 10K-10K (10K CT to 10K CT)
The Dogzilla amp I mentioned has a single ended compressor system that used a 6K7 tube for the variable gain stage. There are dozens of 7 and 9 pin alternatives that could be used, and my favorite from the "make stuff from old radios" days of my youth, the 6SK7.
The Dogzilla amp I mentioned has a single ended compressor system that used a 6K7 tube for the variable gain stage. There are dozens of 7 and 9 pin alternatives that could be used, and my favorite from the "make stuff from old radios" days of my youth, the 6SK7.
Attachments
I tried the 12BA6 trick and it does work in the original 570 circuit, but after getting it to work I came to the realization that I was doing it all wrong. The 570 and it's brethren use a rectified and filtered version of the audio signal to drive a rectifier and generate control voltage which varies from zero to 10's of volts. That's why they use a pair of 6V6 tubes to drive the rectifier. I'm working (slowly) on a vacuum tube modular synthesizer where the control voltage range is from -5 to +10 volts. Ordinary sharp cutoff pentodes work best here. I think those are 6KT6's since I have several hundred of them.
So I took some data on the 1st stage, a simple 12AX7 LTP with about 4 mA of current sink at the common cathodes. 1st, since it's been awhile since I look at this, I wanted to see what emerges from my guitar as a signal when I...thrash the strings, everything wide open, neck pickup where the string displacement us greatest;
(Luckily my Fluke scope can print...to postscript. It takes going though a few operations to make this post-able image, including waiting for the data to chug through an optically isolated serial connection at 9600 baud)
Observing this, I set the function generator amplitude "accordingly" - about 6V p-p. I chose ~440Hz for the time being.
Then, looking at the two outputs this stage provides, I'm getting;
This is with the 1st volume control turned all the way down, as driving the next stage's grids distorts these signals. This would be with ~250k load; the value of the dual potentiometer. Now, I'd like to present an REW harmonic analysis, just havent got around to arranging an interface that wont load this high voltage circuit (much) and stay within the input range of that device.
So, here's the signal gain level I have, with the "pre" control metering it into the next stage, the so-called "cold clipper". Which basically lops off bottom 1/2 of the cycle. It seems to be a common signal modification technique used in guitar amps.
(Luckily my Fluke scope can print...to postscript. It takes going though a few operations to make this post-able image, including waiting for the data to chug through an optically isolated serial connection at 9600 baud)
Observing this, I set the function generator amplitude "accordingly" - about 6V p-p. I chose ~440Hz for the time being.
Then, looking at the two outputs this stage provides, I'm getting;
This is with the 1st volume control turned all the way down, as driving the next stage's grids distorts these signals. This would be with ~250k load; the value of the dual potentiometer. Now, I'd like to present an REW harmonic analysis, just havent got around to arranging an interface that wont load this high voltage circuit (much) and stay within the input range of that device.
So, here's the signal gain level I have, with the "pre" control metering it into the next stage, the so-called "cold clipper". Which basically lops off bottom 1/2 of the cycle. It seems to be a common signal modification technique used in guitar amps.
Skipping around just a bit, this is sort of what I'm aiming at for this amp to do. I derived an envelope from the two cathode resistors of the 12AU7 driver stage, briefly; a signal diode from each cathode to a common node, a 6.3V Zener to drop off the DC component, a 1K "attack" resistor feeding a 100uF cap, with a 10K across it for "decay".
This gives a DC envelope waveform of appx 5V amplitude, with the first two volumes wide open. About 100K feeds current into the Vactrol LED, the output of which loads across just one of the "post" volume control's input. I did provide 56k of padding from the driver tube output into each of the 250k control's inputs, after which the Vactrol yanks the signal toward the negative fixed bias voltage of the output tube.
Ultimately, I'd like to make the attack, decay and effect amount have panel controls, but not there "yet" a day after getting the parts...
1st thing I noticed, when soldering to the Vactrol LED side, the heat just wipes out whatever goop they use to hold it in place on that end. I turned down my solder temp to 325 and after attaching leads, the LED part was visibly wiggling in the body. So, glue and overnight dry...I have two more parts.
Looking at the signal driving the effected output tube grid with this arrangement;
Let's pretend the lines I (barely) drew represent the sound envelope of the opposite output tube in the push-pull circuit, which is the ordinary decay of the plucked string. So we have an initial attack where both tubes are operating push-pull, and about a second later, it's running more toward "single-ended" and then ~6 seconds out, back to more of balanced push-pull drive into the output transformer.
It's...something you can hear; the sound gets thinner initially, then fattens up as the un-effected tube's mate comes back on to meet up with it again drive wise. The time constants and depth of effect I assume can be adjusted; I only picked something quick (1k, 10k, 100u, 100k) to explore. Can it be made into a musically useful sound? Without op-amps and +/- 15V power supplies?
This gives a DC envelope waveform of appx 5V amplitude, with the first two volumes wide open. About 100K feeds current into the Vactrol LED, the output of which loads across just one of the "post" volume control's input. I did provide 56k of padding from the driver tube output into each of the 250k control's inputs, after which the Vactrol yanks the signal toward the negative fixed bias voltage of the output tube.
Ultimately, I'd like to make the attack, decay and effect amount have panel controls, but not there "yet" a day after getting the parts...
1st thing I noticed, when soldering to the Vactrol LED side, the heat just wipes out whatever goop they use to hold it in place on that end. I turned down my solder temp to 325 and after attaching leads, the LED part was visibly wiggling in the body. So, glue and overnight dry...I have two more parts.
Looking at the signal driving the effected output tube grid with this arrangement;
Let's pretend the lines I (barely) drew represent the sound envelope of the opposite output tube in the push-pull circuit, which is the ordinary decay of the plucked string. So we have an initial attack where both tubes are operating push-pull, and about a second later, it's running more toward "single-ended" and then ~6 seconds out, back to more of balanced push-pull drive into the output transformer.
It's...something you can hear; the sound gets thinner initially, then fattens up as the un-effected tube's mate comes back on to meet up with it again drive wise. The time constants and depth of effect I assume can be adjusted; I only picked something quick (1k, 10k, 100u, 100k) to explore. Can it be made into a musically useful sound? Without op-amps and +/- 15V power supplies?
Last edited:
I'm pretty unfamiliar with the "common values" of decay times used in compressor devices. Obviously they are adjustable. I know the value I'm using for decay is way longer than 100 ms. What I really need is a circuit that can do ASDR from the guitar signal input - or as many of those letters as I can get - that can be adjusted to taste by a guitarist.
That's why my post volume meters to the output tube grids... Hard to fiddle at performance SPLs.
I got a "working" LTSpice model, based on the Duncan Amp's work as the suggested starting point. Left the output stage as-is, same as I did with the Bogen hardware. A Vactrol model I found and used is suspect, but good enough to get me in the ballpark at least topology wise - and better than soldering wires as I "imagineer" how to make a control envelope of the input signal, where to get the signal for it within the circuit and so on. Here's what it looks like now -
Since what I'm looking for is behavior under transient conditions, I attempted to construct an elementary string pluck and subsequent decay as my input. I'm sure some noise at the outset and some "DC" in exponentially decaying form is present in a real string, which I havent yet included - it's a start in approach.
That goes as well with the 3 potentiometers; I know there's LTSpice pot models available, have yet to get around to including those; the fixed value pairs at some ratio adjustment will have to do for now.
If anyone would like to try, here's the .zip with I believe all that is needed to run the above diagram. If anyone would like to help me out, perhaps with better models for the tubes, a better Vactrol model (I have VTL5C3), a smarter way to effect what I'm trying to do - that would be most appreciated. If someone perceives it's a bit anemic with all those 10n coupling caps, suggest away!
If someone wants to tear out all the 2nd tubes in each diff pair and make it into a more ordinary guitar amp design - I hope this can get you going!
My ultimate goal is to provide a link to a recording of how the amp sounds, when played using electric guitar. Hopefully by someone far more skilled in that art than me ;')
Since what I'm looking for is behavior under transient conditions, I attempted to construct an elementary string pluck and subsequent decay as my input. I'm sure some noise at the outset and some "DC" in exponentially decaying form is present in a real string, which I havent yet included - it's a start in approach.
That goes as well with the 3 potentiometers; I know there's LTSpice pot models available, have yet to get around to including those; the fixed value pairs at some ratio adjustment will have to do for now.
If anyone would like to try, here's the .zip with I believe all that is needed to run the above diagram. If anyone would like to help me out, perhaps with better models for the tubes, a better Vactrol model (I have VTL5C3), a smarter way to effect what I'm trying to do - that would be most appreciated. If someone perceives it's a bit anemic with all those 10n coupling caps, suggest away!
If someone wants to tear out all the 2nd tubes in each diff pair and make it into a more ordinary guitar amp design - I hope this can get you going!
My ultimate goal is to provide a link to a recording of how the amp sounds, when played using electric guitar. Hopefully by someone far more skilled in that art than me ;')
Last edited:
I'd definitely be interested.
Last night, I found and transformed a 5670 model from a TINA to LTSpice, which only required replacing a few characters. Dropping it into my simulation in place of the 12AU7 drivers, I now get why some of these early tube compressors used a differential design. It's not just out of convenience to accommodate a 600 Ohm mic input, using a transformer which can easily be center tapped on its secondary.
Though I'm pretty sure those transformers used contributed to the sound of the device. You read quotes about that Fairchild unit such as "Everything sounds great through it, especially vocals, piano and drums. I like it on my master buss too, even without compressing at all, just for color” and it makes you wonder.
So, putting the phase splitter as the 1st tube in a guitar amp allows subsequent circuitry to do compression. Tube compression. Then you read something supposedly attributable to Pete Townsend, referring to the Fairchild unit "They were rolled out like secret weapons to bring an electric guitar to life". I'm sure Pete doesnt know what he's talking about, regarding electric guitar...
Manley audio says you can replace the 6386 with a pair of 6/12BA6s. The 5670 isnt like $100 to buy and I think it would be fun to try and see if it behaves like the simulation predicts. The Ampeg amp has a floating 70V winding on its OPT. In my mind, that can be full wave bridge rectified and fed to a similar attack decay circuit in pretty much that same way as in the Fairchild, which can then control the tube gain stage. Which - should - net a similar fast attack behavior the Fairchild was most famous for - at one of the many envelope filter selections I see, just after that full wave bridge in their circuit.
Another interesting tidbit comes from a Rane service manual, for their MA 6 amp, which I have weighing down my garage floor. "Field measurements show a minimum 4 dB increase in continuous SPL, which is the equivalent SPL of a 250 watt amplifier without limiting. As many customers have proclaimed, “This is the loudest 100 watt amplifier, I’ve ever heard.” They say their "limiter" is inaudible, other than it makes the amp sound louder.
So tubes, compression / limiting, interstage transformers; the proper mix and match is all about the sound that comes out. I - FWIW - look at these circuits and remark to myself "why would anyone build a guitar amp, any other way - it's all right there...been right under everyone's collective nose all this time!" Well, cost. Instead, you can just follow an op-amp with a fender tone stack, a couple diodes with a switch to ground and a chip amp - done! But it's not going to sound like building the schematic below, with a few mods to accommodate a speaker and 1/4 input jack.
Reminds me about an amp that I found interesting due to the pentode PI.
With "triodizing" plate to grid feedback resistors and cross coupled feedback from the output tube plates, throw in some GNFB and you need a box full of pentodes to get enough gain....it must be a Citation II, a well regarded amp.
With "triodizing" plate to grid feedback resistors and cross coupled feedback from the output tube plates, throw in some GNFB and you need a box full of pentodes to get enough gain....it must be a Citation II, a well regarded amp.
Ok I ordered a 5670 and a 6ES8 and a couple of 15k CT transformers off ebay; we'll see what I can do with them. The transformers dont have a good low end response even for guitar, but I'm not paying $90 for one just to try to see what it sounds like.
The Chinese have discovered such "isolation" transformers have other uses and effect the sound of audio enough to start manufacturing them with symmetrical 10K windings, solely for that purpose, and they go down to 20 Hz, supposedly. It's hard to find a reasonable one otherwise, because it's audio and you pay that bill for a positive effect. You also pay the bill in time it takes for a shipment to arrive; I can get them when I can no longer tolerate the low end response of the $8 jobs I ordered.
The Chinese have discovered such "isolation" transformers have other uses and effect the sound of audio enough to start manufacturing them with symmetrical 10K windings, solely for that purpose, and they go down to 20 Hz, supposedly. It's hard to find a reasonable one otherwise, because it's audio and you pay that bill for a positive effect. You also pay the bill in time it takes for a shipment to arrive; I can get them when I can no longer tolerate the low end response of the $8 jobs I ordered.