In the last post I showed a couple of pictures of my first attempt at a vacuum tube Moog style ladder filter whose cutoff frequency is varied by a control voltage. This is called a Voltage Controlled Filter, or VCF. The filter typically follows a 1 volt per octave (1V/oct) exponential curve with 4 to 7 octaves of range. That board fell woefully short barely covering 1 octave with 20 volts of control.
I also posted a picture of a digital 1V/oct synth breadboard that I have been working with. It is my intention that, in time, the two breadboards will evolve into a modular synth that encompasses analog and digital modules constructed with tubes, transistors, and IC chips, or a mix of these within a given module. A complete synth station with capability for Eurorack and RU (Moog style) modules is under construction.
I will use whatever works best for the chosen task. Vacuum tubes are known for optimum sound under mild to extreme overload conditions, but impose power and heat load conditions on the rest of the system. Modern digital synths can stay in tune under extreme conditions, but can sound rather "clinical." There is some middle ground in between these extremes, and I aim to find it.
A close look at the discussion surrounding the Knifonium vacuum tube analog synth reveals 6AL5 dual vacuum diodes in the VCF. This prompted me to rip the top 3 dual triodes from my ladder and replace them with 6JU8 quad vacuum diodes (two 6AL5's in a 9 pin bottle). This yielded a working VCF. It's still a bit limited in range, and I'm not happy with the sound yet, but it's one of several paths I must explore.
In the weeks to come, I plan to try some unknown elements in the ladder.
I also posted a picture of a digital 1V/oct synth breadboard that I have been working with. It is my intention that, in time, the two breadboards will evolve into a modular synth that encompasses analog and digital modules constructed with tubes, transistors, and IC chips, or a mix of these within a given module. A complete synth station with capability for Eurorack and RU (Moog style) modules is under construction.
I will use whatever works best for the chosen task. Vacuum tubes are known for optimum sound under mild to extreme overload conditions, but impose power and heat load conditions on the rest of the system. Modern digital synths can stay in tune under extreme conditions, but can sound rather "clinical." There is some middle ground in between these extremes, and I aim to find it.
A close look at the discussion surrounding the Knifonium vacuum tube analog synth reveals 6AL5 dual vacuum diodes in the VCF. This prompted me to rip the top 3 dual triodes from my ladder and replace them with 6JU8 quad vacuum diodes (two 6AL5's in a 9 pin bottle). This yielded a working VCF. It's still a bit limited in range, and I'm not happy with the sound yet, but it's one of several paths I must explore.
In the weeks to come, I plan to try some unknown elements in the ladder.
Tubelab, keep up the good work!
I really want to get back to my synth..
But life, you know.. Wife had a baby, etc...
Oh, but you'd appreciate this..
My baby likes sounds.. loves to hear new noises. So, I got out my modular and set it on the floor and patched together a little sequence. He walks up to it (~1 year old at the time), more confident than I've seen him do most things, and starts tweaking knobs!! Never seen a synth before in this life, and seriously walks up to it and twists knobs on the wave shaper, filter, vco, like he knows what he's doing!
That's enough to make me philosophical. Genetic knowledge? Reincarnation? Life's crazy...
I really want to get back to my synth..
But life, you know.. Wife had a baby, etc...
Oh, but you'd appreciate this..
My baby likes sounds.. loves to hear new noises. So, I got out my modular and set it on the floor and patched together a little sequence. He walks up to it (~1 year old at the time), more confident than I've seen him do most things, and starts tweaking knobs!! Never seen a synth before in this life, and seriously walks up to it and twists knobs on the wave shaper, filter, vco, like he knows what he's doing!
That's enough to make me philosophical. Genetic knowledge? Reincarnation? Life's crazy...
I'm certain that his rock star persona and dress style will be miles away from Pampers.
+1
All I have from my childhood are a few black and white pictures of my brothers and I posing for the camera, and an B&W 8mm movie of me picking up a rather unhappy cat by its tail. I was 4 or 5 at the time. Home video was still 20 some years in the future. There isn't a single picture of any of my creations, any smoke, flames or anything cool.
It's been over two and a half years since I, or anyone else has posted in this thread. In that time my own modular synth has been built, then taken apart and rebuilt again, twice. I outgrew the cases, so they were sold and a new BIGGER case is being built with a BIGGER power supply.
To some this may be a crime, but some of my favorite "modules" are Behringer synths. The Neutron and D are both capable of being dropped right into a Eurorack cabinet and powered with the Eurorack power supply. The Deepmind 12 needs a 5U rack case.
I have also been tinkering on and off with vacuum tube based synth modules. They are bigger, and need about 200 volts for the tubes, hence the "universal" power supply and big cabinet that accommodates multiple different size modules.
The 4 tube ladder filter seen in post #20 is in pretty much the same shape that it was back then, except for some parts changes that have made it work a bit better. Then there was this diversion to a Vactrol based ladder. Awesome sound, except for the delay when trying to slam the filter shut. More experiments are planned.
Another element needed for any filter is the VCA, so I decided to clone the best known vacuum tube VCA out there, the voltage controlled gain cell from the Fairchild 670 compressor. This is a most excellent place to start, but there is just one problem with it. The Fairchild gain cell uses 4 identical tubes per channel, GE6386's. They were pretty common right after WWII since they were used in some mitary radio equipment of that era, but were never made since then, current cost, about $125 EACH!!!!!! That makes this little baggie full of glass very valuable......
So, I simplified the Fairchild design down to it's minimum essentials for functionality, and ordered some budget friendly transformers, then built a one tube version.....It worked extremely well on the first try. Dynamic range was over 70 dB, THD in the 0.4% range at full output, rising to about 1% -50 dB and under 100 Hz......not bad but still too expensive.
Over the next few weeks I plan on fully optimizing it, and characterizing it over a wide range of operating conditions. I have 21 of the 6386 tubes, but only one has been tried in this board. It worked, so will the others be better, worse or the same?
After that, I will be working on making it work with a tube or tubes that costs $5 or less. I have several candidates to try already.
For now I'm using a Behringer model D for the VCO signal. I have not yet decided what the final synth will use for a VCO / signal source. They are just too many nice options that can't be done with tubes......vacuum tube wavetable source......I don't think so.
To some this may be a crime, but some of my favorite "modules" are Behringer synths. The Neutron and D are both capable of being dropped right into a Eurorack cabinet and powered with the Eurorack power supply. The Deepmind 12 needs a 5U rack case.
I have also been tinkering on and off with vacuum tube based synth modules. They are bigger, and need about 200 volts for the tubes, hence the "universal" power supply and big cabinet that accommodates multiple different size modules.
The 4 tube ladder filter seen in post #20 is in pretty much the same shape that it was back then, except for some parts changes that have made it work a bit better. Then there was this diversion to a Vactrol based ladder. Awesome sound, except for the delay when trying to slam the filter shut. More experiments are planned.
Another element needed for any filter is the VCA, so I decided to clone the best known vacuum tube VCA out there, the voltage controlled gain cell from the Fairchild 670 compressor. This is a most excellent place to start, but there is just one problem with it. The Fairchild gain cell uses 4 identical tubes per channel, GE6386's. They were pretty common right after WWII since they were used in some mitary radio equipment of that era, but were never made since then, current cost, about $125 EACH!!!!!! That makes this little baggie full of glass very valuable......
So, I simplified the Fairchild design down to it's minimum essentials for functionality, and ordered some budget friendly transformers, then built a one tube version.....It worked extremely well on the first try. Dynamic range was over 70 dB, THD in the 0.4% range at full output, rising to about 1% -50 dB and under 100 Hz......not bad but still too expensive.
Over the next few weeks I plan on fully optimizing it, and characterizing it over a wide range of operating conditions. I have 21 of the 6386 tubes, but only one has been tried in this board. It worked, so will the others be better, worse or the same?
After that, I will be working on making it work with a tube or tubes that costs $5 or less. I have several candidates to try already.
For now I'm using a Behringer model D for the VCO signal. I have not yet decided what the final synth will use for a VCO / signal source. They are just too many nice options that can't be done with tubes......vacuum tube wavetable source......I don't think so.
Attachments
Dogzilla has a cheap varimu compressor that can easily be adapted to modular.
Introducing DOGZILLA
Introducing DOGZILLA
I have not looked into the Elector synth at all.
I have been building music synthesizer stuff since I saw ELP in concert in early 1971. I came home from that show and ordered a PAIA 2700 kit which took about 6 months to arrive. I started on a full blown DIGITAL modular system during that delay. The disagreements between myself and my father turned ugly before it was finished, and I left home with only a car full of stuff. Both synths were lost. I did grab a bunch of boards for the digital modular, which I still have nearly 50 years later.
I got a job at a Motorola electronics plant and within a few years I was making more money than I could spend, so I bought an ARP Odyssey and later a Little Brother expander (one more VCO, a second VCA and VCF). I had also connected up several modules from an Aries modular rig that I bought cheap because it never worked right.
My stupidest move was selling all the vintage stuff for next to nothing at the low point of their worth....the digital 80's.
I know what I need to make a conventional modular, but again I am taking a rather unusual path. I have a rack full of modules, of different sizes and formats, including some MFOS and DIY stuff, but when you can buy a Behringer MiniMoog Model D clone for less than some Eurorack modules, I took that short cut.
I have built a LF sequencer. It's sort of like a conventional step sequencer, but it can step really fast and can be retriggered (started over) mid sequence. It also has a variable slew rate (like a portamento knob) so that I can draw out an ADSR type wave on the knobs and trigger it with the keyboard gate signal. Draw a sine or other wave, and set it on repeat, it's an LFO. All digital until the bits hit the D/A chip...then it's CV, patched to the mod in. Flip the speed switch and it slows down and becomes a step sequencer. Patch it to the CV in on the VCO's, and it's a step sequencer.
I want to design and build a vacuum tube synthesizer just because I'm Tubelab. I know it can be done, and I know what needs to be done. I'm leaving the VCO for last, because I will likely use some digital technology somewhat similar to what Dave Smith does in his modern version of the Prophet synths. A square wave sounds like a square wave no matter what makes it, ditto a sawtooth. Whether it is made by a tube with some digital assistance, or directly from a digital chip, I haven't decided yet.
I have played with at least a dozen different VCF both analog and digital. Nothing sounds like a Moog ladder filter, so that's where I have spent much of my design time. My vacuum tube ladder still needs some work.
That circuit uses a 6K7 tube for it's gain cell. I did not try them, but I have experimented with the newer and slightly better 6SK7 since I was a kid. It is a remote cutoff pentode, and should have a wide dynamic range. Both the 6K7 and 6SK7 are too big for modular synth use. I need to try some smaller tubes.
I built the 6386 breadboard for benchmarking purposes. I took the data on it, then stuffed in some pin compatible 5670 tubes, which did not work well. That was expected, but I have a bunch and they had the same pinout, so I can cross those off my list.
I rewired the tube sockets to fit tubes with the 9AJ compatible pinout. The list is long with about 20 different type numbers to try. I'm about halfway through them now.
After that I will rip out the dual triode socket, and stick in two sockets for some small remote cutoff pentodes and some dual control pentodes. Some, like the 6BA6 are miniature versions of the 6K7 / 6SK7 type tubes. My guess is that one of them will wind up in my VCA. Time will tell.
The UI is where they money is in many synths. Knobs, pots, and switches cost money, often more money than all the other parts. That blue faced digital synth that I showed in post #20 has 49 pots with knobs and 4 rotary encoders. That's almost $150 in UI. The rest of the synth costs about $35. It now has a 1V/oct keyboard attached to it so it's playable, and I have finally got it to track the "D". The whole UI could be reduced to one or two knobs, and a bunch of menus, which nobody likes, but greatly reduces the cost and size......there is a happy medium somewhere.
> That circuit uses a 6K7 tube for it's gain cell.
There's two kinds of "VCA". A radio takes a variable small signal and makes it uniformly big. A volume control (or a synth) takes a known-level signal and makes it as small as desired.
These two applications typically lead to very different implementation. The radio (or music limiter) often starts from a naked device(s) with no NFB. The synth-type would often start from a high-NFB amplifier and use current-steering to divert signal.
A tube current-steer is probably obvious/trivial to you, but here's a sketch for those following:
There's two kinds of "VCA". A radio takes a variable small signal and makes it uniformly big. A volume control (or a synth) takes a known-level signal and makes it as small as desired.
These two applications typically lead to very different implementation. The radio (or music limiter) often starts from a naked device(s) with no NFB. The synth-type would often start from a high-NFB amplifier and use current-steering to divert signal.
A tube current-steer is probably obvious/trivial to you, but here's a sketch for those following:
Attachments
There is a nifty single tube way of doing something similar. The tube is called a sheet-beam tube. The 6AR8 sitting on my desk in front of me is one. There are a few others, but most are rather rare.....I'll test this one any way. It's somewhat like a conventional pentode with a pair of CRT like deflection plates to steer the beam between one of two plates. The beam steering can be gradually done from one plate to the other so that the total tube current, and thus it's Gm remain constant.
These came late in the color TV world and didn't catch on well enough for serious volume production before silicon took over the TV world. Those that were made have often found their way into SSB modulators / demodulators.
These came late in the color TV world and didn't catch on well enough for serious volume production before silicon took over the TV world. Those that were made have often found their way into SSB modulators / demodulators.
Far be it from me to tell you how to skin your cats.
Yes, the deflection tube is the same thing. Solo, there's a DC shift, acceptable for HF Chroma but annoying for audio. The obvious fix is two of them, just like above (this also gives a second audio input for cross-mixing). Now it's up to two odd bottles (I suppose you have buckets of them) opposed to three common twin-triodes. In return for the extra piece of glass, we can shove so much resistance up their cathodes that the tube parameters hardly matter and CV linearity is good (but is that important?).
Yes, the deflection tube is the same thing. Solo, there's a DC shift, acceptable for HF Chroma but annoying for audio. The obvious fix is two of them, just like above (this also gives a second audio input for cross-mixing). Now it's up to two odd bottles (I suppose you have buckets of them) opposed to three common twin-triodes. In return for the extra piece of glass, we can shove so much resistance up their cathodes that the tube parameters hardly matter and CV linearity is good (but is that important?).
I looked at that schematic when you posted it last night. It looked familiar and sort of reminded me of something. The 6AR8 sitting on the bench crying out "Try me" captured my attention, so I mentioned it and went back to rewiring my breadboard to test a long list of 7 pin remote cutoff pentodes in the same circuit that I used to test all the dual triodes.
Later last night while drifting off to sleep, that annoying light bulb in my head that keeps me awake at night lit up and told me ....hey idiot that's the multiplier circuit that you spent about a year optimizing for UHF mixer / demodulator use. That was an all mosfet design which is now buried many layers deep inside one of the chips in Motorola's latest police radios.
I may stick it into LT spice later and see what the simulator says. If there is some light at the end of the tunnel, I'll build one. Three tubes VS one or two, space, heat, and heater power are concerns, but so is performance.
.
CV linearity isn't too important as long as it's somewhat consistent. There will be a microcontroller present to deal with MIDI data. It converts MIDI to CV so some math can be added to the conversion as needed.
Many synth builders view distortion as OK, and even a good thing. Many of the popular Eurorack modules out there today derive their sound directly from the 12 bit DAC in a Teensy board......I do not favor this approach. I aim for a low distortion design to which I can add the sonic coloration of my choosing.
No, I only have one of the 6AR8's and I'm not even sure if it's any good. I do have a few other BDT's and SBT's around here somewhere, but only in single digit quantities. Dual triodes in several flavors, I have lots of.
CV linearity isn't too important as long as it's consistent. There is a microcontroller for MIDI to CV conversion and parameter storage. Some math can be applied to the VCA's CV input can be added as needed.
Last edited:
> It looked familiar ....hey idiot that's the multiplier circuit that you spent about a year optimizing for UHF .... an all mosfet design
If you peel the WhiteOut you will see I snagged a MOSFET drawing and pasted triodes over it.
The linearity can be good-excellent when you have enough supply voltage to use large cathode resistors. Then you can use any darn tubes. Two 3-T Compactrons if you swing that way. There could be price/linearity advantage using MOSFETs for the upper quad, keeping thermionics in the lower pair for "great tube sound".
If you peel the WhiteOut you will see I snagged a MOSFET drawing and pasted triodes over it.
The linearity can be good-excellent when you have enough supply voltage to use large cathode resistors. Then you can use any darn tubes. Two 3-T Compactrons if you swing that way. There could be price/linearity advantage using MOSFETs for the upper quad, keeping thermionics in the lower pair for "great tube sound".
I ordered a box full of tubes from ESRC so that I can try several different designs of vacuum tube VCA's. So far I have toyed with a simple twin triode diff amp with $12 Edcor transformers on the input and output. It's basically a simplified version of the gain cell lifted from the Fairchild 670 compressor. I ran at least a dozen different twin triodes through it, and none come close to the 6386's.
I then rewired the board for with two 7 pin sockets, and was going to try some of them, but to my dismay my large collection of 7 pin tubes is gone...... I now vaguely remember trading it for a guitar at a hamfest in Florida 5 or 6 years ago. So I just ordered a box of small tubes.
So I rip it up and rewire it again. This time for some 6BV11's, a dual pentode intended for color TV demodulation. This works quite well until I get down to some losses in the -30 db range and the distortion goes up. I am in the process of some serious knob twiddling when there is a bang, the distortion drops to below 1% and smoke is coming from my board.
Power off, autopsy.......
My trusty 50+ year old Fluke 407D had decided to stop regulating and crank out about 400 volts. That split a 250 volt cap in half on the board.
Second autopsy, rip apart the Fluke.....OK, it's 50+ years old, I bought it in "untested" condition on Ebay for $25 somewhere between 20 and 30 years ago. I took it apart to fix a loose switch when I got it, saw that it was all original, then put it back together. I expected the worse, but the autopsy had revealed that the loose voltage adjust pot has wiggled around enough in several years that one of the wires had broken off. Solder the wire, tighten up the pot, put back together.....done.
The tube melting mayhem will resume tonight......
I then rewired the board for with two 7 pin sockets, and was going to try some of them, but to my dismay my large collection of 7 pin tubes is gone...... I now vaguely remember trading it for a guitar at a hamfest in Florida 5 or 6 years ago. So I just ordered a box of small tubes.
So I rip it up and rewire it again. This time for some 6BV11's, a dual pentode intended for color TV demodulation. This works quite well until I get down to some losses in the -30 db range and the distortion goes up. I am in the process of some serious knob twiddling when there is a bang, the distortion drops to below 1% and smoke is coming from my board.
Power off, autopsy.......
My trusty 50+ year old Fluke 407D had decided to stop regulating and crank out about 400 volts. That split a 250 volt cap in half on the board.
Second autopsy, rip apart the Fluke.....OK, it's 50+ years old, I bought it in "untested" condition on Ebay for $25 somewhere between 20 and 30 years ago. I took it apart to fix a loose switch when I got it, saw that it was all original, then put it back together. I expected the worse, but the autopsy had revealed that the loose voltage adjust pot has wiggled around enough in several years that one of the wires had broken off. Solder the wire, tighten up the pot, put back together.....done.
The tube melting mayhem will resume tonight......
I did get to try some basic knob twiddling with the 6BV11 VCA today. I found that it works best at the upper limits of just about every spec. Cathode current, screen voltage, plate voltage, screen dissipation, and plate dissipation (but not simultaneously). They also have 4 nice blue stripes on the glass when the gain (and plate current) is maximized.
These are dual control pentodes. You put audio into G1, and vary the gain by changing the voltage on G3.
When G3 is at zero volts, or slightly positive, the tube acts like a normal pentode. A pair in push pull will produce 15 db of stage gain or so, while putting out a 20 volt sinewave at 0.3 to 0.6% THD (10 different tubes tried). Here screen current is low, a mA or two, and plate current is 15 mA per pentode.
Crank G3 in the negative direction and the gain turns into a loss. At -10 volts the plate current is in the 1 mA region and the stage loss is about 40 db for 55 db of dynamic range. THD goes up as the gain goes down. The best tube hits about 1% at the point where my analyzer stops reading, while the worst tube is at 5%. The best tube is nearly on par with the 6386 VCA, but the worst tube beats all other twin triodes I tested. Unfortunately I only tested ONE 6386 before ripping up the circuit to make this one.
A big box of tubes should arrive here in a day or three, so I'll rip it up several times more before it's all over. I'm going to order a few more Edcor transformers so that I van have more than one working circuit at a time.
These are dual control pentodes. You put audio into G1, and vary the gain by changing the voltage on G3.
When G3 is at zero volts, or slightly positive, the tube acts like a normal pentode. A pair in push pull will produce 15 db of stage gain or so, while putting out a 20 volt sinewave at 0.3 to 0.6% THD (10 different tubes tried). Here screen current is low, a mA or two, and plate current is 15 mA per pentode.
Crank G3 in the negative direction and the gain turns into a loss. At -10 volts the plate current is in the 1 mA region and the stage loss is about 40 db for 55 db of dynamic range. THD goes up as the gain goes down. The best tube hits about 1% at the point where my analyzer stops reading, while the worst tube is at 5%. The best tube is nearly on par with the 6386 VCA, but the worst tube beats all other twin triodes I tested. Unfortunately I only tested ONE 6386 before ripping up the circuit to make this one.
A big box of tubes should arrive here in a day or three, so I'll rip it up several times more before it's all over. I'm going to order a few more Edcor transformers so that I van have more than one working circuit at a time.
I just came across this thread and, since I find every Tubelab post something to learn from, and loving the Diy spirit of this forum, I would share some info about 6386 sub.
I lately repaired few RETRO STA-Level rack units, those are used in recording studios and based on the Gates STA-Level.
The original schematic shows 6386 but in the RETRO two cheap and easy to find 6BA6 are being used. Those are remote control pentodes, but in the unit are connected as triodes, g3 is connected to k, g2 goes to plate by 1k ohm resistor.
I hope this will help you
I lately repaired few RETRO STA-Level rack units, those are used in recording studios and based on the Gates STA-Level.
The original schematic shows 6386 but in the RETRO two cheap and easy to find 6BA6 are being used. Those are remote control pentodes, but in the unit are connected as triodes, g3 is connected to k, g2 goes to plate by 1k ohm resistor.
I hope this will help you
Last edited:
The Manley clone compressor can be wired to accept 6386, 5670 or 6BA6.
I find that the 5670 it really bad since it is a conventional triode with a very low bias change needed to go from cutoff to max plate current, maybe 6 volts. At low currents it can only pass a few volts of signal without severe distortion. How they make this work, I don't know.
There are 3 12BA6 tubes and some 7 pin sockets sitting on the bench waiting to be tested. There are 20 or so different 7 pin tubes with similar pinouts such that they can be easily accommodated in the same test setup. I will get to them when the box of tubes I ordered arrive.
I have been tinkering with this Variable Gain Amplifier / gain cell design off and on over a period of a few weeks. Perhaps half of my box full of tubes have been tested....but some new discoveries will probably force me to start over. So what have I learned?
I started out with a simplified clone of the Fairchild 670 made with ONE 6386 instead of 4 in parallel as in the Fairchild, and I used some $10 Edcor transformers. It worked very well at 1KHz with about 60 dB of dynamic range The THD at 10 V RMS output is less than half a percent at 1KHz. It rises at each end of the audio spectrum, but that's not the real problem. The frequency response sucks to put it mildly. Loss at the low end due to transformer saturation was expected. Severe amplitude loss at each end of the spectrum was not.
The -3dB points are 240 Hz and 3.3 KHz at 100 mV output....not even good enough for guitar or synthesizer use, forget HiFi audio. At first I blamed the cheap transformers, and went on testing tubes.
My first breadboard had room for one 9 pin tube socket. I started with the 6386, and 5670 since it has the same pinout. Then I rewired to socket such that I could stuff in one of a few dozen dual triodes with the 9AJ pinout, including some remote and semi remote cutoff triodes seen in other compressors like the 6BC8 and 6ES8. These were all dual triodes, some from each cutoff characteristic were tested. None of these worked as good as the 6386, but the 6BC8 came close enough for a music synth. The frequency response varied with each different tube, and always got worse as the gain went down, really bad at negative gain. In these tubes, the plate current, and thus the Gm is reduced as the grid is made more negative, reducing the stage gain. As the plate current is reduced, the plate resistance goes up. Could this increase, coupled with the parasitics of the cheap transformers be responsible for the bad FR?
I dig through my junk box and find an old UTC interstage transformer, and wire it in with clip leads. Things get better, but still not good enough. At this point I rewire the socket for the 6386 and stuff one back in the socket. The response with the UTC is much better, but still not good, so....I sky wire another 6386 in parallel with the tube on the board, and things get better.
It seems that Fairchild used some pricey transformers that were maybe a wee a bit better than my $10 Edcors. they also use 4 X 6386 tubes in parallel. Could this be needed? I wired up a second 6386, and yes the frequency response got better. Four in parallel would probably get me in the ball park. OK, so to clone a Fairchild 670 I would need 4 X $120 tubes and probably the same $$$ worth of transformers....for ONE channel. I guess that's why old 670's fetch upwards of $20K!!!!
I ripped out the 9 pin socket and wired in a 12 pin compactron socket, as it was the easiest way to test some dual control pentodes because I have 10 NOS 6BV11's. The dual control pentodes work by deflecting plate current back to the screen grid. As you crank G3 more negative the plate current goes down while the screen current goes up. Again these were probably good enough for a music synth, except for that nagging frequency response issue.
The 6BA6 has been mentioned as a possible replacement for the 6386, and has been seen in some Fairchild and Gates clones, so I rip out the 12 pin socket, and wire in two 7 pin sockets. I didn't actually have any 6BA6's. I had 3 used Zenith brand 12BA6's that looked like they came out of old radios. They worked pretty good, but these, and nearly everything else that I tried made 40 to 55 dB of control range with THD's in the 1 to 3% at negative gain, but all had the variable frequency response issues, albeit in differing amounts. So while I have the 12BA6's in the hot seat, lets try wiring them as pentodes. Gain control is better, maybe 70 dB, but the frequency response and THD is MUCH WORSE, especially at negative gain.
Up until this point I had been just blindly testing tubes to find a cheap 6386, but there is clearly another problem that needs to be addressed. Trying the 12BA6 as a pentode made it perfectly clear. The plate resistance and the transformer imperfections are sucking the life out of my signal. I had to remove this variable to move forward with the testing.
Frustration with my test board led me to rip it up, make it bigger and wire in 5 tube sockets. 2 X 7 pin, 2 X 9 pin, and 1 big fat Compactron socket. I also added some mosfet followers to go between the tube plates and the cheap transformers. Variable plate resistance no longer matters.
This time I stuff in some $1 semi remote cutoff pentodes that I have a few hundred of. WOAH, 85 dB of gain control, zero point nuthin for THD, rising to 0.9% at the lower limit of measurement (about 50 mV). and fairly constant frequency response from 30 Hz (transformer saturation limited) to nearly 30 KHz.
I try lots of tubes.....ALL work to varying degree....but now I need to take all the measurements over again. I believe that this can be made into a useful circuit. It's already plenty good enough for the music synth. Some more Edcor transformers are on order, so that I can make another one of these setup for whatever tubes / circuit I chose so that I can get back to the vacuum tube ladder filter VCF.
I never tried the current steering circuit, or the BDT's and SBT's, nor have I run sound through this and put the output through some speakers to listen for thumps / pops and other sonic weirdness More testing is needed, but time is rather short right now.
No, you don't stuff all 5 tubes in at once, but it makes for a better picture, since it hides some of my ugly wiring. Unfortunately, the only tubes it can't test now are the 6386's and the BDT's. Depending on how long Edcor takes to make the new batch of transformers, and how much time I have for this, I'll either rip it up again, or build a new one.
I started out with a simplified clone of the Fairchild 670 made with ONE 6386 instead of 4 in parallel as in the Fairchild, and I used some $10 Edcor transformers. It worked very well at 1KHz with about 60 dB of dynamic range The THD at 10 V RMS output is less than half a percent at 1KHz. It rises at each end of the audio spectrum, but that's not the real problem. The frequency response sucks to put it mildly. Loss at the low end due to transformer saturation was expected. Severe amplitude loss at each end of the spectrum was not.
The -3dB points are 240 Hz and 3.3 KHz at 100 mV output....not even good enough for guitar or synthesizer use, forget HiFi audio. At first I blamed the cheap transformers, and went on testing tubes.
My first breadboard had room for one 9 pin tube socket. I started with the 6386, and 5670 since it has the same pinout. Then I rewired to socket such that I could stuff in one of a few dozen dual triodes with the 9AJ pinout, including some remote and semi remote cutoff triodes seen in other compressors like the 6BC8 and 6ES8. These were all dual triodes, some from each cutoff characteristic were tested. None of these worked as good as the 6386, but the 6BC8 came close enough for a music synth. The frequency response varied with each different tube, and always got worse as the gain went down, really bad at negative gain. In these tubes, the plate current, and thus the Gm is reduced as the grid is made more negative, reducing the stage gain. As the plate current is reduced, the plate resistance goes up. Could this increase, coupled with the parasitics of the cheap transformers be responsible for the bad FR?
I dig through my junk box and find an old UTC interstage transformer, and wire it in with clip leads. Things get better, but still not good enough. At this point I rewire the socket for the 6386 and stuff one back in the socket. The response with the UTC is much better, but still not good, so....I sky wire another 6386 in parallel with the tube on the board, and things get better.
It seems that Fairchild used some pricey transformers that were maybe a wee a bit better than my $10 Edcors. they also use 4 X 6386 tubes in parallel. Could this be needed? I wired up a second 6386, and yes the frequency response got better. Four in parallel would probably get me in the ball park. OK, so to clone a Fairchild 670 I would need 4 X $120 tubes and probably the same $$$ worth of transformers....for ONE channel. I guess that's why old 670's fetch upwards of $20K!!!!
I ripped out the 9 pin socket and wired in a 12 pin compactron socket, as it was the easiest way to test some dual control pentodes because I have 10 NOS 6BV11's. The dual control pentodes work by deflecting plate current back to the screen grid. As you crank G3 more negative the plate current goes down while the screen current goes up. Again these were probably good enough for a music synth, except for that nagging frequency response issue.
The 6BA6 has been mentioned as a possible replacement for the 6386, and has been seen in some Fairchild and Gates clones, so I rip out the 12 pin socket, and wire in two 7 pin sockets. I didn't actually have any 6BA6's. I had 3 used Zenith brand 12BA6's that looked like they came out of old radios. They worked pretty good, but these, and nearly everything else that I tried made 40 to 55 dB of control range with THD's in the 1 to 3% at negative gain, but all had the variable frequency response issues, albeit in differing amounts. So while I have the 12BA6's in the hot seat, lets try wiring them as pentodes. Gain control is better, maybe 70 dB, but the frequency response and THD is MUCH WORSE, especially at negative gain.
Up until this point I had been just blindly testing tubes to find a cheap 6386, but there is clearly another problem that needs to be addressed. Trying the 12BA6 as a pentode made it perfectly clear. The plate resistance and the transformer imperfections are sucking the life out of my signal. I had to remove this variable to move forward with the testing.
Frustration with my test board led me to rip it up, make it bigger and wire in 5 tube sockets. 2 X 7 pin, 2 X 9 pin, and 1 big fat Compactron socket. I also added some mosfet followers to go between the tube plates and the cheap transformers. Variable plate resistance no longer matters.
This time I stuff in some $1 semi remote cutoff pentodes that I have a few hundred of. WOAH, 85 dB of gain control, zero point nuthin for THD, rising to 0.9% at the lower limit of measurement (about 50 mV). and fairly constant frequency response from 30 Hz (transformer saturation limited) to nearly 30 KHz.
I try lots of tubes.....ALL work to varying degree....but now I need to take all the measurements over again. I believe that this can be made into a useful circuit. It's already plenty good enough for the music synth. Some more Edcor transformers are on order, so that I can make another one of these setup for whatever tubes / circuit I chose so that I can get back to the vacuum tube ladder filter VCF.
I never tried the current steering circuit, or the BDT's and SBT's, nor have I run sound through this and put the output through some speakers to listen for thumps / pops and other sonic weirdness More testing is needed, but time is rather short right now.
No, you don't stuff all 5 tubes in at once, but it makes for a better picture, since it hides some of my ugly wiring. Unfortunately, the only tubes it can't test now are the 6386's and the BDT's. Depending on how long Edcor takes to make the new batch of transformers, and how much time I have for this, I'll either rip it up again, or build a new one.
Attachments
> The -3dB points are..........
A transformer needs a "low" impedance for good bandwidth.
A VCA can often run with low max gain (follow it with fixed gain as needed).
My crayon scribbles often include few-K resistors across the plate winding.
This is especially important with pentodes and their infinite rp.
This also swamps the rise of triode rp as tube goes to cut-off (negating some of the drop of Gm).
Obviously the resistor reduces output level and raises THD, so much fiddling may be needed.
A transformer needs a "low" impedance for good bandwidth.
A VCA can often run with low max gain (follow it with fixed gain as needed).
My crayon scribbles often include few-K resistors across the plate winding.
This is especially important with pentodes and their infinite rp.
This also swamps the rise of triode rp as tube goes to cut-off (negating some of the drop of Gm).
Obviously the resistor reduces output level and raises THD, so much fiddling may be needed.