I've been doing some bed time reading of old schematics, and came across an idea I wasn't aware of. I first saw it in the RCA Mi-4297, which uses a 6L7 for electronic volume control. The signal is applied to grid 3, and DC bias on grid 1 controls gain. The advantages are that no signal passes through any pots, and the volume control sees only DC so may be remotely located.
A similar circuit is presented in Rdh4, using the more modern 6BE6. In the instance of the rdh4 schematic, it can also be used as a mixer.
I wonder if anyone has tried this circuit? At a first glance it looks appealing, getting the volume control out of the circuit and all. The down side I see is this circuit won't work with negative feedback, so the harmonic distortion may be appreciable. Also, the curves of any pentagrid converter are fairly hard to interpret.
***note on the rca drawing, for resistor values where it says M, it means k. What we call M is shown as Meg.
A similar circuit is presented in Rdh4, using the more modern 6BE6. In the instance of the rdh4 schematic, it can also be used as a mixer.
I wonder if anyone has tried this circuit? At a first glance it looks appealing, getting the volume control out of the circuit and all. The down side I see is this circuit won't work with negative feedback, so the harmonic distortion may be appreciable. Also, the curves of any pentagrid converter are fairly hard to interpret.
***note on the rca drawing, for resistor values where it says M, it means k. What we call M is shown as Meg.
Attachments
The way the 6L7 circuit works, is it's a tube with two grids - one is remote cutoff (G1), the other is sharp cutoff (G3).
If you set G1 to a specific bias value, it sets the gain of the amplifier, and when you vary the voltage to G1, it varies the gain of the tube. G3 works just like any pentode stage, producing a supposedly linear output for a given input signal.
To control the volume, you adjust only the bias voltage on G1, which is a DC signal. The original intention of the design, was to enable a remotely located volume control to set the gain of a PA system, such that volume could be adjusted from the audience, via a long and unshielded connecting cable but yet not corrupt the audio signal.
It must have worked well, the basic circuit was in the RCA catalog for 20 or so years.
On the other hand, 1930s public address systems weren't really designed with the same goals in mind as a high fidelity system. What it's got me thinking about, is possibly making a two channel volume control for a tube amp, without resorting to an expensive volume control potentiometer. You could presumably use a single pot to control multiple channels of 6L7.
What I don't know about though, is what's the distortion of this tube, used in this way? They're certainly affordable as NOS tubes.
I think 80dB of attenuation isn't really needed for most applications - my volume knob position probably varies over about a 15dB range, when I use my hi-fi system.
I am assuming your preamp uses the volume control conventionally - i.e. passing the signal though the control between the two stages.
If you set G1 to a specific bias value, it sets the gain of the amplifier, and when you vary the voltage to G1, it varies the gain of the tube. G3 works just like any pentode stage, producing a supposedly linear output for a given input signal.
To control the volume, you adjust only the bias voltage on G1, which is a DC signal. The original intention of the design, was to enable a remotely located volume control to set the gain of a PA system, such that volume could be adjusted from the audience, via a long and unshielded connecting cable but yet not corrupt the audio signal.
It must have worked well, the basic circuit was in the RCA catalog for 20 or so years.
On the other hand, 1930s public address systems weren't really designed with the same goals in mind as a high fidelity system. What it's got me thinking about, is possibly making a two channel volume control for a tube amp, without resorting to an expensive volume control potentiometer. You could presumably use a single pot to control multiple channels of 6L7.
What I don't know about though, is what's the distortion of this tube, used in this way? They're certainly affordable as NOS tubes.
I think 80dB of attenuation isn't really needed for most applications - my volume knob position probably varies over about a 15dB range, when I use my hi-fi system.
I am assuming your preamp uses the volume control conventionally - i.e. passing the signal though the control between the two stages.
Considering that RDH4 is a 1952 book, I´d take the "modern" label with a grain of salt
A fair point! But the 6BE6 is a super common radio tube, The 6L7 I think I have one, and it's in a 1936 GE radio. I should have said newer and easier to find, rather than modern.
Interesting stuff there, thanks for the link!
Bah, too complicated.
Bogen had remote volume with simple (dual-)triodes. See CHB50:
http://makearadio.com/schematics/images/bogen-chb50-6.jpg
Put a 1Meg Audio pot (as rheostat) from REM# screw to GND.
Yes, this is for VERY small (mike) signals. Rated sensitivity is 5mV. With panel pots turned down for normal use, maybe 50mV. Much more than that will overload the gridleak bias. Taking the triode hiss as 2uV, S/N may be 66dB-86dB. Distortion will typically be far below 5% but never point-oh percent.
BTW, these things thump. Not when manually dialed, but when used in fast automatic gain controllers. The elegant path is full push-pull, usually with transformers to split the AC from the DC. The program path of the Fairchild 660 is a fine example, and very-very costly to build and balance.
Note that these systems are a class of VCAs which have a maximum input, enough hiss that the source should approach that max input, and they turn-down from there roughly to silence. OTOH the humble $1 pot will take "any" input without overload, and with little hiss issue. Another class of VCAs (often the first class in the tail of a NFB loop) will take "any" input but the max output is limited; this is perhaps what you want with a plethora of sources and an amplifier of known sensitivity. However wrapping a tube VCA in the back of a NFB loop is extremely cumbersome.
Bogen had remote volume with simple (dual-)triodes. See CHB50:
http://makearadio.com/schematics/images/bogen-chb50-6.jpg
Put a 1Meg Audio pot (as rheostat) from REM# screw to GND.
Yes, this is for VERY small (mike) signals. Rated sensitivity is 5mV. With panel pots turned down for normal use, maybe 50mV. Much more than that will overload the gridleak bias. Taking the triode hiss as 2uV, S/N may be 66dB-86dB. Distortion will typically be far below 5% but never point-oh percent.
BTW, these things thump. Not when manually dialed, but when used in fast automatic gain controllers. The elegant path is full push-pull, usually with transformers to split the AC from the DC. The program path of the Fairchild 660 is a fine example, and very-very costly to build and balance.
Note that these systems are a class of VCAs which have a maximum input, enough hiss that the source should approach that max input, and they turn-down from there roughly to silence. OTOH the humble $1 pot will take "any" input without overload, and with little hiss issue. Another class of VCAs (often the first class in the tail of a NFB loop) will take "any" input but the max output is limited; this is perhaps what you want with a plethora of sources and an amplifier of known sensitivity. However wrapping a tube VCA in the back of a NFB loop is extremely cumbersome.
I'm looking at the Bogen circuit, it appears to function by killing B+ to the input tubes. This would push the operating point way down into the curved part of the tube curves. I'm not sure how great this design is, it looks extremely easy to overload, puts significant voltage across the remote control pot, and also operates the tubes in a non-linear way.
The Fairchild 660 looks to operate by using the tubes as a variable resistance stuck in between an input and output transformer, and controlling this resistance via a DC amplifier. This is an interesting approach too, I wonder why they used so many tubes in parallel instead of using a larger one. It looks to me like the Fairchild should be the most transparent of the designs, it really doesn't depend on the tubes being linear to work properly, since they're just being used as a variable resistance instead of as amplifiers.
http://thehistoryofrecording.com/Sc...irchild_660_mono_limiting_amplifier_schem.pdf
Have you checked the price of the 6386? Wow, I was a bit surprised!
I wonder if you couldn't do the same thing with say a 6SK7, or a 6BA6. The key would be to use it only as a variable resistor, instead of as an amplifier.
It's curved ALL the way down (and up).
There is less curvature when fixed resistance swamps tube resistance.
BTW, current-steering can be highly linear. (Not cheap.)
This field has been thoroughly explored for 90 years. Do homework.
The 6386 was used in the Fairchild 670 compressor / limiter. A working 670 goes for somewhere north of $25K today. The world's supply of NOS tubes has been sucked up by the people who bought $25K compressors.
Somewhere on my list of crazy projects is a vacuum tube music synthesizer. I work on this from time to time, but not very often. One of the required building blocks is a VCA, Voltage Controlled Amplifier. This is relatively easy to do in the solid state world.
There are several ways to do this in the vacuum tube world, each with drawbacks. I did several experiments exploring some of the possibilities when I lived in Florida. I have the data somewhere on an archived hard drive, but I have moved twice. I have located some of the test boards and plan to fire them up again, but it's not high on my list right now.
The Fairchild 670 is the king of desirable classics for several reasons, but it looked like a good place to start since I have a hand full of well used 6386 tubes. I built up one stage of the 6386 gain controlled amp using budget Edcor transformers several years ago. I remember something like 30 to 35 DB of usable gain control range, which is why the Fairchild uses multiple stages.
The 6386 is a variable Mu triode (also called remote cutoff triode) originally intended for gain controlled RF amplifiers. The grid is wound with a variable pitch spacing so that at near zero grid bias all of the grid, and hence all of the tube is being used, and gain is maximized. As the grid bias is made more negative the part of the grid with the tightly spaced wires becomes cut off and the tube becomes effectively smaller with less gain. Enough negative voltage will eventually cut the entire tube off. The gain of the triode can be controlled by varying it's negative bias, but it becomes less linear with increasing negative voltage. The distortion created is mostly second harmonic, which is why the Fairchild is completely transformer coupled push pull. The P-P operation cancels the second harmonic distortion leaving 30 dB or so of usable gain control. Cascading multiple stages can reach 70 to 80 db of gain control. The 6386 is over $100 per tube, so lets look somewhere else.
ALL triodes exhibit the ability to control their gain with bias voltage, but tubes like the venerable 12AX7 have a narrow range of usable gain control range without excessive distortion. The control voltage range for this gain control is quite small, in the 3 to 10 volt range, and varies with the level of applied signal.
There are however other variable Mu triodes intended for RF amps like TV tuners which were created for the same purpose as the 6386. You are looking for a triode whose Mu and Gm vary a lot with the plate current, or grid voltage. I tried a bunch in my 6386 test circuit. None worked as good as real 6386's, but some came close enough for music synthesizer use. The 6AF4, 6AN4, 6BC8, 6ES8 come to mind.
There is a commercial adapter board available that allows Farichild 670's to use TV tuner tubes like the 6BQ7, 6BK7, and 6BZ7. These seemed to work in my test circuit as well. I did this nearly 10 years ago, so I don't remember which tubes worked the best, or if any were good enough for HiFi use. I did not try any of these in single ended circuit yet.
Multi grid converter tubes designed for mixer use in AM radios should be applicable, but I had the least success with them. The 6L7, 6SA7, 6BE6, and 6BA7 were designed to make one tube do the work of two. A single tube mixes the RF and is the local oscillator at the same time, and was optimized for this purpose. Tube to tube variation was rather large and linearity was poor, even though the 6BE6 makes a very nice triode when all the grids are connected together. I did not try these tubes in push pull transformer coupled operation since I'm looking for a physically small design, but they might have worked better.
There are also plenty of remote cutoff pentodes designed for gain controlled RF amps. The 6K7 dates to the same era as the 6L7, while the 6SK7 is newer as is the 6BA6. These could be used in the same manner as the 6386 was, although the gain control range is somewhat smaller.
There are a class of dual control pentodes intended for gain controlled amplifiers. I did some extensive testing with the 6AS6 since I have a large box full of them, they are tiny, consume little heater power, and they are $1 each new. They seem adequate for music synthesizer use in single ended operation, although two stages are likely needed for enough gain control range. The 6HZ6 and 6DT6 are similar tubes of a newer design, but I didn't test them at the time since they are larger and hungrier.
There are also gated beam tubes that could do the same thing, but their characteristics seem very sharp with almost on-off control. The 6BN6 is an example.
There are several tubes that are normal pentodes except that their G3 and the plates have been split in half so that there are two G3's and two plates. You can drive a push pull transformer with each plate and get near zero total output when both G3's are at equal potential. Varying the G3 voltages can produce an output since one side has more gain than the other. This seems to work well, but transformer saturation can be an issue at extremes. These these tubes were common in TV sets and are cheap. The 6BU8 and 6HS8 are common types. This is where I was experimenting when I had to pack it all up and move 1200 miles.
There are also beam deflection tubes that took the split plate idea to extremes. The normal G3 is replaced with a deflection plate such that the electron stream can be deflected to one plate, both or the other. I have a couple of 6AR8's to try, but they don't seem to be very common.
Any of these tube types should work better than the 6L7, but I haven't finished my testing, and probably won't get back to it for the foreseeable future.
Thank you for the very detailed and informative post. The application I originally had in mind, was a simple line level preamplifier modeled after RCA's MI-4118 design, but with two 6L7 tubes, one for each channel, being controlled by one common potentiometer. Signal to G3, control voltage to G1.
If tube characteristics vary enough that this will not result in matched left and right output levels, and if non-linearity is excessive compared with a conventional tube gain stage, I will abandon this idea as impractical.
The other solutions you explore do look like interesting and viable options for signal processing and manipulation, but given the complexity it looks to me to be a solution looking for a problem, if all I'm doing is controlling gain in a line level preamp. I suppose I got caught up in the idea that eliminating potentiometers from the signal path is desirable, without considering that the solution may well be worse than the original problem.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.