Push-pull driver with partial grid-leak biasing, for driving PP output stage

[trobbins]

Just checking if anyone has come across any amp using a push-pull driver stage, where the drivers appear to include substantial grid-leak biasing with signal operation. The schematic from a clone that is just being fired up shows 2M2 grid leak, although the original circuit used 22M. The clone idles with about 77V on screens, and 200V on plates and 1.7V on cathodes, for a 335V supply and using 6SJ7's (instead of 6SK7's).

Will try and check what the V6/7 anode voltages shift to when signal provides some additional negative biasing.



Ciao, Tim

[jazbo8]

Are you sure the supressor grid and screen grid connections were drawn correctly? I believe the circuit was traced by a "layman", so its accuracy is somewhat questionable, ditto for the preamp section as well...

[J M Fahey]

Agree and add: what a *weird* tube choice, design type, etc.

Big expensive KT88 ... cathode biased where they won't get as much power as if using fixed bias ?
There's a reason big power amps use it.
It's also way easier to adjust; I'd rather turn a little pot than solder power resistors in/out for any change or fine tuning.

Old pentodes used as PI ?

Grid leak bias which is a *terrible* choice there?

Biggest problem of grid leak bias is that loooooong time constants are involved so it's very easy to get blocking distortion with the slightest overdrive , even a narrow peak will cut those tubes off for long time.

2M2 is bad enough, the claim of 22M "original value" is incredible.

It's barely acceptable on a dated "microphone input" stage where high signal peaks are not expected ... but ... on a PI?
The point where you are guaranteed highest audio voltages in the whole amp?

Is this a commercial product?
They are cloning what?

Weird.

[trobbins]

We aussies like our guitar amp heritage. A circa 1960's amp range labelled MAXIM is not very commonly found, with only a few examples known:
Maxim

One example was checked out a few years ago, and the schematic prepared. Of course changes could have been made, and valve types could have changed along the way, but it appears that another example had the same part values so at the moment the aim is to appreciate what has been found.

An avid builder is cloning the 'as found' schematic - the section of schematic I have shown is from that clone's schematic. The amp includes a 6V6GT screen voltage regulator to derive 300V from a 620V output stage supply, as an example of the 'variety' of circuit techniques exhibited in the amp. The aim of the clone is not to change the 'as found' circuitry to a perceived better circuit or more practical form, but rather to see how it sounded.

Back then, some people obviously went their own way, so I have no issue with the phase splitter, or the KT88 cathode bias, or the use of KT88's. But the driver stage has me intrigued - yes it might be a one off quirky composition, but 'ya never know'

I haven't come across a push-pull grid-leak stage. I'm thinking that each grid's coupling cap should independently bias itself from that sides grid conduction, with no major influence from the other side of the push pull. But if sides move in to cutoff then cathode bias will drop - so a form of alleviation.

[Mike Sulzer]

Biggest problem of grid leak bias is that loooooong time constants are involved so it's very easy to get blocking distortion with the slightest overdrive , even a narrow peak will cut those tubes off for long time.

2M2 is bad enough, the claim of 22M "original value" is incredible.

Never before have I seen a full wave blocking distortion generator. But I do not get around much.

[Lauri]

The reason for high value grid leak resistor and 6SK7 remote cut off tubes is that it's a primitive compressor circuit. Grid current charges the coupling caps negative and biases the tubes more negative. The more you drive the 6SK7 tubes the more compression you get. With 0.022µ coupling caps and 22meg grid leak resistor attack time is really fast and release is about one second.

The schematic from a clone that is just being fired up shows 2M2 grid leak, although the original circuit used 22M. The clone idles with about 77V on screens, and 200V on plates and 1.7V on cathodes, for a 335V supply and using 6SJ7's (instead of 6SK7's).

You'll probably want to change the grid leak to 22M and use 6SK7 tubes. With 2.2M and 6SJ7 tubes it's most likely going to distort a lot and sound nothing like the original.

[J M Fahey]

Well, Lauri got it right.
So Peter Mc Carthy was thinking outside the box / laterally thinking? ... I like it, I do that myself all the time

And being remote cutoff pentodes (I use ECC189 triodes for the exact same reason) the horrible blocking cutoff I expect actually turns into gain loss and still operating linear ... the very definition of compression, fine

An elegant design if I ever saw one. (once you know what it is meant to do)

*Maybe* it was a Cinema/Radio compressor idea from the 30's and 40's , where a lot of fine brain power was crunching compression ideas, because both just created and powerful ($$$$$$$$) Radio and "Talkies" had terrible headroom problems, specially optical track band in movies, very narrow dynamic range and clipping was to be avoided at all costs.

@trobbins, don't worry, I'm not disrespecting Aussie designs, quite the contrary
And (surprise!.... surprise! .... ), I *am* a member of oz valve amps , thanks to our common (and lamented) Friend Roly Roper .

Australia/NZ and Argentina share many common traits, including the push to Industrialize without being major World powers nor having strong export markets to sustain it.
We even made Ford Falcons until the late 80's early 90's !!!!!!!!!
Many of which are still running today !!!!!

And 6DQ6 were more used than all others combined to make Guitar amps

And on a personal curious note: I learnt to speak Australian *before* proper English, to the tune that my Cambride authorized English teachers were horrified at my accent and I had to unlearn it before going on with normal studies, go figure.

[Mike Sulzer]

The reason for high value grid leak resistor and 6SK7 remote cut off tubes is that it's a primitive compressor circuit. Grid current charges the coupling caps negative and biases the tubes more negative. The more you drive the 6SK7 tubes the more compression you get. With 0.022µ coupling caps and 22meg grid leak resistor attack time is really fast and release is about one second.



You'll probably want to change the grid leak to 22M and use 6SK7 tubes. With 2.2M and 6SJ7 tubes it's most likely going to distort a lot and sound nothing like the original.

But does this really work as an audio compressor, that is, a variable gain circuit that provides linear amplification over a wide range of plate currents (that is, variable gm) in a situation where you you need such a large output in order to drive the grids of the KT-88s to full on and off? It would seem that you only get significant gain reduction as you get into clipping, and then you cut down the plate current, reducing the gain, but also cutting down on the p-to-p voltage output capability. That affects the ability to drive the KT-88s to full power, and thus would appear to be a kind of blocking distortion.

[Lauri]

But does this really work as an audio compressor, that is, a variable gain circuit that provides linear amplification over a wide range of plate currents (that is, variable gm) in a situation where you you need such a large output in order to drive the grids of the KT-88s to full on and off? It would seem that you only get significant gain reduction as you get into clipping, and then you cut down the plate current, reducing the gain, but also cutting down on the p-to-p voltage output capability. That affects the ability to drive the KT-88s to full power, and thus would appear to be a kind of blocking distortion.

I've never seen exactly this kind of a compressor circuit before but I see no reason why it wouldn't work quite well. I'm guessing you maybe need something like 35Vrms on the control grids of the KT88's to get full power output. With small signal on the control grids of 6SK7's they have quite a lot gain. As the signal gets larger so does the negative bias voltage and you get gain reduction. With remote cut off tubes in this kind of grid leak bias compressor circuit I don't think you'll be able to cut off the plate current so much that you wouldn't be able to get enough voltage swing to drive the KT88's to full power output.

[Mike Sulzer]

I've never seen exactly this kind of a compressor circuit before but I see no reason why it wouldn't work quite well. I'm guessing you maybe need something like 35Vrms on the control grids of the KT88's to get full power output. With small signal on the control grids of 6SK7's they have quite a lot gain. As the signal gets larger so does the negative bias voltage and you get gain reduction. With remote cut off tubes in this kind of grid leak bias compressor circuit I don't think you'll be able to cut off the plate current so much that you wouldn't be able to enough voltage swing to drive the KT88's to full power output.

Well, I guess it is a kind of compressor in the sense that it can control the amount of overload. I do not think it reduces gain significantly until the input signal is large enough to want to draw a bit of grid current, and thus you have a distorted signal. (Or am I missing something?) As the input signal continues to increase, the gain is further reduced. The effect of this depends on the "remoteness" of the pentode characteristic. A sharp cutoff pentode would turn off. I suppose the 6SK7 might just hold the amount of distortion about constant or something. It sounds a bit Goldilocks to me. But you do have the value of that resistor to adjust, and that should help get the temperature of the oatmeal just right. The chair and the bed, not so sure. I suspect you might need more than one parameter to adjust.

[trobbins]

Good on ya Juan. Yes Roly was an inspiration to many. On a local 'Australia/NZ and Argentina' sidetrack, have you ever come across any South American magazine articles on the Williamson amp back in the late 1940's to 50's ? I'm just getting a NZ article copied for me. and thought there may be more around. Ta, Tim

[trobbins]

Well done Lauri for honing in on the key function. Much appreciated. Now it does make sense!

I can see that the basic mechanism of increasing the negative bias as a function of signal magnitude is very neat and compact compared to the many other more complex compression circuits used to achieve that end outcome. Yes, the attack and decay times would be dictated by the CR coupling cap and 22M, and the attack time would also be dependant on the magnitude of the input signal and how much it is initially pushing through grid current. The screen voltage varies the level of remoteness, so is dictated by the screen dropper. (The schem does show screen and suppressor grids arse-about). And yes the 6SJ7 would be a poor replacement in terms of trying to recreate the original intent.

I'll try and get some 'user feedback' when the amp is properly set up.

[Lauri]

I do not think it reduces gain significantly until the input signal is large enough to want to draw a bit of grid current

This is exactly the whole point of this kind of circuit. Voltage at cathode sets the threshold for compression and when the grids start nearing cathode potential tiny amount of grid current will start biasing the grids more negative and gain reduction starts hopefully before any audible distortion.

With 6SJ7 tubes the amp is most likely at full power output before the grids start to get negative. Even if the girds get negative the tubes will never completely cut off because the screen voltage rises and helps keeping small amount of current going. There won't be much gain reduction and everything will be very distorted. With 6SK7 tubes there's maybe some compression before full power output. If I had this kind of an amplifier I would put a master volume after the 6SK7 tubes so I would be able to get more compression before everyone in the room is deaf.

[J M Fahey]

What for?
The point in 60's and 70's amps *was* to make everyone deaf ... or to put it in a polite way, to be heard in the whole ballroom or school basketball court in spite of the piss poor PA systems available way back then.

Our most successful PA guy way back then used a homemade mixer, and as many 200W tube amps (6 x 6DQ6 @ 800V +B) and narrow shallow 4x12" speaker columns as you wish, generally 4 columns per power amp.

Which had 32 ohms output tap only so he could wire 4 columns in series and spread them around the audience in a 60 meter arc without resistive wiring losses.

A very simple, efficient and LOUD system, only problem it put out nothing below 150Hz (so Bass was non existent and bass drums had a "knock on wood" sound, while lack of tweeters made the finest Zildjian cymbal sound the same as an aluminum pot cover.

Standard Bass backline was 3 x 100W tube amps driven full tilt into distortion (think Jack Bruce sound) , no niceties as compression available outside a studio, which I took by storm with my 400/500W RMS SS Bass amps driving 2 *fat* 4 x 12"cabinets.

I sold so many at high price (it was the only amp which was clearly heard live, with clean sharp sound) that in 2 or 3 years I bought a car and a large home in a posh Buenos Aires neighbourhood (Recoleta) .... cash of course, no mortgage.
Never ever flew that high since.

So I clearly re,ember when loud Bass amps were an absolute necessity.
I bet the Australian scene must not have been very different, way back then.

Of course, nowadays anybody can buy as many China/Korea/India/VietNam made amps as he like and local old pioneers are all but forgotten.

[Mike Sulzer]

This is exactly the whole point of this kind of circuit. Voltage at cathode sets the threshold for compression and when the grids start nearing cathode potential tiny amount of grid current will start biasing the grids more negative and gain reduction starts hopefully before any audible distortion.

That seems like a plausible mechanism. The grid voltage has to want to rise faster than the input voltage does so there is a kind of self limiting negative feedback which keeps it right on the edge of distortion. I would have to build one to be sure; I suppose you could model it with spice; I sure would not want to compute that from scratch.

But here is my concern: Suppose it works as you say: fast attack, slow decay, staying just on the edge of distortion of the peaks in the guitar signal since it is fast/slow. The rest of the note will be quite clean. What guitarist would want to play this? If you are feeding a clean signal to this amp, you are not compressing the signal in the usual sense of increasing the ratio of average to peak power within a note; that is, making it louder as everyone usually wants. You letting the picking transients set the gain of the system, and the average to peak power remains that of a clean guitar signal. This is not usually what someone who wants a pair of KT-88s does with them.