Grid leak/bias splitters

[Dblgun]

Marshall's were sold for a period of time with 6550's in the U.S. market reportedly due to fragility of EL34's at the time. The current required by the two tubes differed and therefore 6550 equipped amps used a 47k bias resistor and 150k Grid leak resistors while EL34's 56k and 220k's with a few exceptions. Both versions used .022uf PI coupling caps.

Grid leak resistors and PI coupling caps form a high-pass filter and halving the resistor value raises -3dB corner point one octave. No doubt amplifiers fitted with 6550's have a different vibe than the EL34 models each with their own fans. Placing the bias /current differences aside it would seem that a change in coupling cap values would have an impact on tone/sound including bass response. KT66 equipped Marshalls used 220k grid leaks with .1uf PI coupling caps as an example. Also, my understanding is that reducing grid leak values will load the PI differently rotating the AC load line clockwise changing its clipping characteristics.

It makes sense to me that Marshall would change as few components as possible to get the 6550's to function. As the amps in this era were ST1 PC board versions it would have made no sense to change PI coupling caps to a more "suitible" value for the 6550. I also think that Marshall felt that amp buyers would convert back to EL34's and most did, not all in the best manner.

Has anyone played with different PI coupling cap/Grid leak resistor values with EL34/6550 equipped amps and what were your take aways? There are calculators and I have fooled with them a bit but they don't really offer any real world tone findings.

 

[neikeel]

Yes I have
Strictly speaking to compensate for the lower grid leaks necessary to avoid grid clamping with 6550 and KT88s they should be 47k (you will find some Marshalls with 68k or 82k and I think that they were from the factory, not just mods) you probably need to go to the 0.1uF PI output couplers to compensate for the shift in corner frequency.
That is why you will see quite a lot of hot rodded Marshalls running 6550s are actually Superbasses (or in some cases Supertrems using the extra tubes). We have two 71s in similar trim (cascaded V0) and the 0.1s certainly give more low end punch, particularly the 6550 equipped one, which is a SB. The SL I have gone back and forth and currently it is EL34s with 220k grid leaks and 0.1uF couplers and switchable extra stage (SIR style). The modded SB is a better all round amp in this case as it is crisper and beefier at the same time.

 

[Dblgun]

Thanks Neil, I have a hot rodded 2204 type circuit with 6550's on the bench right now. It currently has 120k/.047 grid leak/PI caps so I'll tinker with it a while. The idea is to make it more hot rodded JMP and less hot rodded JCM 800 so we'll see what can be done.

 

[neikeel]

I’d recommend 0.1uF couplers with 120k grid leaks. Just my 2c of course!

 

[XTRXTR]

I thought you could have some blocking distortion issues, which is supposedly why the 0.1uf caps were replaced by 0.022uf? I think I read something like that. Also something about the grid resistors can change the tc of the cap. I am not completely sure but I know I read something about this.

 

[neikeel]

With low value of the grid leak resistors, then you're cutting bass response due to another time constant with the previous coupling cap. Also consider that the grid leak resistors AC-wise are in parallel with the PI's plate resistors - so reducing the value of the grid leak resistors is reducing the PI tube's gain.
If you go to the higher 220k grid leak resistors, then you'll get blocking distortion, when you drive the power tubes into saturation.
This especially with 100W amps, where each two tubes are sharing only one grid leak resistor - where the grid leak current of two tubes has to flow through only one resistor.
Most EL34s the max grid leak is 700K - but only then, when NOT operating in saturation, so if you are running two EL34's this means, that max is 350K (two 700K's in parallel). But you have to add to the grid leak resistor's value the value of the 5k6 swamp resistors and the entire resistance through the bias supply (47K or 56K & 25K trim pot)!
So each two output tubes in a 100W Marshall with a grid leak resistor's value of 220K actually are seeing about 300K to ground which adds up to 600K each tube alarmingly close to te max. 700K for clean so not even when fully saturated.
So ideally drop the grid leak size and raise the PI output cap to compensate for loss of bass is my take home.

 

[FourT6and2]

Yeah I've played with these values before. Honestly... not a huuuuuge difference in "tone" when it comes to the frequency cutoffs with the resistor/cap values because by this point in the circuit these frequencies have already been affected to some degree. BUT... changing the resistor values here is similar to a PPIMV. You're gonna reduce the amp's volume and effectiveness of some of the tone/NFB controls as you go lower and lower. I've tried everything from 110K up to the stock 220K value. I like the stock cap value over something like .1uF, which allows too much bass through and can make the amp a bit woofy/loose depending on the rest of the circuit. With the stock cap value, 150K up to 220K is fine, tonally speaking.

Lowering the resistor value to make it easier on tube life is fine. I haven't personally seen a benefit either way. But as more and more tube options dwindle, longevity might continue to drop.