A schematic really would help.
Myself I have used a choke in my last Bassamp I have designed (PP 6xGU50). Here are the reason for that case:
- minimize the voltage reduction caused by the filtering
- avoid screen grid modulation caused by the ripple on the supply by using CLC filtering. Without that filtering the AC ripple at the screen grid has caused amplitude modulation at the anode when driving the powerstage near clipping.
Myself I have used a choke in my last Bassamp I have designed (PP 6xGU50). Here are the reason for that case:
- minimize the voltage reduction caused by the filtering
- avoid screen grid modulation caused by the ripple on the supply by using CLC filtering. Without that filtering the AC ripple at the screen grid has caused amplitude modulation at the anode when driving the powerstage near clipping.
Do you mean a large iron core choke between the B+ node for the power tube plates and the B+ node for the power tube screens? If so, it is simply part of the B+ filter circuit. It offers more filtration than a simple resistor.
The way your question is worded, though, makes it sound like individual chokes for each grid. ANd speaking only for myself, I tend to reserve the word "grid" for the control grid, leaving the screen grid to be called just the screen. SO if you are not just refering to the power supply filter, please provide a link to an example.
The way your question is worded, though, makes it sound like individual chokes for each grid. ANd speaking only for myself, I tend to reserve the word "grid" for the control grid, leaving the screen grid to be called just the screen. SO if you are not just refering to the power supply filter, please provide a link to an example.
I ran into this a few places talking aobut hi-fi, not guitar amps. They made it sound like using a "grid choke" instead of a "grid resistor" (definitely control GRID) was a huge upgrade in fidelity and made a big difference. This is not B+ or B++ or to screens. I first ran into the term while shopping for other chokes.
What Loudthud said.
Gingertubes "essay"
As you drive more current thru' an output tube then grid 1 current increases, this grid current develops a voltage across Rg1 which SUBTRACTS from the bias voltage. Thus you get more anode + Screen current and therefore more grid current which develops more voltage across Rg1 etc. If Rg1 is too high then thermal run away occurs untill the tube "melts". Thus each tube will have a max Rg1 spec. This spec depends upon whether cathode bias or fixed bias is used. With cathode (auto) bias this opposes the bias shift caused by the voltage developed across Rg1 from grid current. With fixed bias that does not happen and that is why max Rg1 values for Fixed bias are smaller than for cathode bias.
With 2 tubes in parallel and with fixed bias then the max Rg1 is typically 50K even thou' most manufacturers use 220K or so.
That is why guitar amps blow up about 10 times as frequently as tube HiFi Amps where proper attention to such things is more likely (but not guaranteed). If you have 10K grid stops then that leaves just 40K for the grid leak back to 0V via the bias supply.
The common 12AX7 phase splitter would have a hernia trying to drive that so it is not done.
If you look at some old tube data sheets, take KT88 as an example, you will see that in fixed bias at 42W Anode + screen dissipation then max Rg1 is 100K. If you take the disspation down to less than 35W then Rg1 can be increased to 220K. This is the maximum value to prevent thermal run away of your output tube.
So for your average guitar amp with max Rg1 RECOMMENDATIONS violated/ignored then biasing so that tubes idle at around 70% of max dissipation makes good sense from both a reliablity point of view as well as tube longevity. I would use 80% of max dissipation as the absolute limit.
UNLESS
You use a good beefy driver (NOT a 12AX7) with low value Rg1
OR
A grid choke. The grid choke gives you lots of AC Impedance so as to not load down the phase splitter / driver BUT very low DC resistance so that max Rg1 values are not exceeded, that would allow you to bias the output tubes right up to the 100% dissipation limit without the large risk of thermal run away.
For HiFi Amps I use a cathode follower driver, direct coupled to the output tube grid with an active current source as the cathode follower load instead of a resistor. You would not believe how that affects the sound, very strong dynamic sound with the characteristic of an amp with 3 or 4 times as much power and dead quiet background, no hiss (except for what is coming in from the preamp). It sounds like this becaudse we have taken firm control of the output tube grid1, it is no longer "flapping in the thermal breeze". That would add another 30 or 40 bucks to the manufacturing cost so of-course no one does it. No reason why DIY'ers should'nt do it though.
Hope this "rave" is of some value to somebody.
Cheers,
Ian
Gingertubes "essay"
As you drive more current thru' an output tube then grid 1 current increases, this grid current develops a voltage across Rg1 which SUBTRACTS from the bias voltage. Thus you get more anode + Screen current and therefore more grid current which develops more voltage across Rg1 etc. If Rg1 is too high then thermal run away occurs untill the tube "melts". Thus each tube will have a max Rg1 spec. This spec depends upon whether cathode bias or fixed bias is used. With cathode (auto) bias this opposes the bias shift caused by the voltage developed across Rg1 from grid current. With fixed bias that does not happen and that is why max Rg1 values for Fixed bias are smaller than for cathode bias.
With 2 tubes in parallel and with fixed bias then the max Rg1 is typically 50K even thou' most manufacturers use 220K or so.
That is why guitar amps blow up about 10 times as frequently as tube HiFi Amps where proper attention to such things is more likely (but not guaranteed). If you have 10K grid stops then that leaves just 40K for the grid leak back to 0V via the bias supply.
The common 12AX7 phase splitter would have a hernia trying to drive that so it is not done.
If you look at some old tube data sheets, take KT88 as an example, you will see that in fixed bias at 42W Anode + screen dissipation then max Rg1 is 100K. If you take the disspation down to less than 35W then Rg1 can be increased to 220K. This is the maximum value to prevent thermal run away of your output tube.
So for your average guitar amp with max Rg1 RECOMMENDATIONS violated/ignored then biasing so that tubes idle at around 70% of max dissipation makes good sense from both a reliablity point of view as well as tube longevity. I would use 80% of max dissipation as the absolute limit.
UNLESS
You use a good beefy driver (NOT a 12AX7) with low value Rg1
OR
A grid choke. The grid choke gives you lots of AC Impedance so as to not load down the phase splitter / driver BUT very low DC resistance so that max Rg1 values are not exceeded, that would allow you to bias the output tubes right up to the 100% dissipation limit without the large risk of thermal run away.
For HiFi Amps I use a cathode follower driver, direct coupled to the output tube grid with an active current source as the cathode follower load instead of a resistor. You would not believe how that affects the sound, very strong dynamic sound with the characteristic of an amp with 3 or 4 times as much power and dead quiet background, no hiss (except for what is coming in from the preamp). It sounds like this becaudse we have taken firm control of the output tube grid1, it is no longer "flapping in the thermal breeze". That would add another 30 or 40 bucks to the manufacturing cost so of-course no one does it. No reason why DIY'ers should'nt do it though.
Hope this "rave" is of some value to somebody.
Cheers,
Ian
Last edited:
Man this is really extremely interesting reading.
Anybody got some relevant schematics to look at?
In the guitar amp I'm fooling around with at the moment, (see attached schematic) with 6 6L6 output tubes, how would such a "grid choke" conversion and conversion to bias pots (instead of "output balance" pot) ideally be done? Replace each of the two 33K resistors with an 8,000H choke and a pot? Or rework the circuit with a choke and pot for each output tube?
Maybe my Sound City project with its 6 EL34s might be a candidate too, though I think maybe the EL34s might be easier to drive than the 6L6s.
Anybody got some relevant schematics to look at?
In the guitar amp I'm fooling around with at the moment, (see attached schematic) with 6 6L6 output tubes, how would such a "grid choke" conversion and conversion to bias pots (instead of "output balance" pot) ideally be done? Replace each of the two 33K resistors with an 8,000H choke and a pot? Or rework the circuit with a choke and pot for each output tube?
Maybe my Sound City project with its 6 EL34s might be a candidate too, though I think maybe the EL34s might be easier to drive than the 6L6s.
Good reading with tons of valuable information, as always.
I definitely appreciate it. Keep it coming!
Cheers,
Martin
In a guitar amp, you're looking for a "sound" and this is one way of tuning it. With RC coupling, the high pass function is a first order filter and thus has no peaking in the frequency response. Now, convert to chokes and you have a second order filter which can have a wide range of response depending on how you choose the values of C, R, and L. A big peak and the resulting whomp (i.e., a high Q at the cutoff frequency) is a disadvantage in a tube amp, but may be what you're looking for in a musical instrument amp. Likewise, a very low Q and some slow LF rolloff may be useful depending on what you're trying to get for a sound.
This is one where I'd spend some time with a simulator juggling values to see the effects on frequency (and hence transient) response.
With a sufficiently beefy driver stage, a grid choke would allow you to stray partway into Class AB2 mode - you could draw some grid current on peaks without upsetting the bias too much and getting blocking due to the coupling cap getting charged up. Think of it as a poor man's transformer drive.
Very interesting. Might have some potential for a bass amp. Sounds like one of those things I'd just have to listen to and futz around with.
Back to my question about the parallel tubes on each side...would I need just one grid choke for the push side and one grid choke for the pull side (to ground, instead of the existing 33K resistor) and leave the individual 1500 ohm resistors to each output tube's grid? Do I 'tweek' with resistors for similar but hotter bias current (compared to stock)?
Back to my question about the parallel tubes on each side...would I need just one grid choke for the push side and one grid choke for the pull side (to ground, instead of the existing 33K resistor) and leave the individual 1500 ohm resistors to each output tube's grid? Do I 'tweek' with resistors for similar but hotter bias current (compared to stock)?
I assume that you refer to the schematic of the Super Twin.
First one remark: the 33K are going to the bias supply not to ground.
To your question: You can use one choke per side as long as you don't want to adjust the bias separately for each tube (pair).
My experience: If you want to go for a bass amp (no guitar overdrive usage) I would insert additional MOSFET follower as driver directly coupled to the power tubes, This has the advantage to enable real AB2 operating mode.
First one remark: the 33K are going to the bias supply not to ground.
To your question: You can use one choke per side as long as you don't want to adjust the bias separately for each tube (pair).
My experience: If you want to go for a bass amp (no guitar overdrive usage) I would insert additional MOSFET follower as driver directly coupled to the power tubes, This has the advantage to enable real AB2 operating mode.
You're right, I'm wrong. Yes, the 33K, thru some other series resistors and caps to ground, is from the 60v AC bias secondary winding that's half-wave rectified.
So where would a grid choke go, in the supertwin schematic I posted? Replacing those 33k bias resistors? Will the 'standard' grid chokes handle the bias current of 3 6L6 in parallel? Do I still 'tweek' the circuit with resistors to achieve some bias spec? There's no spec for current or voltage at the grid in the schematic. Can I still use a small pot on each tube for individual bias adjustment?
I'm really not sure I understand AB1 vs AB2; can't I just bias it hot and drive it even harder to achieve the same thing?
And thanks everyone for your comments, so I can learn.
So where would a grid choke go, in the supertwin schematic I posted? Replacing those 33k bias resistors? Will the 'standard' grid chokes handle the bias current of 3 6L6 in parallel? Do I still 'tweek' the circuit with resistors to achieve some bias spec? There's no spec for current or voltage at the grid in the schematic. Can I still use a small pot on each tube for individual bias adjustment?
I'm really not sure I understand AB1 vs AB2; can't I just bias it hot and drive it even harder to achieve the same thing?
And thanks everyone for your comments, so I can learn.
Last edited:
now, a slight change of subject inspired by this site:
https://sites.google.com/site/yourtubeamp/mods-and-maintenance/-power-reduction/power-scaling
He says he was using the high-voltage regulator "power scaling" kits from London Power. Going for adjustable lower voltage to scale the B++way back, he added a switch to run the B++ from just half the power transformer secondary! That seems like an odd but interesting idea for a more dirty-guitar amp. The regulator would be pretty stiff and hot, but switching one bridge connection from end-leg to center-tap would have more sag. Couldn't he just switch in another dropping rung in the filter ladder (switch only while turned off) and tailor the resistor and cap to his sag tastes? Power-reduction switches to disconnect the cathodes on some output tubes have become popular; how come various B+ switches haven't become as popular? It seems to me that for a low-power mode on a big guitar (not bass) amp I would want one switch that turned off some tubes to reduce current and also lowered the B+ voltage. Then though the volume would not be much less in DB, the power would be lowered so that I could switch in some reasonable-sized power soak resistors at the speaker connection for a 1/4 power switch there. That would get me down to the point where a pair of variable power rheostats becomes practical. And everything would run much cooler than just running a big amp all-out and then trying to burn it off at the speaker output with a huge power soak.
Then he talks about problems with the bias supply when he scales back the B++. I don't understand that. So in a fixed-bias amp, if I want to reduce the B+ don't I want to also reduce the bias?
But I do understand the idea of beefing up the bias supply current and RMS v. via a full-wave bridge on the same transformer taps. That might even be worth doing with an added filter ladder too, just to achieve orignal bias but now with reduced ripple. I understand that the ripple in a push/pull should cancel in the output transformer, but isn't it still worth cleaning up the power? Especially on a grid, where any ripple will get amplified. If that's where the grid chokes go, in the grid bias supply circuit, then not only does the grid choke make it easier for the drivers to push the impedance of the grids for the outputs while keeping a decent DC bias, it also helps filter the ripple out of the grid supply.
https://sites.google.com/site/yourtubeamp/mods-and-maintenance/-power-reduction/power-scaling
He says he was using the high-voltage regulator "power scaling" kits from London Power. Going for adjustable lower voltage to scale the B++way back, he added a switch to run the B++ from just half the power transformer secondary! That seems like an odd but interesting idea for a more dirty-guitar amp. The regulator would be pretty stiff and hot, but switching one bridge connection from end-leg to center-tap would have more sag. Couldn't he just switch in another dropping rung in the filter ladder (switch only while turned off) and tailor the resistor and cap to his sag tastes? Power-reduction switches to disconnect the cathodes on some output tubes have become popular; how come various B+ switches haven't become as popular? It seems to me that for a low-power mode on a big guitar (not bass) amp I would want one switch that turned off some tubes to reduce current and also lowered the B+ voltage. Then though the volume would not be much less in DB, the power would be lowered so that I could switch in some reasonable-sized power soak resistors at the speaker connection for a 1/4 power switch there. That would get me down to the point where a pair of variable power rheostats becomes practical. And everything would run much cooler than just running a big amp all-out and then trying to burn it off at the speaker output with a huge power soak.
Then he talks about problems with the bias supply when he scales back the B++. I don't understand that. So in a fixed-bias amp, if I want to reduce the B+ don't I want to also reduce the bias?
But I do understand the idea of beefing up the bias supply current and RMS v. via a full-wave bridge on the same transformer taps. That might even be worth doing with an added filter ladder too, just to achieve orignal bias but now with reduced ripple. I understand that the ripple in a push/pull should cancel in the output transformer, but isn't it still worth cleaning up the power? Especially on a grid, where any ripple will get amplified. If that's where the grid chokes go, in the grid bias supply circuit, then not only does the grid choke make it easier for the drivers to push the impedance of the grids for the outputs while keeping a decent DC bias, it also helps filter the ripple out of the grid supply.
OK back to fundamentals, if I screw with the bias supply I measure cathode current with no signal...not voltage or current at the grid...
If I solder a low-value shunt resistor in series with the cathode ground for measuring bias current, can I leave the resistor there all the time during operation? It doesn't seem to waste much power if the value is low enough. Does it affect sound at all? Why don't amps come that way, to make it easier to measure the current?
If I solder a low-value shunt resistor in series with the cathode ground for measuring bias current, can I leave the resistor there all the time during operation? It doesn't seem to waste much power if the value is low enough. Does it affect sound at all? Why don't amps come that way, to make it easier to measure the current?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.