Cathode bias regulation

[Chuck H]

In one of my designs I use a zener across the power tube cathode resistor/cap to "fix" the bias at a certain point in the shift. The circuit works fine but I was wondering...

Couldn't one also terminate this zener at the grids rather than ground? Any DC breaking over the zener should then fix the grid cathode differential, Right? So I'm wondering about any advantage doing it one way or the other. The first thing that comes to my mind is that with the zener terminating at the grids there would be less current across the diode (and therefor less heat). The second thing that comes to mind is that any LF not bypassed by the cathode capacitor may be problematic.

Any thoughts?

[pdf64]

Having trouble visualising what you're describing - could you add a schematic?

[Chuck H]

1) is how I've been doing it
2) is just an idea.
Wondering about pros and cons. Obviously 2) is the more involved circuit and would require a new layout, but if there are any advantages.?.

[Dave H]

I’d stick with the zener across the cathode resistor to ‘fix’ the bias. For large signals the second circuit may make the bias shift greater. The diodes will limit the negative swing of the signal which will stop the valves turning off for that part of the cycle making the average current and therefore bias shift greater (I think).

Dave H.

[Chuck H]

Hmmm... I also use the Paul Ruby mod on this model (omitted from drawing) which may not be needed if the regulation is done this way. As I see it, any -V signal loss would need to pass through the zener (+ another volt or so for the standard diode) that sets the bias voltage. The tubes should cut off, right? Any voltage rise at the cathode should be analogous to a positive DC voltage presented to the grid "fixing" the bias relationship between the grid and cathode.

[diagrammatiks]

does the zener provide significantly better regulation then just the resistor and cap?

a truly regulated cathode would just run a mosfet or a regulator to set the exact bias.

it would have the same effect as a resistor with a completely bypassed cap, only it would be regulated completely.

[Chuck H]

The goal with the original setup was to get a cathode bias "sag" but have the amp tighten up under over drive. The zener across the cathode circuit does this just fine. ie: 10V on the cathode at idle, amp begins to OD and the cathode rises to 13V, set the cathode rise limit to 14V with a 14V zener. Now I get the soft attack on cleaner tones but the amp tightens up when I'm wailing on it. Regulating the cathode can't do this. If one steady voltage was what I wanted I would have built in a fixed bias supply and been done.

I've always been a little vexed that I use the Paul Ruby circuit in this amp. My brain won't rest until I find my own equally eloquent solution to the enduring problem of crossover distortion in "18 watt" type amps (since mine is one).

Using the Paul Ruby circuit and the zener across the cathode circuit DOES give me what I want. I would rather have my own circuit though.

[trobbins]

Your second circuit is just the Ruby concept with the clamping of the grid taken to the cathode, rather than 0V. I recall this option being discussed in a few threads over the years.

In my view, I think taking the zener to 0V is in theory slightly better, as it minimises the influence of cathode bias changes - but there is probably not much real difference.

Be mindful that the zener is aimed at allowing a compensating discharge current flow in to the coupling capacitor in each alternate negative half of signal swing, given that the previous positive swing had increased the DC bias voltage across the coupling cap due to overdrive induced grid current flow (ie. charging the capacitor up a bit).

If operation was just class A, then cathode bias voltage would not change, and you would size the zener voltage to be about the same as the idle cathode bias voltage (without taking in to account grid stopper, or the diode in series with the zener).

If operation went in to class B at overdrive peaks, then the cathode bypass cap voltage will increase. This increase, by itself alleviates the coupling cap charging, as the grid has to get to a higher voltage above 0V to cause grid conduction. As such, the Ruby zener voltage would be set higher to match the increased cathode bias level during overdrive.

An example is illustrative. If you tie the zener to 0V and size it to say 14V, with an idle cathode bias of 10V, and introduce an abrupt overdrive then zener clipping would take place at -24V relative to cathode on the negative peak (as the cathode bypass capacitor had not yet charged up). If you tie the zener to the cathode and size it to say 28V, given the same conditions as just described, then zener clipping would take place at -28V relative to cathode on the negative peak.

Ciao, Tim

[Chuck H]

Your second circuit is just the Ruby concept with the clamping of the grid taken to the cathode, rather than 0V. I recall this option being discussed in a few threads over the years.

Well, it was originally intended to equalize the bias relationship between the grid and cathode. The reverse action of bleeding conduction current during cutoff (like the Paul Ruby mod) was a secondary realization.

In my view, I think taking the zener to 0V is in theory slightly better, as it minimises the influence of cathode bias changes - but there is probably not much real difference.

Taking the zeners to 0V is certainly less complicated... Thats always something.

If operation went in to class B at overdrive peaks, then the cathode bypass cap voltage will increase. This increase, by itself alleviates the coupling cap charging, as the grid has to get to a higher voltage above 0V to cause grid conduction. As such, the Ruby zener voltage would be set higher to match the increased cathode bias level during overdrive.

Right. What I've done in the past is use another zener across the cathode circuit. Then as long as the PR zenerV value is above the cathode zenerV value I know the amp is in cutoff when the PR diodes are conducting.

An example is illustrative. If you tie the zener to 0V and size it to say 14V, with an idle cathode bias of 10V, and introduce an abrupt overdrive then zener clipping would take place at -24V relative to cathode on the negative peak (as the cathode bypass capacitor had not yet charged up). If you tie the zener to the cathode and size it to say 28V, given the same conditions as just described, then zener clipping would take place at -28V relative to cathode on the negative peak.

Hmmm... So, somrthing I hadn't considered. I was counting on that big big fat capacitor across the cathode resistor. But anything getting past the diodes isn't AC anymore.

[trobbins]

I have used the zener across the bypassed common cathode as well in a 6BM8 UL PP 1960's stereo amp as a quick fix to manage overdrive.

I don't know about your comment "But anything getting past the diodes isn't AC anymore." To me, the instigating grid conduction from overdrive is a form of DC biasing (which tends to raise the cathode bias level via the cathode coupling cap, and hence minimise the potential blocking distortion effect). The zener is just attempting to reverse the impact of that grid conduction, in the other half of the AC cycle, and hence to minimise the net DC bias shift across the coupling capacitor. In terms of energy flow, the zener is a quick cycle by cycle compensation scheme, rather than waiting for the bleed effect of the grid leaks.

Ciao, Tim

[Chuck H]

"isn't AC anymore" meaning that to achieve the desired secondary action of fixing the bias at a given grid cathode relationship the cathode bypass cap would have to dump any positive swing impressing on the cathode through the diodes...Which of course it can't since any positive swing voltage getting past the diodes has been "rectified". And caps don't pass DC voltage.

More so than heaing the bias I think the "cycle by cycle compensation" dumping voltage build up from the coupling caps is primarily responsible for the elimination of crossover distortion. At least in guitar amps where your driving the piss out of the tubes and grid conduction is a constant battle.

[trobbins]

The level of grid current that can flow via a PI stage is very small when compared to the current level required to change the PP common cathode bypass capacitor voltage (ie. the anode current peaking level). So any zener/diode network that connects to the common cathode, and is used to restrain grid voltage levels, will not really have any influence on the voltage of the cathode bypass cap.

Yes, crossover distortion caused by blocking can be ugly, and the Ruby technique is a good final back-stop if grid stoppers and altered values of coupling cap and grid leak don't give enough benefit.

[Chuck H]

OK. So your saying that any voltage from the PI that presents itself to the top of the cathode circuit has no effect on bias because of it's low current capacity. Good. Then the way I see it it's entirely possible, using the above #2 circuit, to achieve bias shift clamping AND PI negative swing clamping with a low impedance discharge path during cutoff.

[trobbins]

"bias shift clamping AND PI negative swing clamping "

Did you mean cathode voltage clamping when you refer to "bias shift"? Cct #2 doesn't clamp the cathode voltage.

[Chuck H]

I meant bias clamping. Not cathode voltage clamping. Any positive voltage above the zener voltage on the cathode will be present on the grids. If the same positive voltage rise is present at both the cathode and the grid does this not clamp the bias to a fixed grid/cathode relationship?

It does occur to me that I will need to add another zener to seperate the two -PI swings for accuracy. Otherwise the two halves pass through what amounts to a full wave rectifier prior to clamping making the voltage seen by the zener 1.4X the desired clamping voltage. The -PI swings would clamp before the power tubes were in cutoff.