Bias without standby switch

[Keppy]

Anyone care to school me on bias voltage during startup in a non-standby situation? I've seen some pretty good arguments against using standby switches, but I've never seen this aspect addressed.


The bias supply often seems to use higher value filter caps than HT, so it seems intuitively possible that depending on the diodes/resistors used, the bias caps could take longer to charge than the HT caps, causing insufficient negative bias and excess current through the power tubes during startup.

• Am I wrong about the relative timing of the development of bias and HT voltages?
• If I'm not wrong is it a problem? Does the answer depend on the component values?
• If it's not a problem, why not? Is heater warm-up time long enough that cap charge-up time doesn't matter? Maybe the excess current doesn't last long enough to matter? Something else?

[Latole]

1- yes you are wrong, value of bias filter caps is not that high to be issue.


Short answer ;
I put standby switch most of the time on the amps I built from scratch.
I understand why some people don't like Standby switch.

To me is like Fender or Gibson

[pdf64]

…
The bias supply often seems to use higher value filter caps than HT, so it seems intuitively possible that depending on the diodes/resistors used, the bias caps could take longer to charge than the HT caps, causing insufficient negative bias and excess current through the power tubes during startup.
…

Got an example?
It certainly could be a concern, but the time constants in the bias supply would have to be perhaps an order of magnitude larger than usual for that to be the case.

Consider that a decade’s worth of production of Marshall 50W models wired the AC feed to the bias supply after the standby switch. So HT was applied to hot valves with, momentarily, zero bias.
It’s appallingly bad design and the life of the output valves is surely affected, but the valves don’t immediately explode, amps are still running like that 50 years later.
https://www.thetubestore.com/lib/thetubestore/schematics/Marshall/Marshall-JMP-Lead-50W-1987-Schematic.pdf

[Latole]

Wow !

Another Jim Marshall mistake ....after copying the Fender Basmann 

[mresistor]

Get two meters   tie one to grid of a power tube and another to the plate and watch the separate voltages come up when you turn the amp on without using the standby.  I built a single channel deluxe reverb wihout a standby switch and there are no issues. I rarely use the standby switch on my Super I just leave it up, not problems.  Probably millions of Princeton Reverbs out there working fine without a standby switch.


I've not seen a bias filter cap larger in value than a ps filter cap. Nor with higher voltage rating.

[Keppy]

I appreciate all the replies, but I think I'm being mis-read here. I'm not advocating for standby switches or asserting that there's a problem they solve. I'm also not trying to make a decision about whether to put one in a build or not. I'm well aware that plenty of amps are built without standby switches and have no issues.

I'm trying to understand WHY this set of circumstances is not a problem.

Got an example?

This one has 30uF on HT, 100uF on the bias.
https://el34world.com/charts/Schematics/files/Ampeg/Ampeg_g15_gemini_ii.pdf

...the bias supply would have to be perhaps an order of magnitude larger than usual for that to be the case.

This one has 60uF for HT, 1000uF for the bias, and the bias supply has a 120k series resistor before the cap to make the RC constant even bigger.
https://el34world.com/charts/Schematics/files/Mesa_boogie/Boogie_subwayrocket.pdf

Consider that a decade’s worth of production of Marshall 50W models wired the AC feed to the bias supply after the standby switch. So HT was applied to hot valves with, momentarily, zero bias.
It’s appallingly bad design and the life of the output valves is surely affected, but the valves don’t immediately explode, amps are still running like that 50 years later.
https://www.thetubestore.com/lib/thetubestore/schematics/Marshall/Marshall-JMP-Lead-50W-1987-Schematic.pdf

This is what I'm looking for. If I'm getting what you're saying, it's that there is an excess current condition in these situations, but for the length of time in question it doesn't matter, at least in the short term. Please let me know if I'm understanding you wrong.


This set of circumstances might be less common than I thought. I happen to own these two amps and have looked at the schematics a lot, so I'm used to seeing large bias caps. I looked for Fender/Marshall examples and saw much smaller bias caps. Although, the 1987 link posted above shows a 220k resistor in series with the bias cap, so even with a smaller bias cap that's potentially a much larger RC constant than the HT filter.

[PRR]

It's not *just* the cap values. Resistance matters also. So do funky circuits like Marshall's 0.047u to the diode. Also Ampeg's switching in the FWB return. And yes the time constant IS generally longer for the sensitive grid than the insensitive plate.

Radio transmitter builders know that power tubes can be mis-biased for many-many seconds while you dial-in the tank circuits. Some datasheets even have a number, OTOO a minute at a time.

"it doesn't matter, at least in the short term."

[pdf64]

…
This one has 60uF for HT, 1000uF for the bias, and the bias supply has a 120k series resistor before the cap to make the RC constant even bigger.
https://el34world.com/charts/Schematics/files/Mesa_boogie/Boogie_subwayrocket.pdf

I suspect that there may be some components missing from the Mesa schematic? eg a decoupling cap and shunt resistor at node F.
Without them, what’s the point of the 33k resistor?
On your amp, what are the voltages at the 1mF cap, and at node F? If there’s a voltage drop across the 33k, then current is being drawn through it.

… the 1987 link posted above shows a 220k resistor in series with the bias cap, so even with a smaller bias cap that's potentially a much larger RC constant than the HT filter.

I think that according to Thevenin, the bias reservoir cap will charge up via the 220k // (15k + 47k + 25k), so the effective time constant may be a fair bit lower than you’re thinking.