bias vs HT charging time without Standby

[thetubefairy]

trying to determine how i might design a bias supply that reliably charges in tandem with, or before the HT supply, without the use of a DC Standby switch. and ideally without the use of a more complex automatic HT soft start circuit, though i'll go there if i have to.

by the same token, i'd like to be able to evaluate existing circuits and determine the charging times of the bias and HT circuits, in the interest of eliminating or re-routing the DC Standby switch in most designs. i can always measure both voltages with dual meters on start up, but it'd be great to be able to predict it with some accuracy just from the schematic of a particular amp.

i'm not so concerned about a cold start-up, as the normal 10+ seconds before any cathode current can flow should be more than enough time for the bias to charge up, and i pretty much always have a thermistor in the primary anyway. mostly concerned about the occasional hot starts, when the amp is run for a while, gets shut off for only a minute, then switched on again while the filaments and thermistor are still pretty hot. not a super frequent occurrence, but fairly inevitable at some point, and i've seen it short rectifier diodes.



so what factors control those charge time constants, and what factors would decrease that charge time for the bias circuit? can it be designed to have a reliably quicker charge time than the HT without more elaborate slow start circuitry? and can any simple design element reliably ensure the bias voltage charges first during hot starts?

i'm sure it depends a lot on the topology... my guess would be that 'charge pump' type bias supplies that are capacitor coupled to a high voltage winding(s) with no center tap, would have a bit slower charge time than one through a bias tap on a center tapped winding, or auxiliary winding. and that smaller capacitor values in the bias supply compared to the sizes used in the HT supply would help with the charge time ratio. but i'd like to be able to calculate those factors to a degree of certainty.

[Roe]

trying to determine how i might design a bias supply that reliably charges in tandem with, or before the HT supply, without the use of a DC Standby switch. and ideally without the use of a more complex automatic HT soft start circuit

How about a thermistor? or a tube rectifier? Small bias caps does help by charging faster

[sluckey]

trying to determine how i might design a bias supply that reliably charges in tandem with, or before the HT supply, without the use of a DC Standby switch. and ideally without the use of a more complex automatic HT soft start circuit, though i'll go there if i have to.

Just copy Marshall's typical bias supply. It uses 10µF caps that charge quickly.

[R.G.]

i'm not so concerned about a cold start-up, as the normal 10+ seconds before any cathode current can flow should be more than enough time for the bias to charge up, and i pretty much always have a thermistor in the primary anyway. mostly concerned about the occasional hot starts, when the amp is run for a while, gets shut off for only a minute, then switched on again while the filaments and thermistor are still pretty hot. not a super frequent occurrence, but fairly inevitable at some point, and i've seen it short rectifier diodes.

Maybe something like this:
http://geofex.com/Article_Folders/mosfe ... 0Clamp.pdf
MOSFET limits the size of rectifier charging pulses to the size needed for full power, but not sudden surges, like startup. Caps charge at a moderate rate.
If you parallel a transistor output opto isolator with Q2, you get an optically isolated off/on switch for the B+ as well. Usual concerns for high voltage and heating on the MOSFET.

[B Ingram]

trying to determine how i might design a bias supply that reliably charges in tandem with, or before the HT supply, without the use of a DC Standby switch. ...

Forget the HT supply, and focus on the heater warmup time (something like 10-25 seconds?).


If the output tube heaters are cold, the cathode doesn't emit electrons. You have probably >90% chance your bias supply will be substantially non-zero-volts before the cathodes get warm enough to pass plate current.

Give that a shot with any fixed-bias amp you already have.

[thetubefairy]

great suggestions all, thanks!

R.G., that FET current limiter is a really cool idea... correct me if i'm wrong, but this circuit would limit startup pulses even during hot starts, correct? as in power blips, or any occasion where a hot amp gets turned off then rapidly on again? that's the main thing i'm after here, as that's what i figure is about worst possible conditions for the amp to go through, and it's likely to happen to most amps at one point or another.

one thermistor more or less takes care of the job for any cold start, as has been mentioned, but aren't much help on hot starts. sometimes i'll include a thermistor rated for several times the amp's current, so that it actually stays pretty cold during operation, and that works well to protect during hot starts, but is only practical in amps with fairly low current where dropping a few volts off the primary is beneficial anyway.

besides that, i'm hearing that keeping the bias filter caps small is the best method. why might some amps use more than 10uF caps, then? perhaps it's just needed to filter out enough hum with some bias circuits?

i'm in the habit of minimizing bias circuits resistance to ground, to minimize risk of thermal runaway for the power tube grids under heavy overdrive. it occurs to me that this might possibly lengthen the initial bias charge time upon startup, would this be the case?

[cdemike]

If your concern is handling DC on a standby switch, you could always have a double pole standby switch that (dis)connects both of the HT taps from the B+ rectifier. If your power supply uses a center tapped PT, then the bias supply can then draw from one pole of the standby switch. Obviously if you're using a dedicated bias tap, then changing the configuration of the standby switch solves the problem by itself.

I also try to install a fuse between the output tubes' cathodes and ground as protection against both bias supply failure and inappropriate output section loading.

i'm in the habit of minimizing bias circuits resistance to ground, to minimize risk of thermal runaway for the power tube grids under heavy overdrive. it occurs to me that this might possibly lengthen the initial bias charge time upon startup, would this be the case?

Minimizing the overall resistance between ground and the PT tap generating bias voltage is the criterion for charge time. If I'm not mistaken, the key value when designing to prevent runaway with large input signals is the resistance between the output tubes' control grids and the current source generating the bias voltage (i.e., the power transformer).

If this power supply is for an amp with any DC-coupled stages you may want to also consider installing a diode or neon bulb between that triode's grid and cathode to prevent arcing during startup.

[R.G.]

R.G., that FET current limiter is a really cool idea... correct me if i'm wrong, but this circuit would limit startup pulses even during hot starts, correct? as in power blips, or any occasion where a hot amp gets turned off then rapidly on again? that's the main thing i'm after here, as that's what i figure is about worst possible conditions for the amp to go through, and it's likely to happen to most amps at one point or another.

Correct. It limits peak currents, period. So you have to be careful to leave enough current peak to run at full/max power in steady state. But startup current is even bigger. You also have to do the basics, making sure the MOSFET will survive the current/voltage/power demands. Effectively, it sets the maximum rate at which the filter caps can be charged, and eventually the maximum power that's allowed into the filter caps in steady state.

If you set the limiting level just a bit higher than full/max power current peaks, then there is no limiting at all - the only losses are the MOSFET on-resistance and the resistance of the sense resistor. It only kicks in to limit when the caps are trying to pull even bigger peaks than maximum operating conditions.

Not originally my idea - I adapted it for tube amp charging/rectification from a bipolar circuit in an old copy of the Radio Amateur's Handbook.

besides that, i'm hearing that keeping the bias filter caps small is the best method. why might some amps use more than 10uF caps, then? perhaps it's just needed to filter out enough hum with some bias circuits?

i'm in the habit of minimizing bias circuits resistance to ground, to minimize risk of thermal runaway for the power tube grids under heavy overdrive. it occurs to me that this might possibly lengthen the initial bias charge time upon startup, would this be the case?

Smaller grid leak resistors from grid to bias does let more grid current through in certain conditions. You're right to think about it.

I have a mildly different view on the bias charge time issue. IMHO, if you want something to stay off until you're ready, don't ditz around with careful timing ramps using electro caps to do it. Electro caps are always drifting and failing. That's one reason I mentioned using an opto isolator to just shut the current limiter off. You could hack together a circuit that didn't enable the MOSFET until the bias was bigger than some voltage, then enable it. It's another way.

[R.G.]

Typical bias voltage supplies charge to full in less than half a second. There's no need to worry about it and start making things complicated.

Way cool! I'd love to see pictures of the oscilloscope traces of that.

[R.G.]

Way cool! I'd love to see pictures of the oscilloscope traces of that.

Then you better start scoping away to fulfill your heart's desires.

I'll do that the next time I tear into an amp.

[Roe]

Most Fender-style bias circuits charge very fast, so the easiest thing is to just copy one of those and stop worrying. Some Marshalls have bias voltage supplies where it takes something like 5–10 seconds for it to fully charge, but even with those, there's nothing to worry about because the discharge time is slow too. Tube filaments and cathodes don't retain heat and will get cool very fast after the amp is turned off, so it will pretty much always take several seconds before cathode current is flowing when the amp is turned back on.

there is a well-known issue with the bias supply of many Marshall 1970s 50w models. On these, Marshall used the standby switch to switch HT and bias voltages alike. The result is a lot of blown power transformers.

[Astronomicum]

I would agree that shorter power off durations are more problematic than longer shutdowns. Once the bias circuit is charged, slowing the discharge time might be an alternate approach to quicker charging. If you could disconnect the bias circuit lower voltage divider from ground, the bias supply would hold its charge for quite some time, especially with large capacitance. What if you were to use some sort of optically coupled transistor to disconnect the resistor from ground? Driving the LED is easy enough. The circuit would also be fail safe if something happened to the transistor.

[sluckey]

This simple diode/cap addition (red circle) will hold the bias up for a long, long time when power is removed. The voltage on the left side of the diode will decay quickly but as soon as the diode becomes reverse biased the extra cap will remain charged for a long time.

I've never tried it so I can't say if it has any ill effects. I used Mark's drawing because it's drawn much better than any Marshall schematic.