Rectifier tube flash

This horse has been beaten to a pulp but just one more if I may?! I have done the recommended diode mod as described above but last night I saw a flash in the tube. I had turned off the amp for about 30 seconds to switch in some different speakers. When I flipped the switch back on I saw a small but brilliant flash come from within the tube. I'm using a Variac to keep the voltage at about 117 volts and the tube and the cap are new. I thought the diodes would prevent this. Can anyone explain why the tube is prone to flashover in the first place so I can understand the issue? Thanks

PS On the sticky thread there is a picture of a diode mod using four diodes. Two are going to chassis. Are they recommended?

The reason you see a flash is because with the short 'off' time, you're effectively doing a hot start.

When the power is first turned on, the filter cap looks pretty much like a short to the rectifier, drawing as much current as the tube can supply, with this current draw fairly rapidly falling off as the cap charges and the voltage across it rises. The tube provides something of a 'soft start' to charge the cap, as it also takes a few moments for the current flowing in the tube to increase as it heats up. The reduced current capability of the tube coincides with the time that the cap is trying to draw the maximum, and between the two things the cap is brought up to its charged state while the current flowing in the tube is limited by its not being fully up to temperature.

If the amp is turned off for a short time (typically less than a minute or two), then turned back on again, the filter cap has time to be discharged (the output tubes are hot and conducting), and does its 'short' act when power is reapplied. The rectifier tube begins conducting more quickly than normal as it's still warm, and the output tubes are likely to still be in a conductive state as well, adding to the 'short' load on the rectifier. Lightning ensues. (The output tubes take longer to heat up from a cold start due to their construction, and thus the rectifier doesn't see them as a load on initial cold power up - it gets the filter cap charged, and as it's finishing that job the outputs start to conduct and everything transitions easily.)

The series diodes don't prevent this because it's a current draw issue, rather than reverse voltage.

To prevent it from happening, if you accidentally or intentionally turn the amp off, let it sit for five minutes or so before turning it back on to permit the tubes to cool off somewhat. Doing so will go a long way to make the rectifier happy.

-Pat

Cubdriver wrote:The reason you see a flash is because with the short 'off' time, you're effectively doing a hot start.

When the power is first turned on, the filter cap looks pretty much like a short to the rectifier, drawing as much current as the tube can supply, with this current draw fairly rapidly falling off as the cap charges and the voltage across it rises. The tube provides something of a 'soft start' to charge the cap, as it also takes a few moments for the current flowing in the tube to increase as it heats up. The reduced current capability of the tube coincides with the time that the cap is trying to draw the maximum, and between the two things the cap is brought up to its charged state while the current flowing in the tube is limited by its not being fully up to temperature.

If the amp is turned off for a short time (typically less than a minute or two), then turned back on again, the filter cap has time to be discharged (the output tubes are hot and conducting), and does its 'short' act when power is reapplied. The rectifier tube begins conducting more quickly than normal as it's still warm, and the output tubes are likely to still be in a conductive state as well, adding to the 'short' load on the rectifier. Lightning ensues. (The output tubes take longer to heat up from a cold start due to their construction, and thus the rectifier doesn't see them as a load on initial cold power up - it gets the filter cap charged, and as it's finishing that job the outputs start to conduct and everything transitions easily.)

The series diodes don't prevent this because it's a current draw issue, rather than reverse voltage.

To prevent it from happening, if you accidentally or intentionally turn the amp off, let it sit for five minutes or so before turning it back on to permit the tubes to cool off somewhat. Doing so will go a long way to make the rectifier happy.

-Pat

I agree and disagree here.
I agree, yes, if you turn off a tube amp and turn it back on shortly there after, it does not like that at all, causing amongst other things, a flash in the rectifier.
However, 30s is more than enough to discharge the capacitor/s, keeping in mind that everything is under load, so the discharge rate is actually pretty quick. I regularly turn an amp back on after about 10s, no more than that, with absolutely no dire effects. I had my DC volt meter on B+, and after about 10s it is down below 15VDC or less.
It could also mean that your rectifier is nearing the end of its life or it just has an issue.

MontanaWay wrote:
I agree and disagree here.
I agree, yes, if you turn off a tube amp and turn it back on shortly there after, it does not like that at all, causing amongst other things, a flash in the rectifier.
However, 30s is more than enough to discharge the capacitor/s, keeping in mind that everything is under load, so the discharge rate is actually pretty quick. I regularly turn an amp back on after about 10s, no more than that, with absolutely no dire effects. I had my DC volt meter on B+, and after about 10s it is down below 15VDC or less.
It could also mean that your rectifier is nearing the end of its life or it just has an issue.

I'm with you on that the filter cap discharges.  My understanding (and I could be mistaken, but it makes sense to me at least) is that short cycling it causes the somewhat warm (and thus with a lower resistance) rectifier tube to see the load of both the discharged filter cap AND the output tubes, which, in the case of what we're using in the Dynacos are indirectly heated and thus slower to cool and completely stop cathode emission.  

In a normal cold start, the rectifier heats up and the current through it increases as the discharged filter cap voltage rises and the current it's demanding drops as it charges.  The initial low current capability of the cold rectifier limits the charging surge that the cap can draw, and by the time the rectifier's 'up to speed', the cap has charged.  A few seconds later, the output tube cathodes warm up, and they begin to draw current, AFTER the cap has charged.  In a hot start, the filter cap discharges, but the rectifier is ready to deliver full current sooner, and the output tubes haven't necessarily cooled off to the point where they're out of the picture.  So you end up with a low resistance rectifier tube trying to fill a fully discharged filter cap AND handle at least some part of the output tube bias current at the same time, rather than sequentially as normally happens.  The low resistance in the load (cap + outputs) leaves a higher forward voltage than normal across the rectifier, and *crack* - lightning strikes.  As I said, I may have the mechanism wrong, but on the surface at least it seems feasible as an explanation.

-Pat

Pat, that's the most descriptive and complete post I've seen. Your description is dead on. Montana, I never considered the additional load of an operational amp. I use a simple circuit to test and reform capacitors. I am amazed at how long it takes to drain a high voltage cap (much more than 10 sec), but in my case they are not subjected to anything more than a drain resistor.

In any case, I think you are both right as the rectifier is flashing because the cycle time is too short. n3ikq does the flash occur if you wait a few minutes before restart?

Kaner

I am going by my own experiences with working with tube amps. The discharge time of the cap/s in a fully operational amp, in my experience, is quite quick.
The aim of my reply is to point to other possibilities/reason as to why the rectifier tube may be flashing, such as age and/or internal fault of the tube.
But yes, the longer time is given to allow the cap/s to discharge, the less chance of any problems occurring.

Thanks for the really good explanations. The tube is brand new. Well may have about 30 hours on it. It only happened once. Doesn't seem to have effected anything and I don't intend on doing it again. I just thought the diode mod was supposed to eliminate the possibility of a flashover. How does a flashover effect a tube and/or shorten its life? Are there major flashovers and minor flashovers? Has anyone seen a single event destroy a tube? I imagine the worst case would melt metal and short things out. I imagine that a minor incident would burn a small amount of plate.

n3ikq wrote:Thanks for the really good explanations. The tube is brand new. Well may have about 30 hours on it. It only happened once. Doesn't seem to have effected anything and I don't intend on doing it again. I just thought the diode mod was supposed to eliminate the possibility of a flashover. How does a flashover effect a tube and/or shorten its life? Are there major flashovers and minor flashovers? Has anyone seen a single event destroy a tube? I imagine the worst case would melt metal and short things out. I imagine that a minor incident would burn a small amount of plate.

The diode mod prevents reverse current flashover like when the amp is suddenly turned off, caps are fully charged up, and the rectifier cathode gets cooler losing its negative charge.  The conduction path in this special case (no diodes) is back through the heater windings to ground which is very low impedance.   In normal operation (forward) the diodes don't do much except drop the B+ AC voltage less than a volt when conducting.  Sounds like you just had a weak rectifier that could not handle power on.  Happened to me on my first fire up of the ST-120 with a new Sovtek.   Luckily, besides the rectifier tube, the only thing I had to replace was the 5A line fuse.   Also, if you have a Time Delay Relay board it will open on power failure removing the low impedance path back through the transformer windings.