I had a spectacular failure of an EL34. It was absolutely incandescent orange and as bright as a Roman candle. I didn’t know whether to run toward the amp to switch the damn thing off or jump behind the sofa!
This is a PP, UL design. I found the screen grid stopper resistor was open and burnt. It looks like a KOA Speer SPR 3 watt 100R, one of those light green carbon film ones with the color band markings. In order to get the amp running again I pieced in a couple 47Rs I had on hand in series, again carbon film. That together with a fresh tube took care of the problem. I’m probably fortunate the output transformer was spared. I’m using some Winged Cs that test well and sound great yet are long in the tooth; I’d happily replace them if not for the current situation with tube supply.
So I’m thinking about the specification of those resistors. Is 3 watts really the correct value? Is carbon film the best choice from a fire safety standpoint? I’ll note that they performed well, failing open with just minimal signs of charing and the KOA Speer data sheet describes them as coated with a flame proof silicone coating. That not withstanding, would I be better served with a metal film or wire wound resistor? And perhaps a 2 watt resistor would have fused more quickly thus saving the output transformer some stress?
Any coaching on best practices here would be appreciated.
This is a PP, UL design. I found the screen grid stopper resistor was open and burnt. It looks like a KOA Speer SPR 3 watt 100R, one of those light green carbon film ones with the color band markings. In order to get the amp running again I pieced in a couple 47Rs I had on hand in series, again carbon film. That together with a fresh tube took care of the problem. I’m probably fortunate the output transformer was spared. I’m using some Winged Cs that test well and sound great yet are long in the tooth; I’d happily replace them if not for the current situation with tube supply.
So I’m thinking about the specification of those resistors. Is 3 watts really the correct value? Is carbon film the best choice from a fire safety standpoint? I’ll note that they performed well, failing open with just minimal signs of charing and the KOA Speer data sheet describes them as coated with a flame proof silicone coating. That not withstanding, would I be better served with a metal film or wire wound resistor? And perhaps a 2 watt resistor would have fused more quickly thus saving the output transformer some stress?
Any coaching on best practices here would be appreciated.
Dave,
Check the input capacitor to G1 for any leakage, causing a lower G1 negative voltage than required or expected.
If you intend to make the G2 resistor rubust, measure the current and the supply voltage. Then calculate the wattage needed.
Without a schematic diagram we can only guess what is needed.
Regards, Gerrit
Check the input capacitor to G1 for any leakage, causing a lower G1 negative voltage than required or expected.
If you intend to make the G2 resistor rubust, measure the current and the supply voltage. Then calculate the wattage needed.
Without a schematic diagram we can only guess what is needed.
Regards, Gerrit
Know that Mullard/Philips advised 1K as the value of the screen grid resistors in UL operation. See for instance: EL34 Mullard
Although some (or most? or even all?) currently produced EL34's are not true pentodes, I would think it's still wise to stick to their advise.
Although some (or most? or even all?) currently produced EL34's are not true pentodes, I would think it's still wise to stick to their advise.
The tube itself might be to blame here. It might have shorted internally causing meltdown. Or as was said the cap from the driver might be leaking badly so def check it before restarting the amp so you don’t fry a good tube.
Would it not be better for the resistor to stay around as long as possible? If you lose g2 voltage doesn't the plate current spike? I'd go with wirewound personally.
The best case failure would be for the G2 resistor to fuse open, so a very small wattage rated resistor is best here. There are small "fusable resistors" but they're really only for fire protection, so may be too slow to help.
All good fortune,
Chris
All good fortune,
Chris
Post # 2 is correct, check the cap on G1 first. If there is a small R on G1 it may be open but that is unusual.
If the R on G2 fails open the tube current simply drops to a very low value.
The screen & G1 stoppers should be physically small as possible & be CC build. And close to the tube socket, no long leads.
And small physically to avoid skin effect since that reduces their ability to retard RF oscillation. No WWs at all. 😀
The small Rs on G1, G2 & sometimes the plate used to be referred to as swamping resistors since their function
is to swamp any tendency to self oscillation.
Small is better, while in research I managed to build tolerable attenuators with 1/2 W Ohmite CC Rs that got to 2 GHz.
The same thing using 2W resistors had trouble at 500 MHz.👎
If the R on G2 fails open the tube current simply drops to a very low value.
The screen & G1 stoppers should be physically small as possible & be CC build. And close to the tube socket, no long leads.
And small physically to avoid skin effect since that reduces their ability to retard RF oscillation. No WWs at all. 😀
The small Rs on G1, G2 & sometimes the plate used to be referred to as swamping resistors since their function
is to swamp any tendency to self oscillation.
Small is better, while in research I managed to build tolerable attenuators with 1/2 W Ohmite CC Rs that got to 2 GHz.
The same thing using 2W resistors had trouble at 500 MHz.👎
As others have said its better for the screen and cathode resistors to fail when the EL34 goes short to prevent damage to the OPT. I tend to fuse the HT with 1KV fuses but you need a reverse protection diode from the transformer CT to ground to deal with the large negative voltage spike which occurs as the fuse blows. Some UL amps do operate right on the extremes of the screen plate dissipation. You can fit some 5W zeners to the screens which will increase the UL slightly but drop the screen voltage. The K4040 was good for popping EL34's. You cpuld try some KT77 but I don't think they are as linear.
https://www.digikey.co.uk/en/products/detail/bel-fuse-inc/0ADBC0600-BE/8028436
https://www.digikey.co.uk/en/products/detail/bel-fuse-inc/0ADBC0600-BE/8028436
Last edited:
I should have added to my previous post that those attenuators were used in 50R transmission lines.
Doesn't mean much at audio but does shew that physical size is a problem as frequency increases.
But skin effect is a factor in AF transformer design & gets in the way as frequency increases.
Looking at the cross section of a winding the copper at the center of the pack is subjected to more magnetic field than that at the outside.
So the AC resistance of the conductors in that part is higher, That is called proximity effect.
Skin effect is a significant factor even at power frequencies (50/60) Hz where it causes significant losses in long distance power transmission.
So HV DC lines are being built. OTOH, skin effect is a useful property used in starting AC induction motors.
Doesn't mean much at audio but does shew that physical size is a problem as frequency increases.
But skin effect is a factor in AF transformer design & gets in the way as frequency increases.
Looking at the cross section of a winding the copper at the center of the pack is subjected to more magnetic field than that at the outside.
So the AC resistance of the conductors in that part is higher, That is called proximity effect.
Skin effect is a significant factor even at power frequencies (50/60) Hz where it causes significant losses in long distance power transmission.
So HV DC lines are being built. OTOH, skin effect is a useful property used in starting AC induction motors.
I checked the coupling caps and resistors and found no discrepancies other than the ones I described. The amp is up and running well now with a replacement tube. Your comment about the wattage rating is at the heart of my question. A quarter watt resistor should be more than adequate for normal operation and yet the manufacturer (Its a Jolida 302b) spec’d a 3 watt. I’m curious of the rationale for that. I’m wondering if start-up inrush might be a factor.
A good observation. I understand the resistor value and the Miller capacitance of the screen set a pole for a low pass filter against RF interference. Lowering the resistance value to 100 from the Mullard recommendation would raise the pole. I can’t see the value in that unless the pole were to get into the sub 20 kHz region and it doesn’t. There are plenty of designs that don’t use a screen resistor and plenty that use a 2K plus value. There is a lot of bandwidth between RF and hearing such that all these values may be fine.
I’m thinking the OT would likely have been destroyed without the resistor opening.
Is it that WWs act as little antenna? Skin effect is Another good argument for lower power rating. I have read recently about HV DC power transmission. Interesting stuff.
I’ll order some 1k half watt carbon film resistors And give them a go.
Resistors have a voltage rating as well. Because of their physical properties I expect the AB 2W CC have a higher voltage rating than 1/2W carbon film.
I use 1W Allen Bradly on G2 in UL with EL34. For voltages over 600V I keep a lifetime stock of cracked carbon resistors. Although very large these were used in Decca radar systems in the 50's and 60's which probably would not have been used if skin effect was an issue.
If induction is an issue you could use a bifilar wound WW.
Last edited:
This might be the reason the existing screen resistors are so large. Looking at the current data sheet for these resistors, the 1/4 w is 300v, the 1/2 and 1 are 500v, the 2 and 3 are 700v. This amp B+ is just over 500. At startup, a bit higher. These resistors might have been produced 25 years ago so it’s hard to say what the ratings were then.
Under operating conditions the voltage drop across the screen resistors are quite low. Probably higher before the heater is on temperature. Yet another reason for the slow turn-on, or time-delayed B+