issues with a GZ34 and open circuit anode resistors.

[bikerhifinut]

Any ideas gents?
I was running up my EL34 monoblocs prior to running a few tests before I take on the task of the off board PSU. One of them on switch on had an instant fault. The GZ34 rectifier flashed quite alarmingly, fireworks in side the bottle and then although the valve stayed lit up the amp was very obviously dead.
I was fearful that the output transformer had gone S/C to cause this but on examination and measurement all I could ascertain was that the 100R anode resistors on the GZ34 had both gone O/C but no distress marks on them at all. These are fairly cheapo white cement coffin jobs from maplin rated at 7W which should be more than ample. So is it just **** resistors or something else? All I can think is the PSU electrolytics could be to blame as they run very close to their 500V rating and have on the odd occasion gone a bit over when testing and setting the amp up (High mains volts and transformer with only 230V taps). I can find no problems elsewhere in either amplifier. So my instinct is to junk the "Ruby" USA made (I think) PSU caps from Hi Fi collective and source some better branded jobs, maybe wire em as series pairs to handle any voltage surges.
The valve is definitely sick now too so was it the valve failing that caused the resistors to go open circuit or a fault elsewhere that took the valve out too? The thing is that the flashover occured pretty much at switch on so that is a mystery too.
for the record I have gone through a few anode resistors on the rects so that may point to a fault in the PSU? All circuit voltages check out as bang on design values by the way and I am sure cathode bypass caps are good as are coupling caps.

Andy

[bob orbell]

I would say that it is only the valve that was at fault Andy, do you have a variac ?, they are very handy with this game. BOB

[bikerhifinut]

Its an item I am sadly lacking Bob, I can probably lay my hands on one.
however these monos have a history of causing rectifier resistors too go OC, in the past they have usually shown signs of overheating first, not every time but this time there was a fast and obviously catastrophic failure of them. I think its a combination of lousy quality resistors and something else PSU related that's causing the stress, hence my suspicion of the main reservoir and smoothing capacitor. by the time the volts get to the phase splitter and first stage supply the stress on the smoother/reservoirs is bugger all.

I was minded to try a polyprop first cap rated at 630V I don't need 50uF before the choke. 16uF would be fine and 32uF ample. The original design i adapted these from used only 16uF after the rectifier.

A.

[pre65]

What make is the GZ34 ?

Does the "flashing" start immediately the amp is switched on ?

Check the GZ34 data sheet, because there is often a maximum size for the first capacitor.What value were you using ?

Was the value of the anode limiting resistors calculated by yourself ?

Do you use a thermistor on the amp mains input ?

[bikerhifinut]

Originally Posted by pre65

What make is the GZ34 ?

Does the "flashing" start immediately the amp is switched on ?

Check the GZ34 data sheet, because there is often a maximum size for the first capacitor.What value were you using ?

Was the value of the anode limiting resistors calculated by yourself ?

Do you use a thermistor on the amp mains input ?

It's a genuine Blackburn mullard as is its mate in the other amp but even a Chinese one would be adequate at the voltage and current I am using. It's venerable but as far as i could see still perfectly healthy.

It flashed just the once at switch on, after that I assume the anode resistors had failed fairly instantly thus leaving it O/C and therefore no load on it.

Trust me Philip I have been around long enough to know about the maximum capacitance straight off the cathode, I was using 50uF easily inside the 60uF maximum.

The anode resistors are fine at 100R, at that voltage that's the Mullard recommended value off their data book, and with the transformer internal resistance of 47R thats a good 150R so a bit in reserve to limit current inrush. The Leak amps never used any sort of limiting resistor relying on the transformer resistance to provide the load and mine have never been a problem.

I see no obvious reason to use a thermistor on the mains input, on an EI transformer you don't see one on commercial or even the WD designs. it's got a fuse nicely rated at just above the load current which seems to me to be enough. Now if it was a big toroid then no doubt at all use a GEC MOVistor or similar device (varistor) to counteract any voltage spikes on switch on or switch off. This is something i will need to watch when I build the offboard PSU for these amps as its a pretty massive toroid that I have had custom wound by a friend along with dire warnings about correctly protecting it!

The circuit looks and measures fine, no obvious shorts or component failures anywhere. I am starting to wonder if the type of anode resistor is the issue as they were cheapo (except they aint that cheap) maplin sourced white cement coffin jobs. I know JC has a low opinion of them and I now think i can see why. They seem to go open circuit arbitrarily.
The other issue is possibly the type of electrolytic i am using. They are "Ruby" 50 + 50uF at 500V sourced from hi fi collective connected across a Hammond 5H choke which runs very cool so not concerned there. I suspect they are barely capable of withstanding their 500 rated dc volts and a fair bit of ripple not withstanding they were advertised as suitable for PSU use.
My gut feeling right now is to swap the first 50uF out and shoehorn in a 10 to 16uF polywotsit 630V job. If the Rubies are ok still I can safely parallel them to give 100uF south of the choke. I can only assume its either faulty electrolytics or just too much inrush current (the faulty elcos having a similar effect) which either takes out the resistors and/or stresses the rectifier valve. I am minded to try silicon but frankly as I am modding the amps anyway its a waste of time as they will be fed from a good silicon rectified supply once thats built.
No i think the main issue is that I may have used underspecced PSU capacitors and have possibly trashed them when I should have used series connected pairs at the front of the PSU to get the working volts up above 700 or so to give a safety cushion. Not an expensive solution in the scheme of things.

As an aside i checked the other mono as it seemed to be humming a bit too much and the already very noisy mains transformer seemed to be throbbing more than its usual racket and lo and behold one of the rectifier anode limiters/droppers had also gone O/C, no sign of overheating or distress of any sort.
I think its simply crappy wirewound resistors not fit for purpose and possibly the elcos straight off the cathode.
Back in the day Harold Leak used 16 uF here and in the higher powered TL50 8 uF as first capacitors. I always assumed because in those days at those voltages bigger electrolytics just weren't available, at least at any sort of reasonable price.
But now I'm wondering if there were other reasons. I'll dig the data out again and see if I missed something in the small print.
I could always use a 5U4 and take the voltage drop on the chin bearing in mind I'd have to keep the capacitor below 10uF in that case. the transformers can handle the extra amp of heater current.

Phew.

Andy.

[Richard]

Did you check the voltage at cold switch on Andy? Thinking the caps may be going short. My Matamp 5-20s had valve rects straight onto 2 seried 300uF caps (=150uF), no choke or second filter before going to the output, 100Rs on the anodes and no problem or hum. I tried several types without problem inc Mullard GZ34s and preferred the old coke bottle style 5U4 winged C (5U3 Russian).

[bikerhifinut]

I've had some feedback from some electronic engineering accquaintances.

The way the GZ34 went is apparently typical of the beast, they seldom die quietly apparently! it's quite possible that a momentary arc as it flashed caused enough of a voltage surge to take out the resistors and if they were a bit **** they possibly saved the mains and output transformers from injury. Maybe.

All the same the main PSU cap will be suspect and I'll likely just replace it with a single one of 100uF or more (although at 600V wkg I reckon I'll be pushed to find many choices). I think a low value poly like the 4.7uF I have in the parts bin may be useful and as you say I can try a 5U4 as there's enough capacity in the mains transformer to handle the extra amp. But then i'd be tweaking the anode resistors to compensate for the voltage drop.

[bikerhifinut]

If you were sticking the cathode of a 5U4G onto more than 10uF you would be stressing it quite badly. These old directly heated rectifiers can't tolerate much capacitance straight off the cathode even though they are good for more steady state current than a GZ34. I used a 5U4G after I refurbed my Stereo20 before I knew better and it would quite often flash over as the 50uF first capacitor charged up. suffice to say I wouldn't do it now.

A.

[Richard]

Originally Posted by bikerhifinut

If you were sticking the cathode of a 5U4G onto more than 10uF you would be stressing it quite badly. These old directly heated rectifiers can't tolerate much capacitance straight off the cathode even though they are good for more steady state current than a GZ34. I used a 5U4G after I refurbed my Stereo20 before I knew better and it would quite often flash over as the 50uF first capacitor charged up. suffice to say I wouldn't do it now.

A.

The Matamps were built that way and worked fine with a variety of valve rects into 150uF first cap.

It intrigued me at the time so I drew out the non-standard 5-20 PSU and did some reading and found we can run bigger first cap if the supply impedance is increased. The supply impedance is that of the mains TX plus any anode resistors. TX voltage, supply impedance, and first cap value are inter-related.

That was a while back so I've just checked here from Valve Wizard, http://www.valvewizard.co.uk/bridge.html

"In addition to the maximum AC voltage and DC current ratings, valve rectifiers have two other ratings that must be observed: maximum allowable reservoir capacitance, and minimum current-limiting resistance. These two limits are interelated and serve to keep the peak ripple current below a certain (unstated) level. The bigger the reservoir capacitance, the more limiting resistance you need. The GZ34 data sheet quotes a maximum capacitance of 60uF, although you can, in theory, exceed this if you increase the limiting resistance proportionately. However, this incurs extra voltage loss and wasted heat which is why the manufacturer assumes no one would want to do it.
The total limiting resistance (per anode) in the actual circuit is the combnation of transformer resistance, plus any resistance we add ourselves:"

[Richard]

Moving along, the Matamps too only had 100R on the anodes... so the reason for them working fine must lie with the mains TX impedance and voltage.

Here are a couple of pics. The inside shows the green anode 100Rs bottom left. The big first caps are outside in the other pic. (That little choke supplies only the first stages not the output stage btw.)

The drop-through OPTX is one of the same batch as yours so you know it's good and on a big stack but look at the size of the chassis mounted mains TX. It never got hot or hummed.

The point being that mains TX must have had a fairly high impedance and almost certainly had good self-regulation (as in it was wound for the needed voltage and didn't vary much with specified load). A smaller TX would be wound for a higher voltage as the maker would expect it to be dragged down under load. (Load causes heat and humming in a small TX and high voltages at switch-on when there is little load.) Thus at switch-on the larger TX voltages would probably not go as high as from a smaller TX.

As you know your problem may be a variety of reasons from failed valve to failed cap etc. I'd firstly check voltages from stone cold to hot on the good one to get bench mark figures, as it is switch-on when the problem happens. Then change the first cap to new series ones totaling the same uF or even a bit more if you fancy at 700+V. Then use 200R anode resistors (as spec'd for GZ34 running 550V). Then fire it up with a new rect and hope...