Rectifier Flashover and open circuit anode resistors

[bikerhifinut]

Another idea flitted across my brain cell.

I have a pair of 630V rated Polypropylene 4.7uF capacitors languishing in the toybox. What, if any, effect would it have on things if i used these as the first capacitor off the rectifier? If nothing else i could use a 5U4G that I have and if the extra voltage drop wasn't too severe it would be a cheaper option than a GZ34. (Plenty of ex military russian ones about for cheap).

A.

[Ed_Dinning]

Hi Andy, note that some of the Chinese coffins are not actually wire wound inside the case and have a much poorer normal to dying ratio than wire wounds.

Ed

[bikerhifinut]

As I understand it Ed they should be WW up to about 1k and thereafter are metal oxide. But who knows for sure and somethings up for certain. I suspect they just can't cope with high surge currents. I'll be buying a selection of "proper" WW resistors next time. I need a batch of bits so will try RSonline for some decent welwyns or whatever the current equivalent is.
I find it useful to have a few values when prototyping.

Its probably going to be academic once I get the toroid sorted out but it's good to get to the bottom of a problem.

This all started after the second pair of mains transformers turned out to be pigs in a poke. You may remember Mike B inspecting them for me at golborne last year? And the verdict wasn't a good one. It's made me very wary of commercial transformer manufacturers.

A.

[ronbryan]

Re. post 21: There would be increased 100Hz ripple compared to that with a 50uF reservoir capacitor.

Ron

[bikerhifinut]

Granted.

It then goes via a 5H choke and then to the rest of the Psu and I figured to configure the second cap, currently half of a 50uF in parallel as its part of a dual 50 +50 with a common negative terminal, as 100uF. figuring the choke will reduce ripple quite a lot and the second filter/reservoir capacitor will do the rest. as that stage only feeds the push pull output pair, common mode noise will be cancelled out in theory and then the phase splitter and first stage valve are fed via dropper resistors and further 50uF decoupling capacitors so by the time the supply reaches the first stage it's had a further 3 stages of filtering 1 LC and 2 RC stages.
It struck me that I can use a valve like a 5U4G to rectify the HT if I reduce the first capacitor substantially. The added Voltage drop in this case wouldn't be a disadvantage and the transformer is well capable of handling the extra amp of heater current due to my over specifying it. Now a lot of data sheets suggest the 5U4G can take a 40uF capacitor but my Mullard data book quite clearly states 10uF as the limiting capacitance. My own experience in the past with the 5U4 indicates that the Mullard data may be the accurate one as I have seen a 5U4G flash over with the inrush current with a 47uF capacitor. It's directly heated and I reckon that could be part of it as it starts to conduct within a couple of seconds of turning on. The reason why I plucked 4.7uF out of the air was because i have a pair of polypropylene foil/film 630V ones and they'd be well capable of handling the ripple current.
Just a thought.

A.

[joebog1]

You will need UF rectifiers Biker!!! Rated at LEAST 6 amps, and DONT cut the leads off!!!! They are the heatsinks for the rectifier "chip" itself. That toroid will happily supply 1 amp, but as I told you it is low flux and rated at about 600 mA forever.

For the heaters, I would be using a UF 25 amp bridge mounted on a BIG lump of heatsink.
Ship the ~ 12 volts DC up to each amp and fit a heater regulator there. For heater umbilical wires I would be contemplating 3mm automotive wire. (The twin stuff in a black outer sheath, with red and black inside) Its cheap and will carry the load. I use LM338K TO3 cased 5 amp regulators in my amps. They cost about 30 shillings, are overload protected, over temp protected. and only require a few bits that are easily attached to a length of 4 way tagstrip. They are also adjustable so that any "losses" due to volt drop within the umbilical can be catered for. DO NOT FORGET the BIG GEMOV across the secondary HT winding, IF you cant get them in GB I can send some over, they cost a few pence each. You should easily be able to source 240 volt rated GEMOV's as every switchmode power supply has 2 or 3 fitted, in computer service. Fit this across the primary and hope very close coupling and normal transformer action will clamp the secondary also!!. ( NOT as good as having them BOTH sides, primary AND secondary).
Great to hear that the cold has retreated and your brain defrosted after winter, when boys will turn to toys and play with amplifiers.
Oh !!! the mains switch on the new power supply should be rated at LEAST 25 amps!!, and have capacitive snubbers fitted.

Cheers

Joe

[Ed_Dinning]

Hi Andy, time to start winding your own!

Ed

[bikerhifinut]

Ah ed i wish I had the means and the skills!

But it does strike me that your good self and Mike B, and joe bog amongst others on this forum have more knowledge and PRACTICAL skill than many of the commercial transformer manufacturerd, many it would seem to me are incapable of the simplest requirement viz winding a SILENT transformer given a reasonable and realistic spec.

Anyway Joe, yes UF are the way to go I hadnt even thought of anything less.

I need to get hold of or make a decent sized and ventilatable case for the toroid first and then set about planning layout and earthing arrangements

[bikerhifinut]

Although Joe, I have been looking at the parts I have and have been thinking about what sort of monster I would have if I put the toroid and the existing opt transformers (cos so far they have been superb if potentially flaky but I will always be running them at about 50% of their rating) into a single chassis? it would save a bit of head scratching and the hard bit will be where to heatsink and mount the dc voltage regs.

A.

[bikerhifinut]

Originally Posted by joebog1

You will need UF rectifiers Biker!!! Rated at LEAST 6 amps, and DONT cut the leads off!!!! They are the heatsinks for the rectifier "chip" itself. That toroid will happily supply 1 amp, but as I told you it is low flux and rated at about 600 mA forever

Joe, why 6Amp? My Kt88 amp uses UF5408 rated at 3 amp 1000V 75nS and they have been 1000% reliable. my maths tells me even at the max 1 amp of thattransformer each diode in a bridge is only passing 500mA so surely 3 amp is enough?
I ask because the availablility of the UF5408 is good and at a reasonable cost (pence). I usually bypass with a high voltage 10nF cap also.

[bikerhifinut]

A little dim light just went on.
these resistors i have been blowing, I don't know what voltage they are rated at. Now the standard welwyns are rated at 500V which I reckon just scrapes under if I am applying 410DC, but what about AC is it 1.4x 410 = 574V? In that case I am thinking that for a short time, just as the rectifier starts conducting the voltage across the anode resistor is in excess of its rated value. So is it any wonder they go "pop" and with no outward indication as the failure may be very rapid? These are the cheap white coffin jobs I am on about.

Andy.

[Radio Wrangler]

OK,

AC mains has a sinewave shaped alternating voltage. And that's what comes out of the secondaries of a good transformer. Your reservoir capacitor holds a fairly constant voltage with a (relative to the average) small amount of ripple.

Silicon diodes start to turn on at about 0.6V and by 1.0v they are carrying several amps.

So the diode charges the reservoir up to about 0.7v less than the peak voltage coming out of the secondary and the diode turns off as the transformer voltage falls faster than the reservoir voltage. The reservoir voltage ramps down gently. The amplifier is taking a fairly continuous current and the capacitor gives a downwards ramp rate of I (amps) over C (Farads) volts per second. Meanwhile the transformer secondary is far below the capacitor voltage and going negative. At some time the other secondary end is coming positive. The diode on that side stays off right until the voltage gets a little bit above what the reservoir has got down to, and then it turns on and stays on just up to the peak as the other one did.

So the diodes spend most of their time off.

But the amplifier is draining the reservoir all the off-time, and the diodes have to get it charged back up again all the way in just that small fraction of time. So the diodes see a lot of time at no current, then have to conduct hard for a short time, then they get a rest again.

If you fit a bigger reservoir, the voltage it discharges in the off time is proportionately less. This leaves less time for recharging, so the recharging current must be greater.

Rectifiers with capacitor input supplies get hammered.

So your diodes have to work at several times the mean current your amplifier draws.

This is why the GZ34 has a spec for the max reservoir value you may use before the peak current density at the cathode strips it. The same issues affect solid and vacuu state devices.

David

[bikerhifinut]

have you ever thought about writing a textbook David?
What you say corresponds absolutely with my MJ "bibles", except I actually got what you wrote in one pass. The graphs and pictures now make sense.
So by using a small capacitor before the choke and a bigger reservoir after it might I be giving any rectifier a smoother ride?
I'm looking at options here and using a smaller first capacitor gets me into the opportunity to use a 5U4G which is these days a relatively cheap valve if you are prepared to hunt for a russian military job.
I'm serously looking at options for the big Joe bog toroid and for all the messing about with umbilicals etc I think I may as well rebuild the amp on a stereo Chassis with a few design tweaks.

If I wanted arc welding power I'd really need to look at silicon. But in all honesty I think its more cost effective to look for a commercially made power amp with the bugs ironed out for me......................................... I just love fiddling with valves for me they are alive, a bit like older motorbikes, imperfect but I can work on them with workshop tools and a bit of inspiration.

However I now think my troubles may well be down to inferior resistors, it's cost me a valuable GZ34 but at the age it is and its been very well used, I think it died with honour. And its convinced me that there is absolutely NO REASON to persist with thermionic rectification for its own sake. Once a PSU is smooth and stable how can the other valves know where their electrons came from?

A.

[Radio Wrangler]

I wrote part of one. If you ever want to see my views on oscillators and frequency synthesisers, I wrote the chapter on those things for the ARRL handbook (it's in every year since 1995). It's a bit more difficult writing for that sort of publication because it will be read by a far wider spectrum of people than a normal textbook will ever see.

If you have no input capacitor, you get a choke input filter and they have two modes. If you take very little current, the diodes do peak rectification and charge the capacitor to peak volts. If you take enough current that the ripple current is always positive, then the choke filter makes the capacitor voltage equal to the mean of the rectified waveform... a lot less than the peak.

So the choke input filter runs with a lower output DC voltage than a capacitor input filter would give, so you need more secondary volts on your transformer, but if the DC current load drops, the voltage on the capacitor goes rocketing up. So you wind up with the bizarre requirement that a choke input filter gives good regulation, but you must ensure a minimum current load and do the maths to relate this to minimum choke inductance.

Do a C-L-C filter with not enough C on the input and you have a supply which slides between capacitor input and choke input characteristics. The regulation is lousy and the minimum current for choke input voltage is raised.

Guitarists love valve rectifiers because their softness allows HT to droop when they start really expressing themselves, and they like the effect.

For hifi, you don't want to mutilate the music, you want to hear what was put on the record. If it includes a drooping guitar amp, you want to hear it, but droop in the hifi amp means any one loud instrument droops everything as well as itself. So there's good reason for silicon. It seems silly, but some cults want valves everywhere as a matter of religious purity or something like that.

Cheers
David

[joebog1]

So what David said!! Thats why I sent you the two 680 uF capacitors!! Put them in series ( with voltage sharing resistors of course)
I have forgotten the DC ohms of the toroid but 3 ohms or so on the secondary rings a bell. It wont work as a welder, but for the first five or six cycles it will supply HUGE current. Hence I ask for 6 amp diodes. Especially into those massive Rodestein caps, which have super low ESR.
BUT, if you fit an inrush limiter, you could easily get away with 3 amp diodes.
The reason the toroid and output transformers were second hand, BTW, is that the amp comes out at about 40 kilograms in single chassis mode, with choke regulated power supply. So I now change the order of components you need!!
FIRST component you require is a truss. A BIG TRUSS!!!!
If you dont believe me you can have a very neat second hand chassis that JUST happens to fit those components!!

Best of luck

Joe

[bikerhifinut]

Thanks chaps.
Joe I think I'll take the rest of our communication ref the toroid etc off forum so that I don't bore everyone with my inane questions.
I'm not sure how much mileage is left on this thread ref the Resistors and rectifier failure, but I think now ist been a coincidence maybe of a failing old GZ34 and my repeated use of resistors that really aren't "fit for purpose" and in any future PSU apps I will use decent branded resistors from reputable suppliers.

Many thanks to those who gave advice, all of it appreciated.

A.

[julie_m]

If the resistors were fine in normal operation but went O/C under fault conditions and nothing else was harmed, that sounds like a good result to me. Manufacturers usually charge extra for fusible resistors, too; even although they are really little more than ordinary metal film resistors running a bit close to the edge. It's the piece of paper that says they won't set anything else on fire that is the expensive bit, and that ceramic encapsulation looks non-inflammable enough for me.

[Biggles]

I have had the metal clad heatsink type power wirewound resistors (the gold coloured ones that R-S supply) go unexpectedly o/c before. I put it down to surge currents, which to be honest is not something I would expect to cause a wirewound to fail. I assume they are wirewound!
Alan.

[bikerhifinut]

whilst doing a bit of online research I came across this formula. Now I am sure when I think back that David (RW) and others mentioned something about it?

Its basically the power rating formula for series resistors in a PSU.

P = (1.1 × Idc)^2 × R (from the valve wizard site)

So assuming a maximum DC current per anode of 100mA (It's closer to 80 but i figured round it up for a bit of headroom) and a resistor of 100 ohms we get this:

P = (1.1 x 0.1) x 1.414 x 100 = 15.4 watts. I assumed ^2means sq root of 2?

If that's the case no wonder the anode resistors are failing as I'm overloading them. This now suggests to me that I should be using 20W rated wirewounds here but thats one big resistor and I am not sure I have ever seen anything that big used on equipment.

The Mullard 5-20 circuit specifies 6W resistors here and if the ^ symbol means divide by 2 then the resistor works out at 5.5Watts which then makes more sense.

have I misunderstood something?

A.

[bikerhifinut]

Originally Posted by Biggles

I have had the metal clad heatsink type power wirewound resistors (the gold coloured ones that R-S supply) go unexpectedly o/c before. I put it down to surge currents, which to be honest is not something I would expect to cause a wirewound to fail. I assume they are wirewound!
Alan.

I wonder if they get a "hot spot" perhaps near the termination to the leadouts and that combined with a bit of tension perhaps as the resistor cools on switch off is enough to break the resistance wire? That might explain why I don't see evidence of "burning up"?

A.