Maximum Filter Capacitance with Valve Rectifiers

[percival007]

Hello Folks,

I am looking at an Amplifier which uses two 5U4G's as Rectifiers for the HT.
The Anodes of each Valve are connected together and each 5U4G is therefore used as a 'Diode' connected to each of the two Transformer O/P wires.
(Not the best description but I hope that is clear).

My question is that, the datasheet for the 5U4G states that the maximum Filter Capacitance should be 40uF when used with a Capacitor Input Filter.
If the 5U4G is used in the way I describe above can this be increased?

I ask because the owner complains that it has 'blown' two Rectifiers.
Someone has added some 'extra' Capacitance but as far as I can tell, just 2.53uF (one 2.2uF and one 0.33uF Film types (an 'Audiophile 'Upgrade' ?)).

The failed Rectifiers have actually failed with O/C Heaters. Their Heater Supply is a good solid 5 Volts AC.

I am doing more tests but thought I would ask my question now.

Thanks in advance,

James.

[GrimJosef]

As is so often the case, it's quite complex. The 'max capacitance' specification is designed to stop rectifiers being killed by excessive peak current. They are more vulnerable to this when they are also being run at maximum average current and (to a lesser extent) maximum voltage. The spec assumes that they are. If the actual average current is less than the maximum that the rectifer is rated for then you can get away with a higher-value capacitor. The peak current will also be limited by the resistance in the rectifier's anode circuit and (probably to a much lesser extent) by the capacitor's ESR. If those resistances are higher than the minimum specified then again you can get away with a higher-value capacitor.

I'd be extremely surprised if 2.5uF of additional capacitance made any difference at all. The only proviso is that the reason for paralleling additional lower-value capacitors is to use a different dielectric and geometry to lower the overall ESR. I suppose in some circumstances that might raise the peak current somewhat.

To answer your question about two 5U4Gs vs one then yes, the pair will have twice the peak current capability that one would have had and so could charge a higher-value capacitor.

Cheers,

GJ

[Radio Wrangler]

If current-sharing was perfect, then two rectifiers could handle twice the load capacitance of one, but matching won't be perfect.

Electron emission with high density of current per unit area has an etching effect on the emitting surface. Under the right circumstances you can actually see an extreme example of this happening in real time.

If you TIG-weld aluminium, you set your machine to give AC current. For most other materials, you use DC with the electrode negative and the work positive. This puts most heat into the work and doesn't overheat your tungsten. The reason for AC to weld aluminium is that the period when the aluminium is negative and emitting electrons (like a cathode in a valve) is that it etches the oxidised surface off of the aluminium, leaving cleaned aluminium surface to weld to. This is quite violent and dramatic to see. You see tiny flakes cracking off and vanishing at speed. If you look at a finished weld, you will see a roughened zone outside the weld bead, following alongside it. This was electron-emission etched. In milliseconds.

The capacitance limit for rectifier valves is to limit the peak currents they run to and thence limit the current density of electron emission from the cathode surface. Run them too hard and the life expectancy of the valve before its emission falls too far is reduced.

A valve failing through this mechanism, will suffer progressively worsening current limitation as its emission decreases, and your HT voltage will drop and soften with load. It shouldn't cause premature heater failure, so I think there is a different problem there. As said above, the capacitance increases are fairly mild, so a good quality valve should just see a reduction in the time before it is 'worn out'

David

[ms660]

The largest current is the hot switch on current, if it's not limited to the valve manufactures specs then arcing can happen with bad consequences for the valve and anything else in the circuit.

Lawrence.

[snowman_al]

Unfortunately new 5U4s are not as robust as the originals. Flash over is likely to damage the filament as they are directly heated valves, so there's no cathode to 'shield' it.

Is this a WAD or World Designs amplifier by any chance?
Has this one got the additional four 47Ω anode load balancing resistors fitted?

Also there is a mod to use efficiency diodes in place of the 5U4s, have a look here: http://www.world-designs.co.uk/forum...ead.php?t=7971

Good point re the hot switching, worth mentioning to the customer?

[Sideband]

I had a problem with a Marshall Guitar amp where the user complained of rectifier failure on a regular basis. That used a 5U4. I can't remember the reservoir value but it might have been 32uF. The problem was that the amplifier had a 'standby' switch which was basically an HT switch in the cathode of the rectifier. As seems to be the custom with such amplifiers, the user manual recommends that the 'standby' switch is on when the mains is switched on which effectively means that only the heaters are on to start with. The standby switch is then operated which means the poor old 5U4 which is at full operating temperature then has to charge a fully discharged 32uF capacitor. The 5U4 was never designed for 'surge' operation and I actually watched the cathode connection 'sag' as it passed the charge current....no wonder it was eating rectifiers! When I looked at the circuit, there were no surge limiters fitted either. The user insisted that he should continue to use the 'standby' feature so I fitted a couple of surge limiter resistors in each anode of the rectifier....I think I used 47 ohms which made up the recommended limiting resistance.

Marshall were obviously not worried about the short life of the rectifiers but certainly after fitting the limiters, the rectifier was happier. Dreadful design though.

[joebog1]

snowman_al has it!! If the rectifiers are modern. i.e. Russian or Chinese, they fail with great regularity. BUT, try to buy an origional made in USA, Australia or Great Britain variety made in the 50's, expect to pay stupid prices for them. I like valve rectifiers too, but don't use them as a pensioner I just cannot afford those prices.
I have bought up a whole lot of 6AU4 damper diodes to try but I have to wind a new power transformer to supply the silly current the heaters require. A pair of 6AU4 will happily supply a surge, and fairly large currents ( up to 500 mA per pair) at amplifier voltages we use.

Just my experience.

Joe

[trobbins]

The reason for an open 5U4G heater is still 'open' (pardon the pun). I reckon there is no likely failure mode that would cause that except a gross mains voltage swell or old age. James hasn't provided any information on age of amp or parts, or if both failures occurred at the same time, so I'd suggest old age seems a more likely cause. At least it was a failsafe type of fault.

If at all possible, I'd recommend inserting two series connected 1N4007 in series with each 5U4 anode. That is a form of belts and braces safeguard as not all valve diodes fail in a failsafe manner, and it can prolong the life of a valve diode because it stops internal arcing (whatever the cause - such as for 'modern' valves, or poorly balanced valves, or aging gassy valves).

Although the use of balancing resistors can have some benefit, that is more during early life, as aged 5U4 anode 'resistance' can vary by quite a lot - I measured anode resistances ranging from 200 to 400 ohm at 24Vdc drop in a batch of 30 valves (ie.60 diodes).

[percival007]

Thanks for all your replies folks.

It's a relatively recent Canary Audio amplifier and it doesn't have anything in series with the Anode connections on the 5U4Gs.
I am not sure when all the failures happened, separately or all at once. The most recently fitted ones have been JJ brand.
It's a beast of a thing, two massive mono-block with 4 300B's as O/P Valves, I think it is push-pull with two parallelled 300B's in each side of the O/P TX. I don't have a circuit yet.

The chassis that hasn't destroyed any 5U4G's Hums a little and three of the four 300B's are glowing with a slight blue-ish hue. I've read that this may not be anything to worry about but it concerns me a little as it's strange that it's only 3 of the 4(?) and the Amp hums.
More investigation work necessary.
I shall think about the addition of either some resistors or diodes with the 5U4G's. Thanks for the tips.

James.

[GrimJosef]

Is the hum electronic (through the speaker) or electromechanical (from the chassis itself) ? These are US amps and unless they've had a UK-specific mains transformer fitted they may be running with one designed for 60Hz and 234V (two 117V primary windings in series). Our 50Hz and 240V (could be more) may be enough to tip a transformer on the edge over into humming.

Cheers,

GJ

[percival007]

@ Grimjosef

Hello,

no, it's through the Speakers. Mechanically they are quiet.

James

[joebog1]

This may be due to leaky filter caps!! The amp that hasn't blown any rectifiers hums and that's a dead set give away. It may also be due to the caps being too large for the rectifiers to cope with especially if it has soft start or delayed start fitted.

Just thinking.

Joe

[Jez1234]

I was given 5 NOS boxed Edicron branded GZ34's and the first 4 lasted between 30 seconds and 20 mins fitted to a Leak Stereo 20! Strangely the 5th has been reliable for a couple of years now...

[G6Tanuki]

I would say you need to add current-sharing resistors between the strapped anodes of each valve - if you do the math then these can also be the specified 'minimum series resistance' for each valve.

[If the manufacturer specifies, say, 100 Ohms minimum resistance, use two 200-Ohm resistors per valve].

As noted upthread, the problem of applying a load when the filaments are up-to-temperature can be a big issue: this can happen if there's a power-outage of half a second or so - the 'load' valves are up to full emission and when the power goes-off they drain the filter-capacitors - then when the power comes back the rectifier filaments are still hot, the 'load' valves have their filaments hot-enough to provide a significant current-drain, and you get a few cycles of AC where the rectifiers have to drive into a rather-low load-impedance.

Ouch!

[Double-Plus-Ouch if mercury-vapour rectifiers are involved]

Personally, I was glad to see the back of vacuum-tube rectifiers; I'm sure that if the designers of 'vintage' amplifiers [whether for domestic audio, PA or guitar] had had access to good silicon diodes in the 50s and 60s they would have viewed them with delight!

[TV designers forced to use the crazy U801 rectifier - four anodes, one cathode - sometimes used both to rectify mains for HT and also to provide energy-recovery for the LOPT - would no doubt agree with me! Mazda's flaky cathode-coating really did make those things a firework-display-in-a-bottle!]

[percival007]

Originally Posted by trobbins

If at all possible, I'd recommend inserting two series connected 1N4007 in series with each 5U4 anode. That is a form of belts and braces safeguard as not all valve diodes fail in a failsafe manner, and it can prolong the life of a valve diode because it stops internal arcing (whatever the cause - such as for 'modern' valves, or poorly balanced valves, or aging gassy valves).

Thanks again for all the replies.

If fitting two Diodes per Anode as you recommend, wouldn't these then be performing the rectification and the Valves just 'passing'' the already rectified DC?
Or am I missing something?

James.

[ms660]

Yes, that's correct.

Lawrence.

[trobbins]

Originally Posted by percival007

If fitting two Diodes per Anode as you recommend, wouldn't these then be performing the rectification and the Valves just 'passing'' the already rectified DC?

Like many a hybrid configuration of parts, each part may contribute a certain dominant characteristic.

If the characteristic of interest was the series resistance of the 'diode' then the valve diode's characteristic dominates and provides the nicest form of soft turn-on and off of any diode, which then has a dominant effect on diode current waveform, and typically relegating the power transformer leakage inductance related commutation noise to a benign level.

If the characteristic of interest was leakage current when reverse biased, then it depends on the aging of the valve diode and whether the valve or the ss diode has the lower/higher level of leakage. The key reason for the mod is that valve diodes do age, and can be hammered under certain conditions, and the result can be arcing when the diode anode is reverse biased - but the ss diode saves the day (and saves the valve's cathode) and allows the valve diode to keep performing its job for what can be a lot longer (and in a safer manner for other parts).