Paper in oil capacitors.

[mpegjohn]

Hi,
I am starting to get together the parts for a Mullard 5-20 amp.

I have opted for 50uF + 50uF reservoir caps as opposed to the 8uF + 8UF in the original which were paper in oil.

This got me thinking what are the benefits of using non electrolytic caps in this position, if any.

you seem to be able to buy large polypropylene caps 8uF 500v, would they be any better here?

To be honest I can't see what the difference would be, but I would be interested to know.

John.

[HamishBoxer]

Stick with 8uF or you will take the rectifier valve out.

Those you mention should be fine (8uF poly)

[G8HQP Dave]

Film caps will last longer. Electrolytics may only last 10-50 years.

[ms660]

Also note that the no load reservoir voltage would be approx. 580 volts according to the schematic.

Lawrence.

[Electronpusher0]

Are you building the full Mullard design including the valve rectifier and choke or are you using silicon rectification?
If using Silicon you can use much larger value capacitors and do away with the choke.
If using valve don't forget that the rectifier valve will have a maximum rated value of smoothing capacitor.
The 5-20 specifies a GZ34, which has a maximum value of capacitor of 60uF.

As to the question of Electrolytic vs Polypropylene the only advantage I can see of Polypropylene is long term stability (in this application)

Peter

[Electronpusher0]

The other advantage of silicon is that you can use voltage doubling which means you could use a 1:1 isolation transformer. However that would give a no load voltage in excess of 600V, the capacitors would be in series though so with voltage sharing resistors (a good idea anyway to discharge the caps when switched off) the voltage across each capacitor will be half.

Peter

[turretslug]

As always, economics rears its head as well as purely technical consideration. When electrolytics first came out, high voltage types with appreciable capacitance value were large, expensive and not altogether trustworthy- hence high-quality equipment would often use paper block capacitors despite their bulk and expense and high-value series chokes (large, heavy and filled with lots of copper wire and quality strip steel) were necessary whatever capacitor type was used. Progress meant electrolytics became smaller, cheaper and more reliable and as electrically tougher indirectly heated rectifiers replaced directly heated types, designers got used to specifying larger value electrolytics and smaller, cheaper chokes. 50 + 50uF makes good sense in the context, rectifier permitting.

It's possible that modern polypropylene types would have lower ESR than equivalent capacitance/voltage electrolytics (and modern electrolytics can be much better here than the ones of half-a-century ago) but an application like this isn't as critical in this respect as a high current, low voltage, high frequency SMPSU.

The modern ubiquity of SMPSUs means that high voltage electrolytics are cheap, widely available, compact and have excellent all round performance but 60 or so years ago when the folk at Mullard were writing the applications, the case for them wasn't as clear-cut.

[bikerhifinut]

Originally Posted by HamishBoxer

Stick with 8uF or you will take the rectifier valve out.

Those you mention should be fine (8uF poly)

No you won't.

Not if you use the specified GZ34. 50 + 50 is fine, and there are plenty of options out there that can handle the Ripple current off the rectifier. And a lot cheaper than polyprops not to mention the size of large capacity polys. In fact I really cant think of any practical reason not to use suitable electrolytics. Just be careful about the working voltage as it can initially rise quite high. You'll get away with 500V jobs but I would be a little bit twitchy about this and if you can find 600V wkg they'll have a better headroom. RS sell a few choices at 600V and they aren't expensive compared to some of the "Specialist" suppliers.
Later down the thread another recommends you can do without the choke.
Perhaps so and use something like a 100 Ohm power resistor as part of a CRC filter, you'll still need the DC smoothed. Leak used this approach successfully on the first TL25 amps and the early stereo50. They also used EL34 in a very similar output stage to the 5-20. But I wouldnt, you can get a nice Hammond Choke from the likes of Bluebell Audio (other suppliers and manufacturers are available) for not a lot of money, theres a nice 5H x 150mA one that does nicely here. I Know because I have used them and they work a treat.
By all means if you wish use silicon rectifiers, you'll need to specify a different secondary voltage, 325 to 350V gets you right in the ball park. I'd stick with 325V as you will be safe using 500V electrolytics and trust me on this, you dont get any appreciable difference in usable power output with the slightly lower Ht, this is one of the aspects of fairly basic valve amps like these.
I wouldnt go anywhere near a voltage doubling circuit, theres no need and its not really in my view a good practice at all.
I'd be happy to share my experiences with my builds of a similar EL34 circuit, both Valve rectified and the current silicon version. As I am sure lots of other members who have built one from scratch will share their experiences.

A.

[bikerhifinut]

Originally Posted by ms660

Also note that the no load reservoir voltage would be approx. 580 volts according to the schematic.

Lawrence.

Sorry Lawrence, in practice it doesn't because of the Voltage drop across the GZ34 and also the use of the current limiting resistors on each anode of the GZ34, R26 and R27 on the circuit diagram.
If my memory serves correctly, the voltage got within a whisker of 500V with 100 ohm anode resistors with the generously rated transformers I used, and it dropped down to around 450V on load. This was with transformers wound with a 230V primary used on a 250V mains voltage.
I'd be very careful about the choice of mains transformer and go for one with a tapped primary that will allow for areas with high or low mains Voltage. More of an issue for your heaters than HT admittedly.

A.

[bikerhifinut]

A quick look at the Mullard circuits book, 2nd edition, clearly specifies the 8uF main reservoir and smoothing capacitors as Electrolytics, 500V DC working.
So Mullards must have either always specified electrolytics or post 1959 changed that part of the spec.
I'd forgotten just how detailed the description and advice was in the book, You'll find just about all you need to know in regard to the PSU there.

Andy.

[Electronpusher0]

[QUOTE=bikerhifinut;1126485]

Originally Posted by HamishBoxer

I wouldnt go anywhere near a voltage doubling circuit, theres no need and its not really in my view a good practice at all.

A.

I respectfully beg to differ.
Armstrong used voltage doubling for their valve amps and I have used it a couple of times including on my homebrew valve amp based on the Mullard 5-10 very successfully

[mpegjohn]

Thank you everyone for all for your help and advice.
I will carry on down the path I took, 50 + 50uF with the GZ34 rectifier.
I will also have a 10H 200mA choke.
John

[bikerhifinut]

You'll not go wrong with that John, best of luck.
I take it you have a copy of the Mullard Valve circuits book? It's freely available on the internet as a download from various sources. And invaluable for referring to when building it up.
If you already know this I apologise.

A.

[HamishBoxer]

I admit I was not aware of the spec of the rectifier valve and thought if 8uF was original it was a heck of a leap to 50uF.

Very odd that it was made with such a low value cap.

[bikerhifinut]

Originally Posted by Electronpusher0

I respectfully beg to differ.
Armstrong used voltage doubling for their valve amps and I have used it a couple of times including on my homebrew valve amp based on the Mullard 5-10 very successfully

I see where you are coming from, and it certainly opens out the choice of transformers as you'll easily find an inexpensive toroid with 2 x 60V secondaries that will give you 120v and then use another one for the heaters.
On the other hand a 1:1 isolating transformer would give you 330V at a nominal 240V mains with a silicon Bridge, depending on its regulation etc etc.
My personal concerns with Voltage doublers are the hammering they give the PSU capacitors, I will need to refresh my memory but I think a voltage doubler needs csapacitors that can handle very large ripple currents? That I guess can be mitigated by being able to use lower voltage rated caps or to get a better margin? And again, I may have the wrong end of the stick, but does a voltage doubler put a bit more strain/load on the mains transformer? i.e you have to halve the current rating (Ohms law?).
On balance I'd go for a conventional Full wave rectifier, and these days that means a silicon bridge based on cost (pence versus tens of pounds for a GZ34 type), you don't need a 2 amp 5V winding either. And you aren't constrained on reservoir capacitance off the rectifier.
But you made me have a rethink on voltage multipliers, thanks for that. There may be situations where its not economically or otherwise feasible to get a high voltage secondary.
A.

[bikerhifinut]

Originally Posted by HamishBoxer

I admit I was not aware of the spec of the rectifier valve and thought if 8uF was original it was a heck of a leap to 50uF.

Very odd that it was made with such a low value cap.

I think it was designed in the early 50's Hamish, when the EL34, 84 etc were launched, replacing the earlier valves as a marketing tool along with the 5-10, 3-3 etc amps. At that time I guess a high voltage capacitor of 50uF would have been very large and very expensive so they will have gone with what was available.
There's another consideration too, the leak TL50 which ran at a high HT of over 500V and also a higher current through the Valve as an example, used 8uF reservoirs. The data sheet for the GZ34 explains why as the maximum capacity off the cathode goes down with increased voltage and/or current.
So you weren't entirely off the target.

[ms660]

Originally Posted by bikerhifinut

Sorry Lawrence, in practice it doesn't because of the Voltage drop across the GZ34 and also the use of the current limiting resistors on each anode of the GZ34, R26 and R27 on the circuit diagram.

A.

Sorry but it would (small leakage current aside)….voltage drop only applies when current is flowing...no load = no current....no current = no voltage drop across the rectifier etc.....therefore approx 580 VDC across the reservoir using the specified transformer secondary voltage given in the schematic I'm looking at.

Vrms*1.414

Lawrence.

[GrimJosef]

The increased transformer stress as a consequence of using a large reservoir cap and a solid-state rectifier has already been mentioned. Another feature which can be important is that compressing the recharging current down into ever shorter and more intense pulses can increase the amount of high-frequency noise which is generated. Our ears are relatively insensitive to 100Hz noise compared, say, with 500Hz or 1kHz. So the shift in the noise frequency spectrum might be even more objectionable than simple maths would suggest.

Cheers,

GJ

[kalee20]

Originally Posted by HamishBoxer

I admit I was not aware of the spec of the rectifier valve and thought if 8uF was original it was a heck of a leap to 50uF.

Originally Posted by GrimJosef

The increased transformer stress as a consequence of using a large reservoir cap and a solid-state rectifier has already been mentioned. Another feature which can be important is that compressing the recharging current down into ever shorter and more intense pulses can increase the amount of high-frequency noise which is generated.

Both are true - though, given sufficient series resistance (eg transformer winding resistance), peak current and duration of charging pulses hardly changes beyond a certain point.

The change from 8μF to 50μF is huge, yes, and with the typical DC load current of this amplifier, there would be a significant increase in peak current drawn (accompanied by narrowing of pulses). But it could be that you might increase further to 5,000,000μF and see hardly any change in charging pulses. The ripple would obviously be very much less, and the switch-on surge would last very much longer, that's all.

O H Schade did the 'classic' work on this, and his curves give pretty good estimates of currents and voltages. And you can see if, for a given set of conditions, the situation is heavily or slightly dependent on reservoir capacitance.

[julie_m]

Originally Posted by Electronpusher0

I respectfully beg to differ.
Armstrong used voltage doubling for their valve amps and I have used it a couple of times including on my homebrew valve amp based on the Mullard 5-10 very successfully

Nearly every PC in those parts of the world using 110V is using a voltage doubler; as are nearly all microwave ovens with a traditional copper-and-steel transformer in the magnetron supply.

The disadvantage of a voltage doubler is that the ripple voltage is twice what it would be from a half-wave rectifier (since a doubler is literally two half-wave rectifiers in series). This is less important in a switched-mode power supply, and a magnetron supply only has any smoothing on one half anyway.