Dual mono though surely? There’ll be two of those.
And then the filament supplies.
3 Likes
I suspect they will be transported by rail and then Pickfords.
I love this bit of film.
4 Likes
I did mention rack dimensions to Pete, but he informed me that the power supply would be floor mounted. 
5 Likes
Best thing I’ve seen today 
1 Like
I recently picked up a very dusty but tidy Leak Stereo 20 and Varislope Stereo Pre as part of an auction lot. I’ve long had a soft spot for the Stereo 20, a working sample can sound very enjoyable with one or two caveats about what it matches well with.
With a bit of dusting, this one is starting to look ok.
The idea with this was to get it going, listen to it for a while & then move it on. I do like EL84 based amps & it’d be a handy benchmark to have around.
The problem with these old amps is that despite being beautifully built, the passive components in them do drift away from the intended spec, both the carbon resistors and the various types of capacitors they carry. I might set out with the intention of changing as little as possible but you soon realise everything needs to go! I tested the valves it came with, all Mullard, One pair of output valves, all the ECC83s and the GZ34 rectifier seemed fine. The other pair of EL84s tested ok-ish but did appear to have had a ‘moment’ on closer visual inspection.
I put in some new valves, Soviet outputs and some bog standard Chinese 83’s and gradually brought up the HT very slowly from an external regulated power supply. But it was evident that things were amiss even at 50V with the output valves’ bias voltages varying hugely partly because the bias resistors were so far from spec & partly no doubt because the coupling capacitors were leaking DC to their grids. So, where do you start & where do you stop?
I replaced the bias resistors with new metal oxide items of the correct value but the voltages around the output valves were all still adrift.
Here are the innards at this point.
and looking closely, while referring to the schematic, here’s what they actually are.
So I decided to replace with some fairly prosaic items; carbon film resistors, Russian paper in oil capacitors and then for a few of the tiny capacitor values polystyrene.
This isn’t an Ongaku so doesn’t need or warrant maximum foo but also but it doesn’t feel right to fill it with polypropylene caps or metal film resistors which wouldn’t be in keeping either.
Next step to strip out the old stuff.
Will report again once more bits are re-fitted. Caps have to come from Ukraine!
One of the features of the Leak is a 100 Ohm resistor that acts as a fuse in case one of the power valves does misbehave and draw excessive current.
It’s the fat green item bottom right on my picture of the complete board. The idea was that it would heat up and drop out in such an event thereby cutting the high voltage feed to the amplifier.
It’s evidently been producing a fair bit of heat at some point although hasn’t detached itself.
At the moment I’ve left it in place but I need to decide what to do there as I’d rather have a proper fuse for that job.
20 Likes
A project ! Great !
You’re right about almost all the resistors though. They will be absolutely all over the place. I see you’ve kept the 100R W/W. If it hasn’t burnt to death it’ll be the only one which should still be fine (it will have started to burn the paxolin to death though, so there’s something to be said for trying to shift it sideways so at least some of the hot air misses the tagboard).
When it comes to replacements, modern metal film resistors costing pennies measure close to the noise limits set by quantum mechanics. The only problem with the physically smaller ones is that their leads are just too short to stretch between the tags for the 2k2s. Which is bloody aggravating.
You could try carbon films. That technology was around in the time of the amp. But it’s getting increasingly hard to find the 91ks needed for the phase-splitter anode loads. Not nearly as hard as finding NOS Erie high-stabs in that value though …
As you’ve found, most of the caps will have gone west. The HT smoothers can probably be reformed which is nice if you want to preserve the paintwork matching. They’re only 32uF though, and you might well find they struggle to make even that. It’s a change, I know, but adding some more microfarads to the one after the 100R can lower any 100Hz hum significantly. This can help if you’re having to work with unbalanced old EL84s. The wax-covered 200pF chaps across the feedback resistors would probably have been OK. They’re silver-mica under all that goo and they seem to last for ever.
The loom looks in great shape. When amps have been worked on it’s often the case that a clumsy solderer has melted bits of the insulation. But you haven’t .
The layout of the wiring around V1 does matter a bit. If it’s changed too much then it’s easy for 50Hz hum to creep in there (the amp is so ridiculously sensitive, and the grid circuitry is outside the corrective feedback loop of course).
The last thing that comes to mind is that the little plugs which select the speaker impedance on top of the output transformers can play merry hell with the distortion. Not only do they and the sockets they plug into tend to corrode, but the actual metal bridging piece that connects the two plug pins together is only a press fit. And that junction corrodes too. Leak spec’d 0.1% distortion at 10W out and that’s a bit of a struggle to get to. But the joints in the selector plug can easily turn that into 0.3-0.4% of third harmonic (again all this is outside the feedback loop). I’ve taken to soldering a wire between the pins now. You have to be v quick as the plastic of the plug body is v low melting-point.
Onwards and upwards !
EDIT - just noted your comments about the 100R. Yup. The drop-out protection never really worked, apparently, and Leak gave up on it eventually. As you say, fuses make better fuses than resistors do.
VB
2 Likes
Always marks the board. TBH there is little difference in price between redone and unmolested in nice condition so we rarely ever get into refurbishing. On the odd occasions we have the buyer invariably has fired over questions or wanted to change certain caps etc anyway. I do see the merit in it from the listening perspective however - They can sound decent and there is something very attractive about the design IMHO.
I think the only commercial amp (in the world ?) where all the small-signal valves are ECC83s. If you wanted the definitive test-bed for an ECC83 shoot-out then this is it. Even its monoblock sibling, the TL/12 Plus, uses an EF86.
Thanks for the tips Graeme, particularly about the input wiring and the impedance setting plugs. I was thinking of changing the values of the potential divider at the input ie outside the feedback loop, and cut the input sensitivity to some more usable amount such as 600mV to full output rather than the 150mV it has in stock form. Not sure about changing input sockets and output terminals either. This unit still has the buff coloured factory tag attached on some string indicating manufacture in 1963. It’d be easy to get precious about the details but I gather they made some 250000 of them so they aren’t exactly rare.
I notice there’s a company making a full repro unit now but also that they’re asking £4K for it which seems a bit strong. Especially as they appear to be using a PCB.
Re reservoir caps, JJ do a 32+32 and a 50+50uF so I might get one of each, take off the pvc sleeves & prime then spray to match the gold if the fitted units aren’t satisfactory.
Incidentally it seems Kiwame do some 2W carbon resistors at 91 KOhm (HiFi Collective do them) but the legs may be on the short side so may have to decide how best to neatly install them.
If you’ve run it then the original smoothers are almost certainly OK. They fail catastrophically or not at all. I think people just tack 47uF or so across C12. The 100R limits any resulting strain on the GZ34 a bit.
The Ohmite ‘Little Rebel’ carbon film range include a 91k 2W which seems to be 72mm total length https://www.mouser.co.uk/datasheet/2/303/res_little_rebel-1265497.pdf, which is more than long enough. Mouser have them in stock but there’s a minimum order value needed to get them from the US without paying postage. I may also have some Neohm (or were they Multicomp ?) ones somewhere which I bought before Farnell cut those back to E12. The problem is that if you want the full aesthetic impact then you really need all the Rs to match, and ranges do change. The Kiwames would give a very distinctive look !
Up in the loft I have a pretty rare near-contemporary Stereo 20 copy made by someone like AEI, I think. It used a pcb too.
1 Like
If I were making a repro, I’d have a tag board made and wire the other side as they did originally. It’s not ‘that’ complex.
True. I’d want a second look at the V1 wiring layout though. There’s quite a big loop under the board in the g-k wiring for one half of the valve (can’t remember if it’s R or L) and I suspect there’s 50Hz hum pickup there.
Fusible resistors are A Thing, for example these should be in-spec for the purpose.
1 Like
The numbers are a bit iffy, I think.
The nominal average current through that resistor, in the absence of music, is 175mA or so. The power dissipated in it normally is therefore just over 3W. So a 2W resistor might degrade with time. Also, if we wanted it to ‘blow’ in 60 seconds (which is quite a long time for a failed valve to be drawing a large overcurrent through a delicate output transformer primary winding) then we’d have to raise the power in the resistor by a factor of 16 over its rated power i.e. we’d need it to be dissipating 32W, which means it would have to be passing 565mA. The other three valves wouldn’t be passing any more current (in fact they’d be passing less because the HT voltage would drop) so the dying valve would have to handle all the excess, which means it would have to pass around 430mA, or almost exactly 10 times its normal current. That would raise its cathode voltage to 110V which would fry the cathode resistor. The cathode capacitor is probably not rated for that either, and they go bang quite quickly if they’re over-volted. But the big worry is still the output transformer. The type 8778’s half-primary resistance is either 72ohms or 63ohms, depending on which valve goes west. Driving 430mA through that would dissipate either 13.3W or 11.7W in it. The wire is thin and the wax is not a very good thermal conductor. I’d be nervous.
If we were going to try fuse protection then I’d probably put one 50mA fuse in each EL84 cathode circuit. Littelfuse’s type 217 one claims to blow within 5s at less than 140mA and that might be a good place to start.
VB
2 Likes
To be honest, thinking about it, fuses aren’t a very satisfactory solution. The best plan I could come up with would put an extra 15ohms (the resistance of a 50mA fuse) in the cathode circuit and for this to protect the capacitor it’d have to be outside the RC combo in that circuit. So it’d be applying a bit of negative feedback there. Which would change the amp’s properties, albeit not by very much I suspect.
Almost certainly it’d be better to swallow the solid-state pill and build a voltage monitor which would keep an eye on the average cathode voltage at each valve. If any of them went above a safe value (TBD, but maybe 16-18V ?) it’d just open a relay across the ‘switch’ contacts in the mains transformer primary. A ‘Start’ push-button might be needed to get the power on in the first place.
VB
1 Like
I’ve always found this passage from the Spicer book about Leak quite interesting.
This amp appears to have made it from 1963 without self immolating and I’d expect the Russian 6P14’s I like, to continue being robust enough for the application. The output stage actually runs at very similar static conditions to those I have in the little A10 amplifier & I don’t see particular issues with the longevity of that or indeed the Chinese 84’s it generally goes out with.
1 Like
1 Like
I agree about the Russian valves. They’re built more strongly and some are rated more highly than Mullard’s first try. And there are few amps out there which haven’t had the circuit revisions done. I do see ones which have destroyed their cathode components. But the ones that come to me are, by definition, bust. I only see the good ones if I’m looking at ones for sale.
Granted. I was suggesting the device as a means to remain as close to the circuit’s original appearance and function as possible. I would assume more suitably-rated examples exist…
There are foolproof methods, of course - indeed, compact current-limiting solid-state circuit-breakers are on the way apparently, but until then it gets pretty intrusive. Perhaps a simple solution lies in combining a suitably-rated transil to catch overvoltage situations with a suitably-rated PPTC thermistor to catch overcurrent/overheat situations? (Plus a fuse of course!).