Golborne Vintage Radio

Back in December 1999 I replaced the 2n3055/MJ2955 o/p pair of transistors and the two 4a(T) fuses in the Mains Transformer secondary in the above amp., the o/p transistors having failed short circuit.  The other day I was asked if I could repair a NAD Amplifier with one channel faulty.  It tuned out to be the same amplifier I'd repaired 15 years ago, and the same fault! - i.e. LH channel toatlly dead because the o/p transistors had failed s/c and blown the two 4A(T) fuses.  I know nothing of the amps. history over the intervening years, but it seems to be quite a coincidence.  The two transistors I'd fitted back then weren't a matched pair, just two devices of the right types but different makes from my spares box, which would have been sourced from reputable suppliers such as CPC, Willow Vale, etc.  As I currently only have MJ2955s 'in stock', would it be wise for me to order, if available, a matched pair this time?  I've also read somewhere that some 2N3055s are 'not all they claim to be'.  Naturally I'll check the other semiconductors in this DC coupled amp before fitting new o/p transistors,  though all the others are still the original devices.  N.B. The RH channel has never given any trouble, and all components in it are original and untouched since the day the amp was made! When, after removing the S/C Transistors in the L/H Channel, and replacing the blown fuses, I fired the amp. up, the RH channel worked perfectly, and the supply rails measures +39/0/-39v to within a few mV of each other.

There's nothing stopping you from putting beefier transistors in - I recommend using MJ15003 and MJ15004. Farnell stock them for a not-unreasonable price.

As to what caused the failure, quite a lot of NAD amps lack active current limit protection, relying instead on the fuses, so it's possible that an intermittent short in the 'speaker wiring might be a factor (surprisingly common in my experience).

Or perhaps it was thermal runaway? Some of the NAD amps I've serviced had frighteningly small heat sinks, and didn't seem to pay attention to close-coupling the VBE-multiplier transistor to the heat sink. Definitely make sure the quiescent current is correctly set, and is reasonably stable over a few hours...

Have a really close look for dry joints. Again, a common problem area with these in my experience. I use a x7 magnifier to be absolutely sure. I often re-do joints that are hot-running, even if they look OK (remove old, put on new). Basic, I know, but easily overlooked.

What sort of pre-set resistor are they using for the Iq adjust, and if it were to develop a poor contact, what would be the effect? On a sensible amp, the preset should be placed between base and emitter of the VBE multiplier, so the worst-case scenario is crossover distortion. But others use arrangements whereby a poor contact can lead to the VBE multiplier being biased off, which clearly will cause the output pair to conduct heavily with the inevitable results.

Does the owner like the music loud from time to time? A good trick is to retro-fit a bimetallic cut-out to the heat sink - perhaps one set to 70C - wired in series with the mains (well insulated!). When the heat sink gets hot, the power is switched off. After a few minutes, it'll come back on again. Hopefully no damage results, and the owner gets the message that he needs to turn it down, or remove all the stuff that's blocking the vent holes! It's a trick that saved my bacon on several occasions - something like this might do the job: http://uk.farnell.com/multicomp/t23a070a...dp/1006843

Oh, and it's probably at the age where some of the capacitors must be past their best.

Apologies in advance if I was stating the obvious in the above. Good luck with the repair

Mark

I suspect that not all the problems were fixed first time round, hence some time later, the same issue coming round to haunt you

Always hated working on NADs. Enclosed is the service manual for the 3130 - the replacement for the 313 - hope its useful...

(13-02-2015, 10:34 AM)Mark Hennessy Wrote: [ -> ]There's nothing stopping you from putting beefier transistors in - I recommend using MJ15003 and MJ15004. Farnell stock them for a not-unreasonable price.

As to what caused the failure, quite a lot of NAD amps lack active current limit protection, relying instead on the fuses, so it's possible that an intermittent short in the 'speaker wiring might be a factor (surprisingly common in my experience).

Or perhaps it was thermal runaway? Some of the NAD amps I've serviced had frighteningly small heat sinks, and didn't seem to pay attention to close-coupling the VBE-multiplier transistor to the heat sink. Definitely make sure the quiescent current is correctly set, and is reasonably stable over a few hours...

Have a really close look for dry joints. Again, a common problem area with these in my experience. I use a x7 magnifier to be absolutely sure. I often re-do joints that are hot-running, even if they look OK (remove old, put on new). Basic, I know, but easily overlooked.

What sort of pre-set resistor are they using for the Iq adjust, and if it were to develop a poor contact, what would be the effect? On a sensible amp, the preset should be placed between base and emitter of the VBE multiplier, so the worst-case scenario is crossover distortion. But others use arrangements whereby a poor contact can lead to the VBE multiplier being biased off, which clearly will cause the output pair to conduct heavily with the inevitable results.

Does the owner like the music loud from time to time? A good trick is to retro-fit a bimetallic cut-out to the heat sink - perhaps one set to 70C - wired in series with the mains (well insulated!). When the heat sink gets hot, the power is switched off. After a few minutes, it'll come back on again. Hopefully no damage results, and the owner gets the message that he needs to turn it down, or remove all the stuff that's blocking the vent holes! It's a trick that saved my bacon on several occasions - something like this might do the job: http://uk.farnell.com/multicomp/t23a070a...dp/1006843

Oh, and it's probably at the age where some of the capacitors must be past their best.

Apologies in advance if I was stating the obvious in the above. Good luck with the repair

Mark

Points noted, Mark.  Dry joints and/or thermal runaway, plus external causes like faulty speaker wiring may all be factors in the failure.  Uprated transistors and a bimetallic cutout may also be worth considering.  I'll give it a thorough check over before fitting new transistors in any case, though 15 years before it failed again is quite good going, IMHO, and perhaps external problems are the cause, given that the RH channel has never failed.
The amp is actually a 3130, btw, and I don't see a pre-set for iq adjustment on the circuit diagram.  The resistor between the base & emitter of the Vbe multiplier is actually two fixed parallel types.

Ah yes, the 3130 is one of their better efforts. We had one at school, and I liked it

I agree - no Iq preset, so that takes out contact problems. Makes the adjustment procedure more tricky (manual page 4), but at least it'll stay once done.

Again, good luck!

Mark

PS: I made a minor edit to correct the quoting. With the new post editing software, it's quite easy to get the reply in the wrong place. It's worth clicking the "view source" button to see the old way of editing posts where you can see all the code in square brackets. I really must write a quick guide to all this. One day

What I forgot to mention is that the 3130 is already fitted with 3.5A Circuit Breakers in series with the speaker connections.  These, at any rate in theory, should trip out  if the current through the speaker(s) and speaker wiring is excessive.

Yes, I'd spotted them on the schematic. If I recall correctly, they are thermal breakers, so fairly slow to operate relative to how quickly the transistors might pop. Not really sure how effective they would be if the output was shorted while the amp was playing at high level. A more sophisticated amp would use VI limiting on each output transistor, usually instantaneous, but perhaps with a very short time constant built in.

Cheers,

Mark

Mounted where they are, on the PCB some distance from the heatsink, those circuit breakers are surely not going to respond too much to temperature, other than the ambient level within the amplifier, but mainly to rising current caused by temperature rise, so, as you say, Mark, probably unlikely to react quickly enough to, e.g. thermal runaway, which they certainly didn't in this case!

Hi,

The term "thermal breaker" a bit ambiguous, so I should have been clearer in post #7. Apologies

Yes, you're right - they are "thermal breakers" that respond to a localised temperature rise caused by the current that is flowing through them. Conceptually, think of them as being a low-value wire-wound resistor and a bi-metallic strip that opens/closes contacts in series with the resistance. The surrounding ambient temperature shouldn't really influence them all that much - which is why they are in a plastic case that provides thermal insulation.

This sort of breaker can be integrated into switches - here's an example of a type that I've used: http://uk.farnell.com/schurter/ta45-abtw...dp/1214776 - they're expensive, but luckily the ones I've collected were "recycled"

Thermal breakers have the advantage of slow operation, so peak in-rush currents won't cause "nuisance tripping". Just like a regular fuse, the time to trip is a function of how severe the overload it - there's a nice graph on this datasheet: http://www.farnell.com/datasheets/1684939.pdf

So in the case of an audio amplifier, the temperature rise inside the breaker is a function of the average current delivered to the loudspeakers. Now, as we know, decent music has a very high "PMR" - peak to mean ratio - so an average-responding thermal cut-out seems like a good idea in theory... But, while the transistors are interested in averages because that's what determines the heat sink temperature, they are especially interested in the peak currents, as that's what tends to kill them

I've never been 100% sure why manufacturers like NAD are so fond of these breakers. They don't protect the transistors from peak currents, and surely it would be better to monitor the heat sink temperature directly? A direct short on the output won't trip the breaker in time to save the transistors. They are hoping that the conventional fuses will open before the transistors fail, but we all know that designers put the semiconductors in to protect the fuses

So that gets us back to the other type of thermal breaker - the type I linked to in post #2. With these, there is no series resistance to speak of - the bi-metallic strip responds directly to the temperature of the surface it's attached to. Personally, I believe that all amplifiers should use these - if nothing else, there are strict rules about how hot a case is allowed to become...

I do think that thermal cutouts (the series-resistor type) might have a place in protecting loudspeakers, where here the average current will have a close relationship to the voice coil temperature. I can't think of a loudspeaker that has used these. Perhaps the worry is how the users might respond to the sudden lack of sound? I've seen halogen bulbs in series with loudspeakers (very common in PA applications), and I've seen "polysilicon" devices (I think it was Mordant-Short who liked those, back in the 1980s IIRC).

To digress slightly, you might be aware that a typical MCB uses a combination of approaches. For the required use in a consumer unit or similar, a thermal breaker is rather too slow in action, but a magnet breaker is too fast. But you can combine both principles in the same unit, and hopefully avoid nuisance tripping. You can get MCBs with different response times, of course... Having said all that, I've never got around to dissecting an MCB to see the mechanics - has anyone tried?

Anyway, have  you had a chance to make any progress with the amp yet?

Cheers,

Mark

I've ordered replacement output transistors and a Thermal cutout from CPC, so have put the NAD on one side until the parts arrive, which should be Tuesday or Wednesday.  Hopefully, when these itema are fitted, and the rest of the circuit checked/serviced as needed, the amp. will be restored to full working order again, and, if it lasts another 15 years, it will then be over 40 years old, an, as for me, well let's just say I won't be too concerned by then!!