SA9500 intermittent protection issue

[oz-2]

Dear AK'ers

I wonder whether anyone could give me some advice on how to fault find an intermittent protection issue on an SA9500 (mk1)? I will be very grateful for the usual excellent advice and help.

Just to give background.

The amp has been professionally recapped. I splashed out a couple of years ago. Amp, in general very good condition. I think it must be a late version of the Mk1 as the wrap pins are fixed to a clear perspex kind of former.
I couldn't afford to do the known iffy transistors at the same time.
The amp goes into protection intermittently and for very short periods on an irregular basis.

I thought, initially, it was during loud passages of music but not always after observing over a while.

I thought the input level might be too high from CD. Ive turned that down.

I recently noticed a kind of 'rustling' sound at, or just before the amp went into protection. I suspected the 4 2sa726's on the pwr amp board. I replaced them and Q7 and Q8 whilst I was in there with KSA992's and KSC3503's respectively even though Q7 and Q8 were not the dreaded 1451's. I also swapped out the trim pots and replaced those with MTF's recommended Mouser alternatives. The amp set up to 0v balance and 20mv bias beautifully and stably over a long period of watching it. The occasional rustling has gone completely. but every now and then the amp slips into protection and out. Sometimes repeated a lot of times before settling down again. Sometimes it never does it!

Another suspect area is the pwr supply. It has been recapped but not had any transistors replaced. The trim pots have been replaced and adjusted up to set voltage levels.

I use an 8ohm set of spkrs. (Kef Q100's)

So thats whats been done.

My question is really, in view of this, how do I fault find? How can I monitor something to see what's happening? I'm a bit stumped for ideas short of wiring something in to the protection board so I can monitor that but I have no real idea what would work there.

As always, Im grateful that you've even read this far and if you offer any advice even more grateful.

Best wishes

David

 

[chazix]

Given that there have been no takers so far, and I think it's a difficult situation with any amplifier, and especially difficult with the SA-9500 because the protection card is so difficult to work on, I'm going to make a fairly radical suggestion. It will only help narrow down the cause - it won't pinpoint a component. Please use your own judgement about the sanity of the suggestion.

The idea is to selectively disable the fault-sensing abilities of the protection circuit, to help determine whether it is reacting to a real fault input, or is giving a "false positive".

For example, if you connect protection board terminals 5&6 to each other, and also connect 7&8, the circuit will lose the ability to detect an overcurrent condition in the power amp outputs.

Disabling the ability to sense excessive DC offsets is more difficult. You would need to identify the junction of R9 and R10, and connect that junction to ground. It's in the middle of the protection card, so access isn't easy - you'll definitely need the side panel removed. (Alternatively, you could simply remove the connections to terminals 2&3, but I hate disturbing beautifully wire-wrapped connections myself.)

If you try this, don't have speakers connected. (The DC offset sensing, in particular, is mostly for protection of speakers.) Just listen for the protection relay. If it still cuts in and out when the protection circuit is "blind", you know that there's a fault in either the protection circuit itself, or (much less likely, I'd say) in one of the power supplies that service it.

If nothing else, maybe my crazy idea will provoke a saner one from someone...

Cheers,

chazix

 

[oz-2]

Hi Chazix
As you say, there's not much to go with at the moment in advice terms. I appreciate your time and interest and will consider what you say. Its very logical and doable although I'm anxious about shorting anything together! I have a habit of shorting the wrong things!

The only other thought I had, but it needs a bit of further advice around a number of things was to replace the diodes with led's. I'm not even sure a led has the same or similar qualities to a normal diode but I wondered whether it might show up when one of the fault channels was activated. Either by extinguishing or coming on. Perhaps Im not reading the circuit right either.

But thanks - its very kind of you. AK is a great place.

 

[chazix]

replace the diodes with led's

I think that's a clever idea in principle, but I'm not sure I know which diodes you mean. If you mean the "OR logic" diodes that are connected to the base of Q5 on the protection card, I don't think they will pass enough current to noticeably light an LED when active.

I didn't try to figure out if this would be relevant in this particular context, but for general information: While LEDs do act as diodes, they develop a larger forward voltage when they're conducting than any ordinary solid-state diode that I can think of - in the 2-3 volt range for LEDs, vs 0.7-ish for ordinary silicon diodes. Also, LEDs tend to be quite limited in how much reverse voltage they can tolerate. I think some LEDs are limited to single digits of reverse voltage, while ordinary silicon diodes are usually good for at least 50V.

Now, you might be able to use LEDs to implement inductor's suggestion of monitoring DC offset. (A good suggestion in general, but you'd mentioned that the DC offset was low and stable, so I'm not sure it's a primary suspect.) You could connect a pair of LEDs in parallel with opposite polarities, and connect them with, say, a 470 ohm restistor in series, wired between a power amplifier output point and ground. I'd expect any persistent DC offset of much above 2V to noticeably light the LED. (The LED would also respond to audio signals, and would get too bright at higher volume, so I think this setup is only good for testing with the volume all the way down.)

Here's another possible tactic: A can of freeze mist can sometimes - given a healthy dose of luck - help in tracking down intermittent faults. The next time the amp is acting up, you might try giving the protection board a broad spray of freeze mist. If the amp's behavior suddenly gets much better or much worse, you could be pretty confident that you're dealing with a flaky component on that board. With enough patience, you might even be able to tell which component by spraying them individually.

(Best to not try using freeze mist on the power amp board. Depending on what component gets cooled, it could result in either excessive current in the output transistors, or a substantial shift in DC offset.)

chazix

 

[oz-2]

I think that's a clever idea in principle, but I'm not sure I know which diodes you mean. If you mean the "OR logic" diodes that are connected to the base of Q5 on the protection card, I don't think they will pass enough current to noticeably light an LED when active.

I didn't try to figure out if this would be relevant in this particular context, but for general information: While LEDs do act as diodes, they develop a larger forward voltage when they're conducting than any ordinary solid-state diode that I can think of - in the 2-3 volt range for LEDs, vs 0.7-ish for ordinary silicon diodes. Also, LEDs tend to be quite limited in how much reverse voltage they can tolerate. I think some LEDs are limited to single digits of reverse voltage, while ordinary silicon diodes are usually good for at least 50V.

Now, you might be able to use LEDs to implement inductor's suggestion of monitoring DC offset. (A good suggestion in general, but you'd mentioned that the DC offset was low and stable, so I'm not sure it's a primary suspect.) You could connect a pair of LEDs in parallel with opposite polarities, and connect them with, say, a 470 ohm restistor in series, wired between a power amplifier output point and ground. I'd expect any persistent DC offset of much above 2V to noticeably light the LED. (The LED would also respond to audio signals, and would get too bright at higher volume, so I think this setup is only good for testing with the volume all the way down.)

Here's another possible tactic: A can of freeze mist can sometimes - given a healthy dose of luck - help in tracking down intermittent faults. The next time the amp is acting up, you might try giving the protection board a broad spray of freeze mist. If the amp's behavior suddenly gets much better or much worse, you could be pretty confident that you're dealing with a flaky component on that board. With enough patience, you might even be able to tell which component by spraying them individually.

(Best to not try using freeze mist on the power amp board. Depending on what component gets cooled, it could result in either excessive current in the output transistors, or a substantial shift in DC offset.)

chazix

That's very helpful and informative Chazix. Cheers. I learned something about LED's and how it might help. As it happens (confusingly) since I posted my original question the amp has been on best behaviour and not a sign of acting up! I wondered whether that might mean a bit of a brown out was happening to the mains supply. (the amps just connected straight to ordinary mains - no conditioning) and I had noticed that the setting of the trim pots on the PS board was very sensitive for the protection card. A bit high or low trips the relay straight away. Who knows but I now have some extra pointers to help me fathom things out and I like the advice for patience too!! Always a good thing for these old amps, I guess.

Many thanks David

 

[markthefixer]

Your "professional recapper" probably didn't change out the transistors in the power amp.

There are ticking time bomb transistors in the power amp. Q7 and Q8 are listed in the BOM as 2sc1628 BUT 2sc1451's are listed as alternatives and used a lot.
also all the 2sa726 transistors.

somehow matt's list got it right with: AWS-082: Q6: 2SC1451: 512-KSC3503DSTU
(he used my list as a starting point, but I don't know how thoroughly I was able to vet the results)

but somehow missed AWH-041: Q7: 2sc1628(/2SC1451): 512-KSC3503DSTU

so, you need 6 transistors on the power amp board, 4x ksa992, 2x ksc3503

oh, and for "subtractive troubleshooting" of the protect board, just disconnect D5 for over current inputs, and d6&d7 for dc offset inputs

 

[oz-2]

Your "professional recapper" probably didn't change out the transistors in the power amp.

There are ticking time bomb transistors in the power amp. Q7 and Q8 are listed in the BOM as 2sc1628 BUT 2sc1451's are listed as alternatives and used a lot.
also all the 2sa726 transistors.

somehow matt's list got it right with: AWS-082: Q6: 2SC1451: 512-KSC3503DSTU
(he used my list as a starting point, but I don't know how thoroughly I was able to vet the results)

but somehow missed AWH-041: Q7: 2sc1628(/2SC1451): 512-KSC3503DSTU

so, you need 6 transistors on the power amp board, 4x ksa992, 2x ksc3503

oh, and for "subtractive troubleshooting" of the protect board, just disconnect D5 for over current inputs, and d6&d7 for dc offset inputs

Hi MTF
very nice to hear from you and thanks for responding as always. Thanks for the advice especially with the diodes.

I did change those 6 transistors as my first port of call after the protection started acting up. I followed another AK'ers advice and ordered 20 ksa992's and checked the hfe and split them up into groups of similar or identical hfe and then fitted four of the best and the trim pots and the 2sc1628's with ksc3503's whilst at it.
To be fair to my recapper I only asked him to recap nothing else so he's in the clear - but i wish I had.

If I had 1628's on the main pwr amp board am I likely to have them throughout the amp judging by your experience or are there always 2sc1451's in the other areas?
hope alls well with the estate work. If US law is as convoluted as UK it is never pleasant.

David (oz-2)

 

[markthefixer]

Throughout the amp is not a problem, only in the power amp driver stage are they dangerous, and failure prone.

Due to the difficulty of accessing the board, when I have problems in the protection area, I do the whole shebang.
most are ksc1845 q1-q4, one ksa992 q5, one ksc2383 q6 and one ksc2690 q7. A few caps and diodes (512-1n4148)
and an extra flyback diode (512-1n4004) rounds out the load-out.

The protection SHOULD stay out for a MINIMUM of the power on delay (4 seconds?) - if not it's a relay drive problem.

If it still does it, then the power supply should get new silicon,seeing as the power amp is pretty much new silicon.

Also suspicious is the D8 negative power sense, a new d8 1n4004 and check the ac - sometimes a fuse gets dodgey...

Thanks for the well wishes, not a large estate and I do have a good attorney, things are still very busy, but ARE progressing

 

[oz-2]

Thanks for that MTF. I will open her up and look at the Protection board - its a real pig to get to those lower components but it can be done!

Would you happen to have a list of the transistors needed for the power supply board? I would be very grateful as I could do both at the same time.

Many thanks
David Oz

 

[oz-2]

Throughout the amp is not a problem, only in the power amp driver stage are they dangerous, and failure prone.

Due to the difficulty of accessing the board, when I have problems in the protection area, I do the whole shebang.
most are ksc1845 q1-q4, one ksa992 q5, one ksc2383 q6 and one ksc2690 q7. A few caps and diodes (512-1n4148)
and an extra flyback diode (512-1n4004) rounds out the load-out.

The protection SHOULD stay out for a MINIMUM of the power on delay (4 seconds?) - if not it's a relay drive problem.

If it still does it, then the power supply should get new silicon,seeing as the power amp is pretty much new silicon.

Also suspicious is the D8 negative power sense, a new d8 1n4004 and check the ac - sometimes a fuse gets dodgey...

Thanks for the well wishes, not a large estate and I do have a good attorney, things are still very busy, but ARE progressing

Hi Mark and other AK'ers.

I have bitten the bullet and stripped her down and fitted all new silicon to the protection board (except diodes) and to the Power Supply board. Quite surprised at the different shapes of the new transistors compared to the older ones especially in the PS but also the relay driver in the Protection Board. I was so chuffed that the voltages came up straight away on the PS without any tweaking but I let it idle for a while and then set +/-48v dc which seemed quite nice and stable. She has been running with Radio Paradise for about 15 hours now and no protection issue.

I measured in a no signal condition the dc at the speaker terminals and found about 17mv. Left and Right. Is that bad or ok? I never thought about setting Power Amp idle and bias while I had the chassis open. I guess I should have done.

Best wishes to all.
OZ-2

 

[markthefixer]

any amp or power supply work means starting back out at 0 idle current, DBT tickling the dragon's tail, then setting the idle current on straight line power.

 

[oz-2]

any amp or power supply work means starting back out at 0 idle current, DBT tickling the dragon's tail, then setting the idle current on straight line power.

yes. stupid to not pay attention to that. ive now adjusted the power amp bias and balance. all good and sounds better too. thanks for alert! best wishes.