Sansui BA-5000 Oscillation

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Working on my brother in law's new (used) Sansui BA-5000 amp. Bought it knowing there were a few issues, but I told him I've fixed a number of these amps, so if the price was right to go for it.

The main issue was simply some worn-out caps in the input board (the power supply for the small reed relays and temp monitoring circuitry & such). Once that was done, it would engage the soft-start properly, and the small reed relays would fire giving you the 'green for go' LED. The transformer was also making an ugly noise, but recapping the small +/-65V regulated supply (and installing new transistors & zener diodes) cured that.

The problem now is that it is oscillating at the output (both channels!!)...about 10V peak-to-peak, at about 500KHz. I figured there were damaged semiconductor junctions with a few of the outputs, so all 16 of them, and all four drivers, were replaced with new MJ21193 & MJ21194 devices (all emitter resistors and base resistors were checked for proper values). That made no difference.

I also swapped out every transistor & semiconductor on the driver boards (both boards). Depending on the device chosen, the output amplitide and frequency would change (a little, as I got the frequency down to about 400KHz once and amplitude to about 8V P-P), but the oscillation never goes away.

At this point the amp is fully recapped except for the four large power supply caps. One of the driver boards has mostly new ceramics in an attempt to kill the oscillation.

Like I told my brother-in-law, I've fixed a bunch of these amps, but I've never seen anything like this. I'm at a loss as to what to try next. I've reinstalled many of the original devices on the driver boards just to make sure that I wasn't imagining things with the original oscillation problem and perhaps caused it by installing some of my replacement devices (same devices that I have used in the past for several BA-5000's). Problem is still there. The oscillation from the output goes everywhere...back into the regulated supply and from there back to the front end of the amp. It's like a contamination that gets everywhere once unleashed.

Internal wiring looks good, except that a previous tech had done away with a couple of connectors and soldered wires directly to some pins rather than rely on the stock friction-fit connector (I desoldered and crimped on small friction-fit female pins so the boards could be removed without desoldering, as it is supposed to be). Grounds look good.

What's the possibility of the large PS caps or somehow a damaged tranformer playing a role here? (yes, I'm grasping at straws) Anyone think of something else I might try?

PDF Schematic (good till 6/8/09 and 100 downloads...if this expires and someone wants to look at it, let me know and I'll upload it again).

http://www.yousendit.com/download/MnFqK0dzckl6RTkzZUE9PQ
 
Talking about C05? It has been replaced. So has C03.

Ceramics are pretty reliable, and since both channels are oscillating (when each board is installed by itself, that channel oscillates) I keep looking back at the power supply, but I've never seen this kind of thing...
 
more possibilities

on the F-2518 boards, check C15 (0.047 uF) and the 10 Ohm resistor in series with it...R31 I think, but it's a bit hard for me to read the schematic. If R31 (are there actually 2 in parallel?) has gone open, the Zobel network loses its effect, and oscillation can result.

If that doesn't do it, then there are some diagnostic tests we can do with the feedback loop that might help track this down.
 
For Sansui, the '2' in parentheses means 2 watts, so that's a 10 ohm 2 watt resistor. This resistor measures OK. The stock C15 was a cheap little greenie Mylar, which I replaced with a nice 0.047µf 400V polypropylene. Made no difference.

When considering an attack plan, remember that both channels oscillate, with either or both of the driver boards installed, so it seems to me that a focus on the individual components of the driver boards (F-2518) might be self-defeating.

But, if you have some ideas my friend, I'm all ears. :xfingers:
 
EchoWars said:
When considering an attack plan, remember that both channels oscillate, with either or both of the driver boards installed, so it seems to me that a focus on the individual components of the driver boards (F-2518) might be self-defeating.

So its either the power supply or a change you have made to both boards.

Could it be you have changed a component to a wrong value on both boards ?
 
Right now, both driver boards have all new electrolytic caps (Panasonic FC and Elna Silmic for the input DC blocker), and a new 2SA798 input differential transistor. One board has all the rest of the semiconductors as original, the other is using Fairchild KSC2682 for TR02 & TR03, and KSA1142 for TR04. The bias servo transistor (TR05) is stock (and was temporarily replaced as a test with no change), and the drivers are On-Semi MJE15032 and 15033. The rest is all stock except for some ceramic cap changes to one of the boards as a test.
 
EchoWars said:
Talking about C05? It has been replaced. So has C03.

Ceramics are pretty reliable, and since both channels are oscillating (when each board is installed by itself, that channel oscillates) I keep looking back at the power supply, but I've never seen this kind of thing...


It is possible for bad supply caps to cause oscillation. You could try, as a test, to solder in fresh capacitors in parallell.
 
Re: more ideas...

djoffe said:
have you separated the output board from the input preamp? What if the input preamp (F-2520) were oscillating, or passing through oscillations from the LV power supply (F-2521)?

If you can separate the two and the output oscillation goes away...you've narrowed your search considerably...
Yes, there's a connector that runs from the input board to the driver board. I can break this connection and it makes no change in the oscillation.

megajocke said:



It is possible for bad supply caps to cause oscillation. You could try, as a test, to solder in fresh capacitors in parallell.
I thought about that, but there isn't a lot of room. I do have some 35,000µf 80V caps here somewhere. I might install them and see what happens.

Bigun said:
don't count me as vet'...

Is it possible there is some interaction with the load (assuming you have one) ?
No load right now.
Can you isolate pieces, one at a time by taking out their supply etc a few resistors (e.g. R13) ?
If I take out R13, the oscillation will stop because I essentially disable the whole driver board. I'm not sure what that will prove...
 
I may have misread the schematic, but it looked like R13 just provided power to the LTP, removing that leaves the output stages powered up. What I'm really suggesting is that with the oscillations everywhere you can't see the wood for the trees. If you can disable pieces, systematically, you might narrow down key pieces that are essential to sustain the oscillations and get closer to finding the source.
 
payloadde said:
no load at all ? just the probe tip, is that what you are saying ?
if so, I'd suggest to connect a dummy load, 8 or 16 ohm resistor (several watts, no input signal of course)
Well, it's never supposed to oscillate, load or no load. I do see what you're saying, but I don't think the lack of a load matters one way or the other. No oscillation, ever.

Bigun said:
I may have misread the schematic, but it looked like R13 just provided power to the LTP, removing that leaves the output stages powered up. What I'm really suggesting is that with the oscillations everywhere you can't see the wood for the trees. If you can disable pieces, systematically, you might narrow down key pieces that are essential to sustain the oscillations and get closer to finding the source.
I can remove R13, if you think it will prove something, but I'd say that oscillation will quit because the feedback for the amp is now dead. I'll try tomorrow and see what happens.
 
Seeing as you did quote a fair amount of changes to the amp, and seeing all the caps installed into the circuit to stabilise it, I'd hazard a guess that there's some serious changes made to the open loop bandwidth/gain of the amp.

Just as a test, install a 150pF cap across the VAS transistor. That would be C5 across TR3 if I'm reading it right. This would be a pretty severy measure, but would eliminate so many variables. If the amp does staibilise, we can maybe try tweak it from there. If that doesn't help then there's bigger issues at work here.
 
gbyleveldt said:
Seeing as you did quote a fair amount of changes to the amp, and seeing all the caps installed into the circuit to stabilise it, I'd hazard a guess that there's some serious changes made to the open loop bandwidth/gain of the amp.
Well, that's the way things started out...lots of changes. But after identifying the oscillation, I have put one of the driver boards (I'm troubleshooting with only one of the driver boards right now) back to totally stock configuration...only the output transistors are changed.
Just as a test, install a 150pF cap across the VAS transistor. That would be C5 across TR3 if I'm reading it right. This would be a pretty severe measure, but would eliminate so many variables. If the amp does stabilize, we can maybe try tweak it from there. If that doesn't help then there's bigger issues at work here.
Makes sense. I'll try this first.

But, assuming that lots of compensation at the VAS stage stablizes the amp, the question is: what made it unstable when I've worked on so many of these amps using the same parts and had no problems at all?

Well, let's see what happens first.
 
It's hard to say at this point, but looking at the schematics they tried some fun things there to keep the amp stable without limiting the closed loop HF. I also think (just my opinion) that back in those days the transistors were a lot slower than the new fancy stuff we have now and that may wreak havoc on a design that was borderline to begin with. Again, just my opinion, but let's try and tame it first before jumping to conclusions;)
 
Well shoot...I thought I could use the 35,000µf 80V caps I had here to swap out the four large power supply caps. That won't work, as the originals have an oddball mounting method for the terminals that requires the solder lug terminals. Michael Percy has the perfect replacement, but the cost to roll the dice on this fix is $200 ($50 for each cap). If I knew that four new power supply caps would fix the problem, I'd buy them in a heartbeat. But I'm not big on spending $200 on a longshot.

I'll try swapping out C5.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.