Welcome to Solid State Guitar Amp Forum | DIY Guitar Amplifiers. Please login or sign up.

March 28, 2024, 09:38:59 AM

Login with username, password and session length

Recent Posts

 

Fixing this old stereo amplifier

Started by sim0n, August 07, 2012, 04:45:47 AM

Previous topic - Next topic

sim0n

Hey guys, I guess I'll try this here as well cause I know there are a bunch of knowledgeable folk around. My friend gave me this amp that had been sitting under his bench in his workshop for the past 6 or 7 years. The enclsoure made of MDF was all warped and had mould growing over half of it. The amp was supposedly a DIY job he and a friend of his made (the actual circuit seems to be from some kit). According to him, one channel didn't work, the other was allegedly ok.

Anyway, the wiring was absolutely horrid, nothing in the enclosure was actually screwed down other than the transformer and the enclosure itself was probably a biohazard so I dismantled the entire thing.



I've managed to draw myself a schematic of the circuit http://img840.imageshack.us/img840/9194/schematictw.png (and it also functions in LTSpice so I guess I got it right)

As you can see on the photo above, theres some black charring on the circuitboard so I'm guessing that this is the channel thats shot. Though the resistor that has turned black measures ok (R20 in the schematic), the base connection on the transistor Q11 (as numbered in my schematic) is ripped out of the package (not visible on the photo, the driver transistors are mounted directly on the heatsink with the main output transistors).
So, how would I go about fixing that issue, since usually when one things dies it likes to take other things with it.  Is there anything else that I should check? How close does the replacement for the transistor need to be? I've got some TIP32C and a multide of common TO220 package transistors, would any of those work as a replacement or should I order something better?

And, if I do fix it, what am I aiming to get when adjusting the trimmers? (I've got a scope and a handful of multimeters for measurements).

J M Fahey

Start by building a lamp bulb limiter (search this Forum), you'll need it.
Excellent schematic recovery, looks 100% good.
Post a picture of how the power+driver transistors are mounted on the PCB.
Remember if the 2N3055 is bad, replacements probably will be fake ... even if bought in a "serious" shop.
What Country are you in?

Roly

Quote from: sim0n
the enclosure itself was probably a biohazard so I dismantled the entire thing




Congrats on the circuit trace - looks sensible.

On your rough circuit (posted on diyaudio) I'd guess that P1 (10k) is to set the output offset to zero, P2 (2k2) is the idle bias current setting, and P3 (100k) looks like it sets the VU meter to full scale (100% or 0dB) at the onset of clipping.

On the Spice trace, as you can see R20 feeds a sample of the current through R15 to the protection circuit, Q13.

How could R20 come to be burned?

It seems unlikely that it was due to a large voltage across R15 unless R15 has gone open, and what seems more possible is that the protection transistor Q13 has failed shorted C-to-B-open-emitter and provided a path from the half-rail to the negative supply via the VAS Q4.  In any case all the semicons are going to need a close look.

My personal favorite drivers are the BD139/BD140 pair.  You could stick with the original 2N3055/BDX18 output pair, but apart from cheap, the 2N3055/2N2955 didn't have a lot of bandwidth so it wouldn't be hard to do quite a bit better these days; say MJ15003/4, TIP35/6, or TIP 41/2.

I'd go over every semicon on the board with an ohmmeter, and anything that looks even slightly suspicious in-circuit I'd have right out for serious checking.

I'd also isolate the amp from the power supply at the power supply end, and check for reasonable voltages.  It's not unknown for an amp blowup to take the rectifier with it.

Once you are sure you have found and replaced all the faulty silicon, the burned resistor, and anything else that turns up, you power up the amp via your limiting lamps with the speaker disconnected.  The lamps should settle to a dim glow, the supply rails should take on reasonable, if somewhat reduced, values, and the half-rail should be very close to zero volts.

If all goes well you can then try injecting a sinewave signal and see if the output voltage behaves accordingly.  All being well you can remove the limiting lamp.  You can then try setting the respective trim pots for zero DC output, and around 20-30mA of idle current by measuring the voltage drop across R14 and R15.  Finally hook up a speaker for a listening test.

HTH
If you say theory and practice don't agree you haven't applied enough theory.

sim0n

Thanks for the replies :)

Here is the transistor mounted on the heatsink (the TO3 package is visible on the other side, the 150ohm resistor soldered directly between the emitter and base with the driver transistor screwed on nearby). This is NPN side, the other side is identical, the circuitboard is sandwiched between the two heatsinks with standoffs.



Among other visible casualites, along with the Q11 and R20, I also noticed that Q12 has disintegrated (again, one pin completely disconnected from the cracked pacakge) and Q5 is chipped away on the side as well if the sheds any light on the subject. The junction between the base and collector seems to be shorted.
Among resistors that have iffy values are R13 and R19 (around 200ohms instead of 1k), R10 (700ohms inteas of 3k3), R11 (also around 700 instead of 1k). I'm getting conductivity over C6 too. I wrongly read the colour bands, R16 and R20 are actually 180ohms, the burnt R20 measures 140ohms.
And if I measure from the positive to the negative power rail it shows 370ohms.
On the other channel of the amp all these components measure what they visually indicate with any abnormalities. So it looks like the output transistors are shot too.

I'm still in the process of building the lamp limiter and rebuilding the power supply (the bridge rectifier measures OK with my multimeters continuity test). I'll try and get the other channel working before I tackle this.

Roly

I'm beginning to wonder what you need us for, since you seem to be getting on just fine on your own (heh heh   ;) ).

In your latest pic I can't see any sign of a mica washer or other insulator under the driver transistor on the heatsink.  Anodising is nominally insulating, but I certainly wouldn't depend on it.

With resistors and caps, if there is any doubt just unsolder one end, then measure.  Generally speaking it's the active components that are troublesome, the passive (unless obviously broken or reduced to a charred ruin) seldom cause problems.

As you think this was a kit build I'd be inclined to go over every solder joint too, just to be sure they are properly made.

Getting the other channel going if you can is a good idea (but I'd use the limiting lamp on such an unknown quantity); often the nice thing about servicing stereo amps is that you have a going channel alongside the dead one to compare.  Good luck.  :tu:
If you say theory and practice don't agree you haven't applied enough theory.

sim0n

Quote from: Roly on August 08, 2012, 12:08:10 AM
I'm beginning to wonder what you need us for, since you seem to be getting on just fine on your own (heh heh   ;) ).

:cheesy:

I checked the heatsinked parts for continuity, its not visible on the photo but apart from the ugly wiring, the output and driver transistors are actually properly isolated from the heatsink with a mica pad and those plastic washer/spacer things.

I finally rigged up a test jig and got everything connected together. The secondary of the transformer measures 22-0-22. The power supply unloaded measures +-30 volts.
I hooked up the working channel with the lamp limiter and the input grounded. The supply voltage drops to +25. I managed to measure a set of votlages and they look OK for the most part, or at least they roughly match up to the spice model.

The 2k2 trimmer for the bias current functions, the sweep is a bit finnicky but if I turn it too far the lamp noticeably lights up so it seems that that works (and I managed to measure the voltage drop over the 0.39ohm resistors and set it precisely). However, the 10k trimmer, will not adjust the dc offset into the negative range, it doesn't go below 30mV at all.
Any ideas for that issue? My intuition says to check electrolytic capacitors (specifically C2 which is a 10uF bipolar).


J M Fahey

50mV offset is *nothing*. :tu:
Besides, that 10 K resistor offset adjustment is quite poor.
I'd set it on max value (10K) or, say, 4K7 (half value) and leave it there.

sim0n

If its supposed to adjust the offset to zero then it should do that!!  :duh
This falls in the category "I dont know exactly what I'm doing and how it works" but I changed R2 from 3k3 to 2k2 and that shifted the range of adjustment to let me got 0mV (well, I know its got something to do with Q3 being a constant current source but I've got no idea what the rest of the stuff is doing).
I hooked up my scope, everything looked fine on the output, hooked up a speaker and it works too :D so...yay! It hits 57volts peak-to-peak on the output. It also appears to be quiet apt at amplifying the MHz noise my computer puts out along with the audio signal  :-\

But I've begun to suspect the transformer is a bit iffy too, as it also looks like a homemade job (or its just that old, theres brown butchers paper between layers of windings). After an hour or so of reasonably quiet playing it got too hot to touch, way way more than the transistors. I'm suspecting that whoever was stacking this didn't bother to do any kind of isolation between laminations as the core gets hotter than the windings. In the original enclosure the fan was pointed at the transformer, not the heatsinks.

So I finally get to use all that knowledge about how to measure transformer parameters I learnt at school  :lmao:

Roly

You seem to be right on top of it.  Well done.  :dbtu:  It's very satisfying to start out with some item of discarded junk and end up with something useful (I've just brought a couple of computers scavenged at the local tip back to life).

Power tranny - try disconnecting the output rails from the power supply filters and leaving it run idle for an hour or so.  The tranny should be warm but not hot, and you should certainly be able to hold your hand on it without discomfort.  If it is too hot to hold your hand on then you have a problem, however I'd give it another run isolated from the rectifier and filter caps just to be sure that it's the tranny (you could have leaky old electros, but then they would also be getting hot on idle).

The next thought is the idle current.  Don't get too carried away adjusting the 2k2 idle setting 'coz it looks like it could crank the idle current up to quite a high value at the extreme settings.  At the moment you look to have 0.06V/0.39ohm = 154mA which is more than a bit high.  I'd guess at something more like 30 to 50mA which would be E=IR = .03*0.39 ~= 12mV - 20mV.  The idea here is to start out with a low idle current (minimum voltage across Q5) and gently slide it up until you don't have any crossover distortion (which ideally needs a CRO or Noise and Distortion meter, but ears will do).

Output offset - Like JM I'm not surprised you can't set the DC offset to zero, nor do I think 50mV offset is a problem.  The origin of this offset is the base current flowing out of Q1 through R1 (you'll notice the 40mV here and the 50mV at the output are almost the same).  A better way to adjust this offset IMO is to suck the base current of Q1 out of that node using a very large resistor to a largish value pot across the +/- rails.

The way they have done it also changes the amp loop gain, which is not really best practice.  "If its supposed to adjust the offset to zero..." is a fairly large "if".  Given the 2N3055 in the output (and the position of the 10K) it is more likely it is only intended to reduce the offset to a "reasonable" value.

I have never been keen on bipolar electrolytics, so if C2 is oldish it might be a good idea to renew it.

Now R2, Q3, R3 and zener D1 form a constant current source for the Long-Tailed Pair (LTP, aka Differential pair).  The current is set by R2 at the zener voltage minus the BE drop of Q3 (and if that really is a zener it's drawn upside down BTW).  Taking your voltages, there is a 1.4 volt drop across the zener meaning there is around 0.7V across R2.  As this voltage is reasonably constant the collector current of Q3 will be I=E/R = 0.7/3k3 = 200uA, or 0.7/2k2 = 300uA now.  This is actually the tail of the Long Tail Pair, but drawn "upside-down" in PNP format it might be a bit hard to recognise as such.

And now you've got lots of volts of AC on the output - bravo!   :dbtu:

Yeah, 'puters are pretty noisy audio sources.

It may also be because you have an earth loop; both the computer common is connected to mains ground and the amp common is also connected to mains ground.  One thing you can try is put a simple low pass filter at the input, like a few k-ohms in series with C1 and a few pF across R1.  Let's say 2k2 in series to give a 5% reduction in mid-band signal in combination with R1, then we need a cap that will give a rolloff starting about 20-25kHz.  A cap that has a reactance, Xc, equal to the series resistor, 2k2, at 25kHz will give a -6dB or half voltage point at that frequency.

Xc = 1 / 2 Pi f C

or

C = 1 / 2 Pi f Xc

uF = 1/(2*Pi*25*2.2) = 0.00289373uF or around 2.7nF (ballpark, select to taste)

If the problem seems to be due to an earth loop try inserting a small resistor (47 - 220 ohms) in the common connection between the amp and computer.

HTH
If you say theory and practice don't agree you haven't applied enough theory.

sim0n

#9
Hah, finally getting some real world usage of the stuff I learned at university.

After getting a set of measurements for the transformer's parameters with the scope and a multimeter, I then remembered I had one of those $15 wall socket energy meter things :lmao: (if it wasn't summer right now I could probably take it to the lab and hook it up to a shiny $5000 power analyzer)

Running open circuit its has 15W of losses in the core (and a power factor of 0.3). My own measurements gave a result of 20W. Judging by its size (the core is 10.5x10.5x5x5cm) it seems to be somewhere around the 200VA range and according to the tables in my EE handbook, 20W is nothing odd for the core losses for a transfomer of size.


edit: and yeah, I'm puzzled about the zener too buts the way its installed on the circuit and the voltage seems to be right too. It just might be a case of strange component markings.


edit 2: I've desoldered all the dead components and checked their health out of circuit

Q4 and all the silicon ahead of it is dead (short between two terminals with various ailments to the third one) except the BDX18 (the PNP output transistor) and the BD137 (the NPN side driver transistor). D4 was also shorted out and the 0.39ohm resistor on the PNP side has turned into 50kiloohms.

I soldered a couple of things in, but just realised that the largest voltage they'll be seeing is 60V, so if the maximum collector-emitter voltage is 50V that might not be good and I can't find anything in my drawers rated for higher than that thats not a full out power transistor.

edit 3: my logic has lead me to the conclusion, that the burnt out resistor probably saved the PNP output transistor from death

Roly

I do love to watch somebody having fun, and this one certainly sounds like it has had a good dose of the galloping silicon cancer through it - explains why it was unloved under the bench.

Vcer is generally the voltage rating you need to watch.  It's unlikely to be too picky about the exact transistor types you fit, just as long as they have "enough" of what they need, voltage, current, power, and gain.

I used to work on large Variable Speed drives and it was a standing joke that expensive huge SCR's would fail to protect the fuses.  ::)

Talking of fuses, a suitable fuse in each supply rail at each amp is a good idea.
If you say theory and practice don't agree you haven't applied enough theory.

J M Fahey

QuoteRunning open circuit its has 15W of losses in the core
Sounds reasonable, now reconnect the transformer to the amp, no signal, and remeasure.
Amp at idle should eat no more than 30 or 40W.
If more, it's probably overbiased and will overheat on its own.
Of course, the test is valid only *after* the amp is repaired.
Until then, always go through the bulb limiter to play it safe.

sim0n

Huzzah, it works! I wasted two hours yesterday wondering why it was giving me -18V on the output until I finally realised I put a PNP transistor in for Q4 instead of an NPN  :duh

I replaced all the smaller transistors with MPSA06/MPSA56, the driver transistors for BD139/BD140 and just replaced the dead NPN output transistor with another 2n3055 that I already had.

Here one channel hooked up on my temporary test jig:


You can see the age of the transformer  :-\ but it functions. The power supply only has 2x4700uF, I'll probably redo it with some bigger caps and proper fuses and such.
Now to just get an enclosure for this thing and turn it into a useable piece of equipment  :cheesy:

Thanks for the help guys!

Roly

Not to worry, we all have those {duh!} moments, and it's really good to hear you've finally got it going after its long rest under the bench, well done.   :dbtu:

Tranny looks okay to me, nice and chunky - as long as it keeps transforming it's fine.

Yeah, much bigger filter caps; be driven by the ripple current rating and go for equal or in excess of the total load current; and yep, fuses for each rail, for each side (4).

(oh that poor lil speaker!)  xP

Looks to me like it should be capable of a good 50 watts a side all day.

Now all you need is a case, nice preamp, pair of speakers ...  8|
If you say theory and practice don't agree you haven't applied enough theory.