2x200W Audio Amplifier Repair

[plugbark]

A friend gave me a 2x200W audio amplifier composed of 2 PCBs (providing 200W each), which keeps blowing inner fuses.
I don't find any documentation related to this amplifier (Pro System PS400): The PCBs aren't named and the brand doesn't seems to be very popular.
He bay this amplifier to a flea market so the previous owner is not reachable to be asked questions.

There are 4 fuses sockets for main DC voltages (corresponding to +40VDC and -40VDC supply voltages for each card).
When I dissembled the case for the first time, I notice 4x F3.15A 250V fuses placed on their sockets. However, I'm not 100% sure these fuses are the original ones (maybe the previous owner tried to change the fuses and chose a wrong reference?)

I verified the following things:
1. I retried by replacing F3.15A by a slow type T3.15A 250V fuses: FAIL.
2. Transformer: Secondary winding disconnected from the PCBs: 230VAC (France) -> 30VAC measured [marked 220VAC (formerly in France) -> 27VAC].
3. DC voltage: ±40VDC after the diode bridge (while the 2200µF capacitors are only 35V DC rated!)
I desoldered one of these capacitors (4 in total) and tested it: the charge/discharge curves are OK.
Conclusion: The AC/DC conversion seems to be fine although I'm surprised the capacitor 35V DC rating is exceeded.

To prevent fuses from blowing, I used a light bulb current limiter (75W light bulb). The light bulb brights immediately after I switched on the amp.
I performed this test on both cards (only one card connected at a time) and it appears that the short circuit problem seems to be the same on both cards while they are independent. I didn't see any hard short circuit on both PCBs.

What tests could I do to isolate the fault?

[glebert]

Have you tested the output transistors for shorts? Regarding the capacitors, I think it is possible for a device to test OK with a meter but fail in the circuit because the meter uses very low voltages to test. Also, if you tested one out of the four caps you may have happened to pick the good one.

Sometimes with a circuit that is shorting out I have been able to put it on the light bulb limiter and see some very small AC or DC voltages and look for any voltages that seem lower than the rest to try to find the short. Also check for things heating up.

If you pull those internal fuses do you still get a bright bulb? I am wondering if those are on the DC rails since they aren't right next to the diodes and caps.

[nevetslab]

A friend gave me a 2x200W audio amplifier composed of 2 PCBs (providing 200W each), which keeps blowing inner fuses.
I don't find any documentation related to this amplifier (Pro System PS400): The PCBs aren't named and the brand doesn't seems to be very popular.
He bay this amplifier to a flea market so the previous owner is not reachable to be asked questions.

There are 4 fuses sockets for main DC voltages (corresponding to +40VDC and -40VDC supply voltages for each card).
When I dissembled the case for the first time, I notice 4x F3.15A 250V fuses placed on their sockets. However, I'm not 100% sure these fuses are the original ones (maybe the previous owner tried to change the fuses and chose a wrong reference?)

I verified the following things:
1. I retried by replacing F3.15A by a slow type T3.15A 250V fuses: FAIL.
2. Transformer: Secondary winding disconnected from the PCBs: 230VAC (France) -> 30VAC measured [marked 220VAC (formerly in France) -> 27VAC].
3. DC voltage: ±40VDC after the diode bridge (while the 2200µF capacitors are only 35V DC rated!)
I desoldered one of these capacitors (4 in total) and tested it: the charge/discharge curves are OK.
Conclusion: The AC/DC conversion seems to be fine although I'm surprised the capacitor 35V DC rating is exceeded.

To prevent fuses from blowing, I used a light bulb current limiter (75W light bulb). The light bulb brights immediately after I switched on the amp.
I performed this test on both cards (only one card connected at a time) and it appears that the short circuit problem seems to be the same on both cards while they are independent. I didn't see any hard short circuit on both PCBs.

What tests could I do to isolate the fault?

[ATTACH=CONFIG]54950[/ATTACH][ATTACH=CONFIG]54951[/ATTACH]

Do you know how to check transistors & diodes with a multimeter? Most current models have a Diode Test mode, which normally provides a 1mA constant current source with a 2-3 volt potential behind it. That's enough to test for B-E, B-C junctions and anode-cathode junctions, though can't test for zener diode breakdown voltages....it will, of course, verify in the opposite direction that the junction is good or not. Test ALL of your xstrs and diodes on the board. As you're blowing fuses, the suspect parts are the two heatsink-mounted power xstrs and the two driver xstrs, which have a metal tab (TO-202 package towards the middle of the PCB's).

Follow the connection path of the two fuses. I'm guessing they are between the output of the filter capacitors (2200uF/35V) and the collectors of the power xstrs, and all the other circuitry ahead of the power xstrs.....ie, power supply fuses. If they're blowing, it's a good bet the power xstrs and driver xstrs have shorted (C-E, B-C, B-E shorts), and will have to be replaced. If the driver xstrs are also shorted, you're probably going to have to make a part substitution for those, as that package (TO-202) is no longer a currently used package. If so, provide us with the device numbers on them (silkscreen), and we can find a substitute).

The two large power xstrs have to be mounted to the heat sink surface with insulators. Mica with thermal grease will work the best, though messy. If there are rubberized substitutes, chances are they've been damaged from long-term termoset, and breaking that bond from the heat sink peels away the insulation, so they cannot be reused. The heat sink is grounded, so if all else was good, and you attach them to the heat sink again with faulty insulators, it will short the power supply to ground every time you power up.

As for the power supply caps not having sufficient voltage rating.....once you've found the rest of the damage on the PCB, you'll want to replace those. I'm guessing the lead spacing is 10mm, snap-in style, though might not be. Looks like you have some space around them, so in the replacement, you could probably go one size larger in diameter to give you more selection.

[dmeek]

Not sure how they get 200W from a +/- 35V power supply. Can you post a photo of the foil side of the PC board?

[plugbark]

Have you tested the output transistors for shorts? Regarding the capacitors, I think it is possible for a device to test OK with a meter but fail in the circuit because the meter uses very low voltages to test. Also, if you tested one out of the four caps you may have happened to pick the good one.

Sometimes with a circuit that is shorting out I have been able to put it on the light bulb limiter and see some very small AC or DC voltages and look for any voltages that seem lower than the rest to try to find the short. Also check for things heating up.

If you pull those internal fuses do you still get a bright bulb? I am wondering if those are on the DC rails since they aren't right next to the diodes and caps.

I desolder both MOSFET power transistors (= output transistors) of one PCB: it was indeed these shorted MOSFETs that caused the overcurrent which blew the fuses. Before replacing them I guess I have to verify the voltage on the gate and drain/source pads to validate these parts of the circuit.

[glebert]

Not sure how they get 200W from a +/- 35V power supply.

Instantaneous Peak Power? Or just making stuff up?

[J M Fahey]

200W no way.

With +/-40V you can get 100W into 4 ohm and that is a common rail voltage in MI amplifiers.

100W RMS can be written ad 200W (Peak Power)) and is probably the case here.

[g1]

I'm concerned about the 40V showing on 35V caps. Especially if running on lamp limiter.
Can this be a meter that can't correctly discern between AC & DC and lots of ripple?
Or a unit designed for 120V plugged into 240VAC power?

[J M Fahey]

From OP:

2. Transformer: Secondary winding disconnected from the PCBs: 230VAC (France) -> 30VAC measured

Check.

marked 220VAC (formerly in France) -> 27VAC.

check

3. DC voltage: ±40VDC after the diode bridge (while the 2200µF capacitors are only 35V DC rated!)

Semi-check.
27+27VAC means some 37V DC **using 220VAC mains** and are ... um ... "usable" ; at least for sime time so you can sell it andit might even go ost 6 Month or 1 Year warranty.
I have seen such things many times in dirt cheap products.
It "works" only because modern parts are good and usually have some reasonable safety margin built in.

If, you overvoltage caps by, say, >30% they will quickly die in an explosion or at least filling chassis guts with stinky corrosive fluid, buy my exparience is 10% , maybe up to 15% extra just shows up as premature cap drying up , which will be really annoying maybe a couple years later, well beyond warranty.

But inbthis case main culprits must be dead power transistors, most probably their drivers, and associated resistors.

Please post a list of transistor codes and board closeups, maybe split in 2 or 3 pictures, so values are readable.

The amp itself is a generic one made by the millions by unknown subcontractors and sold under hundreds of brands.

Tech data fully unavailable.

[plugbark]

Can this be a meter that can't correctly discern between AC & DC and lots of ripple?

My multimeter is an Extech EX330, there are separate VAC and VDC modes.

Or a unit designed for 120V plugged into 240VAC power?

Based on the transformer marking, the unit seems to by designed to be plugged on a 220V wall outlet.

Please post a list of transistor codes and board closeups, maybe split in 2 or 3 pictures, so values are readable.

Transistor codes:
IRFP240 (N-channel)
IRFP9240 (P-channel)

I can't give the TO-202 transistors codes because the marking is covered by green (N-channel according to the silkscreen) and violet (P-channel) paintings...


Closeups:

[J M Fahey]

1) weird PCB layout

2) looks European, although it might be manufactured anywhere

3) about 20 years old .

4) in principle those power MosFets are shorted or at least compromised, replace all.
As suggested above, use fresh mica and grease.

5) drivers might be fine, but for added peace of mind, get some and be ready to replace them too.

Prior to that, confirm voltages, that they are actually dead, and, most important, **pinout** since available replacements will use another case.

I suggest BD139/BD140 so get 2 pairs, just in case, but again: case is different, pinout may also be different.

6) to test functionality and pinout:
* through bulb limiter, turn amp ON,
* NO power transistors/Mosfets connected to the board,
* first check you have power rails (+/-40V or thereabouts)
* start measuring voltages at all unknown transistors, pin to pin, both ways.
Voltages will both tell you :
- correct pinout
_ whether said transistor is open-shorted-etc. .... how´s that?

this is the generic TO202 pinout, using numbers but nothing else, voltages will tell us which is which.



disregard the bottom right table which applies to SCRs and Triacs.

In a bipolar transistor I expect full VCE voltage so here about 40V and about 0.6 VBE in a healthy , powered transistor,so for an NPN transistor: +40V CE and +0.6V BE ; for PNP transistor -40V CE and -0.6V BE

Yes, it´s that simple, faster and more precise than possible 6 "Diode tests".

Also it will show 0V CE in a shorted transistor (and probably smoke in any series resistor) .

IF transistors are shorted, then pull them and remeasure, again with amp powered.

Voltage between CE pads will tell you polarity (NPN/PNP)
Voltage between BE pads will tell you bias.

**REVERSE** BE voltage will tell you that transistor may be good, but there is a short somewhere else on the board and NFB is desperately trying to compensate for that ... unsuccessfully.

As you see, functional/voltage testing is better and faster than pulling parts and measuring them trying to *guess* the bad one ... wonder why I meet so much *resistance* (pun intended) when I suggest that.