STK0040 module in JVC JA-S22 amplifier

I'm trying to repair a JVC JA-S22 amplifier. It's a class AB amplifier with Darlington output transistors. The power transistors are blown in one channel.

The transistors are contained in a module called an STK-0040, which has an internal circuit of this:-



It's TR4 and TR5 that have blown and are shorted collector - emitter.

There are lots of STK-0040 module on ebay, but I am fairly sure that they are all copies. The original STK-0040 has semiconductor dies bonded onto a circuit board, so it would be quite easy to make an equivalent with a very different technique, but it's likely that the copies don't perform properly.

This video ( which is in Spanish) show someone repairing the same amplifier and uprating the transistors on a copy STK-0040.

I have opened up the blown STK-0040 and think everything except the big transistors is working. I think that I could cut the base and emitter connections on the big transistors, and mount equivalent NPN and PNP transistors in T0220 packages somewhere else on the heatsink, and run wires to the module or the circuit board.

So I've got a choice of:-
1) use existing module with the big transistors fitted remotely.
2) use a module from ebay that is probably a copy and hope for the best.
3) get a module from ebay and change the output transistors, as in the video.

Has anyone got experience of these or any suggestions?

These sorts of modules are a nightmare, as they were rarely available for long - but at least this is a fairly 'minimal' module, so less to replace. In the past amplifier manufacturers often produced 'repair kits' to replace a duff full module, with an alternative type - these were often fairly complicated, consisting of a new (different) module, a small PCB, and a good number of components. They usually came with a very poorly photo-copied instruction leaflet.

As the module is so simple, I'd suggest making your own - use darlington transistors to replace the drivers and outputs (four for the price of two ), and a small NPN for the Vbe multiplier. Add a pot in the Vbe multiplier, so you can adjust the quiescent current.

Even if you replace just the output transistors (either for the existing module, or a copy) I'd still add a pot, as it's unlikely the quiescent current would be correct with different transistors.

Well I tried assembling a circuit, and it hasn't worked. I used a pair of TO220 Darlingtons, BDX33C and BDX34C and a TO126 NPN transistor bolted to the same heatsink as TR1.

The problem with that idea is that there is no way of adding an resistor in the R5 position, and the thermal coupling between the two transistors in a darlington pair is far, far better than the coupling to TR1. The quiescent current wasn't stable and varies very quickly when the temperatures changed. There was a thermal run-away when I tried to put some audio through it.

While adjusting the quiescent current, I had a thermal runaway, but I was monitoring the current and I had my hand on the power switch so I manage to catch it that time.

I also couldn't find very generously rated Darlingtons. If I use discrete transistors I have a much wider choice as well as being able to have a resistor in the R7 position. I'll probably use the power transistors in TO247.

Diver300 said:

Well I tried assembling a circuit, and it hasn't worked. I used a pair of TO220 Darlingtons, BDX33C and BDX34C and a TO126 NPN transistor bolted to the same heatsink as TR1.

The problem with that idea is that there is no way of adding an resistor in the R5 position, and the thermal coupling between the two transistors in a darlington pair is far, far better than the coupling to TR1. The quiescent current wasn't stable and varies very quickly when the temperatures changed. There was a thermal run-away when I tried to put some audio through it.

While adjusting the quiescent current, I had a thermal runaway, but I was monitoring the current and I had my hand on the power switch so I manage to catch it that time.

I also couldn't find very generously rated Darlingtons. If I use discrete transistors I have a much wider choice as well as being able to have a resistor in the R7 position. I'll probably use the power transistors in TO247.

Click to expand...

R5 isn't needed, as for darlingtons, TIP142/147 would be fine, and plenty more power than required for a 40W amplifier. The Vbe multiplier mounted on the same heatsink as the output transistors should be fine.

I don't know if the emitter resistors should be larger to use with Darlingtons. The resistors are 0.22 Ohms each.

I had a transistor Vbe multiplier on the same heatsink, but the quiescent current was certainly not stable. There is quite a time lag between the Darlingtons heating and the Vbe multiplier transistor heating.

Are there any particular changes you would suggest to make the circuit stable? I don't want to make what would essentially be the same circuit again if it's just going to run away again.

The original transistors were in a darlington configuration anyway, but you could certainly try increasing the emitter resistors - however, they will be thermal lag between the transistors heating up and the Vbe multiplier receiving the heat.

The Velleman 100W amplifier uses 0.47 ohms emitter resistors:


Schematic at the end of the User Manual.

I have now fixed the amplifier. Here is what replaces the module:-



These are the components that I ended up with:-

The CD13003D and the BD238 are both TO126 size and I used that so that I could attach them to the heatsink. I don't know if that helped.

I had originally used a CD13003 for TR1, so that it could be bolted to the heatsink, and I now think that was what made the previous attempt fail. The effective resistance of the voltage limiter (D1, D2, R1, R2 and TR1) is approximately the resistance of R1 and R2 divided by the gain of TR1. The gain of the CD13003 was far too low, which is why I used the BC182. It is placed in contact with the heatsink with some heatsink compound on it.

I also tried a 2N5551 for TR1, but it didn't work. There was a difference of around 1.5 V between the voltage across R1 and across the fixed 1kOhm part of R2. That showed that that there was about 1.5 mA of base current, while the collector current was only about 6 mA. I can't explain the lack of gain, as both the data sheet and a component tester also showed a gain in the hundreds.

Nigel's suggestion of a potentiometer to adjust the quiescent current was very useful.

I didn't try using darlington transistor, as I was worried that the lack of R5 might make a difference. I did look at the Velleman circuit diagram and it seems to show a separate current limit circuit, so there is less risk of the shoot-through current getting out of hand, so I wasn't confident that it would work without a limiting circuit.

When initially assembling the amplifier, I put 10 Ohm resistors as emitter resistors. Having them meant the fuses wouldn't have blow if some mistake were made.
The repair manual, including circuit diagram is at https://elektrotanya.com/jvc_ja-s22.pdf/download.html which was also useful.