MJ15012 Replacement

[uliano]

I have this 40 amps current amplifier, schematics refer only to half of it, there is another bank of bjts. I had a led signaling unbalance in the banks so I opened it and I found fuse F8 blown and T5 shorted.

The amp is considered obsolete and irreparable by the factory but as this is just a little part of a very expensive scientific instrument (a new replacement would be several hundred of thousands) I would give it a try.

I've seen that the BJT is (unsurprisingly) obsolete too.

Apart from the obvious question about finding an equivalent (in the same package) still in production I'm worried about replacing only one BJT in the PNP side of this bank.

Can it be that some not perfect match between different part can lead to disaster or, at least, future troubles?

[Kleinstein]

The transistors are effectively in parallel. Much of the resistance for current sharing is from the fuses - so a bit hard to tell how much resistance is actually there.
It helps if the transistors are reasonably matched, to have a similar BE voltage at the operating current. Because of the extra diodes it would also help to have a resonable similar hfe to be on the safe side.
How critical it is also depends on the voltage. The higher the voltage drop on the transistors and the smaller the fuse resistance, the more critical the matching.
It can make sense to replace the whole bank.

Replacements to look at are high power audio transistors. ONSemi is still producing a few similar types.  As such transistors are prone to be sold as fakes, make sure to get them from a reliable source.

[uliano]

I would choose MJ15025 which have most characteristics matching, "major" difference is DC current gain hFE (MIN) :

MJ15012
20 @ IC = 2Adc, VCE = 2Vdc
15 @ IC = 4Adc, VCE = 2Vdc

MJ15025
15 @ IC = 8Adc, VCE = 4Vdc
5 @ IC = 16Adc, VCE = 4Vdc

the rails are +/-60 V

I understand that, to be on the safe side, you'd suggest to change all 10 BJT on PNP side, right?

[Kleinstein]

The MJ15025 should be OK up to some 70 V. This would mean it is OK with a more resistive load, but it could be an issue with an highly reactive load so that the transistors would see more than 70 V while driving a high current.

Yes ideally one would replace all 10 PNPs. The question is how similar the BE voltage is between the types. One could try just replacing 1 transistor and check with a weaker load if the current sharing is good - it may be ok but could also well be different enough that one would need to change the rest too.

[uliano]

Oh the load is definitely inductive: this is driving a coil to create a magnetic field gradient in an old MRI instrument.
However the VCE of MJ15025 is exactly the same as the original, so I wouldn't be particularly scared about that.

So, I decided to change all the 10 PNPs as they aren't particularly expensive ~$5-$6.

Next problem I could face is how to de-solder all those guys, for just one I could have optimistically hoped to manage with just a 80W solders tip and an hand-charged mini vacuum pump, but I'm afraid that all those BJT firmly connected to the huge heatsinks can wreck havoc with the de-soldering process. Maybe it is also better to look for a more professional de-soldering device?

[Kleinstein]

The desoldering should not be that bad. It is just 10 x repeat of the same process. The heat sink is attached to the collector side, the pins are for base and emitter.

The weak points is not so much the voltage rating, but the SOA curve. So how much current the transistor can withstant at something like 100 V or 80 V. Up to some 70 V they should have a similar limit from the power. For higher voltage the MJ15025 has a lower limit from 2nd breakedown. With the old transistors this change over is supposed to be at some 100 V.
The SOA limit is shown for DC and the transistors may withstand more for a short time, as this a thermal type of breakedown.

[wraper]

I would not mix different transistors. NOS MJ15012 are still available on ebay. I would avoid Chinese sellers and listings with pictures showing bright blue glass insulators. Nearly all counterfeits look like this.

[floobydust]

For the remaining power transistors, I would measure V_BE and V_CB on Diode Test and compare to see how matched they are, or if some are damaged.
I do this with big audio power amplifiers to check if the other parts line up.