Possible counterfeit: TSD4M250AV

[guilleDMUPM]

Hi everybody!
I´m repairing a big ClassD amplifier from the 90s and I needed to replace four old TSD4M250V mosfet isotop modules. I tried with modern parts (IXFN140N20P from IXYS), without fails or smoke troubles, but they not working correctly (the control board says "overcurrent fault", probably because they are giving more output current than spected). Now I bought the (supposed) originals from https://www.IC-Components.com and they arrived. They look too modern, too shiny and have with a totaly different aspect. My first impression was of course that they are counterfeit parts. But I really don´t know when ST stops the production... Can they be from another batches from, i.e., the 2000s or something like that?
I am afraid of putting them in the amp (a 5kW specimen) and create a total disaster burning lots components.
Can you help me? Are they fake? Can I use another equivalent part?
Many thanks in regards!

[Peter BENSCH]

Welcome @guilleDMUPM, to the community!

Just how trustworthy the dealer's product website is can be seen from its description, in which it refers to the device as an IGBT. You were almost certainly sold counterfeits because these devices have not been produced since the early 90s of the last millennium. Since you can't refer to any data sheets for such counterfeits and can't even estimate the behaviour, I can only advise against using them in such an infernal power machine.

Unfortunately, I am not aware of a direct replacement. At first glance, the IXFN140N20P appears to be a pin-compatible replacement, but it differs significantly in some respects:

• Gate Threshold Voltage: 2.5-5V instead of previously 2-4V
• Rdson max 18mohms instead of the previous 14mohms

Incidentally, the power amp MOSFET in a Class-D amp cannot supply too much current, but at most be driven incorrectly or insufficiently. The gate threshold voltage in particular, which may be at the upper end of the IXFN140N20P, would then lead to inadequate switching of the MOSFET. However, the gate resistor also plays a decisive role and must be adapted to the parameters of the MOSFET for such power classes. However, without detailed schematics of the output stage and data sheets for the ICs, which have presumably also been discontinued, it is difficult to make a recommendation.

Regards
/Peter

[guilleDMUPM]

Many thanks for your answer and your time. I´ll think what to do, but for the moment, should i return the fake chips? They cost fifty bucks each one, this is not a lot in the context of my company, but surely i want to do something against that pirates and their practices. Do you recommend me to make a complaint?

[Peter BENSCH]

The lack of marking alone is something that should make you suspicious. You can ask the supplier for proof that the delivered goods are original goods, e.g. the unique "BulkID", which STMicroelectronics can use to determine the origin, or a proof of origin. I am quite sure that they will either ignore this or offer a refund of the purchase price.

Regards
/Peter

[guilleDMUPM]

There is no a single thing that should not make me suspicious...
In fact, i did not want to buy them, but it was a desperate last move.
I will tell my experience in forums and customer feedback platforms, because IC Components has good reviews in general and people need to be careful.
I appreciate so much your support and don´t want to waste much more of your time, but i want a last favour: can you indicate me a good substitute for the TSD4M250V? Or failing that, some indication about the most critical parameters i should look at. As you mentioned, equal gate threshold voltage and equal max RdsON?
I can do some reverse engineering to try to know the topology of the amp, if that could clarify things. But this will probably exceed your patience, and i will be happy just knowing your "drop-in" candidate(s).

Thank you very much again. I've never seen such a good support as yours.

[Peter BENSCH]

Well, as mentioned, it will be very difficult to find or adapt an alternative without schematics. The performance class requires close consideration of all relevant parameters in order to assess the switching behaviour and draw conclusions for the replacement. If the MOSFET is controlled by a DSP or similar programmable device, reverse engineering the hardware is probably not sufficient either.

The risk of destroying the MOSFET is also too great for simple testing in this rather high performance class of the amp. I am not aware of any other MOSFETs in this package anymore, but theoretically you could get the IXFN140N20P to work by analysing and measuring how the gate is controlled (timing, voltages) and, if necessary, increasing the driver voltage for the gate. You should also determine how the overcurrent fault is detected and find out the reason for this while using the IXFN140N20P.

In the first approach, you could assume that you can replace this fat MOSFET with several MOSFETs connected in parallel, although you will not achieve the thermal resistance of the original one. However, if you can get hold of the amp's documentation, there is potential for success by adapting the output stage. The only question is whether it would be worth the effort.

Good luck!

Regards
/Peter

[guilleDMUPM]

Thank you one more time! This forum is better than Reddit! hahaha :)
We have limited time and probably studying all the circuit to infer a schematic is too much for us. I´ll look for new mosfets, including your idea of using some in parallel (TO-247 probably), or make deeper investigations about the origin of the overcurrent fault.
In any case, thanks for the support!

[Peter BENSCH]

We haven't even talked about the reason for the failure of the previous MOSFET. It is possible that the replacement from IXYS would work, but the reason for the failure still exists and you therefore get an overcurrent fault?

[guilleDMUPM]

You are right...

The failure implies burned power mosfets, big flyback SOT-227 diodes, other clamp / snubber diodes, and even some resistors... a good mess, but with no signs of having reached the "preamp" or the BJT driver. My suspicion is that the failure was due to excessive heat somewhere and a chain reaction blown that components. I have substituted all, with a smokeless turn on that seemed successful at first. But the problem is that the failure LED can means "overcurrent" OR "module failure"... so i I can't conclude anything in my favor... this is the reason i am dropping-in components and "taking shots in the dark"...

[guilleDMUPM]

So yes, the IXYS may be innocent...