Audio Power Transistors - Buying in Bulk

Looking at the data sheets for the parts I would sy that there are definite causes for concern:

The features section of the Multicomp data sheet seems to be for a different part.

Multicomp features from data sheet.

The gain is wrong, the specified current range is wrong, what does "diode protection' even mean ?, there is no base emitter resistor in the Motorola/On Semi parts.

The figure for Hfe is specified more reasonably in the data section fo the data sheet:

There is, as Shabaz has pointed out very little information in the Multicomp data sheet (most of the graphs on the Onsemi data sheet are omitted).

So if the data sheet is both brief and self-inconsistent you have very little idea of what you are buying.

On the way home last night I was thinking about the set up needed to test these parts for performance over a reasonable working range ( maybe not needing to go as far as the full specified rang).

You would need a dual channel SMU or equivalent, capable of 10A (@ 2V) and 140V(@1A-pulsed), and most difficult, 5A at 50V

Some way of measuring the thermal resistance - this is challenging - you need to put serious power into the device. keep the case temperature constant and measure the chip temperature.

I guessed it would take a couple of weeks work to set up a rig to do this (non automated and over a limited number of set points) using a suitable lab power supply and standard electronic components with some kind of low power pulse generator and a scope. Or you could just buy the proper gear but expect no change from £50k.

Is this a good use of time ?

I would think that it isn't because the parts are not well enough specified to be sure that you would be getting a consistent product, regardless of the economics.

The real thing from Mouser cost £4.53 if you buy 100, the Multicomp part is £3.26 - so you save £1.27 on each one and as Shabaz says, almost all your customers would rather have the Onsemi parts.

MK

Looking at the data sheets for the parts I would sy that there are definite causes for concern:

The features section of the Multicomp data sheet seems to be for a different part.

Multicomp features from data sheet.

The gain is wrong, the specified current range is wrong, what does "diode protection' even mean ?, there is no base emitter resistor in the Motorola/On Semi parts.

The figure for Hfe is specified more reasonably in the data section fo the data sheet:

There is, as Shabaz has pointed out very little information in the Multicomp data sheet (most of the graphs on the Onsemi data sheet are omitted).

So if the data sheet is both brief and self-inconsistent you have very little idea of what you are buying.

On the way home last night I was thinking about the set up needed to test these parts for performance over a reasonable working range ( maybe not needing to go as far as the full specified rang).

You would need a dual channel SMU or equivalent, capable of 10A (@ 2V) and 140V(@1A-pulsed), and most difficult, 5A at 50V

Some way of measuring the thermal resistance - this is challenging - you need to put serious power into the device. keep the case temperature constant and measure the chip temperature.

I guessed it would take a couple of weeks work to set up a rig to do this (non automated and over a limited number of set points) using a suitable lab power supply and standard electronic components with some kind of low power pulse generator and a scope. Or you could just buy the proper gear but expect no change from £50k.

Is this a good use of time ?

I would think that it isn't because the parts are not well enough specified to be sure that you would be getting a consistent product, regardless of the economics.

The real thing from Mouser cost £4.53 if you buy 100, the Multicomp part is £3.26 - so you save £1.27 on each one and as Shabaz says, almost all your customers would rather have the Onsemi parts.

MK

michaelkellett said:

almost all your customers would rather have the Onsemi parts.

"No one gets fired for buying IBM" was the IT equivalent for that.

A friend asked for a Bluetooth adapter.. even though off-the-shelf ones only cost £20 or so, I built him a custom one (off-the-shelf Bluetooth board), and even non-technical people want to actually know their product isn't being skimped on cost (this one cost them £100 in parts) and proper external antenna since no-one wants poor signal quality. The filter caps are 105 deg C since this is going to be fitted in a car and the longer life will help. The audio connectors are expensive these days, but I wouldn't feel comfortable fitting the cheaper plugs that become a loose fit over time. The screened cable is really high quality (StarQuad cable originally intended for mics).

The black stuff providing the strain-relief inside the case, is Polydoh! : ) Surprisingly effective.

Next time I win a shopping cart, I'm going to add a few of these. Not because I need some, but for the lab experience.

I agree.

The ONSemi parts are only slightly more expensive than the Multicomp on Newark when you buy 10 pieces. May as well just get ONsemi.

One practical way to test the Multicomp devices would be to take a stereo amplifier and replace the transistors in one of the two channels, then run it for a while to see what happens. Or, run it with a dummy load connected and stress it out. 

There is a pretty good diversity of opinions on here and I think I have all the info I need. Thanks to everyone who chimed in.

The dummy load vs specs approach would be the safest one.

But do we know if the original amplifier would actually have survived a bench test with a dummy load vs specs?

Places where a device with a more restricted SOA might fail are usually not exercised with a dummy resistive load.

But if you try more difficult loads you are into the region of Jan's conundrum where you don't know if the full spec part would have done better.

So to be sure that the parts of equivalent robustness  you would need to test both types to failure - which might be fun but rather time consuming and expensive.

Comparing both channels with different parts is a reasonable compromise but most audio amplifiers are not that well cooled so you may be mainly comparing the susceptibility to overheating - which in turn will be influenced by how well the parts are contacting the heatsink etc. etc.

Please let us know how you get on - and if any parts fail in interesting ways !

MK