Measuring hfe on power transistors.

[Diabolical Artificer]

I'm working on a big power amp, a SA900C, one channel works, the other works but not at full power, I suspect the driver board section. Anyway whilst testing my probe slipped and took out a pair of driver Q's. These are 2SA1494/2SC3858, the amp has 5 pairs, one pair as drivers, four as OP Q's.

After a lot of effort I've managed to source some trannies of the correct hfe "rank" - see data sheet. I suspect the ones in the amp are all of the same batch so are pretty well matched. Therefor I need to knock up a testing rig, preferably simple, to match the existing to the new. I have plenty of suitable tfmr's so am able to make a PSU from 20 to 80v, 80v being the amp PSU and have a reasonable stash of parts. Thinking about it I have a variac/ISO tfmr.

Here's a schematic of such a beast from ESP - http://sound.whsites.net/project106.htm , but as I don't really want to make a permanent device, or one to test a variety of hfe's can I not just connect a DUT, resistor on C, R on B and emmitter to -/+ rail. I am worried about protecting the DUT's, as they ain't cheap. LL to limit current or a big R?

Not sure how near I'll need to match them, but would think 5% or better would be about right.

Any ideas/advice most welcome.

Andy.

[Mark Hennessy]

I'm not convinced you need to match them at all.

The amp should have emitter resistors - usually 0.2 ohms or thereabouts. These drop enough voltage across them to "swamp" the small differences in Vbe and promote equal current sharing.

Remember - it's the Vbe that matters, not Hfe. Just look at how much Hfe varies with collector current and temperature. No sane designer comes up with products that require specific values of Hfe.

Even if the sharing was really unequal, a single pair of those transistors would be good for around 200 watts into 8 ohms.

If you want to measure Hfe, in terms of gear you need nothing more than a DC power supply (a bench power supply is best), and two multimeters. A power resistor forms a collector load, in series with one DMM, while a potentiometer and a resistor or two provides the base current (second DMM measures this). This approach gives you complete flexibility about which collector current you choose to measure at. No need to go higher than an amp IMHO.

You don't need to put a high voltage across the transistor - just a couple would do, so that you know you're not saturating it. And your bench power supply will have a current limit (perhaps adjustable), so you won't blow up the transistors. Just keep an eye on their temperature if you're using currents measured in amps - a heat sink might be wise.

But like I said, I wouldn't bother. Just bung them in and test (via a lamp limiter first of all - the PSU in the amp will give a lot more energy than a typical bench PSU). Once all is well, do some power testing and make sure they are all running at about the same temperature. A cheap IR thermometer is all you need for that. Yes, they are probably all on a common heat sink, but the top of the plastic case will be much hotter than the heat sink, and will be a closer approximation to the junction temperature. Job done

Good luck - amplifiers like that are not for the faint-hearted

[Diabolical Artificer]

Thanks Mark, here's a copy of the schematic of the power amp stage, it has as you say emitter R's and from what I've gleaned so far this takes care of any hfe discrepancy.

The amp has protection circuit's and I'll run it up on a variac and LL. I'm pretty sure it's the pre amp section that is the problem but yes it's the biggest amp I've worked on so is a bit scary. I have the working channel to compare it too which helps.

Andy.

[Mike Watterson]

Any push pull Audio amp or RF amp that actually needs "matched" transistors is not a viable mass production item.
The gain varies massively between low level and high power too.

Make sure all bias parts check out OK before powering up and no shorts or leaky caps on output etc.

[Mark Hennessy]

It's worth saying that the protection circuits are all very well when everything is working correctly, but they might not protect new output devices when there is a fault prior to them.

If in doubt, in addition to using a lamp limiter, perhaps try the initial power-up with only one pair of devices fitted (minimises the cost of any blow-ups!). Or even use a "sacrificial" pair of transistors - something that has sufficient voltage rating, but costs a lot less than the correct originals. They only need to work well enough to establish the correct DC conditions are present and deliver a sine wave a low power. Once past that, a full set of "proper" devices should work with no drama, and full-power testing can be done at that point.

Can't see the schematic, BTW. Would love a copy - if you can't get it uploaded on the forum, email it to me and I'll see what I can do... If it's a PDF, we have a very generous allowance on the file size for those.

[Refugee]

That circuit in post#1 looks very like the transistor I made when I was an older teenager in the late 1970s.
It is the right tool for the job. Mine runs on an HT of 10 volts and measures the HFE at a fixed collector current set by a selector and fixed collector resistors and an opamp that flashes a couple of LEDs to aid setting VCE.
All transistors are tested at 5 volts for VCE.
The original collector current is 2ma to 100ma but recently without modification I have run it up to 2 amps with a bench PSU without issue.
It did a champion job of sorting out all those unmarked transistors that were sold in big bags all those years ago.
It still lives and works to this day.

[Diabolical Artificer]

Away with the faeries again Mark, here's the schematic and I'll send you the PDF , I'd be interested in your opinion of it's design.

That's a good idea to try a pair first, the channel works ok at present with 3 pairs of OP Q's. I'm using "Q2 as short hand for transistor of coarse.

Andy.

Oh and here's a link to a video i made - https://www.youtube.com/watch?v=pRKDfStE_E8