Leach Amp: Matching MJ15003 and MJ15004


I'm currently building a Leach Amp and are about to purchase the transistors MJ15003 and MJ15004. Theese needs to be matched, and cost about $3.5 a piece, just about everywhere you can find them. (Well I found them for $6 at www.farnell.com)

My problem is that I've done some measurements in a local shop here in Gothenburg and found that it wasn't easy to find matching pairs of MJ15003 and MJ15004.

So, my question is: Where can I find 5 matched pairs of transistors _cheap_, without buying a truck load of transistors?

Any Ideas?

The question comes up about once a month, and to my knowledge, no one has ever found such a source.
Part of the problem is that the devices should be matched at a particular current. This means that a supplier would need to stock MJ15003's matched at, say, .5A, 1A, 1.5A...whatever. Multiply that by the number of available devices (don't forget MOSFETs) and life gets unpleasant quickly indeed--you'd need a very large warehouse. All for a niche market that wouldn't buy all that many parts. Keep in mind that you're going to have to pay people to sit there and match them, etc.
Your best bet is to match them yourself.
Keep a few matched spares in case of problems later, then use the rest in voltage regulators and whatever else you need. (Ever notice that Nelson uses a lot of IRF610s in other parts of his circuits?)



2000-12-31 9:31 am
I have matched a few MJ15003 and 04 in the recent past, I think that with 10 pnp and 10 npn you should be able to pull it off. I matched mine to within 0.01V over 2.5 VCE and it was pretty easy to come up with pairs.
Sorry for sneaking this in but I have a TRUCKLOAD (for my standards, which means probably 20 each) of motorola, brand new, real deal, MJE15030-31 brand new for which I have no use in the foreseeable future.
If anyone is interested they are for sale.
I've never bothered to match output transistors mostly because
I wasn't certain what current levels to test at. And at any rate
the amplifier seemed happy with random sets. I have a curve tracer rated up to 1 amp at 40 volts, but I normally just use it for signal transistors and verifying that the parts I'm using in a
project aren't defective or marginal.

But I'm more interested now in matching techniques anyone has
to share:

Having recently upgraded my Leach amplifier to the ver 4.5 board
and being >very< pleased with the sound, I'm now going the
extra step of using MJ21193/4 outputs. I will match these
devices to get the best results from their improved linearity.
(That assumes I'm able to actually >get< them; my local supplier
is reluctant to sell in less than quantities of 100) Since I'm using
4 ohm Acoustic Research AR-11s, I hope for a worthwhile improvement.

I wish Motorola had gone ahead and marketed the TO-3 versions of the 2SA1302/2SC3281 that appeared in their 3/95 catalog.
Eventually I plan to build a version of the Leach amplifier using
this family of devices, hoping to construct an amplifier that not
only sounds good, but looks semi-professional rather than a
junk box special. :/

One of my future projects is to try testing various types of
output transistors to see how random and matched sets perform.
Motorola/ONSemi indicates that matched MJ21193/4 pairs have
1/10th the distortion, and they actually specify distortion for these
devices, which most semiconductor manufacturers don't.
So, what I've learned from today's messages in this posting is that if I _just_ want the darn thing up and running, I just have to go ahead and buy a set of transistors, and fiddle with the matching some time after christmas.

Thanks a lot, guys!

I don't know for you, but this forum really helps most of the time. I work as a UNIX tech. and none of the guys at my job is into homebrewed electronics.

Have a good night!


Nelson Pass

The one and only
Paid Member
2001-03-29 12:38 am
Usually the Vbe is reasonably matched even if the
gain is not, and this is the most crucial aspect of
getting the devices to share the load. Even small
values of resistance will do a good job of covering
the equalization, but larger values of emitter resistance
are often necessary to stabilize the bias against
therm drift.