The power of Leach


I'm going to build a Leach Amp with a 84VCT transformer for 2 channels, and wonder how many watts this will give in 8/4/2 ohms. I'd also like to know if I'd want more than two pairs of mj21193/4 transistors per board, and how many farads of filter caps to throw in.

tnx in adv,


I am finishing up my Leach Amp which also has an 84VCT transformer. I am expecting about 125W into 8 ohms; double that for 4 ohms. The protection circuit will limit the output into 2 ohms to protect the output transistors, but I'm not sure about how much power it will push into 2 ohms.

As for the caps, I'm using 50,000uF split between three channels. If you dont have any caps yet, take a look at He has some surplus 26,000uF caps at a great price. I just ordered a bunch from him. 52,000 should be plenty for 2 channels. At that price you can afford to buy some extras just in case anyway.

I don't know about adding more output transistors to the circuit. If you want to do that, you should probably go up to the Leach SuperAmp.

Hope this info helps,

The MJ21193/94 (and -95/96) are more rugged than the original
MJ15003/4; you should be fine with the supply voltage you plan to use and no additional pairs will need to be paralleled unless
you expect to operate at high power into sub-4 ohm loads. In
that case I would also recommend using a separate power supply transformer for each channel unless it's already got a very high current rating.

The more capacitance the merrier, though turn-on surge above
30,000 uF per rail can endanger your bridge rectifiers. Leach's
design has a high power supply rejection ratio (PSRR) and doesn't
really need very high power supply capacitance.

I'm using the MJ21193/94 and ver 4.5 boards with a dual mono chassis configuration and 60,000 uF per rail for a total supply capacitance of 240,000 uF. I use a double-throw center-off AC
switch so that in the "Start" position the current goes through a
25 watt 6 ohm resistor to limit inrush current. It's worked well for
20 years (I rebuilt the amplifier last year for the 4.5 boards).

The new board is audibly superior to the older 2.something boards I'd been using, even though the circuit design really hasn't
changed a lot.

I've been thinking about using a massive forced-convection heatsink I've had for years to build a 500 watt/channel amp based on Leach's board and five output pairs of MJ21193/4s, but
I haven't gotten to the prototyping stage. Some modifications
to the basic circuit may be needed for the higher rail voltages; I
wanted to avoid the complex 'totem pole' design of Leach's "SuperAmp".

>I ordered some of those transistors at no cost as samples from OnSemi through their samples program. They are in a standard TO-3 style case and not in the TO-247-like TO-3BPL.

Well, they are not easy to find. Today I realizer there are two 21193/4 types: MJ which is TO3 and MJL which is plastic.

Specs are quoted as the same, but the metal type is probably more rugged.

Can you tell me the URL to go ask for samples?

skaara said:
2ohm = somewhere around 300w, maybe little more, but it wont sound good, because current will be limited..

Leach just changed the values of the protection circuit resistors to increase the current before limiting kicks in. Looks like full power into 1.5ohms should be possible, as long as SOA and heat is accounted for. With +-50VDC rails and a resistive load, Circuitmaker gives me about 420W at the onset of clipping into 2 ohms, 230W into 4 ohms, and 125W into 8ohms. You could expect similar, perhaps slightly higher powers due to the slightly higher rail voltages.

According to my Circuitmaker2000 sims (with 3 output pairs of MJL21193/4), the current version of the LeachAmp will drive a 1.25ohm resistive load to clipping before the current limiter kicks in. To increase the current further, increase the values of R37-R40. But beware the SOA and thermal dissipation of all output devices - drivers and predrivers included!


Mark Broker
Thanks alot for all the replies!

My transformer is 800VA (actually it's only 780 cause 20VA is for a lower voltage). This means it will be good for 200wpc, but start to push it's limits around 250, right? I'm planning to use the amp for 8 and 4, but probably not for 2 ohm loads, so I guess it'll do. This also seems to be near the limits of my double pairs of transistors.

For filter caps, I'll probably go for something like 30mF per supply rail. To have better channel seperation there should be different rectifiers and filter caps for each channel, right? For a stereo amp this would mean four seperate rails. Am I correct?



Retired diyAudio Moderator
2002-01-21 2:42 pm
near Atlanta, GA
63v caps will work just fine. Most 63v caps have a max surge of 75v. You could even get away with using 50v caps if that was all you had. Dr. Leach has been using 50v caps with his leach amp everyday for the last 25 years. 58v into a 50v cap isn't necessarily a bad thing, although I would imagine that it would decrease the lifespan on the cap.

On Semiconductor has also been known to screw up their sample orders. Last time I ordered some 10 MJ15003 samples, I opened the box and found that they has sent me a flat of 100 :D

I also odered some small single transistors for the Leech, instead of 10 of each, I got boxes of 5000 :D :D :D , made matching real easy. Not sure what I'm going to do with 10,000 transistors though.
Power of leach

I would strongly advise against 63v caps, the idle voltage will most probably be higher than 59V (which will sag under heavier load of course), possible higher than 63v; Dr Leach may be lucky, but it is a bit like saying you know somebody who was a smoker and died at 96 years of age, therefore smoking is harmless (and no, I will not let myself be drawn in a discussion on smoking, just an example!!), so the life of the caps will be very much lowered because there will be more leakage and also more heat due the extra leakage, you loose on two fronts, so keep the 63v for the next project, get 75v caps (if you ever had to clean an amp after an exploded electrolytic, even a small one, you would be convinced).

Howzat for a single-scentence post:cool:

Jan Didden