Rod Elliot 100W amp suggestions?


2001-06-01 4:53 pm
I'm building it too, and have used substitutes for all of the parts. I'm using 2N3055/MJ2955 output transistors.
Rod's construction notes that he supplies with the project are very good. Use oversize heat sinks if you can.
I've built and tested the modules, and am in the process of mounting into an enclosure along with a headphone amplifier.
I suggest using a 10-turn pot for VR1.


2001-06-01 4:53 pm
Sorry, should have said all the transistors. It wasn't a matter of upgrade, just ones I had available. I don't see too many BD139/140s around, for instance, I think I used MJE340-type devices there.
I do recommend using high quality passives, and possibly bypass (parallel) the electrolytic caps that carry audio (C1, C3, and I think C5) with lower-value polypropylenes, say 0.1 uF. I'll try that next.
Substitute parts for Rod Elliott's Project 3A

I used BC550C/560C transistors for the small signal transistors, 2SB649A/2SD669A for the drivers, a 3.3uF MKT input cap, MKT or polycarbonate bypass caps, and as the Toshiba transistors suggested by Rod has been discontinued, I used their new equivalents (also Toshiba; can't remember the part numbers). And the sound? Fabulous! Simple to build too, too me only an hour to stuff the board and install the heatsink.


2000-12-31 9:31 am
If you go through all the trouble to make your own amp why not spend the additional 20 bucks and buy the recommended power transistors instead of using crappy 2n3055-mj2955 (which by the way are undersized for a 100W amp)?
The design has been optimized enough to really benefit from the exceptional performance of the 1302-3281.
Toshiba and Sanken should be the only types of transistor used in power amp this day in age. Everything about these transistors but in particular bandwidth and hfe linearity is far superior to any of the motorola mj150xx and 2n-whatever available.
Don't take my word for it, all the major audio designers Dean, Clausen, Aloia, Self, not to mention the commercial hi-end brands are using almost exclusively toshiba and sanken transistors.
I emailed Rod and asked him about the 2955/3055 and he said that's actually what he uses in his own amp. If you want high power, especially into 4 ohm loads or are bridging into an 8-ohm load, use the expensive ones, otherwise I would use the others. Cheap output transistors was one of the big attractions of this amp for me. There's enough feedback in this design I bet I would be unable to hear the difference.
An alternative would be to use pairs of output transistors in parallel, I believe Rod mentions this somewhere in the 3a or its predecessor, project 3.

Mix 'n' match: I intend to use this design to drive woofers, a Pass design for midrange, and an IC amp to drive the tweeters. Op-amps in the crossover, just need vacuum tubes somewhere to complete the circle. Maybe I'll just light up some filaments without actually having them in the signal path.

But how expensive can transistors possibly be? The extra power headroom will probably be worth it, not to mention the potential for additional linearity. I guess feedback makes the latter feature less significant. Today, they may be driving the midranges, but tomorrow they might be bridged into a subwoofer.

Thanks, Won
Good point. I haven't checked the 2SCxxx transistors out, but assuming they're the same package you could build the thing with the cheapies, get it working, then plug in the more expensive ones. I do this with opamps.
I'll save the money towards my NEXT project. I always have a NEXT project.
Having blown up some expensive parts, I'm a little gun-shy of them in an untried circuit.
Vivek, I'm not sure what a matrix board is but the amp itself seems very stable and will probably be fine.
Make sure you let the amp warm up for a good while during bias adjustment, temperature affects it.
Note that the MJ3055/2955 pair is cheap but also suffers from poor bandwidth - about 800KHz as compared to the 2SC3281/2SA1302 whose bandwidth is some tens of Megahertz. The MJ/TIP/MJE3055/2955 transistors come in plastic TO220 package and are rated at 90watts Pd, whereas the Japanese ones have a rating of 150watts and come in a larger plastic pack, probably called TOP3. 2N3055E TO3 metal transistors reportedly have a bandwidth of 4MHz but the complimentary may not be available.

The base bias (for a particular hfe) required for the Japanese transistors is 1 Amp and for the 3055/2955 pair it is 4 amps.

Douglas Self's Load Invariant Amplifier showed that the THD produced by 3055/2955 is about 3 times greater than for the Japanese transistors.

Sure, the cheaper one's will work but disatisfactorily. The bandwidth and dissipation factors alone should deter any of you from using the cheaper variety in this project. Ofcourse, it would be prudent to check out the correct operation of the amplifiers with two pairs of the cheaper variety and after a while to replace with better parts. Don't forget to turn the bias setting trimpot to minimum Iq before replacement with better ones and then do the quiscent current setting again.

Having used both varieties in another design, I can vouch that the 2SC3281/2SA1302 output devices sound a lot cleaner and transparent than the alternative cheaper variety.

Good luck all of you trying out Project 3A.
I find that people are talking about 2SC3281/2SA1302 pair at the output. My friend have dicksmith 100w kit amp. He wants some ungradation. They have parallel pair of T03 2N3055/2N2955 at the output. Is it better to replace them with one pair of 2SC3281/2SA1302. Guidance welcome for the experiment.
Mahendra Palesha