Would it be possible to use 2SA1386A and 2SC3519A transistors for Q18 to 21?
I'm building a leach amp and I still have 8 pairs of the Sanken transistors lying around from an old (broken) amplifier.
I think it should work and give good results, any reasons why I shouldn't try this?
Thx!
I'm building a leach amp and I still have 8 pairs of the Sanken transistors lying around from an old (broken) amplifier.
I think it should work and give good results, any reasons why I shouldn't try this?
Thx!
I see from figure 4 on Leach's website that the THD is quite high at 20KHz. Is there a reason why you want to build one?
The Leach amp was very influential, but several decades have passed since it was state-of-the-art.
Ed
The Leach amp was very influential, but several decades have passed since it was state-of-the-art.
Ed
I have to build it for a study assignment, several decades have past since my teacher was state of the art also sigh🙄
I built my Leach Amp with fast, modern power transistors NJL3281/1302. The Leach Amp was designed about 1/2 century ago and had to work with power transistors available at the time, that are much slower than what we have nowadays. Its negative feedback network was designed to have a crossover, so that the driver and the final stages are not encircled in the feedback loop at higher frequencies, in order to ensure stability. This means, however, there is intentionally less loop gain available to reduce the distortion produced by the driver-output stages at higher frequencies, where loop gain is most wanted.
I found that with modern power transistors that have high Ft, such crossover arrangement has become un-necessary. I have since simplified the feedback network in my build to have the feedback loop encircle all stages, and the amp would maintain a healthy stability margin (in simulation). I also experimented with Transitional Miller Compensation feedback topology with my build, with success. It seems to me that fast, modern power transistors greatly open up the potential of the Leach Amp.
I found that with modern power transistors that have high Ft, such crossover arrangement has become un-necessary. I have since simplified the feedback network in my build to have the feedback loop encircle all stages, and the amp would maintain a healthy stability margin (in simulation). I also experimented with Transitional Miller Compensation feedback topology with my build, with success. It seems to me that fast, modern power transistors greatly open up the potential of the Leach Amp.
One could argue that’s it’s not a Leach amp anymore when those mods are made. Taking high frequency feedback earlier reduces the output impedance of the VAS at high frequencies as well. Driving an output triple from a lower impedance reduces distortion, including crossover distortion (because charge suck out can be made to happen faster). So the HF crossover distortion is not as bad as one would think. It doesn’t work anywhere near as well with an EF2 where one needs all the gain the OPS can muster.
High fT (that is, 20 MHz) output pairs were indeed available at the time. Just NOT from Motorola. You could pay out the wazoo for Sanken or Hitachi if you were importing directly. Toshiba hadn’t introduced theirs yet (their TMe process was only good for 4-5 MHz at the time, aka 2SD424), and until distributors like MCM came along, GOOD LUCK GETTING THEM.
High fT (that is, 20 MHz) output pairs were indeed available at the time. Just NOT from Motorola. You could pay out the wazoo for Sanken or Hitachi if you were importing directly. Toshiba hadn’t introduced theirs yet (their TMe process was only good for 4-5 MHz at the time, aka 2SD424), and until distributors like MCM came along, GOOD LUCK GETTING THEM.
Of course there a better designs nowadays, but even in the original setup the Leach is still quite a pleasant amp to live with. THD figures aren't telling everything.
Last edited:
I retrofitted a dead Hafler DH220 with the version 4.5 Leach board and 4 pairs each channel of the NJL (I think? -- the 5 pin version with the built in bias diodes) and cranked up the bias. Verified with scope that no oscillation is there with the original frequency compensation values. Currently driving a set of MMG and I have to say I agree with maudio, it probably is not the best but really very nice to live with -- nothing glaring and it seems with the higher bias the bass got better.
The THD will be quite a bit lower at normal listening levels.
I read Leach's article a long time ago. The Leach Amp was the inspiration for my amplifier design.
The Leach amp underwent several revisions. Leach moved away from his original approach - wide open-loop bandwidth (>20KHz) to a more conventional integrator (dominant pole < 20KHz). His goal was stability, but the version 4.5 amplifier does not have enough loop gain at 20KHz to achieve low THD. Taking high-frequency feedback from the pre-driver removes the biggest source of distortion (the output stage) from the feedback loop.
In contrast, my (unpublished) amplifier has sufficient loop gain to keep distortion low at 20KHz. It also remains true to Leach's original goal of wide open-loop bandwidth. I used the same MJ15003/4 transistors as Leach, but modern 2SC5200/2SA1943 transistors would enable even higher loop gain.
I personally feel the wide open-loop bandwidth approach has merit. It does not reduce THD by enough to win the THD wars, but the THD can be made inaudible. My wide-bandwidth amplifier sounds identical to any modern integrator amplifier.
Ed
I read Leach's article a long time ago. The Leach Amp was the inspiration for my amplifier design.
The Leach amp underwent several revisions. Leach moved away from his original approach - wide open-loop bandwidth (>20KHz) to a more conventional integrator (dominant pole < 20KHz). His goal was stability, but the version 4.5 amplifier does not have enough loop gain at 20KHz to achieve low THD. Taking high-frequency feedback from the pre-driver removes the biggest source of distortion (the output stage) from the feedback loop.
In contrast, my (unpublished) amplifier has sufficient loop gain to keep distortion low at 20KHz. It also remains true to Leach's original goal of wide open-loop bandwidth. I used the same MJ15003/4 transistors as Leach, but modern 2SC5200/2SA1943 transistors would enable even higher loop gain.
I personally feel the wide open-loop bandwidth approach has merit. It does not reduce THD by enough to win the THD wars, but the THD can be made inaudible. My wide-bandwidth amplifier sounds identical to any modern integrator amplifier.
Ed
Leach is an oldie but goodie, I believe a renowned audio magazine used it as a reference amp, I made the old super, and a newer version with the original leach front end and modern output transistors, worked like a charm and it saved alot of cabeling to to3 transistors, which was quite a nightmare
Attachments
I agree completely with the last posts.
A reliable, solid and good sounding amplifier...
It means a lot to me - with this amplifier, I restarted the DIY audio that I had abandoned since childhood... I even managed to contact the professor by e-mail - a nice person remained in my memory.
I completed the first Leach 4.5 in 2009, the second in 2011.
Both still perform well and I like the look of the TO-3 transistors.🙂
A reliable, solid and good sounding amplifier...
It means a lot to me - with this amplifier, I restarted the DIY audio that I had abandoned since childhood... I even managed to contact the professor by e-mail - a nice person remained in my memory.
I completed the first Leach 4.5 in 2009, the second in 2011.
Both still perform well and I like the look of the TO-3 transistors.🙂
Attachments
Here is mine, not a Leach amp but an Ed amp: http://www.edgrochowski.com/electronics-museum/stereo4.html
The symmetrical layout reveals its origins. 🙂
I estimate 0.015% THD, and one-third that at normal listening levels. That is low enough to be inaudible. Most critically, my amplifier has a dead silent background at any volume setting.
Ed
The symmetrical layout reveals its origins. 🙂
I estimate 0.015% THD, and one-third that at normal listening levels. That is low enough to be inaudible. Most critically, my amplifier has a dead silent background at any volume setting.
Ed
Last edited:
In a young age pre internet I spend hours upon hours in the public library reading hifi magazines, and the leach article is what got my addiction started, it was a monster back then .
Several years later I build one, before that I had consumer audio, but the good stuff, luxman 430 was my first, the I got a setup with marantz esotec sm6 driving my jamo concert eight and subwoofer were stereo each with 4 8 inch woofer in sealed cabinet run by a luxman m4000, turntable was a transcriptor skeleton riaa was a creek, it had this setup at age 19.
But still the most warm feeling is about the leach.
Several years later I build one, before that I had consumer audio, but the good stuff, luxman 430 was my first, the I got a setup with marantz esotec sm6 driving my jamo concert eight and subwoofer were stereo each with 4 8 inch woofer in sealed cabinet run by a luxman m4000, turntable was a transcriptor skeleton riaa was a creek, it had this setup at age 19.
But still the most warm feeling is about the leach.
To take the topic further off track -- I met Prof Leach in 2009 when my son was looking to go to Georgia Tech in Engineering. He was very graceful and we chatted an hour or so. He was not condescending at all and was open to discuss any ideas. A gentleman and a scholar.
Back on topic, since I used the NJL1302/3281 pairs with my ver 4.5 build with no issues, I think your using the Sanken transistors should have reasonable chance to succeed.
Back on topic, since I used the NJL1302/3281 pairs with my ver 4.5 build with no issues, I think your using the Sanken transistors should have reasonable chance to succeed.
Built mine over 25 years ago and it is still in daily use. It's built to original specs but I omitted the overcurrent protection and added a dc servo, so I could bypass the electrolytic cap in the feedback network. It blew away quite a few other commercial 80's and 90's amps that we compared it to, Rotels, Nad, Arcam, Marantz etc.
It's been incredible reliable, never opened the case since it was built. It has proven rock stable in even the most demanding loads such as big plate stator ESL's.
It's been incredible reliable, never opened the case since it was built. It has proven rock stable in even the most demanding loads such as big plate stator ESL's.
It seems the leach amp still means a lot to a lot of people 🙂
It seems to work with the sanken transistors. I only tested it with a load resistor and an oscilloscope. I have some ringing with a square wave and there is a dc offset of 190mV on the output, which lowers to 90mV with load. There is also a 20mV dc offset on the input. Any ideas about the ringing and offset?
I want to use the leach amp as a power amp for a guitar amplifier. For the preamp I would use a DR. Trebor preamp by PCBMania. This board works with a power supply from +9 to +15V. I have two options to do this:
Thx!
It seems to work with the sanken transistors. I only tested it with a load resistor and an oscilloscope. I have some ringing with a square wave and there is a dc offset of 190mV on the output, which lowers to 90mV with load. There is also a 20mV dc offset on the input. Any ideas about the ringing and offset?
I want to use the leach amp as a power amp for a guitar amplifier. For the preamp I would use a DR. Trebor preamp by PCBMania. This board works with a power supply from +9 to +15V. I have two options to do this:
- I could use a buck converter to convert the +49V from the main power supply to +15V. Would this lead to an imbalance in the + and - rail of the main power supply? The main power supply has 10000µF capacity, so my guess is that this is enough for the leach amp and the preamp.
- The other option is to use a separate 9V power supply and connect the grounds of this power supply and the main power supply. Is it OK to connect the grounds of these power supplies via a star point grounding?
Thx!
Post a scope photo. A small overshoot is okay. Ringing should not happen. 190mV offset is high. I don't know how much effort you want to put into debugging.
Both power options will work. I would use a discrete transistor linear regulator to reduce 49v to 15v. The current draw should be negligible.
Ed
Both power options will work. I would use a discrete transistor linear regulator to reduce 49v to 15v. The current draw should be negligible.
Ed