@Brownlow, You shouldn't make anything that looks like a generic Hitachi mosfet amplifier with those transistors, it is a waste of time. Of the sonically good amplifiers of that 80s era, except for Hafler amplifiers with four pairs, I still remember just the Elektor Mini Crescendo with two pairs of output transistors. I recently listened to it at a friend's house, after a number of years of use, and it plays well even for today's terms. I have two pairs SJ50/SK135, I don't know what I would make either, maybe a Tumos amplifier/mosfet follower without voltage amplification from Doede Duma, with a tube SRPP/MU Follower front end/gain stage or something like that.
http://www.dddac.de/op06.htm
http://www.dddac.de/op06.htm
Last edited:
I've had good results with the Hitachi circuit, and am currently modding it to make it even better
I'm still soldering, but the limitations are the MOSFETs going into class B, not the driver section - according to my work in LTspice
. So maybe worth a closer look, it's quite a nice circuit really, well respected. https://www.diyaudio.com/community/...fier-ga28f-construction-thread.257488/page-14
The Maplin board I mentioned above can do 25W class A too - you simply wind up the bias.
After then it goes into clas B rather than clipping, but that's a good thing
I built two amps from Hitachi schematics back in the day, and they sounded mediocre. Also listened to the Magnum MF125, the same schematics, and same thing. Then we did Crescendo and Mini Crescendo, and that was much better. There was only a problem with oscillation, but it was being solved. I ended up doing the Leach LOW TIM and that stayed in the system as the best of all. 8 years ago I did a refresh and new wiring, and I still listen to it.
Perhaps it depends up on which schematics?
I applied some of the Douglas Self suggestions to the Maplin design, and thus added a CCS to the input pair and balanced it a little better, the class A performance (Ok, it's a simulation, but still ) is difficult to fault.
https://www.diyaudio.com/community/...ga28f-construction-thread.257488/post-7635217
I've always liked the sound of these, as have many, but the PSU is important - perhaps with that CCS it will be a little less so. I'm currently soldering up my mods in the above post, so i'll find out in a week or two when it's tested
I applied some of the Douglas Self suggestions to the Maplin design, and thus added a CCS to the input pair and balanced it a little better, the class A performance (Ok, it's a simulation, but still
) is difficult to fault.https://www.diyaudio.com/community/...ga28f-construction-thread.257488/post-7635217
I've always liked the sound of these, as have many, but the PSU is important - perhaps with that CCS it will be a little less so. I'm currently soldering up my mods in the above post, so i'll find out in a week or two when it's tested
my build threads .. the Maplin works a treat have been building these for the last 40 years
https://www.diyaudio.com/community/threads/jlh-mosfet-power-amplifier-construction-thread.323017/
https://www.diyaudio.com/community/threads/maplin-mosfet-amplifier-ga28f-construction-thread.257488/
https://www.diyaudio.com/community/threads/jlh-mosfet-power-amplifier-construction-thread.323017/
https://www.diyaudio.com/community/threads/maplin-mosfet-amplifier-ga28f-construction-thread.257488/
One was a Winton amplifier (from Practical Wireless), with a simple preamp, just one NE5532 (OPA2604 later, much better) without tone controls. That was my student amplifier, made in 1983/4. Phono preamp was with HA12017. The other was some kit, similar schematics, I don't remember exactly which one. I made that to my friend. Maplin I never made or heard.
Last edited:
With the original Hitachi Lateral Mosfets you could try First Watt F7 if you can figure out the schematic; it is not in public domain yet. Another choice would be Fab's USSA-3. Yet another easy one would be Juma's Amp here https://www.diyaudio.com/community/threads/lsk-pre-baf-2013.244106/post-3710957
You could search the net for Randy Sloane's amplifier, like the OPTI-MOS.
I have built and listened to all but the USSA-3. They are all great. I do have the original boards for USSA-3 and all semiconductors for the 3A version. Thinking of getting by soon.
You could search the net for Randy Sloane's amplifier, like the OPTI-MOS.
I have built and listened to all but the USSA-3. They are all great. I do have the original boards for USSA-3 and all semiconductors for the 3A version. Thinking of getting by soon.
Just a thought about a class A amp with these MOSFETs, I guess there is two ways:
CCS and one transistor - like the tube SEs
A regular complementary output ( class AB with the bias turned up.. )
So for +/- 20V swing into 8 ohm (25W RMS) you'd need about 2.5A, so am I correct in thinking you'd need a bias of 2.5A at idle?
With 20V rails you get 40x2.5 = 100W dissipation ?
I'm not sure if this is 100% correct, but it leads onto the point about driving gates
- there's a 10V Vgs drop at the top, so you'd need +/- 30V for a single PSU, which leave 60x2.5 = 150W dissipation, so my advice if going for the class A approach is to run the driver at a 10V advantage, and the MOSFETs are the lowest voltage you can, to keep the dissipation at idle down.
And perhaps a slow 120mm fan, they are pretty silent at low speeds.
CCS and one transistor - like the tube SEs
A regular complementary output ( class AB with the bias turned up.. )
So for +/- 20V swing into 8 ohm (25W RMS) you'd need about 2.5A, so am I correct in thinking you'd need a bias of 2.5A at idle?
With 20V rails you get 40x2.5 = 100W dissipation ?
I'm not sure if this is 100% correct, but it leads onto the point about driving gates
- there's a 10V Vgs drop at the top, so you'd need +/- 30V for a single PSU, which leave 60x2.5 = 150W dissipation, so my advice if going for the class A approach is to run the driver at a 10V advantage, and the MOSFETs are the lowest voltage you can, to keep the dissipation at idle down.
And perhaps a slow 120mm fan, they are pretty silent at low speeds.
I think that with one pair of SJ49/SK134 you should not go more than 10-15W in A class (push-pull output stage).
In push-pull, calculate that the peak to peak current (P-P) is equal to 4x the quiescent current (for 1A Q.C. you have 2A in the plus and 2A in the minus side). So for 1A quiescent current you have effective current of 1.41A AC or 16W RMS/8om - 8WRMS/4ohm in A class. When the 2A peak is reached (plus or minus), one of the transistors closes. Everything above that goes to class AB.
If I were to make something like that, I wouldn't bias it so hard in Class A, let it be a few W for normal listening, and above that let it go to AB class.
For single ended, P-P current is only 2x bias current, no other transistor, and everything always remains in A class.
In push-pull, calculate that the peak to peak current (P-P) is equal to 4x the quiescent current (for 1A Q.C. you have 2A in the plus and 2A in the minus side). So for 1A quiescent current you have effective current of 1.41A AC or 16W RMS/8om - 8WRMS/4ohm in A class. When the 2A peak is reached (plus or minus), one of the transistors closes. Everything above that goes to class AB.
If I were to make something like that, I wouldn't bias it so hard in Class A, let it be a few W for normal listening, and above that let it go to AB class.
For single ended, P-P current is only 2x bias current, no other transistor, and everything always remains in A class.
Last edited: