Some time ago I found a schematic by Pavel Macura. I cannot find the exact thread for original though. As far as I know, Pavel did not build this amplifier. Well, I did, and I have to admit, it sounds and works great. I attached schematic of my amplifier. I did not use a lot of capacitance, for stereo only 8x680uF per rail and 4x680uF close to mosfets (one for each voltage rail), but it is really hum free. 47uF soldered to opamps pins. Voltage is around +-36V, unknown trafo, seems to be kW or more, really heavy and big, it barely did fit into old PC's case. 10uF input capacitor, 680uF for NFB. I soldered this in P2P way, at first one channel, and when I powered it up, adjusted resistors - I understood that it sounds fantastic, and I am gonna build second channel too.
It do sound really great, but unfortunately I do not have proper equipment to prove that. I simulated this amp, and TINA-TI says - Harmonic distortion - 8.7178E-0.06% at 1V input @10kHz.
I attached original schematic, the one I built, and amplitude/phase graphs. Also, I hope that my English is understandable 🙂
It do sound really great, but unfortunately I do not have proper equipment to prove that. I simulated this amp, and TINA-TI says - Harmonic distortion - 8.7178E-0.06% at 1V input @10kHz.
I attached original schematic, the one I built, and amplitude/phase graphs. Also, I hope that my English is understandable 🙂
Attachments
Not really, fused rails, and now I use back-to-back series capacitors which protects speaker from DC (this is only for testing phase, because I might leave room, and I don't want to see burnt speakers when I come back). I know this is not the best kind of protection. Not too reliable. I never have used any output protection, because I have only used:
A)chipamps (lots of versions) and I never pushed them too hard. None of them failed.
B)Single-Ended class A amps, cap at output, few have failed, but never did damage.
C)Testing some different circuits - using pair of old speakers, and capacitor in series, if needed.
I agree - protection is required, because something might fail even for best amps, but, I am risking. My bad 🙂 But I am planning to build one, with relays indeed, similar to one in ESP website
A)chipamps (lots of versions) and I never pushed them too hard. None of them failed.
B)Single-Ended class A amps, cap at output, few have failed, but never did damage.
C)Testing some different circuits - using pair of old speakers, and capacitor in series, if needed.
I agree - protection is required, because something might fail even for best amps, but, I am risking. My bad 🙂 But I am planning to build one, with relays indeed, similar to one in ESP website
Last edited:
Wow, I have completely forgotten this one 🙂
I later built something similar with error correction, that was inspired by Nelson Pass:
PM-AB1 error correction amplifier
I later built something similar with error correction, that was inspired by Nelson Pass:
PM-AB1 error correction amplifier
Pavel, that's indeed similar. Anyway, you can be sure that even basic version is great, I can't even decide, which one suits me better, PowerFollower, or this one, with the power I sometimes missed, despite the fact that 90% of listening time I used maybe 4wpc, or less.
Now I can feel the drum riffs of deathmetal, not only hear them 😀
What can I say, it sounds excellent, bias adjusts fine, it is stable. And it is simple, no problem to P2P it at all. Thanks to Pavel for schematic.
Now I can feel the drum riffs of deathmetal, not only hear them 😀
What can I say, it sounds excellent, bias adjusts fine, it is stable. And it is simple, no problem to P2P it at all. Thanks to Pavel for schematic.
Hi Artūrs,
thanks for your feedback.
What is the quiescent current per one mosfet pair? You can measure it as a voltage drop across 0.1R source resistors.
The distortion should be actually low and low order, if the quiescent current per mosfet pair is about 200mA and more. Then, you should consider a good heatsink, like the one I am attaching. With low quiescent current, the higher harmonics would appear.
Regards,
thanks for your feedback.
What is the quiescent current per one mosfet pair? You can measure it as a voltage drop across 0.1R source resistors.
The distortion should be actually low and low order, if the quiescent current per mosfet pair is about 200mA and more. Then, you should consider a good heatsink, like the one I am attaching. With low quiescent current, the higher harmonics would appear.
Regards,
Attachments
But what is stability of bias current? 540 or 240 mosfets are not the same as laterals. Do You believe that large heatsink solve problem? Or may be Q2 and Q4 have thermal contact with heat sink?
Tempco of vertical mosfets up to about 150mA quiescent is typically 6mV/C.
This can create thermal runaway with low thermal inertias and low bias condition, under 100mA, a typical figure for AB amps.
A bias generator can be made up of a combination of Vbe multiplier and resistor. The Vbe multiplier should be on the heatsink rather than the output device, otherwise its effect is too steep. I suggest 3k3 from collector to base, and 1k from base to emitter, with a fixed series resistor on either collector or emitter to make up the full bias voltage.
This will give a tempco around (3.3 + 1)/1 x 1.8 = 7.7mV/C, which will be about right for two vertical mosfets in complementary pair. A balance of static and dynamic tempco is needed; this is a compromise.
Hugh
This can create thermal runaway with low thermal inertias and low bias condition, under 100mA, a typical figure for AB amps.
A bias generator can be made up of a combination of Vbe multiplier and resistor. The Vbe multiplier should be on the heatsink rather than the output device, otherwise its effect is too steep. I suggest 3k3 from collector to base, and 1k from base to emitter, with a fixed series resistor on either collector or emitter to make up the full bias voltage.
This will give a tempco around (3.3 + 1)/1 x 1.8 = 7.7mV/C, which will be about right for two vertical mosfets in complementary pair. A balance of static and dynamic tempco is needed; this is a compromise.
Hugh
There will be no thermal runaway in this circuit. Q2 and Q4 of the CCS reduce CCS current, thus also Vbias, with temperature.
Pavel: Current per one pair is: Max - 1A, when amp is cool, with forced cooling (though there is no need for it), and it falls down to about 400mA when there is no forced cooling, and amp has been running for a time. So basically - 200mA per mosfet, when warm. For heatsink I am using aluminum case of "1200W MOSFET amp" for car, which was actually based on bipolars, TIP35C and TIP36C, nothing close to 1200W. It doesn't get really high, just warm.
I am currently using LME49830 mosfet drivers, wired as opamps, because I had no other HV opamps available at the moment. I know, that some might say that I am wasting LME's potential, but I had no other choice, because I wanted to power it up, without waiting for some OPA454 or similar. Now I am not sure if I am gonna even replace them, they sound great.
I am currently using LME49830 mosfet drivers, wired as opamps, because I had no other HV opamps available at the moment. I know, that some might say that I am wasting LME's potential, but I had no other choice, because I wanted to power it up, without waiting for some OPA454 or similar. Now I am not sure if I am gonna even replace them, they sound great.
Artūrs,
this is as expected, the quiescent current should decrease with raising temperature. It is a great idea to use the LME49830. Much better than the OPA454, which is too slow for this application.
Regards,
this is as expected, the quiescent current should decrease with raising temperature. It is a great idea to use the LME49830. Much better than the OPA454, which is too slow for this application.
Regards,
Well - in fact a lack of power of the 'Power follower' and its high distortion at moderate and higher output power makes its utilization limited to very sensitive and high impedance speakers. With the circuit shown here, one could build a 'serious' amplifier. And with total quiescent current of 400mA, as Artūrs amplifier has, it would sound good. The 5532 was only for simulation purposes.
Not wrong, but there are definitely better choices. The 49830 that Artūrs have chosen is great for the reason of high speed, high output voltage swing and sufficient current drive capacity.
Yes, if summer wouldn't be so close, I would have increased bias, though I didn't hear any audible difference between 200mA per FET or 500mA per FET 🙂
Jay, I am not sure if I might hear a big difference between ops, but what LME offers - high voltage swing and current, that was more important for me. I can only hear the difference between bad opamps and good ones 😀 I swapped some generic RC opy for LME49720, and yep, audible difference. But when I swapped OPA1642, OPA2143 - I didn't hear difference, maybe because it took some time to swap them, and it is hard to compare small details in music without proper ABX test. Anyway, LME49830 is great, although I didn't use it as mosfet driver but opamp.
- Status
- Not open for further replies.