1N4001 diodes works well in this circuit, better than 1N4148, and they have plastic case, can be clued to Q2 and Q5. Thermal contact between D1/Q2 and D3/Q5 gives better thermal stability. No need to use 1N4007 or "something better", there is low currents and low voltages and diodes are always ON like discussed before. C1 for sure is a overkill, can chance to 10µF. When reading last answer I understand, that a 10µF low ESR cap could be a good choice for C1 ?
After more than 12 hours continuous power on amp still keeps output stages current stabile and there is no thermal contact between diodes and driver transistors.
Just ordered some small heatsinks. These one :
FK 230 SA L1 : these are absolutely minimum size and gets quite HOT when in use, but no more than 50 degrees Celsius.
https://www.fischerelektronik.de/we...l;jsessionid=949C1AFF55AC37D249F2A330B7A53017
Maybe not all members knows what is SOA for bjt, so it's good to take a look to Toshibas website.
https://toshiba.semicon-storage.com...for-individual-power-transistors-contain.html
After more than 12 hours continuous power on amp still keeps output stages current stabile and there is no thermal contact between diodes and driver transistors.
Just ordered some small heatsinks. These one :
FK 230 SA L1 : these are absolutely minimum size and gets quite HOT when in use, but no more than 50 degrees Celsius.
https://www.fischerelektronik.de/we...l;jsessionid=949C1AFF55AC37D249F2A330B7A53017
Maybe not all members knows what is SOA for bjt, so it's good to take a look to Toshibas website.
https://toshiba.semicon-storage.com...for-individual-power-transistors-contain.html
And 4001’s have a lower forward voltage drop than 4007’s. Yes, they are just graded for voltage, but they do switch processes between the 4002 and 4003 (or was it the 4004?). So it has an additional “jump” in addition to the usual Vf/vbr correlation. No one misses 50 mV in a power supply application, but here an extra 50 mV might put you out of trimming range.
10 uF is more than enough. In the olden days of capacitors a 10 uF might have had 20 or 30 ohms of ESR, but there good “low impedance” types available these days. Most of the regular brands are. When driver/predriver junction capacitances are small, I’ve been known to use a .022 uF ceramic capacitor.
If the temperature rise in the driver transistor is small and enough emitter resistance is used, direct thermal contact is unnecessary. Elevate the operating voltage and it won’t be. I had a really nice little 15 wpc amp that ran off a 25.2 VCT transformer. CFP outputs, diode bias and thermally stable with just the diodes and drivers co-located on the PCB (well, perfboard actually). The same circuit ran off more voltage had to go to “giant TO-92” or TO-126 drivers with the diodes epoxied to them. And wouldn’t survive with D44/D45 outputs because of the SOA problem discussed in that Toshiba link.
10 uF is more than enough. In the olden days of capacitors a 10 uF might have had 20 or 30 ohms of ESR, but there good “low impedance” types available these days. Most of the regular brands are. When driver/predriver junction capacitances are small, I’ve been known to use a .022 uF ceramic capacitor.
If the temperature rise in the driver transistor is small and enough emitter resistance is used, direct thermal contact is unnecessary. Elevate the operating voltage and it won’t be. I had a really nice little 15 wpc amp that ran off a 25.2 VCT transformer. CFP outputs, diode bias and thermally stable with just the diodes and drivers co-located on the PCB (well, perfboard actually). The same circuit ran off more voltage had to go to “giant TO-92” or TO-126 drivers with the diodes epoxied to them. And wouldn’t survive with D44/D45 outputs because of the SOA problem discussed in that Toshiba link.
Once more updated schema, maybe last update. Also thanks for comments and help for all.
All 10uF caps are Panasonic low ESR. IR led is L53SF4C from Kingbright, available here, but Digikey stocks LTE-4208 from Lite-on which should work as well. C4 can be bigger one to get better low frequence response.
All 10uF caps are Panasonic low ESR. IR led is L53SF4C from Kingbright, available here, but Digikey stocks LTE-4208 from Lite-on which should work as well. C4 can be bigger one to get better low frequence response.
Those Q2&Q5, BC639/640 seems to be more or less obsolete. Will try later MPSA06/MPSA56 instead on those BC:s. They seems to be better available and should work as well.
Also instead of IR led can be used simple current mirror using one 1N4001 and 220ohm resistor in series.
Also instead of IR led can be used simple current mirror using one 1N4001 and 220ohm resistor in series.
How do you calculate the gain in an amplifier like this? I have it in a simulator but I can't figure out how the gain corresponds to the resistor network. I thought it would be R8//(R9+R10) divided by R11.
Thanks!
Thanks!
Yes, measured gain is for sure smaller than R8/R11 and I want to keep it simple. Later can measure gain if important. More loop gain means at least one more transistor and more parts.
BTW, I have "one more" updated amp almost ready, still simple and only minor upgrades.
A real PCB could be a big uprage, have to think about this.
BTW, I have "one more" updated amp almost ready, still simple and only minor upgrades.
A real PCB could be a big uprage, have to think about this.
I think it works like this. Others may correct if I am thinking wrong way.
Just build some simple phase inverter and try bridge.
24V or 36V with bigger output trannies is a challenge, say maybe 10W class A output power, but I don't want to mess good thread with these. This is also good learning project for all using approximately same schema always.
24V or 36V with bigger output trannies is a challenge, say maybe 10W class A output power, but I don't want to mess good thread with these. This is also good learning project for all using approximately same schema always.
Some updates, Q2&Q5 are now more industry standard transistor and better available. No IR led anymore, but a current mirror using one more 1N4001. More gain, R8 is now 220K.
1kHz clipping, 8ohm load, 2V/div.
And something more, all these are 1V/div. First 1kHz, then 10kHz, last one 50kHz.
This could be interesting, load is 8ohm with parallel 1uF cap. Freq is 10kHz. Only very small overshoot, but no ringing.
Scope is Tek535A with CA plug-in, if this is interesting.
1kHz clipping, 8ohm load, 2V/div.
And something more, all these are 1V/div. First 1kHz, then 10kHz, last one 50kHz.
This could be interesting, load is 8ohm with parallel 1uF cap. Freq is 10kHz. Only very small overshoot, but no ringing.
Scope is Tek535A with CA plug-in, if this is interesting.
i managed to reach simmilar levels of performance with 2 transistors less, although thats just a simulation
I am looking forward to see your results. These power supplies can be noisy as hell. You might be able to filter out much of it with the 12V regulator and associated capacitors. However these laptop supplies have a huge HF coupling between the secondary and the mains. HF starts at 50 Hz or so. While the DC voltage might be relatively clean anything you connect to your amplifier might suffer from the leakage between the DC and mains.
I am not sarcastic. I really am looking forward what success you have. Many times I wanted such a power supply for a small project but rejected the idea because of the noise.
I am not sarcastic. I really am looking forward what success you have. Many times I wanted such a power supply for a small project but rejected the idea because of the noise.
Of course not sarcastic.
This maybe needs two regs in series, but will try some 4 transistor design first.
This maybe needs two regs in series, but will try some 4 transistor design first.
Aren’t 19V laptop supplies regulated? Make a nice 12V linear regulator to run a discrete preamp to go with the power stage.
I have an 19V (almost)5amp asus unit from an old laptop, relatively clean power source without much noise. definitly a fully switching power supply