Thanks for the suggestion, I am not familiar with the F7. If my memory serves me right, reducing the current would essentially move the quiescent point down, and send lower half of the sine wave into clipping when playing loud, please feel free to correct me. Thank you.
🤣 I was thinking along the line of having 2 amps housed together, a bit like Siamese twins, the 2 amps share the PSU, output stage MOSFET (and heatsinks) casing, etc. through some switches that can be switched from one to other. Kind of like the Mac having IOS and Windows running concurrently.
it would be ideal if amplification and driver stage are the same, it is okay if it is not.
The main purpose of doing this instead of having two separate amplifiers is cost saving obviously, as PSU and heatsinks are main cost of class A amplifier, especially when it is single ended.
Last edited:
I have succumbed to the fact that it may be too complicated to share the output stage for what it is worth. I will just build a class AB amplifier, mount the output stage on the Aleph 5 heatsinks, and share the PSU.
Any suggestions for a class AB clone? Is the F7 a class AB amplifier?
Unless some genius has a brilliant idea 😊
Any suggestions for a class AB clone? Is the F7 a class AB amplifier?
Unless some genius has a brilliant idea 😊
if heat is a problem, use another amp, do not fret with finished one
main question is - how much dough (W) you need?
main question is - how much dough (W) you need?
I am trying to use it to drive my Martin Logan Quest (nominal impedance 6 Ohm, but can get to as low as 1.5 Ohm) in a small room. The speakers are rated as 200W, maximum input voltage of 99V(p-p).
That essentially translate to ~35Vrms max. input for the speakers.
So yes, I am going to build another amp. Just mount it in the same casing and share heatsinks and PSU with the Aleph clone.
That essentially translate to ~35Vrms max. input for the speakers.
So yes, I am going to build another amp. Just mount it in the same casing and share heatsinks and PSU with the Aleph clone.
Dear all,
I see very similar, ~200 MHz oscillations on both channels (15-20 mVrms / 60-100 mVpp - quite a lot), see attached. Any idea where they may come from, or how to systematically track them at all? The sreenshots were taken with an idle source and identical 4R7 dummy loads. The source is clean, only the baseline noise of the scope is seen at its outputs. I would also rule out pick-up 'from the air' because lid on or off, cable positions etc do not matter. Is this perhaps what C9 is meant for to cure? If so, how much capacitance shall I try?
Thank you in advance.
I see very similar, ~200 MHz oscillations on both channels (15-20 mVrms / 60-100 mVpp - quite a lot), see attached. Any idea where they may come from, or how to systematically track them at all? The sreenshots were taken with an idle source and identical 4R7 dummy loads. The source is clean, only the baseline noise of the scope is seen at its outputs. I would also rule out pick-up 'from the air' because lid on or off, cable positions etc do not matter. Is this perhaps what C9 is meant for to cure? If so, how much capacitance shall I try?
Thank you in advance.
Hello -
I have never encountered the problem that AB100 would oscillate - even with various DIY PCBs. Did you use the original NP PCB or a PRASI PCB? Do the output base stopper resistors have correct values ? How high is the gain set?
I have never encountered the problem that AB100 would oscillate - even with various DIY PCBs. Did you use the original NP PCB or a PRASI PCB? Do the output base stopper resistors have correct values ? How high is the gain set?
i use 22K/1000 (1k) its lower gain from 36db to some 26 or so if i remember correctly
Last edited:
Dear osscar and Loudthud,
thank you for your input. Very unusual, indeed. Which is why I performed more systematic tests and found that those ~200 MHz oscillations are actually injected by my solid state output relay control circuits. If I use a clean, external 10.5V to supply the V_GS of the relay, everything is beautiful, I merely see the baseline noise of my 8 bit scope.
This is surprising, because in the relay control circuit I use Si7851, a widely praised ASIC. The relay itself consist of two FDP083N15A MOSFETs. DC detection is based on an RC filter + window-comparator, whose output is latched by an LM555 and AND-ed with another threshold comparator output watching that the positive rail voltage is within 10% where it should be.
Otherwise I use my own PCB where I have on-board output relays, on-board PSUs (active rectification, 3x10mF/63V / rail), capacitance multipliers instead of R16/R17, KSC3503 / KSA1381 and 2N5551 / 2N5401 (at +/-50V rails I would not feel safe with ZTX450 / ZTX550 rated at 45V V_CEO). C9=0.pF Other than these, everything is as specified by the way more experienced builders at this thread. Gain is set to 27dB (22k1 / 1k), bias is 20 mA / transistor (to begin with).
Currently I am re-designing the relay control circuit to better separate its high / low voltage lines, eliminate everything that is not strictly necessary, and making the board smaller so that I can bring it as close to the relay on the amp board as possible and, thereby, minimize possible pickups.
It goes without saying that I will share the whole design, once it is stable.
thank you for your input. Very unusual, indeed. Which is why I performed more systematic tests and found that those ~200 MHz oscillations are actually injected by my solid state output relay control circuits. If I use a clean, external 10.5V to supply the V_GS of the relay, everything is beautiful, I merely see the baseline noise of my 8 bit scope.
This is surprising, because in the relay control circuit I use Si7851, a widely praised ASIC. The relay itself consist of two FDP083N15A MOSFETs. DC detection is based on an RC filter + window-comparator, whose output is latched by an LM555 and AND-ed with another threshold comparator output watching that the positive rail voltage is within 10% where it should be.
Otherwise I use my own PCB where I have on-board output relays, on-board PSUs (active rectification, 3x10mF/63V / rail), capacitance multipliers instead of R16/R17, KSC3503 / KSA1381 and 2N5551 / 2N5401 (at +/-50V rails I would not feel safe with ZTX450 / ZTX550 rated at 45V V_CEO). C9=0.pF Other than these, everything is as specified by the way more experienced builders at this thread. Gain is set to 27dB (22k1 / 1k), bias is 20 mA / transistor (to begin with).
Currently I am re-designing the relay control circuit to better separate its high / low voltage lines, eliminate everything that is not strictly necessary, and making the board smaller so that I can bring it as close to the relay on the amp board as possible and, thereby, minimize possible pickups.
It goes without saying that I will share the whole design, once it is stable.
Hi
can someone help me where it is written to use higher bias. i want to read through that.
thx
chris
Dear chris,
see e.g. #29 (osscar) and #469 (NP) in this thread or the "Leaving Class A" article by NP here:
https://www.passdiy.com/gallery/articles/leaving-class-a
Best,
Miklos
see e.g. #29 (osscar) and #469 (NP) in this thread or the "Leaving Class A" article by NP here:
https://www.passdiy.com/gallery/articles/leaving-class-a
Best,
Miklos
Dear all,
I was wondering how much asymmetry can be tolerated in the gain of the npn and pnp output darlingtons?
I happen to have some Sanken 2SD2390 / 2SB1560 transistors on my shelf which I would be curious to try in my +/-50V rail version of the amp, especially because of their more appealing 150V V_CEO as opposed to the 100V V_CEO of the TIP142 / TIP147 devices. Only, the npn's are "P" rank (6500 to 20000) whereas the pnp's are "O" rank (5000 to 12000).
Thank you for your input,
Miklos
I was wondering how much asymmetry can be tolerated in the gain of the npn and pnp output darlingtons?
I happen to have some Sanken 2SD2390 / 2SB1560 transistors on my shelf which I would be curious to try in my +/-50V rail version of the amp, especially because of their more appealing 150V V_CEO as opposed to the 100V V_CEO of the TIP142 / TIP147 devices. Only, the npn's are "P" rank (6500 to 20000) whereas the pnp's are "O" rank (5000 to 12000).
Thank you for your input,
Miklos
One of the BAF videos details the impact of gain mismatch between n & p devices on harmonic distortion. Might be 2016 - a discussion on followers but I think there is another presentation somewhere with scope traces. But it is most recently detailed in the ACA Mini article by Nelson
In the ACA Mini reducing the value of the source resistor on the p channel part increases the gain of that device and provides a better balance between top and bottom - reducing the 2nd harmonic distortion.... Presumably increasing the gain further would generate increasing amounts of +ve phase 2nd harmonic. I guess the suggestion is to look at the relative transconductance of devices and play with the source resistor values within limits (or using the parallel capacitor trick to maintain some DC bias current stability), to get what you want to see / hear - I think a 2 channel scope would be helpful.
In the ACA Mini reducing the value of the source resistor on the p channel part increases the gain of that device and provides a better balance between top and bottom - reducing the 2nd harmonic distortion.... Presumably increasing the gain further would generate increasing amounts of +ve phase 2nd harmonic. I guess the suggestion is to look at the relative transconductance of devices and play with the source resistor values within limits (or using the parallel capacitor trick to maintain some DC bias current stability), to get what you want to see / hear - I think a 2 channel scope would be helpful.