Hello there,
I'm playing around a CFA headphone amplifier. I was inspired by Accuphase amplifiers to try with CFA design. My friend from https://muzgaudio.com/ helped me to get going (thanks Konrad!), but he can't give me all his free time (which I understand). I'm a beginner analogue designer and a hobbyist.
The concept is to deliver 2W@64Ohms with THD+N below 0,001% for 1kHz. Currently, I'm at 0,004%.
The compensation is taken from the famous article by Mr. Alexander: https://hifisonix.com/wordpress/wp-content/uploads/2012/09/The-Alexander-Amplifier.pdf
The amplifier is not considered stable yet! It is not calculated, just a placeholder.
The topology consists of a diamond buffer in Marantz HDAM 3 topology, a Widlar current mirror and a 2EF.
Operating currents are described in the sim. Please feel free to download and play around with it.
There is a weird distortion in the current between first pair of emitter followers (Q7, Q9):
Can somebody advise? The sinusoidal current in VAS side, degeneration resistors for current mirrors also seem to not be symmetrical (e.g. I R22).
Still to do:
Baker clamps
Bias generator
Compensation
DC Servo
Output stage bias tuning
I'm playing around a CFA headphone amplifier. I was inspired by Accuphase amplifiers to try with CFA design. My friend from https://muzgaudio.com/ helped me to get going (thanks Konrad!), but he can't give me all his free time (which I understand). I'm a beginner analogue designer and a hobbyist.
The concept is to deliver 2W@64Ohms with THD+N below 0,001% for 1kHz. Currently, I'm at 0,004%.
The compensation is taken from the famous article by Mr. Alexander: https://hifisonix.com/wordpress/wp-content/uploads/2012/09/The-Alexander-Amplifier.pdf
The amplifier is not considered stable yet! It is not calculated, just a placeholder.
The topology consists of a diamond buffer in Marantz HDAM 3 topology, a Widlar current mirror and a 2EF.
Operating currents are described in the sim. Please feel free to download and play around with it.
There is a weird distortion in the current between first pair of emitter followers (Q7, Q9):
Can somebody advise? The sinusoidal current in VAS side, degeneration resistors for current mirrors also seem to not be symmetrical (e.g. I R22).
Still to do:
Baker clamps
Bias generator
Compensation
DC Servo
Output stage bias tuning
Hi Lineup, thanks for reaching out! It's a dead project sadly. I'm not smart enough to do it on my own. Slapping a couple of components together is one thing, but making it work is another. My biggest problem is the compensation. For a working project, please see: https://www.diyaudio.com/community/...a-design-for-oscillation.401335/#post-7403445
I have a soft spot for CFA since I'm in love with the sound of Accuphase amps, but it is much less documented than VFAs. Also, I'm still gathering my equipment (I have a crappy scope in ADALM2000).
I have a soft spot for CFA since I'm in love with the sound of Accuphase amps, but it is much less documented than VFAs. Also, I'm still gathering my equipment (I have a crappy scope in ADALM2000).
@SomekPoland
OKey.
The issue I can see from the voltages is a mismatch in the input stages.
This causes the output offset -1.319V.
This comes from the different currents in input stages.
What is important in CFA amps is to have some adjustment for the offset..
NPN and PNP transistors are different. Diffrent Vbe and different Hfe gain.
OKey.
The issue I can see from the voltages is a mismatch in the input stages.
This causes the output offset -1.319V.
This comes from the different currents in input stages.
What is important in CFA amps is to have some adjustment for the offset..
NPN and PNP transistors are different. Diffrent Vbe and different Hfe gain.
Last edited:
I tried trimming the R4/6 and R16/17 but couldn't get the offset to 0 with any reasonable changes. The standard solution is to replace R4/6 with a pot, isn't it?
I think that for an input stage like this a matched transistor array like THAT340 might be a good idea: https://thatcorp.com/that-300-series-low-noise-matched-transistor-arrays/
The transistors in it have different betas between NPN and PNP of course, but being matched and thermally coupled should make it very easy to trim in a real-world design.
I think that for an input stage like this a matched transistor array like THAT340 might be a good idea: https://thatcorp.com/that-300-series-low-noise-matched-transistor-arrays/
The transistors in it have different betas between NPN and PNP of course, but being matched and thermally coupled should make it very easy to trim in a real-world design.
Last edited:
@SomekPoland
What is your experience with headphones driven by cfa amps?
Does it have sonic benefits compared to voltage driven amps?
I found this one. A very simple TDA2003 amplifier in cfa mode to be converted.
https://www.diyaudio.com/community/...fication-to-current-drive.389985/post-7552419
What is your experience with headphones driven by cfa amps?
Does it have sonic benefits compared to voltage driven amps?
I found this one. A very simple TDA2003 amplifier in cfa mode to be converted.
https://www.diyaudio.com/community/...fication-to-current-drive.389985/post-7552419
My experience is kinda weird to be honest. Most if not all headphone amps I heard were VFA. But, as mentioned I really like the sound of Accuphase and Marantz which utilise CFA.
Also, one of the best-sounding headphone I heard was Sennheiser Orpheus (the OG one) with an energiser and it was connected to an A75 class A power amplifier from Accuphase. Curiously enough, we tried A36, P4300, P7300 and the differences between each were huge.
Also, one of the best-sounding headphone I heard was Sennheiser Orpheus (the OG one) with an energiser and it was connected to an A75 class A power amplifier from Accuphase. Curiously enough, we tried A36, P4300, P7300 and the differences between each were huge.
Yes, replace R4/6 with a pot is a way to adjust the output offset.
Could be 50 Ohm pot or so.