Hi, I'm in the process of designing a high-quality audio interface. I'm already planning on using Polak-Groner composite op-amps for the ES9039Q2M I/V converter, (possibly using lower cost OPA1612s instead of the OPA2211s and the newer OPA891s instead of the THS4031s) and was wondering what you guys thought of using the same composite op-amp buffered with a current-boosting push-pull stage as mentioned in this article as an ultra-low distortion headphone amp?
I've posted the proposed schematic below, which with 10pF of parasitic load capacitance is simulated to have a phase margin of about 40° with a 32ohm load, though that becomes a dangerously low 15° at 16ohms. With the selected transistors the amplifier will do about 24Vpp into 16ohms, which is frankly complete overkill for all but the most demanding headphones, but I wanted to design it to power pretty much anything under the sun, including planar magnetics. The distortion is simulated as -180dB, though I suspect in reality that will be significantly higher. The gain is a fixed at 3.39, mainly because I can't figure out an easy solution for digitally-controlled gain switching that doesn't significantly impact distortion.
My questions (besides thoughts about this amplifier in general) are as follows:
I've posted the proposed schematic below, which with 10pF of parasitic load capacitance is simulated to have a phase margin of about 40° with a 32ohm load, though that becomes a dangerously low 15° at 16ohms. With the selected transistors the amplifier will do about 24Vpp into 16ohms, which is frankly complete overkill for all but the most demanding headphones, but I wanted to design it to power pretty much anything under the sun, including planar magnetics. The distortion is simulated as -180dB, though I suspect in reality that will be significantly higher. The gain is a fixed at 3.39, mainly because I can't figure out an easy solution for digitally-controlled gain switching that doesn't significantly impact distortion.
My questions (besides thoughts about this amplifier in general) are as follows:
- Is it necessary to include gain switching with modern DACs' more advanced digital gain control and if so, is there a reliable, low-component count way to do it while maintaining these distortion figures?
- Am I absolutely insane for using composite op-amps because SNR and THD+n are limited by the DAC itself?
- Am I missing anything besides Aol/β and phase-margin for stability criteria?
- Is the maximum Vpp I have for line-level (before the headphone amp) of 8.272V way too much? I know +4dBu is 3.472V, but also heard that most pro-audio has around +10dBu of headroom.
I'm using Groner-Polak op amp as I/V stage in my ES9039Q2M DAC. I doubt such DAC will limit the performance of your headphone amplifier.
For the I/V stage, do you have the composite op-amps in the same configuration as the reference design?
My apologies, I mean the ES9039Q2M reference design–are you similarly using a dual op-amp configuration with one taking its feedback from the CM voltage, just with the composite op-amps instead of the OPA1612s?
Also, what do you think of the STM32 as opposed to XMOS? I was thinking about going XMOS for lowest latency, but is that really an issue with how powerful the H7 series is?
Also, what do you think of the STM32 as opposed to XMOS? I was thinking about going XMOS for lowest latency, but is that really an issue with how powerful the H7 series is?
Last edited:
No need for such complication. Composite op amps are used as I/V op amps. Differential output is taken directly from I/V stage. Additional summing op amp is used for SE output as differential +/- legs have DC.
I can recommend Omicron which also is based on composite op amps. I doubt Groner-Polak would be an improvement over that.
https://www.diyaudio.com/community/...amp-with-140db-distortion.387033/post-7043157
https://www.diyaudio.com/community/...amp-with-140db-distortion.387033/post-7043157
