I started this thread because I was curious how much a chip amp can be different by having different power supply and other circuits.
Today I finished my Audio Sector parallel 4780. When first listened to it as is (minimalistic approach), I honestly couldn't tell the difference between 4780 and 3886 amps. I felt like 4780 drove Magnepan MMG (4ohm) with a little bit of ease, but the difference was very small.
I disconnected the power supply, which is basically a fast recovery bridge diode and 10uF (yes, 10) capacitance.
And then I connected CRC with total of 20,000uF bank. It has normal slow bridge rectifier.
The difference was immediate and huge, so I didn't even have to listen to MMG carefully. Sound was so smooth and bass was well controlled. Improvement in bass was somewhat expected, but I don't know why mid-high also became smoother.
So now I believe that power supply can make huge difference.
Will it make another obvious difference if I use regulated power supply like Akitika? What about composite approach like Parallel 86? I don't know.
As I said in the OP, I would really appreciate if someone chime in who actually compared different approaches directly as I did today.
Today I finished my Audio Sector parallel 4780. When first listened to it as is (minimalistic approach), I honestly couldn't tell the difference between 4780 and 3886 amps. I felt like 4780 drove Magnepan MMG (4ohm) with a little bit of ease, but the difference was very small.
I disconnected the power supply, which is basically a fast recovery bridge diode and 10uF (yes, 10) capacitance.
And then I connected CRC with total of 20,000uF bank. It has normal slow bridge rectifier.
The difference was immediate and huge, so I didn't even have to listen to MMG carefully. Sound was so smooth and bass was well controlled. Improvement in bass was somewhat expected, but I don't know why mid-high also became smoother.
So now I believe that power supply can make huge difference.
Will it make another obvious difference if I use regulated power supply like Akitika? What about composite approach like Parallel 86? I don't know.
As I said in the OP, I would really appreciate if someone chime in who actually compared different approaches directly as I did today.
I arrived at the same conclusion as you back in Post #2. I ended up further concluding that to get stellar performance (get rid of the somewhat strained highs of the LM3886) one needed a better architecture, hence, I developed the Modulus-86 and subsequently Parallel-86.
Tom
Having a clean power supply with filtering in the right freqs can make a huge difference in SQ imo.
Last edited:
Are you serious about having a power supply with just ten microfarads of reservoir capacitance powering a chipamp? Or does your chipamp PCB have additional res capacitance on it?
Curious - were these 'somewhat strained highs' determined by THD+N measurement or from listening?
That (10uF) is what comes with basic kit. Total of 40 uF in total.
But there are 1,500 uF capacitors placed right next to the chip. 6,000 uF total.
With this basic setup, the kit has no hum/buzz and has been very satisfactory listening to 8 ohm bookshelf speakers, mostly in low volume. It revealed some limitation driving MMG at decent volume level.
Tom,
Now I have idea that there are a lot of rooms to improve 3886/4780 sound quality.
Thanks,
Doug
Thanks for the clarification. The PCB layout would need to be done carefully for the tracks to the GNDs of those two 1500uF caps or their charging pulses will appear on the signal ground. An AudioSector PCB I looked at a few years ago did look to have layout issues under control.
With 2x10uF in the power supply PCB for one channel, I could hear very low noise from the speaker. It is not 60Hz or 120Hz, but it is more like hiss. The noise at the output was 2-3 mVac.
When I add 2x10,000uF in the power supply, there was absolutely no noise heard, and DMM measures 0.0 - 0.1 mVac.
Yes.
When I first listened to the LM3886 I found it okay but not high end. In particular, I found the highs bothersome as they sound stained to me. I have noticed this with other amplifiers as well. What these amplifiers have in common is that the THD+N rises at higher frequencies. In two of the cases, the rise in THD+N began already at 500 Hz!
My DG300B exhibits relatively high THD (0.2 % @ 1 W, 8 Ω) but does not sound strained. Its THD+N is flat vs frequency. I've noticed the same with other tube designs. Now, these tube designs don't sound precise to me. They do add a little something-something to the sound. That's why people use tube amps. They want the sound (and pretty tubes to look at).
The Modulus-86 and Parallel-86 sound precise, neutral, and transparent to me. The "LM3886 sound" is gone and all you have is a wire with gain (and current drive capability). The THD+N of the Modulus-86 and Parallel-86 is practically flat vs frequency. There is a measurable rise in the THD+N vs frequency above 5 kHz, but the THD+N is a few orders of magnitude below the level of audibility.
I believe, based on my own subjective experience as well as my objective measurements, that to sound good, an amplifier should have flat THD+N vs frequency. If the amplifier is to sound precise, it should also have low THD+N, low IMD, low mains hum, flat amplitude response within the audio band, etc.
Having arrived at this conclusion, I am now able to design amplifiers that sound open and natural. They need to perform well on an array of measured parameters and they will also sound good. This approach has yet to disappoint me.
Tom
So that's an additional qualification. THD+N does not need to be flat with frequency provided any THD+N is below the level of audibility.
But with most music its been known for decades that IMD swamps THD. Doug Self mentions this fact in his book I recall. So the THD+N won't be audible anyway because the IMD masks it. THD+N audibility on single tones is just of academic interest.
But with most music its been known for decades that IMD swamps THD. Doug Self mentions this fact in his book I recall. So the THD+N won't be audible anyway because the IMD masks it. THD+N audibility on single tones is just of academic interest.
I don't recall Douglas Self mentioning anything about how anything sounds in any of his books that I have. I also don't recall seeing anything about THD vs IMD. I count four of his books on my bookshelf. Would you mind providing a complete reference (title, edition, and page number preferred)?
Furthermore, I already said that THD+N wasn't my only criterion for good amplifier design. Let me highlight the relevant bit:
Tom
Last edited:
Oh perhaps I didn't express my meaning clearly enough. I agree he didn't mention anything how any of his amps sound. But he did mention that in any amp the amount of THD would be swamped by the amount of IMD. If you need a reference to that I would have to dig out the book but I'll be happy to do so.
Yes I did read what you wrote before. You feel I missed something there in making my point?
Since I'd overlooked that my technical books are all boxed up for an impending move, I went over to Doug Self's website to see if that mirrored any of his statements in his book. And indeed it does, here :
[FONT=arial,helvetica]THD measurements, taken with the usual notch-type analyser, are of limited use in predicting the subjective impairment produced by an imperfect audio path. With music etc, intermodulation effects are demonstrably more important than the harmonics themselves.
Since he also cites Belcher, perhaps he sourced this knowledge from the origin, which I believe to be (as quoted in Belcher's 1978 WW article) Brockbank and Wass. You can find Belcher's article here and check for yourself : https://www.researchgate.net/publication/271501230_A_new_distortion_measurement The relevant bit is in the third column, first page.
[/FONT]
[FONT=arial,helvetica]THD measurements, taken with the usual notch-type analyser, are of limited use in predicting the subjective impairment produced by an imperfect audio path. With music etc, intermodulation effects are demonstrably more important than the harmonics themselves.
Since he also cites Belcher, perhaps he sourced this knowledge from the origin, which I believe to be (as quoted in Belcher's 1978 WW article) Brockbank and Wass. You can find Belcher's article here and check for yourself : https://www.researchgate.net/publication/271501230_A_new_distortion_measurement The relevant bit is in the third column, first page.
[/FONT]
So basically you support my point that it's not just the THD+N that makes an amp sound good. Excellent. I'm glad we agree on something.
The IMD of the Modulus-86 and Parallel-86 is vanishingly low as well, measuring 0.00069 % (38 W, 8 Ω, SMPTE) and 0.0017 % (110 W, 4 Ω, SMPTE), respectively.
Thanks for the reference. I'll go check it out. Do beware that the science could have moved a bit since 1978, though.
Tom
The IMD of the Modulus-86 and Parallel-86 is vanishingly low as well, measuring 0.00069 % (38 W, 8 Ω, SMPTE) and 0.0017 % (110 W, 4 Ω, SMPTE), respectively.
Thanks for the reference. I'll go check it out. Do beware that the science could have moved a bit since 1978, though.
Tom
Looks like you're putting your own words into my mouth here so what in what I wrote gave you this impression please? Perhaps I need to tidy up my form of words there.
I think we probably agree on a few things here - like the importance of good PSRR. But talking about agreements gets rather boring.
That (SMPTE) is not the IMD that Brockbank and Wass were talking about, so its rather irrelevant.
Whatever dude. I was trying to extend you an olive branch. I do my best to be accommodating of different views and allow others to get their point across. I do my best to be non-confrontational. I'm sorry the fact that I'm being a nice guy seems to ruffle your feathers.
They do mention it as an acceptable method for measuring IMD, though. First page, top of the third column of the article you liked to above. They go on to expand the IMD to a multi-tone test.
The APx525 does support multi-tone testing and I'll be happy to take the measurements. The question then becomes how to interpret them. If I read the results of Belcher (1978) correctly, an amplifier with multi-tone IMD products 35 dB below the amplitude of the test tones should sound good. Is that your interpretation as well?
Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.