Out of curiosity, can you use the 1212 daughter card at the same time as using an attached 1820? For a couple more I/O?
LOL, looks like I bought the datasheet hype: "The OPA1656 operates over a very wide supply range of ±2.25 V to ±18 V or (4.5 V to 36 V) on 3.9 mA of supply current to accommodate the power supply constraints of many types of audio products."
and buried in the spec:
IQ Quiescent current (per channel)
IO = 0 A, VS = ±2.25 V to ±18 V 3.9 4.6 mA
IO = 0 A, TA = –40°C to +125°C(2) 5.0 mA
One of those is wrong, we'll see tomorrow or the next day when they show up...
Last edited:
I'll try (probably not today) and let you know... I've not seen anything documenting it, and given how the 1820m doc works, it would likely only do 48KHz max if it will work at all.
Take a look at Panasonic FR-A series for power supply. I'm staying true to the capacitance but going up one notch on voltage where possible. ESR and ripple current is almost twice as good as the stock G-LUXON LZ/LU, and rated endurance around 4X.
Last edited:
Just FYI: The E-MU 1820 and 1616 contain some nice ultra low noise Micrel (now Microchip) LDO regulators for the 5V supplies. So if SNR is degraded perhaps you need to look at the SMPS that feeds the LDO and then the FILT caps.
Off the top of my head I think the part number is MIC5209‑5.0:
https://ww1.microchip.com/downloads/en/DeviceDoc/20005720A.pdf
They are the SOT23-5 packages such as the one next to the last channel CS4398.
Off the top of my head I think the part number is MIC5209‑5.0:
https://ww1.microchip.com/downloads/en/DeviceDoc/20005720A.pdf
They are the SOT23-5 packages such as the one next to the last channel CS4398.
I want to introduce another aspect in op amps that makes a great difference in transient mode.
Op amps have open loop BW of only a few hertz, maybe a few hundred, this means the signal arrives at the output at -90° shifted and the NFB will adjust it to near zero. This doesn't happen instantly, the higher is the NFB the less time is lost for phase alignment. When in music several similar sounds start together, as choral or large orchestra burst, the transients in slow amps go distorted.
For comparison, take TL082 at 0db gain. It offers at 20khz 45db NFB. The OPA627,<60db. The OPA1611 80db. Of course the higher is the NFB at 20khz, higher is the frequency response.
How about having opamps with high open loop BW. On this thread mentioned on post 5 Parallel opamp: an extreme audiophile amplifier made of parallel multibrand op-amps, The casting list to assign the opamp for the roll of the master, all opamps have their OLBW above 10khz, from which AD826 was chosen. With such opamps the signal reaches the output with less phase shift, hence, less time needed for phase alignment. The best opamp I found for this aspect is the ultra low distortion, ultra low noise ultra high frequency 450Mhz, the AD8008 CFA opamp with OLBW 100khz.
Op amps have open loop BW of only a few hertz, maybe a few hundred, this means the signal arrives at the output at -90° shifted and the NFB will adjust it to near zero. This doesn't happen instantly, the higher is the NFB the less time is lost for phase alignment. When in music several similar sounds start together, as choral or large orchestra burst, the transients in slow amps go distorted.
For comparison, take TL082 at 0db gain. It offers at 20khz 45db NFB. The OPA627,<60db. The OPA1611 80db. Of course the higher is the NFB at 20khz, higher is the frequency response.
How about having opamps with high open loop BW. On this thread mentioned on post 5 Parallel opamp: an extreme audiophile amplifier made of parallel multibrand op-amps, The casting list to assign the opamp for the roll of the master, all opamps have their OLBW above 10khz, from which AD826 was chosen. With such opamps the signal reaches the output with less phase shift, hence, less time needed for phase alignment. The best opamp I found for this aspect is the ultra low distortion, ultra low noise ultra high frequency 450Mhz, the AD8008 CFA opamp with OLBW 100khz.
If the phase correction from 90 degrees back to zero degrees didn't happen instantly, for audio purposes, then the phase would not be corrected back to zero degrees.
Here's the rub. People basically cannot hear relative phase, especially if it's natural phase due to a 1st-order low pass cut-off. Studies with all-pass filters have shown that phase changes are inaudible. Basically, no orchestra starts together more coherently than a 45 degree phase shift at 20 kHz. It's all in the way the ear works, physiologically. The cochlea has frequency bins, and nerves are selectively stimulated if sound falls in those bins. The brain doesn't have the processing speed to check the correlation faster than a few milliseconds.
I am all about having op amps with stupidly high open-loop BW, however. Just not because of phase response. I just want the high levels of error-correction, and the lack of modulation products that fall into the midrange and lower parts of the audio band. A CCIF-20K measurement tracks two things: the linearity of the amp at 20 kHz, and the amount of error correction available at 1 kHz.
Paralleling op-amps is really only good for lowering noise, and driving heavier loads.
I think the frequency of the amplifier's first pole is irrelevant. I've made IC headphone amps with the dominant pole below 1 Hz, but with 160 dB of OL gain. What matters are how linear is the open-loop amp's transfer curve, and how much loop gain is left at 20 kHz.
Here's the rub. People basically cannot hear relative phase, especially if it's natural phase due to a 1st-order low pass cut-off. Studies with all-pass filters have shown that phase changes are inaudible. Basically, no orchestra starts together more coherently than a 45 degree phase shift at 20 kHz. It's all in the way the ear works, physiologically. The cochlea has frequency bins, and nerves are selectively stimulated if sound falls in those bins. The brain doesn't have the processing speed to check the correlation faster than a few milliseconds.
I am all about having op amps with stupidly high open-loop BW, however. Just not because of phase response. I just want the high levels of error-correction, and the lack of modulation products that fall into the midrange and lower parts of the audio band. A CCIF-20K measurement tracks two things: the linearity of the amp at 20 kHz, and the amount of error correction available at 1 kHz.
Paralleling op-amps is really only good for lowering noise, and driving heavier loads.
I think the frequency of the amplifier's first pole is irrelevant. I've made IC headphone amps with the dominant pole below 1 Hz, but with 160 dB of OL gain. What matters are how linear is the open-loop amp's transfer curve, and how much loop gain is left at 20 kHz.
Some time ago in this forum, a respected member, PMA, performed a test with two identical amplifiers, one with the first pole at 100kHz and the other with an even higher first pole. A recording of music was played through the amplifiers and re-recorded into new wav files. Two people who listened to the files could reliably tell the difference between the two amplifiers by ear. PMA is a skeptical guy, so he ABX double-blind tested himself and could still reliably hear the difference. Guys like Scott Wurcer appeared to accept the result, IIRC Scott said there is still some phase shift in the audio band from the higher frequency poles.
More generally, I personally have some trouble believing all the published psychoacoustic research results, particularly as they tend to be interpreted by EEs. One PhD EE who still occasionally posts in the forum, after getting out of school went to work in medical research. After being exposed the very real and difficult problems of researching humans he remarked maybe a couple of times in the forum that he had become skeptical of all the old research (that he used to believe in) because the number of test subjects was always too low to accurately represent the population. A different forum member with direct experience in human perceptual research, including for hearing, also cast much doubt on the published research but for other reasons than the first guy I just mentioned. The problems the second member was concerned about was the nature of planning and executing research on humans was probably never done correctly by EEs who simply are not experts in that area. Good old engineering common sense is not good enough when it comes to working with human brain processing of perception. Mine own objection to the old research besides agreeing with the first two guys is that equipment used to perform testing was often not SOA by today's standards. When I listen now I do it on better equipment than was available even one decade ago. IME its easier to hear small aberrations in SQ when there is less masking by imperfect equipment.
More generally, I personally have some trouble believing all the published psychoacoustic research results, particularly as they tend to be interpreted by EEs. One PhD EE who still occasionally posts in the forum, after getting out of school went to work in medical research. After being exposed the very real and difficult problems of researching humans he remarked maybe a couple of times in the forum that he had become skeptical of all the old research (that he used to believe in) because the number of test subjects was always too low to accurately represent the population. A different forum member with direct experience in human perceptual research, including for hearing, also cast much doubt on the published research but for other reasons than the first guy I just mentioned. The problems the second member was concerned about was the nature of planning and executing research on humans was probably never done correctly by EEs who simply are not experts in that area. Good old engineering common sense is not good enough when it comes to working with human brain processing of perception. Mine own objection to the old research besides agreeing with the first two guys is that equipment used to perform testing was often not SOA by today's standards. When I listen now I do it on better equipment than was available even one decade ago. IME its easier to hear small aberrations in SQ when there is less masking by imperfect equipment.
Last edited:
What helps explaining is to quote them in addition to posting your speculations about them.
Do you have a degree in music or acoustic engineering?
Can you hear a difference between 2 solid state preamps?
Didn't he also just use an input filter?
Didn't he also just use an input filter?
As the other Scott said, you need to be careful about what you call BW limit. Taking say a 500kHz closed-loop BW preamp and putting a 100kHz filter in front of it is very different from playing with the open-loop BW and then closing the loop.
BTW I saw a youtuber reviewing DAC's proposing the difference might lie in the tails between notes. Another guy was building an ultra-low noise preamp and played a few seconds of Merzbow as a joke while he fiddled with his instruments. Are they are reading here?
If there were concerns then better to have raised them at the time with PMA rather than to raise them now with me. I don't know exactly what PMA did with any of his listening tests.
I think I asked at the time in PMA's thread if there could be any other explanation besides pole location. He said what I read as more or less the equivalent of: "No, period." IIRC, nobody challenged him on it.
I think I asked at the time in PMA's thread if there could be any other explanation besides pole location. He said what I read as more or less the equivalent of: "No, period." IIRC, nobody challenged him on it.
Last edited:
I measured total quiescent current for the device on several samples and here is what I found:
OPA1656 7.7ma
LM4562 14.7ma
LME49720 13.8ma
All were sourced from Mouser so no Chinese knockoffs. I have a number of 4562 and they all showed similar quiescent current.
For reference, I also checked a few other duals of various grades that see common use in audio:
NJM2068 1.7ma
MC33078 1.4ma
OPA2134 8.6ma
NE5532 9.9ma
OPA2227 8.1ma
All were configured as unity gain buffers with no load or signal applied, and supply voltage was +/- 12vdc.
Really not an issue for a single stereo DAC, but a big issue for retrofitting multichannel sound cards/docks, mixers, etc. containing high numbers of opamps; due to power supply and thermal budgets for the device being retrofit.
Last edited:
The LM4562 (or LME49720) quiescent currents are surprisingly high considering the datasheet specifies typical quiescent current as 10mA and maximum current as 12mA@Vs=+-15V.
Yes they are. And this is from several different batches bought a few years apart from a reputable source. Just goes to show you can't always believe what you read, even on a datasheet...