I think you might be getting the OP stage bias confused with whatever else is driving that OP stage. Class A just implies that both OP devices used
in OP stage of I-V conduct all the time. All other things being equal, this should give superior performance.
If you could get your fav I-V opamp, say OPA1612 and magically bias the OP stage, as in both devices, up to say 12mA, it would almost certainly sound
better, handle RF better and have lower distortion.
Having said this, I'm not sure of the exact implementation of these new rail to rail OP stage opamps, specifically how they negotiate device current
'handover' or if they may even have circuitry to prevent device switchoff.
One thing is for sure, opamps are getting very very good.
TCD
Or that one never conducts because it doesn't need to, i.e. no crossover distortion.
I would agree that the output devices are probably running on the underbiased side when both are being used symmetrically. Also, agree opamps are getting very good.
Last edited:
Because the OP stage, when asked to drive more than a couple of mA is
generally the weakest link. That's where most of your distortion is generated.
TCD
How do you know? We aren't using TL072 for that. With OPA1612 it is possible to get ES9038Q2M THD down to -126.5dBFS (not that it is fully musical at that point). How much lower do you expect to get it with a buffer? How much is even due to the opamp?
If you never try, you'll never know. There are always dead ends in between / before success.
You said the same thing of my suggestion to use HQplayer.
OK - no worries, you can stop using it now!
Just because it's not your idea doesn't invalidate it.
That's where a/ a lot of gains are historically to be had and b/ many people
can exercise their usually significant analog skills.
TCD
As stated, generally speaking, OP stage is the weakest link.
Your ABH2 power amp has error correction for the OP stage non linearity.
But as stated, opamps are on a current budget so they (nearly) always run
lean on OP stage bias. However they just make up for it with smarter design
in other areas.
Mark you have stated yourself that THD numbers don't always tell the whole
story when it comes to subjective results.
My only suggestion is that there may be gains in a well implemented class A
OP stage. Historically this has often been the case.
You can try it... or not.
TCD
Laserscrape, the reason for choosing the LiFePO4 is that it generates less noise than li-lion and it's much more safe regarding shorts and overcharge. It usally doesnt blow up like
Li-ion. And the storage voltage of 3.3V works for this application, Li-ion likes to to be stored at about 4V for longevity. It also have 15-20 year lifespan when kept at a perfect storage voltage. It also has little self discharge and very low impedance so I hope the ripple voltage will be low with this battery topology.
Li-ion. And the storage voltage of 3.3V works for this application, Li-ion likes to to be stored at about 4V for longevity. It also have 15-20 year lifespan when kept at a perfect storage voltage. It also has little self discharge and very low impedance so I hope the ripple voltage will be low with this battery topology.
I mean I/V stage for modern dac chips( ESS,AKM) where output current is high 50-72mApp.
well, actually I meant OPA1612 will be better choice than discrete design for mobile chips like q2m with 4mA output. already at 16mA delivered by 28pro an output buffer is highly preferable. for 38pro there are 64mA out and a buffer, discrete or monolith, is indispensable imho.
IV Stage Load
Hi Jan,
You should also consider the feedback circuitry which is essentially connected in parallel to the load resistance. To achieve lower noise the feedback resistors may be around 1K or even lower. That brings your output current to over 2mA. This is just a general opamp theory. Now, when we are talking the IV function, the opamp output should provide the current to match the DAC output current, which is 4mA in case of Q2M. Now your opamp should handle that current plus the load one.
Hi Jan,
You should also consider the feedback circuitry which is essentially connected in parallel to the load resistance. To achieve lower noise the feedback resistors may be around 1K or even lower. That brings your output current to over 2mA. This is just a general opamp theory. Now, when we are talking the IV function, the opamp output should provide the current to match the DAC output current, which is 4mA in case of Q2M. Now your opamp should handle that current plus the load one.
At peak with ES9038PRO current can get quite high. However, ESS still uses OPA1612 in evaluation boards for that dac chip. Measured distortion remains quite low which they specify at -122dB.
When buffers really become necessary is when multiple ES9038PRO output channels are tied together before I/V conversion. That and or the very low value resistors found in some output stages can place heavy demands on opamps. The intent seems to get noise down low enough so that dac chip noise floor modulation will hopefully become low enough such that even very perceptive humans cannot hear it. Unfortunately, that effort is sometimes pursued to the extent of failing to minimize opamp distortion which other very perceptive humans can hear.
All that said, ES9038Q2M current is down much lower around 4mA as has already been stated.
IME, very few people get distortion of ES9038Q2M down to the lowest possible levels, as it is rather challenging to do without careful multilayer design, use of multiple dedicated voltage regulators optimized for various purposes, etc. Simply adding a class A output stage to an otherwise lacking design will offer no useful improvement in sound quality, IMHO.
Last edited:
AK4499 in DSD bypass mode is lower than the 38mA quoted for PCM or non-bypassed DSD. Either way, OPA1612 handles the job quite well for I/V conversion. Having dedicated voltage regulation for the I/V opamps does audibly help sound quality though. As is often the case, what may be perceived by some as an audible problem with crossover distortion is actually more likely a power quality issue, IME.
Terry,
You have hit on a truth for high end audio designers, but it shouldn't necessarily be pushed onto people who do not have interest or tolerance for carrying through on experiments that require comparing multiple design options.
The gentleman that asked about class A output stages in this instance has been here in this thread before and to my possibly imperfect memory seems more the type that tries one thing and calls it good. If so, probably better to encourage trying a thing that is already known to work well with ES9038Q2M dacs.
I didn't mean to suggest the same would apply to you, someone who has been doing commercial work with dacs for a long time.
right, that would be the case of using this 8th channel DAC in stereo. if using only one channel out of four for a stereo side, then what is the point of 38pro? sort of downgrade to 28pro, since digital part is the same for both, I guess?
Do you have some link to data on noise and impedance?
The LiFePo piqued my interest in battery as power supply, I have read SLA offers similar electrical performance and reliability to Li, they also come in huge capacity at low cost (useful for my purpose of class A amp supply)... but of course one of the most hazardous.
This was the only real data to be found but nothing on SLA or Li : https://tf.nist.gov/general/pdf/1133.pdf
Some reports of certain LiFePo having low ESR than SLA, but then no mention of noise.