This one has an LT1054. Wonder if it is to produce a negative rail voltage for the opamp? Sometimes used for that purpose in guitar effects pedals, but the power I have seen didn't look very clean with a scope, and I was unable to make it much better.
Big 'Golden' type clock in there too. Not low phase noise, IIRC.
Looks like a single OPA2134 used as a stereo output stage. Not many components around it either. Could possibly be the ESS recommended low-cost output circuit option, can't tell from the picture.
Yes, I believe the LT1054 is for the dual rail. The OPA2134s feed into an OPA1622 (tiny black square behind the headphone jack). The clock, yeah, I don't know.. However, right next to it is I think a schmitt trigger inverter ('U20') (?). Curious what that does.
But after lots of comparing, the Hifime sounds much better than the Topping. In fact, the Hifime doesn't sound quite like a typical ESS, reminds me more of an AKM; bolder, more 'musical', mid-centric. Not sure why. THD compensation? Gain compensation? Filter?
Last edited:
Could have something to do with the particular clock jitter. Some people seem to like some kinds of it. That might be a long shot though.
Could be less switching hash from the power supplies affecting the audio. I see some inductors in the corner, but can't see if there are any switching regulators anywhere. That could be a big one.
Could be less switching hash from the power supplies affecting the audio. I see some inductors in the corner, but can't see if there are any switching regulators anywhere. That could be a big one.
Well sometimes things look good on paper until you hear it. I added some capacitance to the Cset in the LT3042. I did not like the results it produced at this point. It reminded me of when I put 2200uF on the AVCC before the SEPC caps I put in last week.
Initially it sounded like there was more low end but the weight and life of the bass was gone. This is after the LT3042 replacing the AMS1117 and replacing the AVCC decoupling with 470uF Oscon SEPC caps. I'll give another day or two and decide if it stays or comes off. Definitely soundstage depth was lost. Everything was brought more forward.
Initially it sounded like there was more low end but the weight and life of the bass was gone. This is after the LT3042 replacing the AMS1117 and replacing the AVCC decoupling with 470uF Oscon SEPC caps. I'll give another day or two and decide if it stays or comes off. Definitely soundstage depth was lost. Everything was brought more forward.
It could be you are just exposing another problem you didn't know you had until you start to get the DAC working well enough. If stabilizing AVCC makes the bass sound worse and less defined, it probably means you previously made the DAC into a compressor circuit with an attack delay that created fake bass dynamics. The bass will come back better than ever when you get everything else fixed.
If you have a soundcard with SPDIF out and an analog line input, or if you have more than one computer and one has a SPDIF out and another computer has an analog line input then you could do a test. Basically, you just need a way to play something through the DAC, even from a CD player, and some way to record what comes out of the DAC back into a digital file.
The idea would be to play some test waveforms through the DAC and record the analog output of the DAC back into the line input of some computer. By looking at the resulting recorded waveform envelopes, you could probably see exactly what it is that you are hearing in the bass with different AVCC circuits.
If you have a soundcard with SPDIF out and an analog line input, or if you have more than one computer and one has a SPDIF out and another computer has an analog line input then you could do a test. Basically, you just need a way to play something through the DAC, even from a CD player, and some way to record what comes out of the DAC back into a digital file.
The idea would be to play some test waveforms through the DAC and record the analog output of the DAC back into the line input of some computer. By looking at the resulting recorded waveform envelopes, you could probably see exactly what it is that you are hearing in the bass with different AVCC circuits.
Last edited:
A little snippet attached of what I have been working on. Almost enough done for taking some measurements.
The experience of doing this particular part of the project does remind me of why I did not want to become a programmer.
The experience of doing this particular part of the project does remind me of why I did not want to become a programmer.
Hm, actually the AVCC seems to draw power from a small MP2359 (or copy most likely) 'step down switch mode converter with 1.2A peak output'. Guessing that's not the best choice.
Just a general FYI: 3.2v Lifepo4 batteries are an excellent option as a low noise power supply. And they have become much more affordable the last few years since A123 no longer controls the patent. I have one powering the Hifime usb receiver and clocks. Needless to say it's quite a jump in sound quality. Less clutter as well since there's no AC/transformer/rectification to worry about. May do AVCC next.
Attachments
Last edited:
Actually, I have been wondering about a similar thing a few days ago. Since AVCC requirements are not for a defined voltage but a very steady one, it does not matter when the voltage slowly drops off. As the battery voltage drops the offset drops with it so everything is good. I could be wrong.
Now I don't know about battery noise profiles but I do remember something said that the Jung Didden Super regulators were actually quieter than batteries in a preamp and shorter than a piece of wire when properly implemented. Again, I wish I could refer back to that but I do remember that imprint in my mind because it was impressive performance.
There should be some Lifepo4 measurements around here -they were very popular for a bit. I see there's one here. I don't know about the Jung Didden, but the Lifepo4 sound much better than other low noise regulators and battery options (alkaline, SLA) that I've tried in the past.
This has been my experience too. LifePO4 batteries beat the default LME49720 based reg, as well as several popular shunt regs, when powering the analog supply of my ES9018. The improvement in transparency was very obvious. But charging solutions can complicate things -they add complexity as well as potentially another noise source.
If using more then 3.2v then yes things start to get more complicated, requiring wiring in series, float charging etc.
Haven't tried LiPO4 batteries myself, but will give them a try. At the very least they may be good for telling you if you have power quality problems with a line powered supply. For that they may be highly useful.
However, probably good not to forget that batteries have internal source resistance themselves, and the voltage will modulate with load variation. In that regard they could be good for making fake bass pumping that is not actually in a recording. One would have to do some careful checking to know for a fact why one thing seems to sound better than another.
Also, I would point out that state-of-the-art DACs such as Benchmark use very carefully designed switching supplies (although they may be resonant or quasi-resonant, not the usual types).
Regarding reports of LME49720 based AVCC regulation not being as good as batteries, I don't know. Could well depend on power quality of the input voltage reference, of the LME49720 rails, and if more current is needed than they can source. One would have to do some experiments and or some careful measurements to be sure of the actual problem before concluding the problem is with the opamp itself, or that batteries will always be better. It has become very clear here that any upstream power problems, especially EMI/RFI, may work themselves down into the lowest level circuitry regardless of attempts at filtering and or successive regulation stages.
However, probably good not to forget that batteries have internal source resistance themselves, and the voltage will modulate with load variation. In that regard they could be good for making fake bass pumping that is not actually in a recording. One would have to do some careful checking to know for a fact why one thing seems to sound better than another.
Also, I would point out that state-of-the-art DACs such as Benchmark use very carefully designed switching supplies (although they may be resonant or quasi-resonant, not the usual types).
Regarding reports of LME49720 based AVCC regulation not being as good as batteries, I don't know. Could well depend on power quality of the input voltage reference, of the LME49720 rails, and if more current is needed than they can source. One would have to do some experiments and or some careful measurements to be sure of the actual problem before concluding the problem is with the opamp itself, or that batteries will always be better. It has become very clear here that any upstream power problems, especially EMI/RFI, may work themselves down into the lowest level circuitry regardless of attempts at filtering and or successive regulation stages.
Last edited:
If you can, get the A123 brand - since they had the original patent and were the goto for a long time. I'm sure Lifepo4s are not the last word in performance, but cost/simplicity to sound quality ratio is pretty darn good.
RC car sized LiFePO4 are very low self impedance. Talking car starter levels. Not gonna be a problem for a dac.
The output Z of the LifePO4 batteries is very low, sub 10mOhm, so equivalent to a good regulator. The improvement I heard was one of a cleaner sound rather than fake bass.
For any regulator, it's certainly true that other factors matter, such as the upstream power quality, reference quality etc. I had already done a lot of work optimizing these things, with much filtering, snubbing, balanced mains etc., so I'm comfortable in reporting my results. I'm also talking about power Avcc of an ES9018, so current draw isn't an issue.
This is purely a subjective report of course, so isn't meant to be definitive. However many others have had similar results, so worth experimenting with I think.
Okay. LifePO4 batteries are on order. Will try.
Has anyone tried running the whole DAC on such batteries? Based on the descriptions, it might well be the best 'gold standard' to compare with any power supplies to find out if there are any power problems and how much effect they may be having on SQ.
Maybe even run AVCC on a couple of batteries, then run the nominal +-15 from five or so in series for each rail. They might discharge fairly quickly on the + rail, but it could be very informative.
In the end however, I would like to think it must be possible to make a good, clean power supply that can run continuously without needing recharging.
Has anyone tried running the whole DAC on such batteries? Based on the descriptions, it might well be the best 'gold standard' to compare with any power supplies to find out if there are any power problems and how much effect they may be having on SQ.
Maybe even run AVCC on a couple of batteries, then run the nominal +-15 from five or so in series for each rail. They might discharge fairly quickly on the + rail, but it could be very informative.
In the end however, I would like to think it must be possible to make a good, clean power supply that can run continuously without needing recharging.
There's only a few critical voltages in the system, for which at least Linear and TI make extremely good LDO's for. A rail for opamps/etc isn't anywhere near as critical. The major benefit I see from batteries is ground loops and RFI injection.
The tougher problem is whether you can get the right parts in the right places on these boards or a whole new layout is needed, which, by and large, is a nonstarter.
Is the entire board drawing any more than ~100 mA off the 15v rail under normal use? Can get a few nights worth of music off that much draw no problem with a decent sized battery.
And one doesn't necessarily need to go to lithium ion for these silly low noise and impedance abilities, good sized gel cell sealed-lead-acid will get you there as well, just not as light or compact (which doesn't matter so much in this application?)
The tougher problem is whether you can get the right parts in the right places on these boards or a whole new layout is needed, which, by and large, is a nonstarter.
Is the entire board drawing any more than ~100 mA off the 15v rail under normal use? Can get a few nights worth of music off that much draw no problem with a decent sized battery.
And one doesn't necessarily need to go to lithium ion for these silly low noise and impedance abilities, good sized gel cell sealed-lead-acid will get you there as well, just not as light or compact (which doesn't matter so much in this application?)