Totally agree all the point you said. I personally did compare my two builds, one using salas shunt reg in all the rails, the other using default setting, but only battery in the shift reg (I de-soldered the two +/-5V LDOs).
I think they sure sounds different, but not day and night difference, also I can't say which one is definitely better than which. So far I just feel that using battery one songs more high definition in sound stage. However, it doesn't sound as musical as the shunt reg version. I need some more comparison when I get back my dac (now it's being lend to my friend)
Are both builds identical except for vref?
Which battery chemistry are you using ?
Are you using battery management circuit?
No.
Build 1: rev1 0.02%, w/vref cap mod, transistor mod, 5v,3.3v,iso3.3 salas shunt reg (four wires to the original LDO position or near with large pad)
powered by custom made 54VA transformer with multiple 9V@1A AC rails winded using OCC silver coated copper wire.
Build 2: rev2-012, only de-soldered the +/-5V ldo and electro-cap after the two LDOs. Powered by talema dual 7V AC 15VA transformer.
Panasonic NCR18650A Li-ion MH12210 two pairs in parallel
I dont think so, just simple +ve to the two vref +ve output pins, -ve to the other two vref -ve output pins, the middle of two pairs of batteries is ground to the dac's ground pin in J7. The wires not directly connected to DAC but intercepted by a relay. The relay is powered by PWR_A+ and GND so that when the DAC has power, it will connect the battery to shift regs.
All I know is that I would be quite surprised if all of the horrible things expected brought about a more relaxed sound along with a reduction in the digital noise infecting every note (from within - not as an add on) in comparison to what I was hearing before. Maybe there is this horrible high frequency noise being generated without the use of bypass capacitors but I would argue there was something even worse being generated with them. Audio is peculiar. We can make all kinds of measurements but we have yet to find many measurements that correlate to sound quality. What if this noise, which I tend to doubt is as great as posited, actually has little/no effect whatsoever on the end result? I doubt anyone has done much actual listening based testing on this. Instead we see something on a 'scope and figure, "there it is, now we know what to do", because in some data transfer product this corrupted the data flowing at a higher rate of speed than is required for digital audio.
We still know so little about digital audio. There is still so far to go.
There are plenty of instances where the conventional wisdom has been found to not work in the very dynamic world of music reproduction.
As a coarse analogy there is the bypass capacitor for coupling and power supply caps that is very popular with many. But there are those who find this is not a good idea for high quality audio. I find it creates a confused sound and after getting over the initial excitement of the extra "sparkle" I always go back to the capacitor un-bypassed. One best use a good capacitor. You are not going to turn an ordinary capacitor into something good with a bypass capacitor. Wish it was that easy. And inexpensive!
We are not using BATTERIES, we are (most of us) using a very specific battery which is why I mentioned it. The A123 26650 which has many unusual qualities. An extremely low ESR and an ability to deliver large amounts of current instantaneously which I surmise is the reason the bypass caps are not needed and if anything, get in the way of the current delivery.
There have been folks within DIYAudio who have measured them and compared them to very highly regarded voltage regulators and there is no comparison in the noise performance. If I had the link available I would post it now. I will have to look for it and post it later.
I would counter that lesser quality voltage supplies require the bypass not the device being "fed". i would bet the current delivery performance of the stock Vref is pretty puny.
I did not write my post to encourage anyone to do what I had done. I posted in response to someone asking what I had done. Those who are not interested will ignore my post and, therefore, will not need any nannies to advice them not to proceed.
If I had a scope I would measure the noise but i do not. No surprise to the engineers on the forum, I have no idea how to operate one, either.
But one must try things to find out what happens. Progress would quickly grind to a halt if everything was done just the way it always was. This is the difference between engineering and science. Not to pretend I am in any way a scientist but when presented with an idea I think has merit I will give it a try.
Yes, it would be better to try something before condemning it. I have heard this board with and without the battery scheme. My system is up to the task of resolving differences and the difference is plain.
I am happy for those who are satisfied with their systems and wish, one day, I could join their ranks but in the interim, especially for digital (I actually am very satisfied with my phonograph) I think it best we keep looking for ways to make something that sounds more like real instruments. Of course, electrified music is somewhat less demanding but still benefits from improvements found while using acoustic music as a reference.
Rev2 and newer already have the vref circuit updated, resulting in about 50 mohm output impedance and 80 uF capacitance, which are fine although other people might have other opinions. Additional capacitance might improve things, but a some point you get diminishing returns....
Rick, the shift registers are operating at 25 Mhz, at that frequency the only decoupling that work are fast ceramic capacitors close to the chips, and close mean minimeters.... Without those capacitors the voltage spikes can actually be high enough to cause long term damage to the cmos parts....
Well, I will be the guinea pig for this! No way I could ever return the caps and the thing sounds better without them. I will let you know when my DAC stops working.
One neat thing about the battery is its ability to absorb power as well as release it. It could be these spikes are absorbed by the batteries even with a couple of inches of wire? Just speculating ...
Wish I had a 'scope. Can you recommend an affordable one that could make a decent measurement?
Take care,
well you could put on those pps caps they dont seem to cause any harm
and its more politically correct
And it will sound even better with the ceramics....
That what the ceramics do, just at high frequencies.... It's correct that some batteries have very low impedance over the audio range, but even just a couple of inches wire work as a good inductor, making the batteries high impedance at higher frequencies....
The key to good decoupling is low impedance over the whole frequency range up to many Mhz.... For power supply decoupling impedance matter, not linearity....
I have a Tektronix, but those are expensive, mine was $8K about 10 years ago.... Today I might take a look at Hameg, but I have also read that Rigol is one of the better low cost manufacturer.... Minimum would be a 100 Mhz one.
Hope we are not experiencing a CULTURAL REVOLUTION at DIYAudio!
Just what we need: those shrieking women walking around with megaphones haranguing us as to what the authorities have deemed proper behaviour
Political correctness is the path to mediocrity!
Problem is I (and you) are not hearing what one would think should be plainly audible IF it is happening.
I stand by my (mindless) assertion that the power supply needs bypassing, not the device.
I will try to find someone to look at this if oscilloscopes are too expensive.
Just what we need: those shrieking women walking around with megaphones haranguing us as to what the authorities have deemed proper behaviour
Political correctness is the path to mediocrity!
Problem is I (and you) are not hearing what one would think should be plainly audible IF it is happening.
I stand by my (mindless) assertion that the power supply needs bypassing, not the device.
I will try to find someone to look at this if oscilloscopes are too expensive.
Dear SOEKRIS,
I listened to the thing for many months with a full compliment of ceramic capacitors! I assure you the sound is superior as it is now configured.
I figured a 'scope was beyond my mad money budget and sure enough it is!
I know an audio tech - I will take the thing to him one Saturday.
Would you tell me what to tell him to do?
I am curious.
Another comment about regulators vs THIS battery - the specs for ESR at high frequencies of regulators is suspect since these specs depend on massive amounts of feedback to get those low numbers. I consider this a form of cheating - though it works just fine for steady state conditions my contention is when music is playing all of it goes awry. I would bet the A123s have a comparatively flat series resistance whose impedance is maintained with dynamic situations. That is how I conceive it ...
I listened to the thing for many months with a full compliment of ceramic capacitors! I assure you the sound is superior as it is now configured.
I figured a 'scope was beyond my mad money budget and sure enough it is!
I know an audio tech - I will take the thing to him one Saturday.
Would you tell me what to tell him to do?
I am curious.
Another comment about regulators vs THIS battery - the specs for ESR at high frequencies of regulators is suspect since these specs depend on massive amounts of feedback to get those low numbers. I consider this a form of cheating - though it works just fine for steady state conditions my contention is when music is playing all of it goes awry. I would bet the A123s have a comparatively flat series resistance whose impedance is maintained with dynamic situations. That is how I conceive it ...
Last edited:
ok ive two almost identical builds
both with +/- 3.3 lifepo4 vref
one with pps bypass caps on shift registers +/- added in where the 4v regulation was
the other has +/- 3.3v wired from cell to each of shift registers with no bypass caps
i only got a couple minutes to put the scope on both of them to get the jist of it
both showed a little switching noise however not a big difference between them
on first build with caps the noise seemed to worsen very slightly at the shift registers that are further away from point of +/- insertion
overall the build with the shift registers powered separately with no caps seemed a little better
id imagine the best way is to wire the power to each shift register and use a good film cap
i remember when i tested it when the builds were identical except for the ceramic caps/no caps
no caps was superior sounding
after that i tested no caps versus the replacement panasonic pps caps and i couldn't recognise any difference
if i ever have another stock board on lifepo4 ill put the scope on the caps
both with +/- 3.3 lifepo4 vref
one with pps bypass caps on shift registers +/- added in where the 4v regulation was
the other has +/- 3.3v wired from cell to each of shift registers with no bypass caps
i only got a couple minutes to put the scope on both of them to get the jist of it
both showed a little switching noise however not a big difference between them
on first build with caps the noise seemed to worsen very slightly at the shift registers that are further away from point of +/- insertion
overall the build with the shift registers powered separately with no caps seemed a little better
id imagine the best way is to wire the power to each shift register and use a good film cap
i remember when i tested it when the builds were identical except for the ceramic caps/no caps
no caps was superior sounding
after that i tested no caps versus the replacement panasonic pps caps and i couldn't recognise any difference
if i ever have another stock board on lifepo4 ill put the scope on the caps
Rick, lets do a little math....
The battery cells you're using is specified at 6 mohm at 1 Khz, which is very very good, except the wiring ruins it at high frequencies....
The self inductance of a plain wire with a diameter of 0.3 mm and length of 50 mm is about 50 nH, you seems to have 8 wires parallel so lets say 6 nH. That don't sounds like much, but the impedance at 25 Mhz is about 1 ohm....
You can do whatever you feel like, after all it is DIY, I just don't want others to think it's a good idea to ruin boards too much....
Just put a scope across vref power, any of the removed small capacitor pads are fine, then take a look....
The battery cells you're using is specified at 6 mohm at 1 Khz, which is very very good, except the wiring ruins it at high frequencies....
The self inductance of a plain wire with a diameter of 0.3 mm and length of 50 mm is about 50 nH, you seems to have 8 wires parallel so lets say 6 nH. That don't sounds like much, but the impedance at 25 Mhz is about 1 ohm....
You can do whatever you feel like, after all it is DIY, I just don't want others to think it's a good idea to ruin boards too much....
Just put a scope across vref power, any of the removed small capacitor pads are fine, then take a look....
It's not clear to me, did you also remove all the small ceramic decoupling capacitors ?