LOL... Rick, we don't disagree in that LiFePo4 cells are the best in that usage. It is just that battery power is not for everyone and I was offering some alternatives.
And I also agree that the Belleson is a great option as a regulator too.
Where we MIGHT disagree is that I think better regulation on the microprocessor and FPGA rails would alo likely show some gains. OTOH, I have not tried that yet, so I cannot say with any level of certainty.
Greg in Mississippi
Removal of e cap
Dear Rick,
Thanks for your reply, I have replaced the 5v power supply by battery, but it sounds almostly the same with DC supply by regulator, I noted that you remove a row of electrolytic cap, can you show me which one is removed or replace by other film cap? A picture will be nice.
Arthur
Dear Rick,
Thanks for your reply, I have replaced the 5v power supply by battery, but it sounds almostly the same with DC supply by regulator, I noted that you remove a row of electrolytic cap, can you show me which one is removed or replace by other film cap? A picture will be nice.
Arthur
Are you powering all three 3.3 volts rails with batteries AND what battery are you using? I hope the A123 26650!
If you have implemented battery power for those rails remove the electrolytic caps near the corresponding, no longer used, regulator.
I will take a picture tonight.
AND we are not talking about BIG changes in quality. At this level, as I am sure you know, we have to be happy with baby steps. The SDTrans is better than anything I had heard with just the 5 volts battery on the main power input.
I am assuming you are using a 5 volts battery for the two FPGA rails.
Another thing to put on the list: getting the selected for low phase noise clocks from this fellow - look here: analog research NDK low jitter clocks for digital audio - now these things make a big difference. Same clock as what was supplied but selected for the low frequency phase noise parameter. Not absurdly expensive considering how many he has to throw away.
Will have more tomorrow.
If you have implemented battery power for those rails remove the electrolytic caps near the corresponding, no longer used, regulator.
I will take a picture tonight.
AND we are not talking about BIG changes in quality. At this level, as I am sure you know, we have to be happy with baby steps. The SDTrans is better than anything I had heard with just the 5 volts battery on the main power input.
I am assuming you are using a 5 volts battery for the two FPGA rails.
Another thing to put on the list: getting the selected for low phase noise clocks from this fellow - look here: analog research NDK low jitter clocks for digital audio - now these things make a big difference. Same clock as what was supplied but selected for the low frequency phase noise parameter. Not absurdly expensive considering how many he has to throw away.
Will have more tomorrow.
Dear Rick,
For NDK clock, I think it already been installed on SDtrans, it is mentioned on previous thread. Or I mix up other thing?
Regards,
Arthur
For NDK clock, I think it already been installed on SDtrans, it is mentioned on previous thread. Or I mix up other thing?
Regards,
Arthur
Hi Rick, when you will have a chance to put up the pictures could you please also publish one with the A123 26650 batteries you use. I am interested in the make and capacity, and where from did you buy these.
This would be really helpful as I recently read some online rumors that there might be fake batteries offered.
Last edited:
About battery voltage noise performance
It has been measured and reported that batteries (different types, including rechargeable ones) have by far much better noise performance and in the range of -165 to - 205 dBV/Hz scratching the limits of the measurement system accuracy.
Triggered by these measurements and results I got curious and, from my own hands on experience trying different batteries in my system, I can confirm that different batteries sound different and in line with the measured noise levels (NB this statement is valid for those types of batteries I had at hand) in the report.
Surprisingly, AA Li batteries did not measure the best (but were excellent by any means anyway). Batteries used by Rick were not part of this research. I am wondering how A123 batteries would compare to others wrt the voltage noise performance.
Last edited:
@Julf
Tapping into your experience in electronics, for batteries as a single power source, what battery parameters would you consider most important for the SQ and how would you rank it. I guess these would be (not in the order of importance) low voltage noise, low ESR, peak current, capacity, etc.
Tapping into your experience in electronics, for batteries as a single power source, what battery parameters would you consider most important for the SQ and how would you rank it. I guess these would be (not in the order of importance) low voltage noise, low ESR, peak current, capacity, etc.
Sorry to be unresponsive.
Took some pictures of the SDTrans with the caps removed but they were too close to show where the removed caps were.
If you are using an A123 26650 pair for input you can remove ALL of the first row of electrolytics - all those along the side where the input connector resides.
After you power the rails individually you can remove the caps after the the now unused caps. It is obvious which ones they are.
Here is the battery I use and recommend. there are clones and many have used them but for the slight additional cost I would go with A123s/
A123 System High Power Nanophosphate LiFePO4 26650 Rechargeable Cell: 3.2V 2500 mAh, 120A Rate, 8.0Wh
These folks are the official US distributor for A123 to consumers. There is no question these are the real thing. They offer clones, also.
This company offers a tab fixing service which I would highly recommend. I attached tabs myself on the first batteries I got and it is difficult. I worried I was harming the battery and wondered if I had a compromised connection. Which I probably did. I am using those batteries but not on the SDTrans. They have been relegated to lesser duties.
The clocks I mentioned are the same brand and model as the ones installed on the SDTrans - the difference is these clocks have been selected for excellent low frequency phase noise. The fellow I got them from says this parameter is what defines jitter and from what I am hearing I have to think he is correct. Many clocks are thrown away to get these good ones.
I installed these clocks and the batteries on the 3 3.3 volts rails in one operation. I had previously used a pair of these batteries in series to power the whole board previously. I know that separate supplies will bring about sound quality improvements but the magnitude of improvement I heard after could only be from the use of the clocks.
I will attempt photographs again.
Took some pictures of the SDTrans with the caps removed but they were too close to show where the removed caps were.
If you are using an A123 26650 pair for input you can remove ALL of the first row of electrolytics - all those along the side where the input connector resides.
After you power the rails individually you can remove the caps after the the now unused caps. It is obvious which ones they are.
Here is the battery I use and recommend. there are clones and many have used them but for the slight additional cost I would go with A123s/
A123 System High Power Nanophosphate LiFePO4 26650 Rechargeable Cell: 3.2V 2500 mAh, 120A Rate, 8.0Wh
These folks are the official US distributor for A123 to consumers. There is no question these are the real thing. They offer clones, also.
This company offers a tab fixing service which I would highly recommend. I attached tabs myself on the first batteries I got and it is difficult. I worried I was harming the battery and wondered if I had a compromised connection. Which I probably did. I am using those batteries but not on the SDTrans. They have been relegated to lesser duties.
The clocks I mentioned are the same brand and model as the ones installed on the SDTrans - the difference is these clocks have been selected for excellent low frequency phase noise. The fellow I got them from says this parameter is what defines jitter and from what I am hearing I have to think he is correct. Many clocks are thrown away to get these good ones.
I installed these clocks and the batteries on the 3 3.3 volts rails in one operation. I had previously used a pair of these batteries in series to power the whole board previously. I know that separate supplies will bring about sound quality improvements but the magnitude of improvement I heard after could only be from the use of the clocks.
I will attempt photographs again.
Just as with other components, there is no simple answer - it all depends on the circuitry you are feeding (and what issue you are trying to address). Digital circuits are in most cases not very sensitive to noise, but generate a lot of noise themselves - something that is best addressed by smallish capacitors as close to the individual circuits as possible, but low ESR is important for the power source.
Dear Rick,
you mean the 1st row of E caps are from 1 to 7 refer to attached photo? if powered by battery (only the main power barrel plug - 5V) those caps can be removed?
you mean the 1st row of E caps are from 1 to 7 refer to attached photo? if powered by battery (only the main power barrel plug - 5V) those caps can be removed?
Attachments
comickit,
When using good battery power for the input for ALL rails I removed:
1, 2, 3, 4, 5, 7 - all of the caps that are between input power and the onboard regulators.
When I separately powered the 3.3 volts rails I removed:
a, d, and e
b and c are after the onboard regulators for the 2 FPGA rails which are being powered by the main power input on my board. I figure those caps are needed for those regulators.
Cap 6, I presume, has something to do with the display, and not knowing anything about displays I left this one in.
Since I am using only REDBOOK files I removed the junper for the right most clock since it is not needed. Figure there is no need for it to be buzzing away.
I have got used to no display, removed that jumper. I will eventually install a switch to be able to turn it on and off.
Much of this is of the "cumulative effect" syndrome. Many of them, individually, do not make a noticeable difference but all together, I think, add to the fine sound I am hearing from this, truly, great device.
I feel sorry for those continually chasing after good sound from byzantine computer setups when, I suspect, they will never hear what this thing does, today. And the tidiness of the thing, even with all of the extra power supplies, is a great benefit.
I have to mention again, I heard a noticeable improvement when using a separate raw supply for the IIS rail. I am using three raw supplies, one for the 5 volts input battery and regulator(fpga), another for the IIS battery and regulator and the third for the clock rail and card/micro-processor rail's batteries and regulators.
I do not doubt a separate raw supply for each rail would be best just not sure it is worth the trouble.
Tried to make it clear. Hope I did not make it confusing!
Take care,
When using good battery power for the input for ALL rails I removed:
1, 2, 3, 4, 5, 7 - all of the caps that are between input power and the onboard regulators.
When I separately powered the 3.3 volts rails I removed:
a, d, and e
b and c are after the onboard regulators for the 2 FPGA rails which are being powered by the main power input on my board. I figure those caps are needed for those regulators.
Cap 6, I presume, has something to do with the display, and not knowing anything about displays I left this one in.
Since I am using only REDBOOK files I removed the junper for the right most clock since it is not needed. Figure there is no need for it to be buzzing away.
I have got used to no display, removed that jumper. I will eventually install a switch to be able to turn it on and off.
Much of this is of the "cumulative effect" syndrome. Many of them, individually, do not make a noticeable difference but all together, I think, add to the fine sound I am hearing from this, truly, great device.
I feel sorry for those continually chasing after good sound from byzantine computer setups when, I suspect, they will never hear what this thing does, today. And the tidiness of the thing, even with all of the extra power supplies, is a great benefit.
I have to mention again, I heard a noticeable improvement when using a separate raw supply for the IIS rail. I am using three raw supplies, one for the 5 volts input battery and regulator(fpga), another for the IIS battery and regulator and the third for the clock rail and card/micro-processor rail's batteries and regulators.
I do not doubt a separate raw supply for each rail would be best just not sure it is worth the trouble.
Tried to make it clear. Hope I did not make it confusing!
Take care,
Removing capacitors of any type, but especially electrolytic and ceramic capacitors improves the sound by getting out of the way of the A123 26650 battery. I am being battery specific here - this is my experience which has been shared by those who have tried it.
I am not recommending this be done with any other battery. I have no idea how well it would work with other batteries. Of course, I am not recommending the use of anything BUT this battery
The caps are there to fill in the blanks that the regulators cannot fill because of their infinite feedback which gives them the pretend low ESR (measured for very specific conditions which have little to do with what is happening when music is playing) but makes them slow and requires the capacitor to keep the current flowing.
The battery has no such limitations so the caps make what is a dynamically capable power supply, as nige2000 likens it: "spongey". No electrolytic can let go of its charge as fast as one of these batteries so you have the conceptual equivalent of tuned port loudspeaker's huffing and puffing trying to simulate real bass. I know that is not a proper analogy but it is similar in audible effect.
The caps become a very unreliable gatekeeper where none is needed or wanted.
I am not recommending this be done with any other battery. I have no idea how well it would work with other batteries. Of course, I am not recommending the use of anything BUT this battery
The caps are there to fill in the blanks that the regulators cannot fill because of their infinite feedback which gives them the pretend low ESR (measured for very specific conditions which have little to do with what is happening when music is playing) but makes them slow and requires the capacitor to keep the current flowing.
The battery has no such limitations so the caps make what is a dynamically capable power supply, as nige2000 likens it: "spongey". No electrolytic can let go of its charge as fast as one of these batteries so you have the conceptual equivalent of tuned port loudspeaker's huffing and puffing trying to simulate real bass. I know that is not a proper analogy but it is similar in audible effect.
The caps become a very unreliable gatekeeper where none is needed or wanted.
I would suggest looking up what decoupling capacitor do especially where digital is concerned. It not just the speed of the supply but the distance a supply is from devices and the parasitic inductance of the power delivery system. There is plenty out there on power delivery system integrity, removing decoupling and the larger reservoir caps is NOT recommended.
The digital also requires a stable supply, batteries without a regulator are not really going to supply this...
Only in audio!
The digital also requires a stable supply, batteries without a regulator are not really going to supply this...
Only in audio!
In what way are the capacitors "in the way"? They are not in series, but in parallel with the battery.
The regulators do not have "infinite feedback". In what way is the ESR "pretend"?
So you are saying that a less stable power supply makes for better bass?
And what, in your analogy, is the gate? And who is trying to get through?
Unfortunately "my faith trumps your science" is not in any way limited to audio...
True...
I am just surprised that such bad advice is been given regarding the power delivery system for a digital circuit...
Some rather interesting links, rather audio orientated from JS, as I am sure he would be aware of the availability of Power Integrity software from most of the CAD software vendors, never mind the wealth of information out there on power supplies and decoupling capacitors... the power delivery system includes not only the actual PSU but also the decoupling capacitors, the PCB layout, whether there are power planes and whether these are in closely coupled power/0V planes that will add planar capacitance etc. etc.
I am just surprised that such bad advice is been given regarding the power delivery system for a digital circuit...
Some rather interesting links, rather audio orientated from JS, as I am sure he would be aware of the availability of Power Integrity software from most of the CAD software vendors, never mind the wealth of information out there on power supplies and decoupling capacitors... the power delivery system includes not only the actual PSU but also the decoupling capacitors, the PCB layout, whether there are power planes and whether these are in closely coupled power/0V planes that will add planar capacitance etc. etc.
Indeed. The problem is that while simple analog circuits are still somewhat manageable based just on "intuition", hearsay and second-hand knowledge, understanding the demands of digital circuits is a different thing.
Hi Marce, if I understood you correctly, you question the results of SQ improvement in a specific system that Rick has developed with theoretical arguments about what works and what does not.
To be fair, do not just believe anybody but could you please do yourself a favour and actually listen to the SQ change (i am politely not saying improvement) in your dam implementation with batteries. If you do not have the dam, feel free to try it with another suitable device. Your hands on experience would we very valuable and welcome.
Not theoretical arguments, facts about how digital works and how critical decoupling is to digital signal integrity...
We are talking digital signals here, how is using a battery going to improve things... How is removing the decoupling capacitors going to improve things... How can to attribute analogue sound changes to a digital transport system, either the digital signal gets through or it doesn't...
As I have said my arguments are far from theoretical, they are based on facts and good engineering practice, its a bit farcical just removing stuff without understanding the reasons why they are there or what there function is....
Of course if you show me the bit stream and how this mod has managed to change only the bits related to the actual audio out of the data package then I will stand corrected, but I don't think that will happen.
We are talking digital signals here, how is using a battery going to improve things... How is removing the decoupling capacitors going to improve things... How can to attribute analogue sound changes to a digital transport system, either the digital signal gets through or it doesn't...
As I have said my arguments are far from theoretical, they are based on facts and good engineering practice, its a bit farcical just removing stuff without understanding the reasons why they are there or what there function is....
Of course if you show me the bit stream and how this mod has managed to change only the bits related to the actual audio out of the data package then I will stand corrected, but I don't think that will happen.