Noise measurements for LiFePo4 and supercapacitors?

im currently questioning whether i should go for Li-Ion or LiFePo4

as far i saw A123 are pretty much the only "really good" LiFePo4 cells where on the Li-Ion end you get a few to compare

did any of you come across some noise measurements between those ?

as i see it:

Li-Ion Pros:

Cons:
- a bit unsafer to handle, LiFePo4 are more forgiving on fault conditions
over/undervoltage/overcurrent/shortcurcuit protection is probably good to have


Now the big question would be if in reality LiFePo4 cells are less noisy... beside this i see honestly more pros for Li-Ion
 
your comparison is too vague especially those "pros", for noise under load look at the internal resistance (impedance) lower the better, regardless of cell chemistry, however 3.2/3.3V LiFe/A123 is much more versatile as it can be used to power digital circuits directly
 
Hi ... A couple of thoughts on your post above (some of which you may already know, of course):

  • I think that in practice both Li-ion and LiFePO4 batteries may be virtually noiseless. That is, depending on their capacity and design, it is indeed very difficult to measure whatever noise they may generate on their own with even very-low-noise capable measurement equipment.
  • One more parameter that might be relevant is that Li-ion battery voltages drop some when discharged whereas the LiFePO4 battery voltage stay almost the same at various discharge levels. So depending on your use LiFePO4 may provide a more stable voltage.
  • Also there's the rated battery voltage to consider - at least in my context the two battery chemistries need be used specifically for some purposes.
  • I have used both types of batteries for many years now and the main difference I notice in (my) practice is that batteries - like IME most everything else audio - have a sound of their own meaning that different models sound differently. E.g. the A123 26650 versions (not the later nanoparticle version which I have not tried) sound very dynamic and "insisting" to my ears whereas some of the Li-ion batteries may be less so (I have not listened to that many models, though) ...

Cheers,

Jesper
 
your comparison is too vague especially those "pros", for noise under load look at the internal resistance (impedance) lower the better, regardless of cell chemistry, however 3.2/3.3V LiFe/A123 is much more versatile as it can be used to power digital circuits directly
for 3.3V this is true, but for anything else i would say we have the same problem with both types

i read this thread trough before posting here, someone suggested using diodes to drop the voltage... might be a good passive solution worth a try

I think that in practice both Li-ion and LiFePO4 batteries may be virtually noiseless. That is, depending on their capacity and design, it is indeed very difficult to measure whatever noise they may generate on their own with even very-low-noise capable measurement equipment.
yup thats also what my research suggests....
i saw a graph where pretty much all battery types were at -160db noise where a LM317 was at -120db

One more parameter that might be relevant is that Li-ion battery voltages drop some when discharged whereas the LiFePO4 battery voltage stay almost the same at various discharge levels. So depending on your use LiFePO4 may provide a more stable voltage.
true i just rechecked.... you can easly get 3.4 - 3.2V with LiFePo4 where with Li-Ion you approx can get 4.1V - 3.5V (while sacrificing approx 30-40% capacity, tho i wonder wether its a positive thing to not completely drain the battery in terms of ESR on low charge)

I have used both types of batteries for many years now and the main difference I notice in (my) practice is that batteries - like IME most everything else audio - have a sound of their own meaning that different models sound differently. E.g. the A123 26650 versions (not the later nanoparticle version which I have not tried) sound very dynamic and "insisting" to my ears whereas some of the Li-ion batteries may be less so (I have not listened to that many models, though) ...
hmm i kinda hoped high ampere Li-Ion would sound on par, tho i have no reference at all beside USB powerbanks which are probably more often than not worse than a AC wallwart

my plan before finding this thread was getting as much voltage as possible out of 2 cells (to keep it somewhat simple) to either power opamps direclty with two supplies or use regulator for 3.3V/5V as this seemed like the most universal approach and the problem with no good 5V source is with both battery types a problem
 
Hi again ... just another couple of comments 😉

i saw a graph where pretty much all battery types were at -160db noise where a LM317 was at -120db

Batteries typically are at a much lower noise level. If - like kinsei mentions - you take the internal resistance/impedance of the battery as a noise level parameter e.g. 10 mohm gives about 12 pV of noise (1 Hz BW). If you then relate this to e.g. the 4.1V of Li-ion batteries then you are at -230 dB levels. For a 20 kHz BW this corresponds to just about - 185 dB. As a practical - and to my memory - example the diyaudio member Gerhard has measured battery noise using his 220 pV noise level measurement equipment, and if I remember correctly the batteries added no noise. So indeed & generally very low noise levels.

with no good 5V source is with both battery types a problem

Lithium titanate, or LTO, batteries have a voltage of nominally 2.4 but in my experience 2.5V per cell is fine. Available in different versions where e.g. the Toshiba SCiB 2.4AH cell has an internal impedance of 1 mohm. Probably not easy to find though.

my plan before finding this thread was getting as much voltage as possible out of 2 cells

Please note that balancing of the individual cells is mandatory as they very rapidly develop voltage imbalances. And regarding placing a regulator in-between the batteries and the audio circuitry in question I personally would not do this. Besides low noise, batteries also have another positive characteristic and that is a high internal capacitance meaning that they can deliver high current levels, typically instantly (depending on the model of course).

Regarding using two batteries for an opamp another option could be SLA batteries (sealed lead acid). Float charged they deliver appr. 13.6V per 12V battery - and the internal impedance is quite low - to my memory on the order of 10 mohms. High level of safety albeit they take up a bit more space.

Cheers,

Jesper
 
Batteries typically are at a much lower noise level. If - like kinsei mentions - you take the internal resistance/impedance of the battery as a noise level parameter e.g. 10 mohm gives about 12 pV of noise (1 Hz BW). If you then relate this to e.g. the 4.1V of Li-ion batteries then you are at -230 dB levels. For a 20 kHz BW this corresponds to just about - 185 dB. As a practical - and to my memory - example the diyaudio member Gerhard has measured battery noise using his 220 pV noise level measurement equipment, and if I remember correctly the batteries added no noise. So indeed & generally very low noise levels.
two series connected Li-Ion would have around 30mOhm (+connectors/holders, so more like 50-100mOhm + 50 mOhm per relay etc...), probably still a good idea to use some larger low esr caps with either battery type in the end

also any IC/opamp/regulator can ruin the low noise characteristic pretty quick, so comparing Li-Ion to LiFePo4 in regards to noise is probably a red hering..

Lithium titanate, or LTO, batteries have a voltage of nominally 2.4 but in my experience 2.5V per cell is fine. Available in different versions where e.g. the Toshiba SCiB 2.4AH cell has an internal impedance of 1 mohm. Probably not easy to find though.
wasnt aware of those! tho i never saw charger IC for it

1mOhm sounds pretty intruiging, i will look em up 🙂

Please note that balancing of the individual cells is mandatory as they very rapidly develop voltage imbalances. And regarding placing a regulator in-between the batteries and the audio circuitry in question I personally would not do this. Besides low noise, batteries also have another positive characteristic and that is a high internal capacitance meaning that they can deliver high current levels, typically instantly (depending on the model of course).

Regarding using two batteries for an opamp another option could be SLA batteries (sealed lead acid). Float charged they deliver appr. 13.6V per 12V battery - and the internal impedance is quite low - to my memory on the order of 10 mohms. High level of safety albeit they take up a bit more space.

Cheers,
i wanted to use 33- 80ohm passive balancing A. "loading" the cell with a bit of current B. balancing at the same time C. only passive on the "clean side"

specifically for the 2 cell series Li-Ion setup (ic also supports LiFePo4..) i wanted to use BQ25171 charger IC + TPS3700 voltage detection + passive balancing

i didnt know 12V batteries perform so well, tho if i remember right you quite a huge voltage range of 13.6V - 10V, might be no problem going with a regulator setup tho...

maybe there are some lead acid projects out there, i will take a look
 
https://www.lifepo4-battery.com/Products/LTO-Lithium-Titanate-Battery/480.html

usual 18650 LTO cells have also something along 15 mOhm impedance, so in theory if you need more cells its going to be worse.. BUT you can charge LTO cells simple with a current limiting voltage regulator, which might be very well preferable over charger ic's with unknown noise levels
also a beefy ldo like 5A is possible..

a range of 4.4V to 5.2V would be possible with two cells which fit the usb standard, but realistically you want to go down to 4V to use up the already low capacitance of LTO's ( 1500mAh for 1 18650 cell ), and overall i feel like with usb devices you dont want to have such large voltage "swing"... so realistically we also end up here at regulator + more cells

one pro for LTO: using a regulator, also allows you to use a beefy one... like 3-5A which makes the battery charged in just a few minutes which the LTO's can whitstand
specially this pro makes me consider LTO ... very little downtime between full isolation battery power

also im wondering whether it would be easly possible to integrate supercaps into the mix (if not large low esr electrolytic caps)
 
LiFePO4 batteries "noise" increases as they age. They also have "burst noise" as they age: Kumarevalu et al, "Burst Noise in LiFePO4 Batteries and its changes after long term calendar aging", Journal of Power Sources 581, 233430. (Open access). There is a burgeoning, staggering field of knowledge emerging in battery noise, factors affecting impedance etc. owing to the advent of EV's and energy storage.

I would also point out that the impedance of the batteries discussed in this thread is a minimum of 1 milli-Ohm. Jung's Super-Regulator beats this by several orders of magnitude (as does Salas in those regulators of his which I measured.)

Gerhard Hoffmann's article would seem to rank NiCad as lowest in noise -- but these are now difficult to source. (probably next to impossible in the EU).
 
Last edited:
Parraleling double layers Supercaps instead (10F, 25F, to acheive low impedance & filtering ? Same behavior than such LiPoEtc cells w/o cons and isolation enough from the main PS ?
yea probably also valid.... big capacitance + wire resistance will work like a big RC filter starting at sub 2 Ohm

i contribute the audible effect supercaps do actually more towards this RC filter configuration then low ESR... since on my usb filter i use a 1,5F cap with like 30 Ohm ESR... and it does still something audible wonderful, for me looking into battery supplies is just a step to 1. not worry about AC at all in my builds (toroidal EMF, EMF trough the grid etc.., 230V..) and start from the "cleanest source" which definitely batteries seem to be

also one should forget in comparison to connectors/relays the ESR of batteries is fairly low, but its the connectors making it rise fairly quick, so i still think the main low esr will still come from local bypass caps

Also i looked at some normal caps, 2200-10000uF is about 40mOhm, so 4 of these give you approx. 10 mOhm and you circumstance the batterie holder contacts, its really a red herring...

.... i might actually go for 4 LTO cells giving me a 8-9.4V which is a pretty perfect 9V range, tho i only found 18650 cells, do you know some 21700 ones (or similar size)

imo a big PRO for LTO's (or a supercap power supply with two banks, one charging one in use) is you can rapidly charge them with 3-5A, which seems key for a small downtime
Li-Ion give you greater runtime, but also the downtime is fairly large, if you dont trigger recharging when you are done you might end up with a complete next listening session in charge mode... exactly this lets me gravitate towards supercaps or LTO's
 
  • Like
Reactions: diyiggy
Super Cap = equal shortest gnd loop : come back to the cap (quasi DC) not the power supply ?
hmm yes, i guess in the end listening tests would need to show if you need/want more than only supercap "buffers"

for me personally its just that you need to start at 230V anway

----

maybe the actual "ultimate power supply" would be this:

two banks of 4x 3.0V supercaps in series, isolated ADC (mcu on charger side), two banks and you set the voltage range you would like
for 5V .... 4.6 - 5.4 .... mcu will set charging voltage/ldo to 5.4V and charge the other bank and switch between load and charge bank, could this be feasable?

question is... how long would the charge last for such a voltage range on "not too big" supercaps, i definitely dont wanna go bigger than 350F, way to costy, preferably smaller.... tho i guess even a 5min load/charge switch cycle would be ok in practice

this would approach would allow for a fairly certain voltage range, any you like actually with no regulator needed!

Edit: maybe one could make it more universal with only two supercaps in series where you wire two pcb in series for higher voltages
 
I have a biasing problem with caps. I do know (dunno?) How to tone with capscaccording a setup (loudspeakers etc). Super caps migth have a so low esr that putting normal lythic cap after even // migth be risky or worse (résonance and all the nasties).
Wonder where is the equilibrium.
I checked 10 y ago that lifepo was worse than à good low z output reg, and it was for a masterclock.
 
Last edited:
@jackinnj ... Thanks for mentioning the article on noise in batteries in your post above 😉 ...

I have actually searched for this kind of information on the internet many times - and over time - however, have never really found anything specific. Yet the article you mention does indeed provide very usable information on battery noise. As far as I can see for LiFePO4s in the main article and then additionally some Li-ion cells in the supplementary information.

To me very interesting to learn that the combination of higher charging voltage, higher discharging current, and aging all contribute to raising the batteries' noise levels. I would have expected this to be the case with the discharging current level, and the aging - but learning that even small changes in the charging voltage may cause quite significant noise level increases is a surprise. Good to know!

Gerhard Hoffmann's article would seem to rank NiCad as lowest in noise -- but these are now difficult to source. (probably next to impossible in the EU).

You may be right about this ... as I mentioned it was to my memory - and I also think that Gerhard in general didn't measure batteries with a larger capacity than around 500 mAH (??). Anyway, the article you mention give more factual information about the noise in battery types that I reckon could also be interesting in an audio context.

And then one question, if I may: During your trials with various regulators for audio have you ever found some kind of "well-defined" audible correlation between PSU/regulator output impedance and the sound quality?

Also a brief follow-up to my post #105 above: I just checked the datasheet of a Panasonic SLA battery, 12V 7AH LC-R127R2PG, and can see that its internal impedance is ~21 mohm. So not 10 mohm, as my memory was, but a little higher.

Cheers & thanks,

Jesper
 
@gentlevoice What was found was that there is a weak correlation to PSRR and perceived audio quality, but it was not statistically significant.

A reader of the article suggested that the perceived audio quality may relate to the THD that the regulator's error amplifier may impinge upon the supply rails, after all it's an amplifier. When we looked at the data again and compared the auditioners comments this was "mostly" born out, with the Jung-Didden and Sjostrom confirming the positive reviews, the LM317 and SuperTeddy confirming the negative reviews. The test was performed by putting a small signal on the supply rail and running an FFT.

I suppose we should have used some batteries as a reference!
 
I would also point out that the impedance of the batteries discussed in this thread is a minimum of 1 milli-Ohm. Jung's Super-Regulator beats this by several orders of magnitude (as does Salas in those regulators of his which I measured.)

thanks for sharing the "burst noise" information!
but i would like some inputs regarding the so called impedance: how do you compare the internal resistance of batteries with the output impedance of regulator circuits? in terms of voltage noise? and by orders of magnitude better than 1 milliohm?
 
@jackinnj: Thanks for your feedback 😉

I reckon there are indeed many parameters involved in the sound of a given circuitry, however, at least from my perspective the power supply is one of the key components in shaping the sound.

In this context I have personally found that batteries have a good bit of merit - and then over time I have also come across the Jung/Didden & Salas regulators here on diyaudio as other approaches to quality active power supply regulators. My "carefulness" here - if I can say it like this - is exactly their being active circuitry, thus potentially injecting their own characteristic noise/distortion. On the other hand I reckon that their active influence on the circuitry in question also tends to attenuate the sound characteristics of e.g. additional decoupling capacitors. With regards to this: in your tests can I ask you if you added additional capacitors to the circuitry you were listening to - or were there no additional capacitors besides e.g. the 120uF HFQ capacitors in the Jung/Didden circuitry?

... Lately I suppose there is also the Didden/bohrok2610 silent switcher which IMHO would be interesting to get an "ear sound assessment" of ... And, well, given the huge success of the GB I suppose this will happen at some point in time.

@diyiggy :
Parraleling double layers Supercaps instead (10F, 25F, to acheive low impedance & filtering ? Same behavior than such LiPoEtc cells w/o cons and isolation enough from the main PS ?

I have not done much listening to Supercapacitors but a couple of years ago I listened to a 3F & a 50F supercapacitor from Maxwell. To my ears their SQ was overall quite bland - greyish, undynamic, & closed-in sounding. My personal guess would be that such capacitors are as differently sounding as e.g. electrolytic capacitors - but maybe without the dynamic "presence" that I experience with many batteries ... (?)

Cheers,

Jesper
 
  • Thank You
Reactions: diyiggy
thanks for sharing the "burst noise" information!
but i would like some inputs regarding the so called impedance: how do you compare the internal resistance of batteries with the output impedance of regulator circuits? in terms of voltage noise? and by orders of magnitude better than 1 milliohm?
Yes, I believe that there is a chart of the Zout for Jung and other regulators in Walt's articles "Regulators for High Performance Audio". I never quite attained the level Walt measured, but got down to around 10 microOhms.