Anyone listened to Li-ion batteries?

[abraxalito]

I have not read the article in Linear Audio (would very much like to, though) but notice that the A123systems battery has an impedance of 6 mohms at 1 kHz. Do regulators go much lower than this?

They can and do go lower than this, normally though they need remote sensing to compensate for the intervening copper to get into the micro-ohm region. However pure Zout may not be the most critical parameter. That's because when feedback regs are used to power classAB circuits (opamps for example) they're fed load-induced noise from the rails. Whether they swallow this wholesale or regurgitate it in some form seems to be quite relevant to subjective SQ.

 

[trobbins]

So are you saying there is only noise floor? I think a schematic of the actual test setup would be de rigour.

ELECTRONICS LETTERS July 1965 Vol. 1 No. 5, Knott, MEASUREMENT OF BATTERY noise and resistor current noise at sub-audio frequencies. This report varied battery current.

There has been a large body of research done on battery impedance measurement over a wide range of frequencies (for battery capacity and health determination), and the resistive element of that will be a noise base level.

 

[gentlevoice]

Hi,

I just found this which I guess is a reply to my question to jackinnj about the conclusions of his/linear audio's PSU test:

http://www.diyaudio.com/forums/powe...-ultra-low-noise-ldo-adm7150.html#post3744968

@abraxalito:

Whether they swallow this wholesale or regurgitate it in some form seems to be quite relevant to subjective SQ.

That's my guess as well and reading the reply in the link above - to me - is very interesting in this context.

@trobbins:

So are you saying there is only noise floor? I think a schematic of the actual test setup would be de rigour.

Hi - thanks for your question & I agree: I've attached a schematic of the test setup. I decided to not include any active circuitry in the setup so as to get as low a measurement noise level as possible (i.e. I could think of).

This report varied battery current.

... If you hover the mouse above the images attached you may see a text string indicating the various currents measured. From "low" (10 mA) to 0.8 A (the SLA).

Best regards,

Jesper

P.S.: The AD7760's input is slightly different as it also includes a built-in opamp, however, in noise terms I believe this is what is relevant (?)

 

[jackinnj]

- @jackinnj:
I have not read the article in Linear Audio (would very much like to, though) but notice that the A123systems battery has an impedance of 6 mohms at 1 kHz. Do regulators go much lower than this? And can you just briefly say (if you are at liberty to do so) what you found to be most significant in the linear audio test?

Jan posted the graphs on the LinearAudio website:
http://www.linearaudio.net/images/letters.pdf/V4 JW F8.pdf

Measuring impedance at microOhms is tricky. Walt Jung did a much better job of it than me!

You can bring the hf impedance of the battery supply down with film caps. With regulated supplies, however, film caps may fritz up the stability.

The correlations weren't conclusive -- noise seemed least important of all -- and PSRR at LF more important. What was concusive were the preferences of the observers for the Jung/Didden, Sjostrom, Burson regs.

The problem with any test like the one for LA is that you are limited to the reproduction equipment at hand. The preamplifier which was powered by the 12 or 13 regulators was chosen because it's PSRR was only 40dB. It's a very nice preamp btw, Borberly's "All JFET"

In hindsight, we should probably have included some samples of batteries.

 

[gentlevoice]

Hi jackinnj,

Thanks for your feedback ... Interesting that you found these PSUs to be the most preferred ones ... Again, without reading the article the Zout (as you sort of hint at by mentioning the PSRR) appears to be less conclusive as well as the Zout varies from quite low (Jung) to quite high (Burson).

Sometimes (if not always ?) a perplexing world, audio ...

Cheers,

Jesper

 

[1audio]

I think you are trying to measure below the input noise floor. This link will show just how hard it is to reliably measure battery noise: http://tf.nist.gov/general/pdf/1133.pdf You could measure an alkaline battery and see if you get similar results.

You need a really low noise preamp to start with. I used a transformer coupled preamp to get the self noise below 1 nV/rtHz. Even with that there are noise mechanisms in coupling caps etc. that are not easy to verify. What NIST did requires two preamps and some software that will do cross correlation (HPWworks HpW-Works does have this, none of the other audio software I know of has it).

Its important to be sure your looking at the right issue when testing. Is it isolation or supply impedance? Is the noise or the internal dynamics the key aspect that's influencing the performance. Batteries are not necessarily linear. They react in nonlinear ways to high and low current and to current in vs. current out.

 

[jackinnj]

Inone of the other audio software I know of has it).

Virtins has a package which allows cross-correlation with a sound-card -- Multi-Instrument - Spectrum Analyzer by Virtins Technology There is also some free-ware written by ham-radio guys.

You can "do" cross-correlation with some of the older analyzers like the HP35760 or the Stanford Research SR-785. But these machines cost $4k!

http://www.thinksrs.com/downloads/PDFs/ApplicationNotes/AboutFFTs.pdf

or with Labview: CrossCorrelation VI - LabVIEW 2012 Help - National Instruments

 

[gentlevoice]

@1audio &jackinnj:

Hi & thanks for commenting & suggesting once more

I think you are trying to measure below the input noise floor.

Its important to be sure your looking at the right issue when testing. Is it isolation or supply impedance? Is the noise or the internal dynamics the key aspect that's influencing the performance. Batteries are not necessarily linear. They react in nonlinear ways to high and low current and to current in vs. current out.

My main intention in this was to find out whether eligible batteries were actually noisy. From reading around this forum (and to my memory, also elsewhere) "batteries" often are considered noisy, something which - based on my subjective listening - is not what I have observed. I've always found that batteries that were sensibly dimensioned to the circuitry in question contributed with "room for" silence, spatiality, nuance, detail etc. on many levels.

That said I've also found that the various battery types - as well as different manufacturers' apparently similar chemistries - may sound differently. So quite likely they are not linear and also respond differently to discharge and charging conditions, load and more.

However, with my objective in mind - finding out if batteries could be considered noisy compared e.g. with a superb regulator - I would say that I have clarified this to the extent that I hoped for. So I'll leave investigating this further for now, yet keep in mind your & jackinnj's suggestions should the future bring me back in this direction.

Cheers - & thanks again to both of you ;-)

Jesper

 

[krokkenoster]

@1audio &jackinnj:

Hi & thanks for commenting & suggesting once more






My main intention in this was to find out whether eligible batteries were actually noisy. From reading around this forum (and to my memory, also elsewhere) "batteries" often are considered noisy, something which - based on my subjective listening - is not what I have observed. I've always found that batteries that were sensibly dimensioned to the circuitry in question contributed with "room for" silence, spatiality, nuance, detail etc. on many levels.

That said I've also found that the various battery types - as well as different manufacturers' apparently similar chemistries - may sound differently. So quite likely they are not linear and also respond differently to discharge and charging conditions, load and more.

However, with my objective in mind - finding out if batteries could be considered noisy compared e.g. with a superb regulator - I would say that I have clarified this to the extent that I hoped for. So I'll leave investigating this further for now, yet keep in mind your & jackinnj's suggestions should the future bring me back in this direction.

Cheers - & thanks again to both of you ;-)

Jesper[/QUOTE
Guys I was taught many years ago that batteries give the best DC for all work but I found that the normal Leclanchè cell is very bad when half discharged. Try to do fine work with only dry batteries and see the irritating flicker of the bulb. Incandescent type but LED's are worst I found the rechargeables better but also with the same characteristics! So for smoothing only non electrolytics will have to be found or made up! Imagine 10 K microfarads paper caps!
Anybody know where a body can find a airplane hanger ennexed to a house for just my pet hi fi albeit that one ear not hearing and the other as noisy as a old type mechanical 2000 type telephone exchange at peak hours!

 

[scopeboy]

I'm guessing that the internal noise of a battery would turn out to be the Johnson noise of its internal resistance. Should be much quieter than a 7800 type linear regulator, and about the same as a precision regulator with lots of capacitance on the output.

One great advantage of batteries is that they are floating. No worries with ground loops or common-mode noise.

 

[AndrewT]

Battery noise is more to do with the discrete "steps" of the chemical reactions.

Otherwise the noise would be so small, none of us would bother trying to measure it.

 

[AndrewT]

.....................One great advantage of batteries is that they are floating. No worries with ground loops or common-mode noise.

"floating" just remove one connection.
Wiring errors will still create loops with unintentional interference pick-up. Common mode noise is still probable with a "floating" supply.

eg.
all our isolated transformer secondaries create a "floating" supply.
All ClassII (double insulated) equipment have floating supplies that omit that one extra connection mentioned in para1 and they too suffer diff & common mode interferences.

 

[vacuphile]

Scopeboy,

In batteries, a chemical process goes on to sustain a flow of electrons, and this process by itself can be noisy. Therefore, battery noise is composed of Johnson noise, plus an excess noise component. The latter is dependant on the battery technology being used.

 

[scopeboy]

Yes, but a battery also looks like a huge capacitor (several farads for a SLA) that will partly shunt said excess noise. Lots of info out there on frequency response of batteries, for example http://www.telepower.com.au/INT95b.PDF

all our isolated transformer secondaries create a "floating" supply.

It is not completely floating, there can be 100-1000pF of capacitance between the winding and the other windings on the transformer. One of which is connected directly to the mains. It still works out more convenient to solve this by Faraday shielding, grounding etc, than to use a battery.

 

[AndrewT]

Everything with some physical size is capacitively coupled to Earth.
That includes battery supplies, isolated secondary windings, medical equipment, equipment Chassis when not connected to Mains PE and all the others.

 

[drewburn]

I use A 30V pack of 18650 cells to power my mk1 Pass Pearl. Previously use 24 V lead/acid. The only difference I can hear is that the 30V is a bit more Dynamic.

 

[trobbins]

scopeboy, the Intelec paper is just about the inherent frequency dependant impedance change of a battery and how that correlates to battery capacity. It has nothing to do with noise. The subtlety of that paper is that a test frequency needs to provide adequate discrimination of change in battery capacity, but not be swamped or aligned to all the other 'noise' generated by battery charger and load equipment that is connected to battery, which is why a particular test frequency and synchronised detection is used.

Did you check out the scientific papers on the subject of battery noise measurement, eg. the one cited in post #122 ?

 

[drewburn]

I use L-ion batteries in my Pass Pearl phono stage, this replaced lead/acid types. All I can say is that both types are quieter than a mains supply power unit.
The L-ion were more dynamic, this may be the result of a 10% higher voltage.