Hi all,
I've been searching around without any joy for evidence of noise measurements of modern batteries or supercapacitors when used as power supplies for digital circuits.
I can see older references and measurements for lead acid and nicad batteries in various articles and in threads like this Battery-powered supplies - diyAudio
But I can't see anything similar for supercapacitors or for example LiFePo4 batteries.
I've heard very good reports from people using lifepo4 with a float charger to power their dacs, but I'd like to understand some of the theory and benefits.
Is anyone aware of any publicly available results comparing lifepo4 to other methods of low noise power supply?
Thanks,
James
I've been searching around without any joy for evidence of noise measurements of modern batteries or supercapacitors when used as power supplies for digital circuits.
I can see older references and measurements for lead acid and nicad batteries in various articles and in threads like this Battery-powered supplies - diyAudio
But I can't see anything similar for supercapacitors or for example LiFePo4 batteries.
I've heard very good reports from people using lifepo4 with a float charger to power their dacs, but I'd like to understand some of the theory and benefits.
Is anyone aware of any publicly available results comparing lifepo4 to other methods of low noise power supply?
Thanks,
James
Don't you think battery cells with a series resistance in the megohms would be kinda... useless?
Also, there is a rather significant amount of ground loop noise to be seen in your measurements, probably because both the DAC and the recording card were connected to the same computer. So your ground goes out one USB port, into the DAC, from DAC to recording card, and back into another USB port. That's a first-rate ground loop if I ever saw one.
You should definitely get rid of that before going into the fine details like power supply noise, either via a good ground loop isolator or a balanced frontend.
The easiest way may actually be using two separate computers for playback and recording, at least one of which must be battery-powered and as such floating.
It is interesting that your comparison did show a difference already, but it would be very useful to know how your external power supply setup is constructed.
And I'm not sure whether you realize, but between a notebook battery and USB +5 V there's about 2 switching regulators.
Actually digital circuitry is very rarely powered directly from unregulated sources. I suppose you could do it for the analog stages though, like in a headphone amp.
Also, there is a rather significant amount of ground loop noise to be seen in your measurements, probably because both the DAC and the recording card were connected to the same computer. So your ground goes out one USB port, into the DAC, from DAC to recording card, and back into another USB port. That's a first-rate ground loop if I ever saw one.
You should definitely get rid of that before going into the fine details like power supply noise, either via a good ground loop isolator or a balanced frontend.
The easiest way may actually be using two separate computers for playback and recording, at least one of which must be battery-powered and as such floating.
It is interesting that your comparison did show a difference already, but it would be very useful to know how your external power supply setup is constructed.
And I'm not sure whether you realize, but between a notebook battery and USB +5 V there's about 2 switching regulators.
Actually digital circuitry is very rarely powered directly from unregulated sources. I suppose you could do it for the analog stages though, like in a headphone amp.
Thanks again, but if you're talking laptop batteries, I guess they're either li-ion or ni-cad.
I'm interested in LiFePo4 and supercapacitors, both of which are relatively new and don't feature on any 'how quiet are batteries' articles or comparisons I can find.
I think you may be confusing your M-ohm with your m-ohm
Certainly the LiFePo4 have an internal resistance of only a few milliohms.
I note that the Red Wine Audio gear used to use LiFePo4, and has now swapped to a new version using 2 banks of supercapacitors which are alternately isolated and charged to keep the mains totally separate from the audio circuitry.
With large storage capacity and incredibly low impedance, I can see how both lifepo4 and supercapacitors would be ideal for audio, but I'd really like to see some technical proof if anyone can shed any light?
I'm interested in LiFePo4 and supercapacitors, both of which are relatively new and don't feature on any 'how quiet are batteries' articles or comparisons I can find.
I think you may be confusing your M-ohm with your m-ohm
Certainly the LiFePo4 have an internal resistance of only a few milliohms. I note that the Red Wine Audio gear used to use LiFePo4, and has now swapped to a new version using 2 banks of supercapacitors which are alternately isolated and charged to keep the mains totally separate from the audio circuitry.
With large storage capacity and incredibly low impedance, I can see how both lifepo4 and supercapacitors would be ideal for audio, but I'd really like to see some technical proof if anyone can shed any light?
No experience with LiFePO4 batteries but some with supercaps. Caps are different from batteries - no chemical reactions going on - so they're totally passive hence inherently low noise. The limitations of supercaps are their low working voltage and their relatively high ESR for the capacitance on offer. To use them in audio they need to be put in series with balancing circuits to ensure they share charge equally.
Vinni Rossi's using 9 Maxwell 350F supercaps in series to get to 24V supply however the ESR of the series combo is relatively high at 29mohms initially. Supercaps degrade in ESR over their lifetime. For the same money (these caps are $10 each on Mouser) I can get 50 off Panasonic NHG 15,000uF/6.3V giving me 750,000uF when paralleled at over twice the working voltage and an ESR of about 1.3mohm (under half that of the Maxwell cap). Of course the Panny caps are going to be considerably more bulky but that's the only downside I can see other than they take time to wire up.
Vinni Rossi's using 9 Maxwell 350F supercaps in series to get to 24V supply however the ESR of the series combo is relatively high at 29mohms initially. Supercaps degrade in ESR over their lifetime. For the same money (these caps are $10 each on Mouser) I can get 50 off Panasonic NHG 15,000uF/6.3V giving me 750,000uF when paralleled at over twice the working voltage and an ESR of about 1.3mohm (under half that of the Maxwell cap). Of course the Panny caps are going to be considerably more bulky but that's the only downside I can see other than they take time to wire up.
hm, i will measure batteries later today for correct resistance but i am pretty sure that it was M ohms not milioms, i am sorry i assumed that LiFePO4 was same as Li-ion battery and now i see it is not (Lithium iron phosphate battery - Wikipedia, the free encyclopedia) . where you got that battery and at what price if i can ask?
it would be nice if everyone would do some actual measurments and post them here as thread starter is not only one with similar questions
off topic:
using same batteries for battery powered drill, dac power, and led flashlight and i never had problem but i maybe measure resistance wrong ,can you tell me how exactly to measure ?!
i will measure direct loop of one usb soundcard today later ,i hope that there will be difference, but i think there will be no difference as my pc uses 2 psus as one is powering graphic card alone and is old power supply i think atx 2.1 and with lot of noise injecting in audio equipment , so i have to use separate 220v circuit for audio and its 100 years old house who knows whats in walls. i would use laptop for measure but my laptop battery is in dac power
it would be nice if everyone would do some actual measurments and post them here as thread starter is not only one with similar questions
off topic:
using same batteries for battery powered drill, dac power, and led flashlight and i never had problem but i maybe measure resistance wrong ,can you tell me how exactly to measure ?!
i will measure direct loop of one usb soundcard today later ,i hope that there will be difference, but i think there will be no difference as my pc uses 2 psus as one is powering graphic card alone and is old power supply i think atx 2.1 and with lot of noise injecting in audio equipment , so i have to use separate 220v circuit for audio and its 100 years old house who knows whats in walls. i would use laptop for measure but my laptop battery is in dac power
i will move soon so i will get rid of all problems
could easiest way be measuring single audio card loop?!
pictures later,maybe on other thread
i am not sure do you realize that main problem in pcs are switching, not linear regulators
i never saw pc digital circuit without at least 2 stages of regulators,inductors,capacitors thats including psus regs.
Cannot offer any pictures of graphs - but ...
nige2000, who has as much, and I suspect more, experience with the A123 batteries than anyone would say do not use ANY capacitor after them.
In his considered opinion they only do harm.
This has the great benefit of saving you money, trouble, and space.
He does adamantly state that one use the A123 cells. All others are boat anchors (remember that from the early days of CD? The comment by the distinguished, if agonizingly largo conductor, von Karajan?)
nige2000, who has as much, and I suspect more, experience with the A123 batteries than anyone would say do not use ANY capacitor after them.
In his considered opinion they only do harm.
This has the great benefit of saving you money, trouble, and space.
He does adamantly state that one use the A123 cells. All others are boat anchors (remember that from the early days of CD? The comment by the distinguished, if agonizingly largo conductor, von Karajan?)
You're funny. How did you take your measurements, with a multimeter in ohm mode on the battery terminals?
There aren't too many ways of measuring resistance. In a multimeter, I'd get a precise current source that's connected in parallel to the output, and then measure what kind of voltage drops over the resistor.
As you can easily tell, any kind of external voltage on the DUT would result in garbage readings.
I would measure battery ESR the same way you do with amplifier output impedance. Actually it's much the same as the current source approach outlined above, just for AC.
For this you will need a multimeter with a good mVAC range. Be sure you are aware of its limits, simple ones are often limited to 400 Hz max and optimized for 50/60 Hz; True RMS would be advantageous.
Get an amplifier that has a good amount of output voltage and flat FR in the frequency range of interest. Speaker power amps are good.
Connect a series resistor of several hundred ohms and a few watts on the output (much larger than expected battery series resistance - this approximately turns our voltage output into a current output). For many integrated amps and receivers these are already present on the headphone output right from the factory, often in the 330-470 ohm range. Verify by service docs or reverse-engineering.
Get a big fat capacitor with low ESR and at least 10000 µF or so, voltage rating not that critical, which you can use as a coupling capacitor. Connect that between the battery output and the cell under test, matching cell and coupling capacitor polarity sensibly.
Measurement procedure:
Step 1: Choose a test frequency. I would start with 50 Hz or so, batteries are kinda slow.
Step 2: Get something to generate test tones. A PC running Audacity would be fine.
Step 3: With the output unloaded, have the amplifier output a defined amplitude, measured with multimeter in VAC mode. Let's start with 2.0 Vrms or so. Once dialed in, turn test tone and amplifier off.
Step 4: Connect DUT cell to output, with multimeter in VAC mode in parallel. Turn equipment back on. Increase multimeter sensitivity if needed - I would expect tends to hundreds of milliohms on an AA cell, so voltage may well drop down into the single-digit mVAC range, and you may have to increase amplifier output level to more than 10 Vrms (unloaded) and put both channels in parallel in order to obtain a decent reading. If you have real low-impedance cells, this may no longer do, and you may have to resort to lower-value resistors connected to the speaker output. Expect coupling caps to be pretty massive at this point.
That tends to help things considerably, as the loop then only runs from output to input, and the direct connection between the two tends to be much lower in impedance so the "long path" doesn't have a lot to say. It would be of no use to you if you want to look at a DAC's noise floor though.
Looks like a second battery for your notebook may be in order. The very cheapest ones around 25€ or below may be somewhat dubious, but somewhat higher up you should be able to get something with decent quality cells.
Yup. If the resistor is inside the amp, it's hard to put it anywhere else anyway. You probably figured out that capacitor + goes to battery +.
That computes to 1.3 ohms at 1 kHz then. Sounds more realistic. (ESR of a decent 1000µ should be <= 100 mOhms or so.) What sort of cell / battery was that?
If you want to get lower in frequency, you'll need to gather some more caps.
Should you run into trouble with lower amplitudes, you can always try to calibrate a recording input (being very careful not to overdrive it - stuff with low-voltage analog stages like typical onboard audio tends not to like that at all, and some people have managed to kill their inputs while testing speaker amps without adequate input protection, like maybe a kOhm in series).
hm i put capacitor + to resistor side and - to + of battery, i used d-class bridged amp capable of almost +-15v at output maybe little more.
should i measure with switched cap polarity ?
i figured 1khz is good for 1000uf cap i have and i think i remember i saw somewhere laptop batteries are tested at 1khz , i hope to learn more maths as my very bad maths or no maths skill at all is making me use aproximations and that leaves me with lot of ????? its excellent thing that there is still people on this forum like you who actualy explain what they say and give real life information and helps others who wants to learn more. thank you for that!
i put single card measurment in zip if someone interested its html format.
this battery are A 0AFA5A code but no information on net , but classic oem laptop li-ion battery i assume
i have old behringer uca202 that could be useful for measurments like this, but now i dont have much time to try that
To measure battery impedance read the articles which Walt Jung wrote in 1995 for Audio Amateur -- archived on his website.
Battery noise can be so low that more specialized techniques are necessary -- cross correlation of FFT's -- there are some free-ware FFT programs which will perform cross-correlation, or you can do the manipulation in MatLab.
I scratch my head as a good power supply regulator will have lower impedance than almost all batteries available -- easily in the low microOhms.
Battery noise can be so low that more specialized techniques are necessary -- cross correlation of FFT's -- there are some free-ware FFT programs which will perform cross-correlation, or you can do the manipulation in MatLab.
I scratch my head as a good power supply regulator will have lower impedance than almost all batteries available -- easily in the low microOhms.
Hi,
If LiFePo4 is a good option for feeding a Clock Board:
What about inductance : which sort of cap would you putt between such 2 mHoms cells and a TCXO/OVXO ?
Something with higher ESR but very little inductance for instance ? smd COG for instance but Risk of oscilations ? (make the ground loop the shortest as possible ?)
Or will you put for instance between the A123 LiFePo4 cell and the clock board a litle more bulky super cap talked here in this thread which has a little more ESR than such cells but able to make a total passive reservoir cap, so a sort of firewall between the active circuit and the LiFePo4 cells?
"High ESR" tantal in smd package could be better ???
thanks in advance
If LiFePo4 is a good option for feeding a Clock Board:
What about inductance : which sort of cap would you putt between such 2 mHoms cells and a TCXO/OVXO ?
Something with higher ESR but very little inductance for instance ? smd COG for instance but Risk of oscilations ? (make the ground loop the shortest as possible ?)
Or will you put for instance between the A123 LiFePo4 cell and the clock board a litle more bulky super cap talked here in this thread which has a little more ESR than such cells but able to make a total passive reservoir cap, so a sort of firewall between the active circuit and the LiFePo4 cells?
"High ESR" tantal in smd package could be better ???
thanks in advance
The only linear regulator to use after a LiFePO4 battery is, perhaps, an LT3042. At this juncture, any other OEM regulator is going to worsen the noise performance. Noise is the most important factor for clocks. for preamplifiers, and line amplifiers seem (to me at least) PSRR and Zout seem most important.
In fact, skip the "after battery" regulator altogether, you can bring down the impedance by bypassing the battery with an aluminum electrolytic and film cap.
In fact, skip the "after battery" regulator altogether, you can bring down the impedance by bypassing the battery with an aluminum electrolytic and film cap.