Battery-powered supplies

[tubenut]

WRT to the Graham Slee phono...

Are you measuring the wall wart supply or the upgrade supply? I would be quite sure the wall art is not regulated and thus under no load conditions would give a higher reading then the voltage would be under load.
There will be regulation inside the Slee. So, I would hook up a "test crib" of 24V battery power as per the specification of the PSU and it will either work and light up the LED or not if it is too low.
Do not go too high on volatage in to the Slee as perhaps the regulators are not designed (heatsink wise) to dump the xtra voltage....

[projectwhite]

Anyone know what a battery's impedance looks like at higher frequencies, around 20khz, 100khz? In particular NiCD or NiMH cells.

The closest info I could find was this: http://tf.nist.gov/general/pdf/1133.pdf. You can find a graph of noise up to about 100khz (figure 3). And since they say noise is predominantly due to Johnson noise, can we infer that impedance remains reasonably constant up to 50khz or so? Or will the battery become inductive (i.e., no Johnson noise, but impedance rises anyway)?

[mikelm]

Like any capacitor, a battery, is bound to to have rising impedance as frequency rises. Even if it is specifically designed for low impedance the will still be some rise with frequency.

[projectwhite]

Are we talking a rise of 50% by 100khz, or a rise of 100000%?

[mikelm]

Originally Posted by projectwhite

Are we talking a rise of 50% by 100khz, or a rise of 100000%?

My guess would be about 1000% but that's just a guess

I'm sure whatever it is it would benefit from a bypass of 1uF + 0.2 ohms

[projectwhite]

Even with a 47u+15u decoupling network the impedance is 0.3R@10khz and 3R@5Khz. I can't add a larger cap since I'm using all film caps. So I need to know if the battery is still going to be close to 25milliR at ~20Khz. Or cross my fingers and hope that PSRR saves the day?

It'd be even nicer to get rid of the 47uF altogether, but I have no idea how the battery performs at 100Khz. I must've gone through 500 papers on battery impedance, all of them dealing in milliHertz and state of charge. Couldn't find a single one showing impedance of a cell higher than 1khz :/

[projectwhite]

It looks like impedance of Li-ion and Nicd cells actually drops or remains constant going from 1khz to 10khz. One datasheet actually suggested using a Nicd cell as a ripple filter like a capacitor. This is looking good.

[Rainwulf]

I wouldnt.
A capacitor has a MUCH higher capability to supply surge current, and to take in charge current. A decent electrolytic cap can easily supply a surge current instantly in the range of 50 amps or more, and that is conservative. For example, a flash capacitor will smack out over 100 amps in microseconds. You wont get a nicad do that, especially a whole lot of them in series which you will need to match the voltage.
Thats why remote control cars have moved to lithium polymer cells, they are much more capable of providing high current bursts, but they to have their issues, specifically cell balance and overcharge/undercharge. If you do either to a lipo cell you risk damaging the cell, or even explosion.

You will also need to heavily regulate the power supply as well, otherwise you will get dendrite formation in your nicads.
As well, you will need to actually have your power supply change its voltage output depending on cell temperature, as their voltage changes with temperature.

Put it this way. I ended up buying a 180 dollar charger just to efficiently and safely discharge and charge nicad and lipo batteries for my remote control cars.

There is no way to replace a good solid well made electrolytic capacitor with respect to surge current capability, and resistance to.. ya know.. exploding when used in a standard PSU circuit.

[projectwhite]

These cells will be used in a supply that draws only 10mA (DAC analogue supply). Rise time is 200ns, and the smaller bypass caps handle everything above 500khz. The first question I think has been answered, i.e., the battery would maintain its impedance within the audio range.

Now the question is just how quick can the battery respond well into the tens or hundreds of Khz range. If it can, there's no need for large reservoir caps.

[Rainwulf]

The question is, why are you trying to remove those caps? Also, at that 10s of khz range, your caps wont be that big anyway. A decent 100uf cap will effectively decouple at that frequency range if the amp only needs 10ma.