Battery Vs conventional PSU

Status
Not open for further replies.
Just to move things along ......
Here's the original question from the other thread:
lagoausente said:
Hello,
I´m going to build a mic preamp, it´s a nice design based on the Green pre. It can be supplied to +/-20v even it would work ok with less voltage. Anyway, four 9V batteries could do the job.
The issue is that I want the mic preamp to be portable, so have the option of the batteries, or DC-DC converter.
As DCpreamp told, batteries would be less noisy than any DC-DC converter, so that´s the way I´ll go, probably.
My doubts here are about the implementation for the battery setup. Only caps? or a linear regulator?
If understood ok, DCpreamp only use caps, isn´t?

Whell, I thought about use Li-ion batteries, that comes with built-in pcb protection inside, see http://www.batteryspace.com/index.asp?PageAction=VIEWPROD&ProdID=2644

or this other new type that are safer, and non explode: http://www.batteryspace.com/index.asp?PageAction=VIEWPROD&ProdID=3071

NiMH, have similar price, and suffer of much higher self-discharge, so those Li-ion seems nicer for a low consumption mic pre.
I would like to now, any information of the Li-ion "noise" if it has, and the Z, aswell if the PCB that comes with the first model, maybe can have an influencia on the noise or Z, since it´s an IC.
The second type, is a new Li type, that has no IC, but it´s more stable and safer, so have advandage of the low selfdischarge, without the problem of the Li-ion ones..
I would like to know your opinions, and any help of what caps would I need to implement the DC PSU.
Thanks a lot, nice thread.


lagoausente said:
What I want to know, is if is better to use a regulator, or maybe worse, with batteries. And what type of filtering would need to decrease the Z as low as posilble, I think with caps can be done.
If I put large caps between + and 0 and 0 and - rails, would get low impedanze? how much cacity? Need to ad any coil?

(PS, Lagoausente: Did you expect everyone here to read that whole thread to find your question?)
 
(PS, Lagoausente: Did you expect everyone here to read that whole thread to find your question?)

Sorry, I was thinking about getting the answer of one of the people who posted there. Just thought that they are not seeing the thread again since in the Solid State section, it comes down very quicly.
Thanks for the point, and for the quote.
Any idea of how can be measured the Z of the battery psu?
 
Hi,
yes, TNT is based on simulation and yet others are interpreting TNT's results as over-riding National's guidance for placing decoupling/snubbing caps on the output pins of the 317/337 regulators. I have asked "who do you believe"?
 
I have been told that the Z can be measured with a Fet and a resistor.
To measure the impedance of the supply, regulator or just the batteries.
Use a power FET as a switch with a non inductive power resistor which will provide a load of say 100mA or possibly more (1Amp).
Use a square wave drive of variable frequency, say 1 Hz up to 50 KHz (more if you are interested) and with a 'scope measure the drop in terminal voltage of the regulator / battery, when the load is switched on and off. Ideally there should be no drop in voltage at any frequency but by using a calculator and scratching your head for a while you can work out what the impedance of the supply is. Try this at low and high frequencies.
You could then see the effect of adding various types and sizes of capacitors to your 'supply' with the aim of reducing the dip in output when the FET switches on.A BUZ11 FET is cheap, mount it on a bit of metal as a heatsink although it won't need much as it should be saturated when on. The Non inductive resistor should be suitable for the possible dissipation and be on a reasonable heatsink. Input about 6 Volts or so of squarewave perhaps through a 1K resistor for safety and to suppress possible oscillation of the FET. Test away to your hearts content!!

Quoted from here: http://www.gearslutz.com/board/geekslutz-forum/120848-eval-board-portable-ad.html
 
Hi,
if the resonant peak (or not so peaky) is truly there as predicted by the simulator, what would the result look like as one changed the frequency of the square wave load?

I guess that the voltage will appear (on a scope) to drop progressively with increasing frequency. But, if the peak is there then the voltage will rise again over a small range of the troublesome frequencies.

Since we would be measuring DC voltage on the supply rail would the normal narrow bandwidth of a DMM detect the voltage drop?
 
IMO, that old paper about noise on 3-terminal regulators, that describes noise peaking, is an important read. It's the reason I don't just slap very low DF caps on regulator outputs, but prefer caps with some finite and predictable DF like tantalum. Another alternative is a small value resistor in series with the cap- I keep some Manganin wire on hand for such things. As for output impedance, it's going to be low regardless, making it tough to measure accurately. For a non-bypassed battery my first attempt would be with something like an old GR 1603 Z-Y bridge, as this can measure AC impedance on components with a significant DC level. The thing is similar in principle to the antenna noise bridges used by hams, and a simple version could be tossed together on the bench- easy to build, hard to calibrate. You might also inject a white noise source (from PC) using a coupling cap and resistor, then do a SA of what's across the resistor- you're just making a network analyzer. There's nothing wrong with the square wave method (it was also described by Jim Williams in an LT databook, and probably an EDN article for looking at supply ringing with various capacitor combinations), but I don't know how one would put numbers on it, other than an FFT of the signal. Seems good enough for comparisons though. For a good high power level measurement, just drive the supply through a resistor and coupling cap using a good stable wideband power amp. Look at the supply output on a scope- the amount of signal and current through the resistor will give you the output impedance. I've only done a couple of these things, as my usual assumption is that with the right bypassing, the impedance will be sufficiently low so as not to matter, at least for my purposes.
 
The question about battery impedance got me curious enough to make a few quick measurements on the Z-Y bridge. It doesn't resolve much below half an ohm, but should give us a clue. First I grabbed some freshly charged Panasonic 1950 mAH NiMH AA cells. The impedance at 10kHz was about .5 ohms for a series pair, so a quarter ohm per cell. No significant phase shift. Then I tried a Gates Cyclon pack of six 2.5AH cells in series (12V pack). This was less then .5 ohms for the pack at 10kHz, so .08 ohms per cell. No significant phase shift. Next was an Eveready 9V battery. Not an alkaline, and not fresh, but still with reasonable life left. 27 ohms at 10kHz, and with a small change in X. At 100 Hz it was 38 ohms, and 32 ohms of capacitive X. Converted to different terms, it looks just like a 49uF cap with a DF of 1.18, a 40 degree phase shift. Last was a 9V Eveready alkaline, also midway through its life. Impedance was 3 ohms at 100 Hz, and 1 ohm at 10kHz, with insignificant change in X. Sorry, but I don't have any lithium ions handy. One should probably test impedance with a DC load, but that's all the time I had. My conclusion is that the rechargables do quite well, with the alkaline in second place. A bit of local bypassing and one should be good to go. I buy a lot of non-alkalines for things like wall clocks and flashlights, as they're cheap, but I'd avoid these where you want a stable voltage under a varying load- like audio equipment! Or, use a regulator.
 
Status
Not open for further replies.