Battery then regulator power supply question

[pepelevamp]

Hiya fellas. Been a while since I posted but I have a noob question for ya'll.

Like lots of us I'm looking at building my own power supply. I'm wanting to build some audio gear and I'm dealing with some real tight SNR margins of like way past -100dB. I'm Talking into -120dBFS range, with reference voltage of around a volt or two. Not sure what that gets us in microvolts.

Aaaaaanyway I'm a poor bogan from New Zealand so I want to make a mechanically switched power supply that simply kicks in or out batteries. Batteries eventually sag though so I want to look at putting a regulator on the business end to get the voltage I want. To hell with parliamentary procedure and efficiency.

My question here is - am I barking up the wrong tree with wanting to put a regulator after a battery? With respect to the noise on the line. I know I'm already asking for trouble with putting series resistors in there to limit my current - but thats another story. Or is it?

Do you lot have any recommendations on good low noise regulators? Say something which has crappy ripple rejection but is ultra clean on its output if given a clean input?

Thanks again & sorry if this went into the wrong category. Admins feel free to move the thread & give me a slap on the wrist. All hail king Dave.

[TimFox]

Batteries typically have low noise, but they do discharge under load and may have mediocre temperature co-efficient.
As a compromise, you could use primary batteries (non-rechargeable) as a voltage reference, and buffer the voltage to your load using low-voltage-noise op amps, readily available, with good PSRR driven from normal line-operated regulated power supplies.  Your noise will probably be limited by grounding and shielding problems, anyway, plus the unavoidable noise from amplifier active devices and resistances.

[pepelevamp]

Hmm. so you are suggesting that i have batteries which are used for the voltage reference, but dont supply current?

And then a second battery tank which does supply all the power, through sexy opamps (good PSRR), using the references from the original primary battery+zener reference?


so the grunt battery would output anger into a regulator, then it would go into a sexy opamp's power rails. the sexy opamp would take the voltage reference from the cowardly yet pristine voltage reference supplied by non-chargeable low current voltage reference zener or something?

my load (say an amplifier im workin on) would get its high quality power from the opamp?

Thanks boss.

[TimFox]

If you use high-capacity batteries to supply the load, then the PSRR of the op amp is less critical, but you need to be careful about the load current.  For your case, I believe you are not sensitive to voltage accuracy but are trying to avoid power supply hum.  There would be no need for Zeners, which have their own noise issues.

[pepelevamp]

Hmm now I think I'm more confused. Could you perhaps describe again the circuit you initially suggested?

Hum's a problem for me yep, but so is general thermal noise on the line. I just want clean battery power that I can select on a dial. Only even need a few values I guess. 12/5/3.3/1.2. Would love some negative rails. Maybe I can put my ground lead half way up my battery tower to get a cheap negative rail.

The ADC I am using can determine -120dB full-scale with AC-couple mode. Looking at around 10^(-12) volts I think. Whatever -120dBu is in volts. Ultimately my SNR or THD margin that I want any device under test to have is 10^(-10) volts clean power.

But you're right - I am risking a lot by having this existing on planet earth and shielding is indeed an issue. The whole thing is going to be shielded more than the starship enterprise tho. That I can do
What can you tell me about zeners and noise? I was unaware of this until now.

[TimFox]

First, -120 dBu is 775 nV (7.75 x 10^-7 V), not 10^-12V.  0 dBu is the voltage (0.775 V) corresponding to 0 dBm (1 mW) in 600 ohms, the traditional reference impedance for audio work.
My original suggestion was to use a primary battery (non-rechargeable) as the voltage reference for a solid-state regulator using a good op amp, stabilizing the voltage from a reasonably clean source.  A larger battery (possibly rechargeable) could supply that power at the current required for your load, in theory avoiding any line-related hum.  A second circuit could use negative voltages.
There is a large literature about Zener diode noise.  The minimum possible noise is the shot noise of the current flowing across the potential drop of the Zener voltage, similar to the shot noise of electrons flowing through a vacuum diode, as opposed to the current flowing through a resistor, which is a different problem.  Laboratory noise generators often use a Zener diode as a controllable noise source.  There are other "parasitic" noise sources on top of the shot noise.

[pepelevamp]

this is really cool mate, thank you. im just reading about shot noise.
i really should be doing my uni assignment but well .. screw it this is more interesting.