messing with voltage doublers on the simulator...

I mocked up a voltage multiplier loosely based on peterc's scheme found at http://1176neve.tripod.com/id26.html

I'm not sure if there's a better way to calculate this with the simulator, but from (virtually) loading the supply, it seems like preregulated I get roundabout 69V with a thevenin source resistance of ~450ohms.

now, still preregulation, once I've pulled 40mA through it, I'm already down 25V!

I guess my question is, what sort of current demands do I need to design for when building mic pre phantom supplies?

The current supply I have on the sim is regulated with a TIP31 and trimpot to set the bias voltage. Not the most elegant solution I know, and if the load impedance drops, the output voltage drops considerably. I'm simulating it with a 3k3 load, which I imagine would be the same load as one mic with a short in it, and the supply stands up to that ok.

What sort of realistic maximum current demands can I expect for 4 hot condensors operating normally?

P48 specs say 10mA max.

47uFd is much too small for a voltage doubler. It might be acceptable for 40mA with a straight rectifier, but multipliers suck and you may need to up-size the caps by about the square of the voltage ratio. Since you are tripling, think 470uFd, not 47uFd.

[quote author="tmbg"]I guess my question is, what sort of current demands do I need to design for when building mic pre phantom supplies?[/quote]

The biggest current demand from any microphone I've ever seen is 11 mA into an Earthworks microphone, so I usually shoot for 12 mA per channel, which will be totally overkill for probably 95% of the time.

Worst-case for all eight lines shorted looks like an 851 ohm R, or 56.4mA. Allow for some idling current for your regulator and its voltage divider, and you could budget for ~60mA and be very safe.

well, I've settled on a design that I'm reasonably happy with... I replaced the crappy single transistor regulator with an LM317, and the simulation claims to be able to supply 60mA with no problems.

I know it's not 100% perfect design, but I made some compromises (particularly on capacitor costs and size on the higher voltage parts, C3 needs to be 100V, and 100V 470uF is ENORMOUS and expensive).

BTW one of the things that we haven't touched on in these PS threads is the importance of ground routing. Sometimes schematics are made to show this explicitly, although working with layout people I find they can be capable of ignoring the most emphatic indications of what goes where.

Anyway---look at the current paths and keep the big pulsed currents from flowing in a conductor that is going back to your "coldest" point. Sometimes this is not that terribly easy to see. Don't cop out and just try to use a ground plane because it's lower Z than you have the desire to calculate (although sometimes in higher frequency work it's hard to do anything else than use the plane).

There's a discussion of the effect routing has on three-terminal regulator performance in this guy's on-line articles: http://www.tnt-audio.com/clinica/regulators_noise1_e.html (there's a part two so far as well).

Brad

what's wrong with ground planes?

I'm lazy and pragmatic, I put ground planes on BOTH sides of my boards

here's an article I found pretty interesting, seems like a simple way to clean up a simple regulator design:

http://www.techlib.com/electronics/finesse.html

tmbg: "here's an article I found pretty interesting, seems like a simple way to clean up a simple regulator design"

Yeah, Norm Thagard put me on to that article a while ago, and I did a fair amount of design and sim afterwards to see how far it could be pushed, in particular extending the idea to a two-pole circuit (amounting to a frequency-dependent negative resistance, with which you can make a 12dB/oct lowpass, with it to ground and a series resistor from the source). I ended up concluding that the two-pole one was impractical but the designs on that site could be substantially improved.

The whole exercise got me to looking at ultimate low-noise voltage references which produced some interesting results. Let's just say that Jung's work is not the final word.

BTW, the article about regulator noise does have an error when it states that the TL431 shunt regulator must be used with a shunt cap to be stable. The actual situation is more complex, in particular the dependence on current. The part is stable below a certain capacitance and stable above another, with a region of instability in the middle, all of these regions being affected by current through the device. This is shown in detail in the TI databook.