Switched capacitor phantom supply

[audiox]

I am designing a switching power supply to make phantom power from +/-15V. A swithed capacitor version looks attractive since there are no (hard to get) inductors or transformers. I am planning to use a simple voltage multiplier circuit (oscillator, few diodes and electrolytic capacitors). That gives 5-10 mA quite easily. The switching frequency is about 100 kHz.

I believe the most critical components are the electrolytic capacitors that are used for switching. They are properly polarised and ripple current is about half of their datasheet maximum. But I am still little concerned. Any long-time experience of that kind of design?

 

[JohnRoberts]

I once designed a phantom supply for a one mic DSP platform using a variant on capacitor charge pump.

I already had a DC to DC switching regulator, running to provide the high current 5V supply for digital electronics from a roughly 20V unregulated voltage. Since I had a HF 20V switching waveform driving the input side of the DC to DC power supply inductor, I just rolled my own two stage cap doubler/charge pump connected to this circuit node. I had a high enough switching frequency that tiny .1 uF caps delivered more than enough voltage and current for one mic. I used a simple transistor emitter follower pass element to clean up and regulate the final 48V feed.

Note: the ripple current will be quite small. I used tiny film caps for pumps, and only used electrolytic caps for the reservoirs.

There was a significant part size, cost, and circuit board real estate savings compared to running this doubler from 60 Hz, vs. the more than 100 kHz.

JR

 

[audiox]

The ripple current is actually quite high. With 5 mA of phantom current the AC current through electrolytic is about 20 mA. I tried it with film caps at first but it didn't work with phantom current higher than 1 mA.

 

[CJ]

ummmm, digi key is littered with =dc to dc, what part of antartica do you live in?


i can get a ski herc out there by mid/,,,,,,,,,,,,,,,,april?

is that ok for you?

 

[gyraf]

http://www.gyraf.dk/schematics/Voltage_multipliers_with_CMOS_gates.pdf

 

[JohnRoberts]

High or low is a relative term. The peak current is much higher for the 60 Hz doubler (average current the same). The IxR losses (the real issue) were small enough that I couldn't feel any heat rise at all after 24 hour testing at full load (one mic). The cheap smt film caps have worked fine, without failure in production, so I'm not inclined to fix something that isn't broke. YMMV.

JR

 

[audiox]

I will try film caps. There is however one problem. Polyester film datasheets specify max voltage rise time (for example Wima 100nF 63V -> 15 V/us). So according to the datasheets 100 kHz square wave can be harmfull to the caps? I am not sure if this is a real problem but it was easy to convert my oscillator from square to trapedsoid...

 

[mwkeene]

A square wave is not a problem for most caps, and its not the frequency, but the rise time that may be a problem for some film caps. Its also a matter of the amount of ripple current passing through the cap. Excessive ripple currents can and will cause caps to fail. Manufacturers generally provide ripple current specs for electrolytics and some films, while it is harder to find numbers for ceramic capacitors (ceramics can generally take the highest ripple current levels vs. size). I would just use electrolytics if size isnt a problem. After you figure out what your ripple currents are, derate by 50% and use that amount of necessary cap. Panasonic told me once that it was safe to use 100% of the rated value, but in my experience its good to derate that value to keep internal temperature rise low. Films can handle higher ripple sometimes, but I wouldnt use them based on cost.

-Mike

 

[JohnRoberts]

[quote author="audiox"]I will try film caps. There is however one problem. Polyester film datasheets specify max voltage rise time (for example Wima 100nF 63V -> 15 V/us). So according to the datasheets 100 kHz square wave can be harmfull to the caps? I am not sure if this is a real problem but it was easy to convert my oscillator from square to trapedsoid...[/quote]

The film caps are being driven by the HF square wave to pump tiny amounts of charge, not terminated across it. IIRC I used electrolytic for the storage caps. If still worried about edge rates you can add a small resistor in series.

JR

 

[12afael]

the tube mic pre from Paia use a Cmos DC-DC converter to supply the phantom voltage.

http://www.paia.com/prodimages/tubehsch.pdf