Power Supply Design

[PhillDubya]

Hey guys,


I am about to build another bench linear DC power supply, one specifically with 3v outputs for my MSP430F uC's, and a couple adj. 12V outputs. Anyway, I was just wondering how important it is to use ceramic caps on the primary before the voltage regulator? Do they really make a considerable difference? What would be your suggestions for a clean voltage output, and decent ripple factor as far as cap values, and especially the types, i.e. ceramic, mylar etc... I was just wondering as I was wanting to make this bench PSU a little better than the last one I through together, where I just threw a 2000uF cap before the NTE 965, and went with that.

The ripple factor equation I am using is: rF= (4*sqrt3*F*C*Rl)

Thanks

 

[dknguyen]

Just use ceramic. Best balance of cost, size, and performance, at least for those capacitances (0.1uF or 0.01uF). It starts to get expensive at 1uF but still doable. Beyond that it gets expensive for ceramic. Unless you are doing RF, it is usually best to stick to ceramics for ICs and electrolytics when you need realy large capacitances.

No need for mylar or other fancy types only used for RF circuits. Good performance but huge and expensive.

 

[PhillDubya]

Allright, went with the ceramic, hopefully it will maintain between (2 - 5)% rF.

Thanks for the response.

 

[Ubergeek63]

you want X7R or NPO materials, and not Y5V.

Ideally every decade from 0.01µF to 10pF in parallel, but generally a 0.01µF and a 150pF is fine. The reason is that each value has a minimum impedance at successively higher frequencies. Of course this has absolutely nothing to do with the power supply and everything to do with decoupling the chips in the load.

Dan

 

[dknguyen]

X5R works well too. It just can't get as hot as X7R before becoming out of tolerance (85C vs 125C).

 

[PhillDubya]

Transients?

Thanks for the help!

I have one more issue, that has just developed though. On my previous power supply that I built, I used a regular two contact, spst switch to cut the hot line coming into the power supply to turn it off, as anyone would do.

When I was in radio shack buying a switch, this time I bought a dpst switch so I could cut both the hot and neutral, I figured why not. However, I am blowing fuses like crazy, and I have taken the circuit board out of the box, and checked it, took the dpst swtich out and just turned it on by plugging it in, and its fine??

Any ideas? Is this normal for a dpst switch? Could it possibly be transient current spikes from the dpst switch?

My first thought was that something was shorted, but as stated, I removed from the box to troubleshoot, and everything is fine.

 

[MrAl]

Hi,


What helps a bit is to estimate the peak to peak ripple voltage,
which is approximately:

Vpp=6000*I/C

where

Vpp is the peak to peak ripple voltage in volts
I is the load current in amps
C is the capacitance in microfarads (uf).
The frequency is 60 Hz and it is full wave rectified.

For example, a 10v rms 60Hz input to a full wave bridge when filtered
with a 6000uf cap and 1 amp load will mean there will be about 1v peak
to peak ripple. You can compute ripple factor from that if you wish.

If there is significant resistance in series with the input to the
full wave bridge (often there is with many wall warts) then the
ripple reduces to approximately 1/2 of the value calculated above,
even for small resistances like 2 to 4 ohms.
This would make the new formula:

Vpp=3000*I/C

If you are after very low ripple then a lot of capacitance helps,
but regulation does wonders and with some regulators there are
additional ways to reduce output ripple (see LM317 for example).

The smaller ceramic caps often used in parallel with the larger
electrolytics are usually there to swamp high frequency components
and are often used on the input AND the output. A typical value
is 0.1uf and that's usually enough.

 

[PhillDubya]

Thanks Mr. Al for the response.
I used .1uF ceramic caps on the primary and secondary of the voltage regulator, while having an initial ripple cap coming straight off the full wave rect. diodes of 2200uF (electrolytic).

However, I think I figured out why I was blowing fuses...

I am assuming that the coil initializing current surge of a transformer is fairly large. I am using a 117 to 28 step-down transformer which is roughly a 4.2 ratio. Therefore, I figured using a 250mA fuse on the hot line coming into the transformer would be enough because that would allow me to pull: (4.2)(.250A) = 1.02A on the secondary which is plenty for my power supply, however when I replaced the .250A fuse with a much larger one, it does fine.