EEVblog Electronics Community Forum

Electronics => Beginners => Topic started by: seb1982 on March 16, 2013, 10:36:43 pm

Title: How much rectification ripple is acceptable?
Post by: seb1982 on March 16, 2013, 10:36:43 pm
Hi everyone  :)

I'm planning to build Amspire's excellent little power supply over on the projects page in the next few days, and will be connecting it to the UK mains.  It needs a 24v DC, 1A input.

I've never done much with AC at all, and when it came to choosing my smoothing capacitor value, I suddenly realised that I have no idea what an "acceptable" ripple voltage is after rectification to aim for

Is 1v peak to peak OK, or should I be aiming for much less than that?  Any help with "rules of thumb" for this sort of thing would be much appreciated!

Thanks.

*Edited for shameful typos...  ::)
Title: Re: How much rectification ripple is acceptable?
Post by: ivan747 on March 17, 2013, 02:07:43 am
1st rule: make sure the lower part of your ripple is high enough to let your regulator regulate i.e. keep it above the drop-out.
2nd rule: if you know the ripple rejection ratio of your regulator (you will have to ask amspire) and the output ripple you are willing to tolerate on the regulator, you can calculate the maximum ripple you can have in the input. It is a logarithmic equation because ripple rejection is given in dB, you can watch Dave's video on decibels for engineers. Highly recommended.

If you are planning to build Mark I, check out this board I laid out. I haven't tested, sadly, but it should work:
https://www.eevblog.com/forum/projects/amspire's-power-supply-in-a-pcb/ (https://www.eevblog.com/forum/projects/amspire's-power-supply-in-a-pcb/)
Title: Re: How much rectification ripple is acceptable?
Post by: ftransform on March 17, 2013, 04:18:16 am
no ripple is acceptable ripple :--
Title: Re: How much rectification ripple is acceptable?
Post by: smashedProton on March 17, 2013, 05:07:28 am
You should be able to have more performance than 1 volt peak to peak.   Use an lc low pass rather than just capacitors.   F=1/(2pi*sqrt (lc))
Hope this helps

www.GarrettBaldwin.com

Title: Re: How much rectification ripple is acceptable?
Post by: Simon on March 17, 2013, 08:38:15 am
I seem to remember a formula along the lines of: 20'000/V/A for a bridge rectifier but that is a rule of thumb formula. I think the number that comes out is uF. Ultimately you do need to check if it's ok with the regultor
Title: Re: How much rectification ripple is acceptable?
Post by: seb1982 on March 17, 2013, 10:11:58 am
Ah, brilliant - thanks everyone - I see - I was just blindly using the formula:

V(ripple) = I(load) / 2f x C

Given a 50Hz UK mains, that was going to give me 0.5vpp ripple with a 20,000uF capacitor.  I'll aim for significantly better, then!  Those are really useful points about the lowest ripple peak being above the drop out (trap for young players, there :P) and I had no idea about ripple rejection, so that's great to know.  As ever, top notch video from Dave!
Quote
Quote
If you are planning to build Mark I, check out this board I laid out. I haven't tested, sadly, but it should work:

I had a look at this last night - absolutely fantastic job you've done there, Ivan!  I've got a couple of quick queries about this - I'll post them here so I don't confuse the  discussion about MkIII on the other thread:

Am I right in thinking that I can just leave JP5 (tracking pre-regulator feedback) unconnected until I'm clever enough to make one?  ;D

So I just put my nice, smooth 24v into both JP1 (power input) and JP8 (LM324 supply)?

I'm slightly confused by JP2 & JP4 (voltage & current inputs).  When you say "1v in = ...", do you mean putting an additional, separate voltage source in, or do you mean simply changing what's there through a multi-turn potentiometer?  If so, what values do you recommend?

I'm ever so sorry for being so stupid - I promise I will learn!  :palm:


Simon

Title: Re: How much rectification ripple is acceptable?
Post by: Thor-Arne on March 17, 2013, 10:40:52 am
I'm slightly confused by JP2 & JP4 (voltage & current inputs).  When you say "1v in = ...", do you mean putting an additional, separate voltage source in, or do you mean simply changing what's there through a multi-turn potentiometer?  If so, what values do you recommend?
I'll answer that as I'm also "playing" with this one.  ;)

The control inputs (JP2 & JP4) can be controlled with a potentiometer, the center tap goes to the input and one of the end taps goes to GND. The last one should probably be connected to a voltage reference of some sort.
Title: Re: How much rectification ripple is acceptable?
Post by: ivan747 on March 17, 2013, 12:13:14 pm
First of all, here's a good article on ripple rejection. It has the formula right in the first page, so you can solve for Ripple[output] if you have the PSRR.


Am I right in thinking that I can just leave JP5 (tracking pre-regulator feedback) unconnected until I'm clever enough to make one?  ;D

So I just put my nice, smooth 24v into both JP1 (power input) and JP8 (LM324 supply)?

I'm slightly confused by JP2 & JP4 (voltage & current inputs).  When you say "1v in = ...", do you mean putting an additional, separate voltage source in, or do you mean simply changing what's there through a multi-turn potentiometer?  If so, what values do you recommend?

JP5 is there to give you the current output voltage even if the fuse F1 is broken (hopefully to make troubleshooting easy), it is useful for tracking pre-regulators because these need to know the output voltage to set the input voltage. You can leave it disconnected.

You can feed JP1 and JP8 straight from the rectified and smoothed power rail you have. Just be careful when the power supply is not loaded with anything because the voltage on JP8 and JP1 will be higher and if gets over 32V it can damage the LM324.

If the output ripple is unacceptable (I would aim for 10mV at 1A) you can try reducing the ripple by adding more capacitance after your rectifier or you can try to filter the voltage only on JP8 using an LC filter like smashedProton said. The second option should be cheaper and smaller.

If Amspire's power supply had -40dB ripple rejection (I am sure it is closer to -60dB), 1V of input ripple would be 10mV on the output. At -50dB it would be 3mV. At -60dB is would be 1mV.
Wikipedia has a handy table on decibels. Use the amplitude ratio in this case: http://en.wikipedia.org/wiki/Decibel (http://en.wikipedia.org/wiki/Decibel)

JP2 and JP4 are used to set the output current and voltage. Let's say you feed 1V on each of those. For now ignore where that volt comes from. The output would be 10V with a 1A current limit. If you feed 0.5V on JP2 and 0.1V on JP4 you get 5V on the output and the current limit set to 100mA.

Where do we get these voltages from? It's your choice. If you want to use a multiturn pot, you connect the wiper pin to JP2/JP4, the first pin to GND and the third pin to 2V if you are using JP2 or 1V if you are using JP4.

You can get the 2V with an LM317. The 1V is trickier to get because the LM317 goes down to only 1.25V. Take the 2V reference and divide it in half with a voltage divider. Buffer that with an LM358 configured as a voltage follower and connect the current pot to that. Feed the LM358 with more than 5V but less than 32V, you can use the 24V you have. If you are not going to use the second op-amp included in the chip, configure it as a voltage follower too, but leave the output disconnected.

The good thing about this power supply is that you can upgrade stuff. If you want, you can build the tracking pre-regulator later.  :-+



Title: Re: How much rectification ripple is acceptable?
Post by: seb1982 on March 17, 2013, 05:00:59 pm
That's utterly brilliant - thank you so much for taking the time to type all that.  I'm learning so much since joining this site (except I'm becoming all too aware of just how much I have yet to learn!)  I'll go the LM317 --> LM358 route for now then for ease - I can always tinker with PWMs and DACs for fun later - I just need to get something working at the moment!

Incidentally, does anyone have a good method of splitting the 24v input wire three different ways to connect to the board(s) in three different places?  I've scoured Farnell and RS, but all I can find are normal terminal blocks, which would need jumper wires connecting the holes in parallel, which doesn't seem particularly clean or elegant...