I have a 5 volt power supply that is from a switched mode type supply from futurlec. This unit gives out +-5V, both which need to be further filtered. I need a really clean output with the lowest AC components possible. I have attached a picture of the signal coming out with 20mV divisions and 5usec time divisions.

I tried using simple filter techniques but may have to move up to higher order active filters or something. I'm also considering a zener diode regulator in which I would lose a few volts which is not ideal. Does anyone know of an easy or accurate way to clean this up. Any input is greatly appreciated!

 

Use an LC filter on the output.

One side of the inductor on the output, the other on the capacitor. Connect the other side of the capacitor to ground.

Is it one output which outputs either +5V or -5V?

If yes, then take care to use a non polarised capacitor.

If there are two independent outputs, then you will need two LC filters.

 

Looks like your supply has transients in the 166kHz range to well over 1MHz.
A couple of LC pi filters (one per rail) should take care of that.

1uF cap to ground, 100uH inductor in series, and another 1uF cap to ground will give you a filter that is ~-20dB down by 53kHz, and better than -40dB down by 100kHz. Keep the cap lead lengths as short as possible. Radio Shack sells 100uH ferrite inductors for a couple of bucks; they're axial instead of toroidal so you should really put them in a metal box to keep the radiated noise down.

 

Great! I have a few unknown inductors laying around and I'll try them out first. By the way, they are two independent power supplies.

 

Get one each of the small and medium size Goldmine surprise boxes. One of the ones I recently received had a lot of 5 mH per side common mode filters and by the size of the wire (appears to be #20 AWG) they're good for 11A. You could hook them up as series or in parallel. A nice cap on the input and output and you'll be set.

Figures out to be around 5.2K of reactance @ 166kHz yet still able to pass over 10A of current. I would put some small disc or film caps around 0.1 uF or better as well as some good electrolytics both before and after the chokes, 470 uF or better @ 16V should be fine as they'll not only add more filtering but help to reduce any transients should you load change quickly.

You can scavenge these common mode filter chokes from about anything nowadays and they're handy to have around. To hook the two coils in series simply run a diagonal wire between the pins, additive I'm reading 18.85 mH due to the mutual inductance, that would equate to a heck of a lot of filtering and still 10A of allowable current.

They all vary by design thus the reason I tend to suggest getting one of those hand held LC meters off of eBay. Not the same one I have but I'm sure any of these would be quite sufficient:
http://cgi.ebay.com/3-1-2-LCD-Induc...lectrical_Equipment_Tools&hash=item3a62d8e773
http://cgi.ebay.com/3-1-2-LCD-Induc...lectrical_Equipment_Tools&hash=item45ecebf1d4

I've checked mine against my B&K bench equipment and they're pretty much spot on as far as accuracy goes.

 

Awesome results with that simple Pi LC filter. Thanks a lot for all your guys help on this. Below is an attached waveform of my (very clean) heartbeat. After I fix up the output a little more and add some calibration stuff I will post this simple ECG schematic in the projects section for a very low cost. My real plan is to make an EEG machine but building the schematic to view a heartbeat is a better indication that everything is working fine before I try to detect brainwaves.

 

Great that you have clean power, but I would ditch it!!! Why? If the power supply fails you could kill someone, low chance, but really bad result. Say switch to batteries and use a couple of regulator chips. Most semi manufactures do not even want you to use their components in these types of devices lest you sue them.

 

I have opto-isolators so that I won't have to worry about that!

 

I have opto-isolators so that I won't have to worry about that!

With just battery power on the human side of the isolators? ( or no power at all )?

Even so are you willing to bet someones life on it? Is 99.99% good enough?

 

I understand, and was going to move to batteries anyway. Especially if I'm going to post the stuff on this site. If there is any danger involved it wouldn't be right to post on here and put myself or anyone else in a dangerous position.