PDA

View Full Version : [Question(s)] 5v power supply cuasing noise on speakers.



Connor
01-01-2009, 04:24 PM
Okay, on my bot, I have a pair of USB speakers I've been using.. They get terrible noise on them.. I've tried plugging it into a USB port on my main machine.. (leavaing the audio plugged into the bot) and the noise goes away.. I've used a external 5v supply.. same result.. noise goes away. I've hooked it up to the 5v rails direct from the power supply on the bot, I get noise. Not motors involved, just a Mini-ITX mainboard with a pico-PSU. (this one here http://www.logicsupply.com/products/pico120wi_25) What can I do to fix this, it's driving me nuts!

Thanks, Connor

Quantum
01-01-2009, 04:53 PM
Sound like the power on the rails is a little noisy. A 0.1uF capacitor between both lines should help to clean this up.

Paul

Connor
01-01-2009, 04:57 PM
Sound like the power on the rails is a little noisy. A 0.1uF capacitor between both lines should help to clean this up.

Paul

Yea, something is odd with it.. the noise on the speakers change dramatically as the CPU increases usage.. I thought I had tried adding a cap, but.. I added a 1.0uF instead of a 0.1uF. Doh!

Thanks, Connor

Adrenalynn
01-01-2009, 05:20 PM
Do you have a 'scope you can put that power supply on?

Check all your ground connections across the entire 'bot.

Connor
01-01-2009, 07:51 PM
Do you have a 'scope you can put that power supply on?

Check all your ground connections across the entire 'bot.

Yes/No.. I have a Parallax USB PC based one.. Which works... I did put it on at the "5v" setting, it looked clean, I turned the voltage setting down to 1v to get better resolution.. and did see some minor spikes and fluctuations in the voltage, not anything very large.. but, maybe enough to cause the noise? What should a perfect "clean" regulated supply look like one the scope? A perfectly flat line?

Thanks, Connor

lnxfergy
01-01-2009, 08:07 PM
Yes/No.. I have a Parallax USB PC based one.. Which works... I did put it on at the "5v" setting, it looked clean, I turned the voltage setting down to 1v to get better resolution.. and did see some minor spikes and fluctuations in the voltage, not anything very large.. but, maybe enough to cause the noise? What should a perfect "clean" regulated supply look like one the scope? A perfectly flat line?

Thanks, Connor

The parrallax unit is likely not going to show accurate waveforms up signals that you could hear (the sampling rate is gonna be too low I think). However, the fact that you are seeing spikes (define "minor" - plus or minus 5% is most likely acceptable, 10% may be, 50% is likely not!) means you could have issues.

A single cap on the power supply is also probably not adequate - Capacitors filter transients at different frequencies - add a few caps 0.1uf and 10/47uF are my typical combination on all of my micro boards - but not sure about audible frequencies.

-Fergs

Connor
01-01-2009, 08:24 PM
The parrallax unit is likely not going to show accurate waveforms up signals that you could hear (the sampling rate is gonna be too low I think). However, the fact that you are seeing spikes (define "minor" - plus or minus 5% is most likely acceptable, 10% may be, 50% is likely not!) means you could have issues.

The software shows a setting of down to 50us.. Which is where I had it.



A single cap on the power supply is also probably not adequate - Capacitors filter transients at different frequencies - add a few caps 0.1uf and 10/47uF are my typical combination on all of my micro boards - but not sure about audible frequencies.

-Fergs



I'm attaching two screen shots of the 5v from the scope..

Thanks, Connor

jes1510
01-01-2009, 09:46 PM
It doesn't look like the spikes you show are the source of the noise. It looks like the noise you show is around 40kHz. the 3rd harmonic would be around 13kHz, it bit high to be considered a "hum". I'm going to guess ground loop.

For future reference you'll get a better view of stuff like this with the coupling set to AC assuming the scope supports it.

Adrenalynn
01-01-2009, 10:30 PM
Yup - looks like it. So I'd go back to my initial suspicion and suggest going back over all the ground connections. Then I might think about a torroidial core try to get rid of some of that low freq.

Connor
01-01-2009, 10:43 PM
That's what I don't understand... This is pretty much a computer on wheels.. Everything was USB based until just recently.. (the sonars) and this problem existed before then.... and thus, grounded to the main board via the USB cable. I'm not arguing, Lord knows you all know more about electronics than I do.. but.. there just simply not that many wires on this thing.. I just don't understand why it go better when I used a separate power supply.

Thanks, Connor

jes1510
01-01-2009, 10:48 PM
I have some clamp-on ferrites that may do the trick. If you don't have it solved by the the next meetup I'll bring them. We can test a few different ones.

Al1970
01-02-2009, 12:12 AM
Hi:

You answered your own question. You know it is the power supply. Audio chips are not like digital chips. Audio chips are made to amplify. So noise on the power rails is going to be amplify. I only took a fast look but it looks like you are using a switcher power supply. They are noise and if not design well, very noise. lf I was you, I would put a linear voltage regulator to where you feed the power to your switcher and use that to power your speakers.

Al
http://www.geocities.com/robotsandyou/