[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: RE: Re: How do I get rid of noise spikes on my Power lines

From: Jeff Mann   jeffjeffmann.com
Date: Mon, 16 Aug 2004 13:09:25 -0700

A! Sent via the Art & Robotics Group mailing list: arg-list@xxxxxxxxxxxxxxx
R! Use your "Reply All" to  reply to the list, "Reply" for private response
G!

Hi all -

Jessica wrote:
>>application.  If I use the +5 V output that does not run when the
>>Power supply's fan is on, my wireless network works perfectly.

I don't quite follow this - the 5V output that does _not_ run when the fan is on? You mean it runs with the fan off, but not when the fan is on? If the fan is not on, there shouldn't be any power coming out at all should there? I've never seen this. Is this on the motherboard connector? I only ever use the disk drive power connectors...

>>As soon as I turn on the other voltages
>>and the fan in the power supply, I get noise spikes on my +5 V
>>supply and I have no wireless communication.

There are good reasons not to use the 5V PC supply for your chips directly; one is that it has so much power that if you mis-connect something or hook it up backwards, the PC supply will happily set fire to all your chips! The other is that it's perhaps more noisy/spikey than eg. a 7805 regulator, so if you have sensitive analog electronics, I would also recommend putting a 7805 or 78L05 regulator on the 12V output instead. Bruce's suggestion to put the regulator after the long wire run is good. Just keep in mind that the regulator will have to drop 12V-5V= 7 volts; multiply that by the amount of current you need for your board, say for example 1/2 an amp, the regulator would have to dissipate 3.5 watts, which means it will get very hot and need a heatsink. But for smaller amounts of current (just powering a few chips) it shouldn't be a problem.

Having said that, I've had good luck running PIC chip circuits directly from the 5V of a PC power supply (with the fan running), even without any filter capacitors at all. I've heard a lot of people say this is a bad idea, and I don't disagree; just telling you that it worked for me. I mean, the whole point of these supplies is to power a bunch of chips on the motherboard, so it must be basically ok for that. I've even been able to run my r/c servo controller board, driving the motors from the same 5V supply as the PIC, although in that case I definitely needed filter capacitors - several .1uF ones you would normally put as close as possible to the power pins of the chips, as well as one 100uF or more to act as a reservoir for the small caps to recharge from. PC power supplies are nice because you can find them almost anywhere, and also most of them will work on both 110/220V, important if you're building it into something you want to take across the ocean. I've been able to find quite a few that will run with no load on them, which avoids the hassle of load resistors that Bruce describes.

Sandor wrote:
I have had great luck in eliminating spikes using 1uF nonpolar capacitors.

Why use non-polarized capacitors on a DC supply? I've never seen anyone do that, at that size of capacitor. At 1uF, I would normally use a regular electrolytic plus smaller ceramics, or else a 1uF tantalum (but I don't like those, they're expensive and tend to explode, and when they do they cause a short circuit...).

Step one. I would certainly suggest placing one across the power
lines that
feed the fan I would start by placing the cap as close as I can to the
actual fan inputs..

I wouldn't assume that it's the fan that's causing the noise; more likely it's the switching power supply itself when it's running normally (the fan being on is a good sign that it's running normally!). I mean, you could test it by disconnecting the fan and seeing if it makes any difference. I bet it won't.

Also, I would place one across the DC power lines
wherever they enter any circuit board after running along a cable.

This is a good suggestion, but again, why use non-polarized caps instead of just normal electrolytics?

Step three. After having done the above, I still had intermittent spike
problems, although they now occurred at different times. I found that
placing a 1uF nonpolar cap across from Signal to GND where those wires
entered a circuit board, completely cleaned up absolutely every spike.
Unfortunately it also cleaned up all my data pulses too. In order to get
around this I substantially lengthened the duty cycle for all my
data pulses
until they were of a long enough duration to get past the cap.
To do this I
just used a whole bunch of NOP instructions.

Well, that's certainly a creative/unconventional approach! Good if it worked for you, but I wouldn't suggest large caps on signal lines for everyone. (b.t.w. I think we've changed subjects here from power lines to signal lines...) yes you'll have trouble sending 5V TTL-type data signals across long wires; this will have more to do with transmission-line effects than spikes per se, and get worse the longer the wire and the faster you try to send data. At some point it makes sense to use a line driver chip, eg. RS485.

cu later <Jeff

A!
R!      messages saved at http://www.interaccess.org/arg/arg-list.html
G!      unsubscribe/help requests to mailto:Majordomo@xxxxxxxxxxxxxxx