extreme ESD protection

Dec 17, 2005 4 Replies
B

You can get shielded ribbon cable, though it costs the earth. Sticking a solid chunk of metal under the board can help, though making the board four layer and devotog the two inner planes to solid 0V and +5V planes works a lot better.

Clamping a ferrite core around your ribbon cable might help - if you've got enough ribbon cable and a big enough ferrite toriod, it can be even better to wrap the ribbon around the torrod once or twice to make balun.

The real trick is to work out where the spikes are getting into your circuit, and how the current spikes get to traverse your board - if it is the current spikes that are the problem, as if often the case. Once you've found a path, blocking it is often relatively easy.

---------- Bill Sloman, Nijmegen

Vote
Twitter Facebook LinkedIn
D

Background: I've got a little board that runs my furnace, replacing a simple relay board that used to be there. The board is controlled by a generic PC sitting next to it, and itself talks to three digital thermostats:



formatting link



The problem I'm having is glitches. The latch that drives the furnace (latch->opto->alternac) occasionally latches in a random state, and the furnace starts up (or shuts down) only to revert a second or so later when the computer sends an update. This isn't rocket science; the signals are nice and slow (parallel port) and all ttl levels.



I've added 0.1uF bypasses across each IC, and a 4.7uF across the power leads, which reduced the glitching but didn't eliminate it. I added a bigger cap across the +5 but it still glitches.



The problem stems from an electrostatic filter in the furnace. Yup, bug zapper. Bzzt, glitch. Bzzt, glitch. Without the bypasses it would glitch randomly, now it seems to glitch only if the filter is sparking (still random, though).



The power for the board is tapped from the +5 in the PC over the ribbon cable. Yeah, not the best of designs but it was only the first design. I'm thinking ethernet or usb for the next one. Anyway, the



+5 is brought to the board, and distributed to the thermostats too. I'm thinking of using the +12 instead, with a 7805 at each board, to clean up the power.

Any thoughts on this? Am I wasting my time because of the ribbon cable? Do I only need more caps? Any ideas about where the glitch is entering the system?


Vote
Twitter Facebook LinkedIn
C

Running your 5v down a cable is a big no no, as is using 5vttl signals. Personally I always use a current drive signal/opto coupler arrangement or you could use a max232 driver. Noise is best supressed at its source so an emi filter on the filter should help.

Vote
Twitter Facebook LinkedIn
G

The best solution I would think is to figure out which sensors / signal paths have this noise, and to change the signal path from single-ended to differential.

When the glitch, known to be external, imposes itself on the now differential signal path, it disappears, 'poof'.

Geoff

Vote
Twitter Facebook LinkedIn
R

...and just slapping on bypass caps is not a complete "solution". One needs RC or Lc or RLC filtering; a lossy inductor can act as a resistor above 100KHz at the lowest end; the loss should be on the noise end or one could R-C-R the line. If the noise spikes are really fast, then the caps could resonate and make things worse; disk capacitors are good at that. So, wrap copper foil around the disk and solder to ground side; that ups the resonance to the GHz region - keep leads very short as they will add inductance...

Vote
Twitter Facebook LinkedIn

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required