parallel port input problem

[stirling]

Hi - I have a NC switch connecting input pin 13 to sig gnd. i.e. an open switch gives me an active high input. Trouble is I get occasional very short duration "phantom" highs on the pin even though the switch is closed.

Other inputs are configured active low with NO switches and they work fine.

Any ideas how I can get rid of these glitches very gratefully received.

Thanks

[Al_The_Man]

Probabally the whole PC is 'floating' above ground, try taking a ground wire connection from the port common to your machine system ground.
Al.

[stirling]

Hi AL - thanks for your reply - much appreciated.

I'd tried your suggestion just before I posted and it actually makes it worse! You're clearly correct (in my opinion anyway) in thinking sig gnd must be floating near logic 1 - but why would that be?

[neilw20]

What software are you using to read the pin?
Try a 100n (0.1uF) small capacitor to ground near the PC socket.
This can catch some gliches.
Caveat: If you need fast response to short pulses, this can kill - delete - erase the required signal.

[stirling]

Hi Neil - Thanks for your reply

I'm using Mach.
Your caveat may be pertinent. This is a touch probe input.

I'm still trying to fathom why sig gnd would be floating so near to logic 1

Ian

[gar]

071019-1034 EST USA


Pins 18 thru 25 are signal ground. Normally these should all be connected together and to the common of the PC power supply and also to the PC chassis.

If at the PC 25 pin connector nothing is plugged into it and you short a signal ground to your input with a very short wire, then you have a very rock solid logic zero into the gate. Very little possible external linkage by electric, magnetic, or radiated fields.

Therefore under these test conditions if you get a change of state it is almost certainly internal to the PC.

You can check for near 0 resistance between the various signal ground lines and also the chassis. Determine the shorted resistance of your meter leads first.

.

[stirling]

Hi Gar - thanks for your reply

Originally Posted by gar 071019-1034 EST USA If at the PC 25 pin connector nothing is plugged into it and you short a signal ground to your input with a very short wire, then you have a very rock solid logic zero into the gate. Very little possible external linkage by electric, magnetic, or radiated fields. Therefore under these test conditions if you get a change of state it is almost certainly internal to the PC.

Did this and left it for a while - no glitches reported. so far so good.

Originally Posted by gar You can check for near 0 resistance between the various signal ground lines and also the chassis. Determine the shorted resistance of your meter leads first.

Tried this but with varied results. When determining the shorted resistance of the meter leads sometimes it was stable around 2 ohms but then it would go all over the place between 5 and anything up to 150 ohms.

occasionally the inter sig gnd resistance would be a steady 2 to 5 ohms or so but then again could go all over the place.
The sig gnd/chassis resistance was the same. Maybe my meter is not so good...

Bit more info - it seems I get the glitches mainly or maybe always when the axis are moving.

edit: just been told the bar's open (nearly 6pm here) - will check back later

[gar]

071021-0525 EST USA

stirling:

Typical meter leads may read about 0.2 ohms with moderate pressure between probe tips. If I apply high pressure and wiggle the contact area I can get a reading of 0.0. My particular meter quantizes to 0.1 ohm.

Try the shorted lead test. If you get the large variations, or moderately high readings, not below 0.2 ohms, then you have lead or meter problems.

If the meter and leads are OK, then you have problems making contact with the connector or internal problems.

Good on the no glitch with shorted input. Is this true when the axes are moving?

.

[neilw20]

As noted by Gar, pins 18-25 are (or should be) all connected to ground.
I have seen el-cheapo leads assume this and only connect 1 or 2 pins on any of pns 18-25.
If all the pins are not connected together, (and maybe you have 2 leads connected in series) it becomes possible that the 0v line is open circuit.
I have seen PC Card connections that have not all been connected.
I have seen cables with the desease.
I have had various combinations of this and been bitten before.
Most problems of this nature are caused by cables, when connected to hardware that the designer decided (for various reasons) not to use all the pins.
Examples:
1: Cable not standard. 20 way cable is cheaper than 25.

2: The hardware can detect that a cable is not plugged in. The designer assumed all cables have 18-25 connected, so used this feature to detect a cable.

3: The cable has all 25 wires connected individually, but one has broken at a db25 connector.

4: Two 'printer' cables are connected in series, but one or both are el-cheapo.

In each of the above examples, a printer will still work. It has all of it's pins connected as the printer manufacturer knows of (or supplies) el-cheapo cables. Even might have got it's 0v connection through the mains cable. Still works, but prints garbage occasionally.

Bottom line. Check continuity of all 18-25 of each cable, at each end, and at every peripheral connector. You would be surprised how many you find that have at least one pin not connected as expected.

[stirling]

gar, neil

Thanks very much indeed - problem solved. I connected all sig/gnd (18-25) together and glitches have gone and I'm getting good stable meter readings at last- but more importantly the probe system is all working nicely (well apart from one small software problem). Clearly some of the sig/gnd pins I was using were floating as you suspected.

I really appreciate your help on this. Many thanks again.

ian

[neilw20]

Originally Posted by stirling gar, neil Thanks very much indeed - problem solved. I connected all sig/gnd (18-25) together and glitches have gone and I'm getting good stable meter readings at last- but more importantly the probe system is all working nicely (well apart from one small software problem). Clearly some of the sig/gnd pins I was using were floating as you suspected. I really appreciate your help on this. Many thanks again. ian

Ian.
You are welcome.
By the way, what's the software problem?

[stirling]

Hi Neil - I'm working on a 2.5D edge finding/contouring probing wizard for Mach3. Please see http://www.machsupport.com/forum/ind...ic,4456.0.html

It seems that there may be a bug in the G31 code. Currently Art is taking a look and hopefully there'll be a fix soon.

Basically after a G31 I back off with a G01 (all in incremental mode). Sometimes Mach doesn't execute the G01 so the probe tries the next G31 but of course it's already touching so etc. etc.

Ian