EMF / RFI radiation issue?

[SoundsDigital]

I have recently replaced a 16F628A with a 16F887 in a circuit in order to gain some extra ports. The new circuit has tested correctly on the bench, but when I move it to the application, it stops responding, despite the old circuit with the 16F628A working fine in the environment. I have tried connecting the new circuit up to a pc and sliding it across the floor towards the old circuit and I can see it lock up at about 2 meters distance.

It is one heck of a long program which is intended to measure items moving along a conveyor belt and fold them in the centre, while also sending information back to a PC via the parrallel port (I know the normal method is the serial port, but this was working on the bench and we have several other interfaces connected to various parrallel ports and want to keep some degree of continuity)

To get to the point, the application has several 3phase AC motors and 2 frequency inverters, which I know can upset some sensitive equipment if mounted very close, although I have taken care to keep the circuit as far from the inverters as possible. The old circuit containing the 16F628A was mounted directly below one of the frequency inverters, yet even at a 2 meter range, the 16F887 gives trouble.

Are there any known issues with the 887 that do not effect the 628A with EMF or RFI and are there any ways to combat it?

Hi, its great to see you visiting our forum. Why not try Proton Compiler for FREE? Download the FREE version of Proton Compiler, Its called Amicus18 and its available from HERE Already using proton Compiler??? Get rid of these pesky messages... get LICENSED USER STATUS

[John Drew]

I remember in the past it was considered unwise to have a parallel port lead that was greater than two metres long although they worked at more than that.
Have you tried a ferrite clamp around the wire near the PC? A downside is that it might have a bad affect on your signals too.
John

[wastrix]

There is no particular reason why a 16F887 would be any more susceptible to EMI than the 628A.

As John some ferrites might help high frequency noise, but "parallel" (that doesn't mean much - is it shielded, what kind of data are you sending?) is really not a good idea for this kind of work. A differential signalling standard like RS485 is much more suitable as it is very noise resistant, RS485 transceivers are commonly and cheaply available which can be transparent to asynchronous serial data.

[pic-ignorant]

I use 16F887's in 2 of my products, and they have both passed EMC testing at industrial level. Is your '887 on a breadboard, veroboard or pcb?
Look at your board.
Any cables leaving the board are antennas for picking up noise. If filtering is not possible because of attenuating data signals, then use screened cables.
Make sure all the IC's are decoupled with 0.1uF caps as close to the vcc pin as possible.
Use a groundplane on the pcb
Use Watchdog timer
Where pullups are needed on port pins that leave the pcb, use a resistor, (and filter cap, not the internal weak pullups.
Compare your config settings between the 2 pics. Maybe you have enabled/disabled MCLR differently on one pic.
Could be one of many things.


John

[SoundsDigital]

Been through the circuit again and discovered that I had connected the MCLR pin to a pull down resistor and not pull up . Still getting the odd hang up, but not like it was. I think I need to check some of my solder joints as it seems to hang up and restart if I move the board.

[pic-ignorant]

What value pullup are you using?
Also, see page 213 of the datasheet for the recommended MCLR circuit.

[SoundsDigital]

I am using a 10K pullup, but I don't have the optional capacitor fitted. Once it is all setup and debugged (still not getting the fold position consistent) I will disaple the MCLR setting and use a separate power source. I am currently powering it through the programmer (PicKit2 clone)

I can't understand how it was working on the bench with the MCLR pin pulled low, but thats not really a concern. I'll try fitting a capacitor to the board if I can find the room, between all the input and output transistors.

[pic-ignorant]

you could use a smaller value pullup, maybe a 4K7.
It would be worth trying a local power supply, such as batteries, to eliminate any loop problems.