Post by thread:[PIC]: Conducted noise causing program to restart

My project is a gas fire controller using a 16F873, during the ignition sequence the program will regularly restart as if a hardware reset has been performed, I do not have the watchdog enabled. The source of the noise is the mains powered spark generator which my customer insists must remain on the same PCB. The thyristor gate drive for the ignition circuit is opto isolated and I'm using a flame rectification technique for flame detection. The PIC is bypassed by 100nF directly across the supply pins and also has 47uF soldered to the same point. All of the PIC ports have at least 330 Ohms in series and its supply is a 78L05 with 100nF and 100uF either side. MCLR has 10uF to 0V and 10K to +5V but I have tried tying MCLR to +5V without success. Generous use of 10 uH chokes is helping but becoming expensive, I'm experimentally trying to reduce the noise level at source but would be grateful for any advice on making my PIC circuitry more immune. Any path into the control circuitry seems to cause the problem. Thanks in advance. Nigel Duckworth -- http://www.piclist.com hint: To leave the PICList spam_OUTpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu>

>grateful for any advice on making my PIC circuitry more immune. >Any path into the control circuitry seems to cause the problem. What happens if you have either the start circuit or the PIC circuit enclosed in a can like happens with TV tuners or RF stages in mobile phones? What I envisage is essentially a flat plate with the corners turned down to solder to on the solder side of the PCB, and a deeper can, but similar mounting method on the component side. This was the sort of shielding the old "wedding cake tins" were good for. -- http://www.piclist.com hint: To leave the PICList .....piclist-unsubscribe-requestKILLspam@spam@mitvma.mit.edu>

Nigel, What about putting Ferrite beads on all signal leads and power supply to and from the PIC. They shouldn't be too expensive, and will act much like the chokes you are using now. Also, they're much smaller. You also might want to try surrounding the PIC in a metal shield of some sort. I don't know if this will help, but it certainly couldn't hurt. I think I'd go with the ferrite beads first, as it's the least expensive, and most expedient. Let me know what you find out, and how you ultimately fix the problem. Regards, Jim On Tue, 01 August 2000, Nigel Duckworth wrote: {Quote hidden}> > My project is a gas fire controller using a 16F873, during the ignition > sequence the program will regularly restart as if a hardware reset has been > performed, I do not have the watchdog enabled. > The source of the noise is the mains powered spark generator which my > customer insists must remain on the same PCB. > > The thyristor gate drive for the ignition circuit is opto isolated and I'm > using a flame rectification technique for flame detection. > The PIC is bypassed by 100nF directly across the supply pins and also has > 47uF > soldered to the same point. > All of the PIC ports have at least 330 Ohms in series and its supply is a > 78L05 > with 100nF and 100uF either side. > MCLR has 10uF to 0V and 10K to +5V but I have tried tying MCLR to +5V > without success. > > Generous use of 10 uH chokes is helping but becoming expensive, I'm > experimentally trying to reduce the noise level at source but would be > grateful for any advice on making my PIC circuitry more immune. > Any path into the control circuitry seems to cause the problem. > > Thanks in advance. > > Nigel Duckworth > > -- > http://www.piclist.com hint: To leave the PICList > piclist-unsubscribe-requestKILLspammitvma.mit.edu .....jimKILLspam.....jpes.com -- http://www.piclist.com hint: To leave the PICList EraseMEpiclist-unsubscribe-requestspam_OUTTakeThisOuTmitvma.mit.edu>

Nigel Duckworth wrote: > > My project is a gas fire controller using a 16F873, during the ignition > sequence the program will regularly restart as if a hardware reset has been > performed, I do not have the watchdog enabled. > The source of the noise is the mains powered spark generator which my > customer insists must remain on the same PCB. I had the same problem with an PIC powered ignition system. The PICs don't like strong EMI near them, so you'll need to shield the PIC or the spark generator or both. -- Best regards Tony ICmicro's http://www.picnpoke.com salesspam_OUTpicnpoke.com -- http://www.piclist.com hint: To leave the PICList @spam@piclist-unsubscribe-requestKILLspammitvma.mit.edu>

Thanks for the advice Alan I'll give it a try but I'm fairly certain the noise is conducted. The prototype consists of two PCBs, one for the mains transformer, 12V PSU, spark generator, flame detection circuit and four relays to operate gas solenoid valves. The top PCB has the 16F873, ULN2003 Darlington driver for the relays, 5V PSU, LED's, switches etc. The two boards are connected by 30mm flying leads and the only interconnections not yet protected by 10uH chokes are the low sides of the relay coils. With the relays in circuit the problem occurs but not when they are disconnected, this is why I believe the noise is conducted. The production unit may well use one PCB but I need to understand what's happening with the prototype first. The obvious thing for me to do is protect these inputs as well but my customer is concerned about the rising cost of the components and I'm hoping to find another solution. On Tuesday, August 01, 2000 4:59 PM, Alan B. Pearce [SMTP:KILLspamA.B.PearceKILLspamRL.AC.UK] wrote: {Quote hidden}> What happens if you have either the start circuit or the PIC circuit enclosed > in > a can like happens with TV tuners or RF stages in mobile phones? What I > envisage > is essentially a flat plate with the corners turned down to solder to on the > solder side of the PCB, and a deeper can, but similar mounting method on the > component side. This was the sort of shielding the old "wedding cake tins" > were > good for. > > -- > http://www.piclist.com hint: To leave the PICList > RemoveMEpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]: PIC only [EE]: engineering [OT]: off topic [AD]: advertisements

{Quote hidden}> My project is a gas fire controller using a 16F873, during the > ignition sequence the program will regularly restart as if a hardware > reset has been performed, I do not have the watchdog enabled. The > source of the noise is the mains powered spark generator which my > customer insists must remain on the same PCB. > > The thyristor gate drive for the ignition circuit is opto isolated and > I'm using a flame rectification technique for flame detection. The PIC > is bypassed by 100nF directly across the supply pins and also has 47uF > soldered to the same point. All of the PIC ports have at least 330 > Ohms in series and its supply is a 78L05 with 100nF and 100uF either > side. MCLR has 10uF to 0V and 10K to +5V but I have tried tying MCLR > to +5V without success. > > Generous use of 10 uH chokes is helping but becoming expensive, I'm > experimentally trying to reduce the noise level at source but would be > grateful for any advice on making my PIC circuitry more immune. Any > path into the control circuitry seems to cause the problem. > > Thanks in advance. > > Nigel Duckworth > > -- > http://www.piclist.com hint: To leave the PICList > spamBeGonepiclist-unsubscribe-requestspamBeGonemitvma.mit.edu The best solution I have found for these type of problems is a good groundplane. I usually try to keep my signal and powersupply traces on the bottomside (of two sided pcb's) and pour copper over the entire topside as a groundplane. Of course I have to make some gaps in the groundplane for some traces that I can't put on the bottomside. Think of it as a one sided PCB with as short jumpers as possible on the topside. In fact, You could get a (virtual) 3 layer board by having all the traces on the topside replaced with jumpers and the middle layer (which is the topside layer of the PCB) with only groundplane and pads for the components and jumpers. Most of my (PIC) boards actually needs very few traces (or jumpers) on the topside. Another thing is the bypass capacitors, make sure they are of (multilayer) ceramic types. And of course, as short traces as possible to the cap (which is very easy if you have the topside poured with copper). I have also found it very usefull to put a small ceramic cap (6.8nF) across MCLR and ground when using an external brownout/reset chip/circutit. ============================== Ruben Jvnsson AB Liros Elektronik Box 9124, 200 39 Malmv, Sweden TEL INT +46 40142078 FAX INT +46 40947388 TakeThisOuTrubenEraseMEspam_OUTpp.sbbs.se ============================== -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]: PIC only [EE]: engineering [OT]: off topic [AD]: advertisements

imho place a 5V6 zener directly across the PIC pins and solder 2 sheets of grounded copper foil (ont on top one on bottom) onto the PIC area. If at this point, assuming that you have proper grounds throughtout the circuit, you still have troubles, then I don't know what you can do ... ;) BTW you probably NEED a guard ground passing between the legs of the OC, connected to the PIC ground (and a second guard ground connected to the mains ground passing in the same space). If you use a transformer type power supply you want to have decoupling across the rectifier diodes and the iron core grounded too. hope this helps, Peter -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]: PIC only [EE]: engineering [OT]: off topic [AD]: advertisements