Susceptibility to noise

[TerryRy]

I'm interested in members experience with problems of reseting processors due the noise/interference.

I have a 20M2 performing a control function for a 18v Ni cad battery charger. The whole circuit operates from 12v DC and uses a 12/26v switching convertor (operating at 10's of KHz) to supply the charging voltage and a 78L05 to supply the processor.

I have installed additional 12v and 5v bypasses (100n and 100u) but the problem remains that when the processor turns the DC/DC convertor on via a relay it immediately resets.

I suspect that it is through the 5v power supply or less likely radiated directly between the convertor and processor pins. I haven't put a scope on it yet but have tied unused inputs to ground via 10k.

Any experiences would be appreciated.

 

[Dippy]

I've had similar things with switchers especially a cheapo when the DC/DC is getting towards it's maximum current spec.
Simply put, the DC/DC was cheap and started producing some nasty switching spikes. serves me right for being a skin-flint.
In my case, extra decoupling cured it.

Without a schematic we're slightly into Crystal Ball Zone. Without an image we're relying on your good wiring techniques. Without 'scope results we're into the unscientific 'try this and see' approach.
('scope tests should have been your FIRST test before posting)


1. Have you included the traditional 'back emf' diode in your relay drive section?
2. Have you tried a capacitor at the relay coil supply side to act as a reservoir and reduce transients in supply line?
3. Have you put a couple of decoupling/bypass caps across PICAXE power pins?
4. Have you used good power and ground wiring/track techniques to minimise load/switching transients in those lines?
5. Are you using nice neat PCB or hairy strip/bread board?
6. Have you tried a snubber across relay output contacts?
7. Have you tried extra caps (such as 100nF X7R) across DC/DC output?
8. Have you tried extra decoupling/bypass at the regulator? (It is important to use good components and HF can go through a 7805 without barely touching the sides).
9. A gentle RC (22R/100nF X7R) suppling the PICAXE may help. It acts as a filter.
10. How are you DRIVING the relay? Schematic please.
11. Anything else I've forgotten

There are so many areas where noise can sneak in; mainly over the ether or through the wires.
Keeping power away from signal is crucial. Sometimes physical screening is required if everything is tight.
And SO MUCH depends on the quality of PSU and your wiring.
Sometimes the solution can be a simple issue and sometimes a long grind.
I hope it's the former , but you really need to supply a good and neat schematic and indicate whether this is a PCB or a pile of wires.

 

[inglewoodpete]

The connections to ground are critical in this type of circuit. Please post a circuit and photos of the physical circuit/wiring.

 

[fernando_g]

Powering up AC mains to a PSU via a relay always results in substantial noise being produced. Both of the radiated and conducted variety.
This thanks to input cap's inrush current coupled to relay bounce.

The best solution is to power up with a TRIAC, of course isolated with an opto-coupled driver.
Choose the zero-crossing variety, and it will switch on at almost zero voltage, minimizing the inrush current.

 

[TerryRy]

Thank you for your responses.

I expected that the problem would most likely be due to power supplies. As a test setup I have been operating the on a 1.5A plug pack. Judging by its weight, it is electronic and is the cause of the problems.

The DC/DC convertor and the plug pack don't like each other, causing fast and probably deep glitches on the 12v output from the plug pack when the convertor starts. Without a scope I cannot be sure how deep they are but there is a pulse according to the multimeter and also on the regulated 5v supply.

Running the circuit from a 12v lead acid battery ( as it will do in real life) fixes the problem.

Lesson - be careful of electronic plug packs even you are using at only 5% of capacity.

All the comments are much appreciated and a good lesson for those constructing projects. I would have found the problem quickly if I owned a scope but the exercise would be of no benefit to other members without the post.