reset microcontroller

[Mahdi]

i use dspic33fj256mc710 and ir2104 to drive half bridge mosfet

i use 12 volt , 150 watt switching power supply and lm2576 3.3 for dspic.

when the mosfet load is 10 watt no problem but when the load is 50 watt ,micro Continuous

reset and power up.the mosfet is irfz44

i pullup mclr pin with 10k resistor and put 100 nf cap for each vcc gnd.

i think when the mosfet goes on the high load, generate noise and the mclr pin reset.

because i dont have Voltage drop on vcc of micro but when i see gnd line with oscop its noisy when the micro is in reset

how to solve problem?

thank you

[temtronic]

You need to eliminate the EMI that the load is generating.
You need LARGE conductors/traces for 10A loads, some sort of caps, diodes,chokes based on the motor's specs.
As well every I/O pin on then PIC may need a cap.
Proper PCB layout and design is needed! Small(thin) traces are bad. A ground plane(checked board not solid) will help.
Power supply lines need filtering..
While you can cut the PIC code in a couple of days,it can take a couple weeks or months to eliminate the EMI noise !

Hopefully you have an oscilloscope to show you where the EMI is. Be sure to read any datasheets, appnotes, etc. from the mfr. of the motor,PSU, H-driver and of course uchip !!

Jay

[Eugeneo]

FYI. So you don't make the same mistake I did. If you really want to know what is happening, make sure you invest in some good differential probes or use a scope with isolated inputs. And I'm not talking about just isolating the power source.

Otherwise you will start talking to yourself thinking the noise is coming from everywhere.

[Ttelmah]

Very true.

Had a lovely one just recently with a board. It was working, but the customer came back saying 'there is a lot of noise on the analog inputs'. Turned out they were grounding the scope to the supply connection, not the 'I/P test ground' test point that was on the board (connected to the PIC's 'Vref-'), for testing what the PIC was actually seeing - wonder 'why' I had put that there....

Remember when designing the ground plane to consider 'where' current flows will actually 'be'. You want a plane section for the motor, which has one nice thick connection 'back' to the supply, and ideally a separate section for the analog circuitry/PIC, which does not cross the other ground at any point, and goes back again to one point. If you draw high currents for the motor section, that flow 'across' the section with the PIC, you may easily be surprised at the spike currents that occur. The biggest current flows will often not be on the 'power' cycle, but on the overshoot cycle. On your MOSFET, what are you using as trap diodes?. Unless the internal diodes are _specifically_ designed to be used as trap diodes, they are often not suitable, with poor recovery, and can result in some of the energy flowing back out of the gate. Haven't look at the data on your FET to see whether the internal ones are suitable on these, but you need to be thinking about this. It's a very common error to assume the internal diodes are 'there to be used'. Generally they are not (they are a 'side effect' of the FET's construction). You can often get better behaviour (and efficiency), using "trap FET's", rather than trap diodes.
Remember the stored energy in the inductors has to go somewhere. You talk about using 0.1uF capacitors on the logic supplies, but what capacitance have you got on the main supply feeding the motor circuitry?.