Micro controller affected by ripple from bldc motor

[Nihaludeen]

Hi friends, I am driving bldc motor(1kw) using stm32f1 micro controller and keil micro vision software as a compiler.
While running motor on no load condition....
- Motor runs good.
- Sensing sensor values, analog values, measuring rpm, etc which are all works good.

While running motor on load condition....
- On load condition the motor runs fine, but the wave forms of mosfet gate and power supply of micro controller ripples lot.
- I can't able to sense any data properly, the keil compiler shows "cannot evaluate" for every 2 or 3 randomly.
- Apart from sensor data, i also declared a variable globally (say int a=10
Keil complier in debugging section show the variable value a =10, at no load condition. But on load condition the constant variable a=10 changes to a random values.
(i,e)
10
10
10
45
10
10
cannot evaluate

Bldc controller.....
- trapezoidal control
- open loop control
- stm32f1 and st link for debugging.

The below waveform is Micro controller power supply at no load


This waveform is at Load condition


This waveform is captured at Micro controller output pin at no load and load condition respectively

 

[FvM]

Welcome to the exciting world of power electronics development!

What you experience is not "ripple" but switching noise disturbing your oscilloscope measurements and debug tools. It may be promoted by an unsuitable circuit layout and signal wiring, but can even occur with well designed systems.

The probably most interesting question is if your controller is still working correctly despite of measurement and debug communication problems. If so, there's still hope to get your design running without redesign.

Regarding oscilloscope screenshots, they are unlikely showing the real waveform, more likely the spikes are measurement artefacts produced by unsuitable probe connection and common mode noise. Don't expect to get reliable measurement with standard oscilloscope probes unless you use toroid cores as common mode chokes. See e.g. this discussion https://www.edaboard.com/showthread.php?371078-Clamp-on-cable-ferrite-for-scope-lead

Similarly you also need noise suppression for your debug signals. Toroid cores for USB and serial cables can work. See picture

 

[Easy peasy]

Also your power ckt is not shown - so hard to offer cogent advice ..

 

[Nihaludeen]

Yes FVM, what you said is correct. If we don't use the proper probes or probe connection, there will be inaccurate waveform.
I connect the probe almost correctly(very near to micro controller's supply). Because if we connect the wire too long there will be a parasitic inductance across the wire. And also i use Key sight EDUX1002a oscilloscope and the probe that they given was really a good quality.

I also checked another bldc controller (newly bought), whose micro controller's power supply with my oscilloscope, which was good. I attached the scopes images for my controller and newly bought bldc controller. I checked the two controller same way, but my controller's power supply only very noisy.
I attached the circuit connection too.
newly brought controllers supply...


my controllers supply...

- The bldc motor drive's micro controller is working fine for long time. But right now only i am noticing this issue.
The circuit attached below is single phase for bldc controller. similarly the circuit is same for other two phases


How to reduce this switching noise in my circuit. guide me guys...

 

[iop95]

If can't use a totally separate 3.3V controller power supply (my recomandation) add LC filters on pins 3 and 9 of IR2110.
Also, may be high dv/dt, so try to increase R1 to 22ohm and add a 10ohm resistor series with D1 to slowdown turn-on/turn-off. Check if switching loss increase too much.
If the noise has a radiant component will be more complicated.

 

[KlausST]

Hi,

I assume it's not a schematic problem, I rather think it's a PCB layout problem.

Most (maybe all) H-bridge systems I've designed work without isolation.

Klaus

 

[Nihaludeen]

Yes iop95, i use separate 3.3v regulator (lm111733v) for micro controller.
Sure i will try that LC filter in ir2110 supply pins. And also increase the resistance accordingly. I cant get this line friend "If the noise has a radiant component will be more complicated "

Also i use gate driver long distance from the fet, i reduce that one. Because that's the loop area right. So i will decrease that one as close as possible.

 

[iop95]

This "If the noise has a radiant component will be more complicated " refer to situation when noise is send by electromagnetic waves (radio).
As you have long distance between driver and MOSFETs, these lines may acts as an antenna transmmiting and noise is "radiant" as radio waves that induce voltages/spikes in all olther circuits.

 

[Nihaludeen]

Thank you guys for your responses and valuable comments. Hope switching noises reduced almost 60 to 70%.
Even if there is any other think to be noted (like PCB design , choosing FET's or any calculations) for bldc drives kindly advice me, it would be really helpful for me to re-design it.
Sorry guys if i am troubling you so much.

 

[Easy peasy]

larger turn on resistors usually a good place to start, 47E or larger ...