IRLZ44N overheating problem

Hi,

I've built the AMC 1.5 and DSMhb. The pcb boards were made by a pcb company, top quality, I use IRLZ44s, a 22 MHz crystal and 10k 1 turn pots. Everything works fine, I can communicate with the AMC board, I can see on the lcd changing both the force-profiler values and the pot values. When I adjust the pot value, it rotates the motor both directions, and the red-green leds are flip-flopping as they should. So everything seems to be ok.
But when I set it up as it is designed, so the wiper motors drive the pots, I got an overheating problem:
When the motor reaches the position determined by the force-profiler, both red and green leds are flashing simultanously (I saw a video that it seems to be normal). And during this dual flashing the irlz44s run extremly hot. (Very first I tried my system with the test-plugin, letting the sliders auto change their values. As only little movement was required by the motors at a time by slow motion, the red-green leds very flashing continously druing the whole test, because the motor position always was near the required. The motor moved well, but after 10 seconds the IRLZ44s were popped.)

I think it is normal, that near the required position both direction leds are flashing continously, as in this phase the program tries to adjust the motor into the correct position, and it needs slight forward-backward movements, if I understand well. At the correct position they should stop flashing and they do stop it in my case also, so it seems to be ok.
But, during this flashing phase should the irlz44s be so extremly hot that they pop? (I tested it with a simple, small cd-rom motor, the irlz44 runs really hot even with this small motor also when both leds are flashing, altough it requires almost no current.)

I have to tell, no heat pipes added yet, because it's a test phase and I'm not sure what is my real problem. I have lost 2 sets of irlz44s already. It's not cheap, so I'd like to be sure, before I try again.

Or the auto slider test-plugin is the worst stress-test as it only moves the motor bit-by-bit? But I think a slow motion change can happen any time, so it shouldn't kill the fets.
Or I (and my board) do something wrong...

The motor is driven back and forth very quickly because of the noise on the analog input and because there is no dead band to stop this constant positioning close to the setpoint.

Does your MOSFET's heat up if you drive them in one direction only?

I tried to simulate it as I turned the pot by hand. Let's assume the required value is 128 or middlepoint. When the pot value is far below or far above 512 it rotates the motor fast, the correct direction led is lighting in both directions, no heat up. When the value is close enough to 512 (+- a few) the motor stops, no direction led lighting, no heat up. But when I set the value near 520 or 500, both leds are flashing and both FETs heat up very quickly. So this is the moment, when the FETs are heated.

With the test-plugin's slowly moving auto sliders, it tries to position itself continously to the new value with small steps. In this case I can see both direction leds flashing almost all the time as it tries to move the motor slowly from one end to the other. The motor moves to the required position, it works as it expected, but with heating up the FETs very quickly (it smells within 10 seconds).

On the video in the video gallery, it can be seen that sometimes both direction leds are flashing up for a moment (seems to be normal), but on the video the motor is driven with rapid direction changes, so I can't see, how should it work, when it is driven with a slowly changing signal.

Maybe something noisy? Should it find the position or setpoint much faster? It might spend too much time to find the setpoint, or unable to find it because of noise on pot input?
I'm using ordinary 1-turn pots with a 4-wired phone cable attached.

Try changing the kP values. What are they now?

Do you see any movement or hear the motor when it should just stop?

If it works fine with the potentiometer in your hand adjusting the speed, it is something firmware/setup related.

First I tried both kp as 3. Then I changed them to 5. Then back to 3. The behaviour is the same. I wasn't brave enough to set any other values

When the pot is in my hand, I'm sure no movement, no sound. But when the pots are turned by the motors interestingly maybe some whinning noise could be heard from one of the wiper motors. But I'm really not sure I'll check tomorrow. What should that mean if I can hear the motor?

If you hear the motor whine when it is stopped, it means it is moving. Trying to self correct the noise that gets through, probably amplifies it and starts to oscillate a little bit. DC motors consume the most power at the moment they start rolling.

Perfectly normal behavior as much as unpleasant.

I don't think this would be enough to blow the MOSFETs, but I really don't know. Try to look at the circuit with an oscilloscope.

So... I wasn't right, my IRLZ44Ns get hot anyway (just a bit later, if not in the adjusting phase) and I'll try to explain what I found.

It might be something very obvious or I don't know what to think...

I checked my board with an oscilloscope. I made the measeurement without load, no motor, only 12V for the H-Bridge. I just connected the AMC board together with the H-Bridge.
On the picture can be seen what I'll try to explain (I use IRLZ44 instead of IRFZ44).
Let's assume, that on the HB0 line there is a PWM signal. In this case the HB1 is ground as (PWM AND M1IN2) = GROUND. On the point A I can see the 5V PWM. But this PWM goes through the H-Brigde GND line (follow the red dotted line) and it appears on the point C, with a bit smaller amplitude (due to the resistors, I think). This signal goes through the 3k6 resistor and meets with HB1 (GND) at point B and it should be grounded also. The interesting thing is that if the driven low-side FET (the right side one on the picture) is taken out, point B is ground, as it should be. But if the driven low-side FET is its place, the measured signal at point B becomes what I painted there, a spiky signal, between -X V and +X V (X is somewhere between 1.6 and 1.9). In this case the HB1 line is not ground anymore and this spiky signal drives the left-side low-side FET and also the right-side upper FET, so all four FETs are open for milliseconds, and the short is ready, and it heats up both the IRLZ44Ns.




As far as I know, the AMC board and the H-Bridge board grounds shouldn't be connected. I can't figure out what happens with the ground at point B.
This spiky signal is there regardless of the 12V. Doesn't matter if it's connected or not.

Is this spiky signal normal? Does anyone have something similar experience? Or did I miss something?
I'm out of ideas, I hope someone can help.

Cheers,
Janos

All of the grounds should be connected.

What do you think is going to happen if the ground is a couple Volts above the other board's ground?
I don't know if this is going to fix it but thats what I would try first.

That spiky signal is causing shoot through for that short period of time. If you get rid of it, I'm confident the heating issue will be solved.

Thanks Frakk,

But I'm a bit confused now. In the DSMdb thread Thanos wrote multiple times:

or

or

It suggests that the ground of AMC board and the ground of motor power supply circuit(s) on the DSMhb board shouldn't have been connected! And of course there' s no place on the DSMhb board where the ground coming from the AMC (powering the 74hc08 chip) should be connected to the grounds of the two motor supply circuits!


I'm sure too, that if I can eliminate that spikes than everything will be ok. But now I don't know how it works this way to the others who have built it successfully.

So you mean that I should connect the ground coming from the AMC board (to the 74HC08) to both grounds of the two motor supply circuits? If so, why Thanos saying not to do that?

Or do I misunderstand something? :?

I don't know why he is saying it, to be honest.

Batteries are floating supplies. Pair them up with inductive DC motors and there will be huge fluctuations of voltage potentials.
If the power side of the MOSFET's are referenced to a floating voltage, and driven by something with reference to ground.... well.... there will be some MOSFET's blown.

I haven't built it, never tested it, so I can only guess what is happening.

Best way to ground it is directly to the 5V regulator on the AMC with a separate wire, not too thin.


ps. Using power supplies that are grounded are a lot better, but there is still a chance for unexpected things to happen.

I think exactly the same. I can't really see how can it work to open a FET with some kind of floating signal.
Maybe this spiky signal is there with the original IRFZ44, but that doesn't open for a 1.6V spike. That's why people using that FET doesn't have this problem. But with IRLZ44N it opens and makes a shoot through.
I would be interesting to hear someone with running IRLZ44Ns. If there's anyone.

I can't test it because I don't have the IRLZ44 ones, but here what you my try to do:
Change the 1.5k resistors to bigger ones like 2k or higher to eliminate that 1.6v spike....

And something else... there is no need for common ground because the each input turns to ground while the other pulses PWM...

Thanos

Thanks Thanos for helping me.

Now you mean just to suppress the spikes while they are small enough to not open the gate on the IRLs.
With the IRFZ44-s there's no spikes there at all? Or I think right, that these spikes there, just does not open the IRFZs?

I wonder if anybody is using IRLZ44Ns on a correctly working DSMhb.


Cheers,
Janos

I'll try to hook the h-bridge in the oscilloscope I have to see if they are present for IRFZ44 also... but I doubt it. The PWM signal is coming very clear from the AMC, so it maybe either IRLZ44 issue or some versions of 74HC08 may introduce this spike as they can't follow correctly the pwm switching (internal delay)!

Thanos

Oh, you're maybe right, as the 74HC08 would be responsible to keep the other line on the ground. That might be possible. My PWM is also very clear.
I would be pleased if you could check that for me.

The IRFZ44N's heat up for a different reason

What happens if the negative battery connection suddenly raises to let's say 5V, with respect to the AMC/74HC08 Ground?
Hint: now your output will be 5V lower... and your 0V will be -5V... with respect to the power side.

And what harm is it going to do if we connect the Grounds? Please expand more on your reasoning Thanos.

Maybe it is a unique problem? Maybe other's connect the grounds? Or use a grounded power supply to charge the batteries?
Opto-Isolators are good practice indeed! However they are not necessary.
Standard MOSFET's like the IRFZ44 (and also P-Ch ones) need proper MOSFET drivers, especially when getting into the 15kHz+ switching range.
A transistor/resistor network is just not going to cut it, in my book at least.

I have started a thread with such an H-Bridge that I got sidetracked from. If anyone interested in building and experimenting with it, please feel free.
My first measurements and scope shots were looking good, but no further testing was done on that circuit.


About the original problem:
From what I can see on your sketches, you get a positive spike on the rising edge of the PWM pulse, you get a negative spike on the falling edge of the pulse, and the 2 waveforms line up perfectly on the scope?
It could be some sort of weird capacitive crosstalk between the two wires, or maybe high-resistance connection (copper left between traces).

I took a deep breath and connected the grounds, as I saw no other solution. The spikes are still there, but they are reduced to 0.05-0.1 V range. No it doesn't effects the IRLZ44Ns and now it works without any heating, as it's suggested. I tried with the wiper motors too and it seems to be ok. (I made some rough initial test with nolimits and my children, the loved it, but it's far from a perfect setup.)
Just for the record, I tried the IRFZ44 in the original setup. With a small motor it does not heat up at all, so the IRF seems to be immune to the spikes. Maybe that's why who uses the IRFs doesn't have this problem, I think.
I would be interested in those, who uses IRLs without common ground and doesn't have any problems.

So it is solved, and I take now the next step.

Good to hear!

:clap: :