Repairing a Thales RM3 Cryocooler Brushless Motor Controller

[MrSheep]

Hi Everyone,

I am in the process of trying to repair a RM3 Stator that seems to not be firing on any of the coils. After inspection of lower PCB board, I see a burnt 1206 resistor. I wasn't able to read the value. It reads 100ohms when desoldered but I doubt that was the original value. After applying rubbing alcohol and trying to clean it up I get 2ohms. So results are unreliable.

I am trying to figure out what the purpose of the resistor was. Could this be some sort of current sense resistor?
I will post more about it as I figure out more stuff by probing.

[MrSheep]

This is the other side of the lower board. It consists of many passives as well as the 6 main Mosfets which are IRF6648TRPBF.

I most likely have to desolder them to check them as the gate and source pins are on the bottom and the outside metal tin can is the drain

[ArsenioDev]

That resistor is likely a current sense shunt, that cooked a lil much.
Large package is familiar for that.
Reason it may have failed is rotor locked, phases saturated and went into current runaway

[MrSheep]

Gotcha, yeah that is my fear. I figured if I repair this I might have a chance of getting it to work. It is definitely cheaper to go this route before just swapping the cooler outright.
What I have discovered while testing is that the piston does spin inside the hermetically sealed casing. When I spin a metal object around the housing and place a bolt/screw on the other side of the housing the screw follows the hex wrench's motion and is always 180 degrees, which indicates the magnets are spinning inside. So I still have some hope.

Now if it is indeed a current sense. I have to figure out the ohms value somehow

[Fraser]

An interesting thread. Thanks for sharing 

Fraser

[MrSheep]

Yes! I will continue to investigate. I think this is the first time I've seen the inside of one of these. This cooler looks similar to a Ricor K508N as that has integrated electronics as well. Most of the time the drive electronics are external. I had to pry using a chisel and hammer to ope this. The top was glued on. Had to have a right balance of hitting hard enough but not too hard.

Just hoping I will have a favorable outcome haha.

P.S. I attached a pic of the main board

[coppercone2]

those round pump motor boards 

If its a current sense then it would need to be a low value. It can be used for direct control or possibly a overload or whatever detector.

I kind of wonder if a reasonable plan could be made to put in reasonable values and test it.


If the resistor is too big, the control loop will say that its getting enough energy when its not getting enough energy. If its too small it will run it at absolute maximum because its not getting feedback.

It seems logical that you can start at a too high value and step down until you get something that makes it work.

With a too big resistor, so long it does not exceed the voltage input of the amplifier, it should just stall or run at way too low power.  If you go way too slow it seems that it could go into some kind of overload condition.

I would figure out what a reasonable guess at the maximum current is, and then put in a part that is reasonably a bit too much voltage drop for the circuit, and see what it does. Then step it down slowly.

They probobly won't have a power supply that is too much more powerful then it needs to be because it looks compact. This should give you a voltage and current maximum, and make it possible to choose a reasonable guess resistor.


People that know motors might be able to tell you a guesstimate about the motor requirements too based on the magnet wire size, inductance of the winding, configuration.


The tricky part might be the inrush. It might be handled by filtering, leading edge blanking, or actually measured. If its substantial then your guess at the shunt might be fortunately way too big. If they actually measure the inrush and respond to it, then you might end up having a rather unusually small shunt value, because it will throw the average operating conditions off. I would figure they are likely to use some trick to get rid of that 'bogus' current reading, because it would kill the dynamic range of control , I don't think anyone would try to measure it for a rotation pump, maybe they would for a precision motion control system (positioner motor), but who cares exactly how it starts if it just needs to spin, its not useful to know that information. Since its all well lubricated and stuff, I don't think it has a requirement to 'break free' too much, meaning they probobly ramp it as slowly as possible to preserve the mechanics, which makes me think a shunt based on maximum control near the setpoint.

[MrSheep]

Thanks for the insight. My plan is to figure out where the resistor connects to. If it connects to a specific chip. I may be able to see suggested values from a datasheet. I know some battery charging IC's have recommended current sense resistor in their datasheet.

[coppercone2]

And a shunt will be directly connected to the winding, or damn close (maybe like some kind of protection switch is after it, or a bridge). I think most shunts are load connected.

[MrSheep]

Just ordered some current sense resistors of various values. 0.1ohm, 0.05ohm, 0.025, 0.01 ohm. Hopefully one of these work.

Did a bit of digging on the chips and found this one to be interesting: LT1491ACS
This Quad Op-amp on the lower board seems to be responsible for the current sensing. And based on some basic info on op amps in brushless motor controllers, I might be onto something:

"In a brushless motor driver, operational amplifiers (op amps) are typically used to precisely measure the current flowing through each phase of the motor, allowing for advanced control mechanisms like current limiting and torque control by providing a high-gain amplified signal from the small current sensing resistors within the motor driver circuit."

[coppercone2]

You might wanna try something too big because if you put something too small it might go crazy. I think right now we know it does not work with 2 ohms, this might mean 2 ohms is too much.

Based on the power handling ability of the resistors, the current draw of the device, are you sure you might not start with one that is too small?

Maybe start with 1 ohm or 0.5 ohm?

the assumption that they will chose one that dissipates 0.01 watts of power because its accurate is often NOT correct, some people might not give a shit that their shunt has a 40c temperature rise and some 'error' on a mechanical part

i.e. despite it being rational to want to keep your shunt cool, they could have designed it with it being normal for a 75c operating temperature.


For some 'less relevant' circuits, I have seen shunts that run at like 100C. There is error but because its wire wound, its way more then accurate enough, and they get a nice big signal from it. I thought it was broken or had the wrong part but when I went through schematics, it turned out thats how its supposed to work.


So if  you have a 500ppm / degree C part, that is 0.05% per degree. At 100 degrees delta, the error is 5%. While the metrologist in you might think the sky is falling, the power designer might only see that the part is still within spec, the error is only 5%, he can't even see it drift on a oscilloscope and he has a giant signal to work with that means no worries about low noise amplification and shit like that. 500ppm/c is reasonable for cheap cement resistors, often built with a thermal fuse, sometimes used for really cheap shunts


power electronics might have a heavy weight PCB too, so parts might be operating under 'bizzare' power levels

noise is a big problem here, so big signals = easy and less likely to break. I could imagine the look on their face if someone tried to use like a incredibly small shunt and it gets burried under all the EMI and stuff, then they send a whambulance when the boss cries that there is unnecessary filter or whatever components because someone made the attempt to keep shunt dissipation under a watt. I noticed that electronics near pumps seems to be pretty bad. Bad like having a real concern about the life time of electrolytic capacitors reaching the end of the warentee bad, Ever see a 135c downhole part (and a backup) specified because someone utterly refuses to put a heat sink and they don't believe you that it might be bad for the fiberglass or just generally cause problems?

[coppercone2]

also the first thing i see in the datasheet is 0.2


a very high chance its a good place to start! charger.. .pump... what is the difference again??

[MrSheep]

Ok yeah that sounds like a plan. I will get larger shunt resistors and gradually work my way down in resistance.

What I found out during probing.

1) There are 3 op amps that map to pins leading from the lower board to the upper microcontroller board. My guess is that it is for each of the phases of the stator.

2) The shunt resistor that burnt is tied to V- of the Quad op amp. However I cannot find where the other side leads to (yet). My guess is it goes under the tin caps to the gate or source pins of the mosfets.

[coppercone2]

You might want to give it some more thought I never actually attempted that, it just seems reasonable to me...

If a shunt fails short, it seems that the supply would attempt 100% power, because its not getting a signal. If it increases in resistance, then the current decreases, which seems to be what you would want to prevent a catastrophic failure... it seems like it naturally 'presses the breaks' .

shorted shunt seems to be the worst case for a control system lol, that seems to be a plausible way in how something gets totally wrecked, keep giving it full power because you think its doing nothing