Inverter drive module overvolt fault

[Eamon]

I'm trying to determine my next steps in diagnosing and repairing an inverter drive module for a compressor in a heat pump.

Pool heat pump Model Astral Pool iHP170

Symptoms:
Shortly after turning on and after the compressor kicks in it stops with a fault; "F10 - Inv. Input Overvolt."

Apart from a range of sensors and relays, there are essentially 3 main components in the electrical side; Touch screen, main controller and the inverter drive module.
There are no obviously damaged components that I can see and no residual magic smoke smell.


The replacement cost from Astral pool for just the inverter module pictured is ~$2500! Repair it is.

I have had no success in determining an OEM for the part or anything related to the marking on it "SA.FNB75GW.1" except for parts listings in the various manuals for this heat pump presumably rebadged under a few names.

The main drive chip is a mitsubishi pss35s92f6-ag, other component numbers can be provided where relevant.

I have determined that the communication between the controller and the compressor Inverter drive module is ModbusRTU over RS485.

After sniffing a bunch of the communication between the boards I have managed to reverse engineer enough of the communication protocol to control the inverter drive board directly from a laptop using a small application I wrote which polls the state of the module every 100ms.

I can instruct the inverter module to control the compressor on/off state, compressor frequency and fan speed and the inverter module will report:
"AC plate voltage", "Inverter DC Voltage", "Compressor frequency", "Compressor Current", "IPMTemperature" along with a bunch of error codes

While powered and with the compressor and fan off the drive module reports the AC voltage ~220v which is about 15v less than I measure on the terminals with my EEVBlog BM786 multimeter
it also reports the DC voltage as ~320v which seems to be about right for the given AC voltage.

When I send the "on" signal for the compressor, a few seconds later I see a jump in the reported compressor current and another few seconds later the compressor frequency begins to rise.
After around 5 seconds with the frequency generally about 25Hz, the reported AC voltage jumps from ~220v to ~320v; pretty much the same value being reported for the DC Voltage.
About 5 seconds later, the compressor cuts out and the "Inv. Input Overvolt" error is reported, at the same time the AC voltage drops to ~255v where it stays for the next 10 seconds or so before returning
to ~220v.  The DC voltage remains around 320v throughout.

I've tried setting the initial compressor frequency to a number of values from 20Hz to 100Hz (all within range of the compressor's specs) and the behavior is fairly consistent.

So that's about as far as I can go on the software side, now I need to get into the electrics but not sure where to start.

The numbers above are all as reported by the inverter module itself via Modbus, not independently measured values.

Using the multimeter I do not see the spike on the AC input terminals reported by the module so I need to determine where the metering circuit might be on the module. 

This is where I need help from more seasoned experts.  Any suggestion on what to check next or where the AC metering circuitry might be would be very welcome.

[capt bullshot]

So there's obviously nothing broken there, just a malfunction of the AC input measurement.
This is good news, as you don't need the expensive replacement kit - which I would suggest to get if something within the power stage was blown.

There's probably very little chance to get a schematic of the inverter module, so one would have to reverse engineer the AC input until you locate the components that are in charge for AC input voltage measurement. Most probably that's just a bunch of resistors or other smallish components. You'd have to follow this path until it ends up in some ADC input of the MCU, and locate the fault within this signal chain.
There's some hazards with this, as I cannot judge from the picture alone whether the MCU is isolated from mains or at "DC-Link minus" potential, and the serial (Modbus communication) port is isolated. I'd expect for this kind of application most of the PCB at live potential.

So this is just some general information, basically how I'd go on with this issue, based on my options and experience. You already did some great work to help locating the fault by analyzing the behaviour of the inverter stage at the communication side. Now it's about time to dive into component level electronics.

Looking at the photo, you can see a string of SMT resistors left hand side of an orange capacitor. This is probably part of the circuit that measures the DC link voltage. The AC voltage measurement might look similar. Look for strings of resistors connected to L and N input. Otherwise, regarding all the cost saving effort that goes into todays products, there might be some totally other circuit that "estimates" (in contrary to measure) the AC input voltage in some other way (ATM I don't have an idea what this could look like). Anyway, I wonder why there's an AC input voltage measurement circuit at all, since it's not required for such kind of inverter to work at all (I've got some experience working together with engineers that designed such kind of power stages, my part was rather at the MCU side).

[NiHaoMike]

I would suggest checking the noise filtering on the AC sense circuit, could be as simple as a cracked ceramic capacitor.

[capt bullshot]

After a while, I've got another thought:
Check the internal (low voltage) MCU and analog supply voltage(s) for excessive ripple

[Eamon]

I would suggest checking the noise filtering on the AC sense circuit, could be as simple as a cracked ceramic capacitor.

Thanks, any idea where the "AC sense circuit" would be from the image?

[Eamon]

ok, I have taken some voltage measurements with the module powered but not with the compressor set "on", I've used the ground pin on the blank fan connector on the left of the board as my 0v/ground reference.


I was surprised to find 380v at one particular point as shown in the image.

Also not sure what the 130vAC is for?  but in addition the bottom right has a second bridge rectifier that appears to be unpowered, the + and - and both ~ pins are all at 0v relative to my ground ref.

The MCU is TMS320F28027PTT, I'll check out the datasheet.  I assume it is likely to be one of the ADC pins that is measuring the AC voltage? 

As for checking for ripple, I have an analog discovery 2 USB Oscilloscope so I'll try to get some more detailed readings.

As for that, this would be a whole lot easier if I could do it at a desk/bench rather that in place at the heat pump unit itself however if I disconnect the compressor from the drive, I can't replicate the fault as it errors with a "Comp. Driver Failure" error instead.  Could I connect something else in place of the compressor?  I have a spare 3 phase motor available, but it is 415v, could I use that?

[capt bullshot]

ok, I have taken some voltage measurements with the module powered but not with the compressor set "on", I've used the ground pin on the blank fan connector on the left of the board as my 0v/ground reference.
(Attachment Link)

I was surprised to find 380v at one particular point as shown in the image.

Also not sure what the 130vAC is for?  but in addition the bottom right has a second bridge rectifier that appears to be unpowered, the + and - and both ~ pins are all at 0v relative to my ground ref.

The MCU is TMS320F28027PTT, I'll check out the datasheet.  I assume it is likely to be one of the ADC pins that is measuring the AC voltage? 

As for checking for ripple, I have an analog discovery 2 USB Oscilloscope so I'll try to get some more detailed readings.

As for that, this would be a whole lot easier if I could do it at a desk/bench rather that in place at the heat pump unit itself however if I disconnect the compressor from the drive, I can't replicate the fault as it errors with a "Comp. Driver Failure" error instead.  Could I connect something else in place of the compressor?  I have a spare 3 phase motor available, but it is 415v, could I use that?

The 380V test point most probably is part of that small auxiliary SMPS - maybe part of the snubber circuit. If so, it's probably not relevant for your issue.
Yes one could connect any 3 phase motor to the output, and the motor should run if you command the inverter to do run. Supposed both the compressor and the spare motor are the same type (e.g. induction motor, or PMSM) Your typical "spare 415V 3ph motor" probably is an induction motor (or asynchronous motor) - but I don't have any idea what type of motor is inside the compressor. If it is a PMSM or similar (permanent magnet synchronous motor) it might not start nor run. But most probably neither the motor or the inverter will die from a short "wire it up and turn it on" experiment - but don't nail me on that. The exact voltages might not match, that shouldn't matter too much for short term operation.

[capt bullshot]

If you connect your "scope" to the 9V and 15V circuits, consider the "GND" potential most probably is at "DC Link minus" in such type of inverters.
This means for the "GND" there's a large negative DC voltage superimposed with half wave line AC voltage in reference to Earth, that'll kill your Analog Discovery in a few milliseconds.
Typically best way to deal with it: Use a proper isolation transformer to supply the inverter board, and connect said reference GND to Earth, then apply the scope measurements for ripple etc.

[capt bullshot]

Reading your first post again (didn't see the additional information at my first reply here) -
I doubt "AC plate voltage" directly relates to the AC input voltage at all, and there might not exist any AC input voltage sensing circuit at all.
Usually there's no need for such kind of inverter to monitor the AC input voltage, especially if it's just single phase (for three phase and higher power you want to monitor the input to detect a lost phase condition).
I've got no idea what "AC plate voltage" means in this context, where's the origin of this term?

Do you have means to operate the inverter board at somewhat lower AC input voltage? Some kind of autotransformer or variac to reduce the mains voltage by some 10 Volts?
One can use an ordinary e.g. mains input and 12V output transformer by some simple but tricky wiring to reduce the mains voltage by 12V - the 12V output winding just has to be able to supply the nominal current of the load, not the nominal power.

[Eamon]

If you connect your "scope" to the 9V and 15V circuits, consider the "GND" potential most probably is at "DC Link minus" in such type of inverters.
This means for the "GND" there's a large negative DC voltage superimposed with half wave line AC voltage in reference to Earth, that'll kill your Analog Discovery in a few milliseconds.
Typically best way to deal with it: Use a proper isolation transformer to supply the inverter board, and connect said reference GND to Earth, then apply the scope measurements for ripple etc.

I was planning on using a pair of 100x probes for this and not connecting the ground leads of the probes to anything.  I believe the analog discovery can handle up to 50v on the probes so with the 100x probes I figured I'd have plenty of headroom?  I don't have access to an isolated transformer, but have considered buying an entry level differential probe if the 100x probes wont work.

[Eamon]

Reading your first post again (didn't see the additional information at my first reply here) -
I doubt "AC plate voltage" directly relates to the AC input voltage at all, and there might not exist any AC input voltage sensing circuit at all.
Usually there's no need for such kind of inverter to monitor the AC input voltage, especially if it's just single phase (for three phase and higher power you want to monitor the input to detect a lost phase condition).
I've got no idea what "AC plate voltage" means in this context, where's the origin of this term?

I mis-typed it slightly but "Inverter plate AC voltage" is the term the touchscreen uses.  When I was reverse engineering the modbus protocol, I had the controller and touch screen connected to my computer acting as the Inverter module and would send back various values for the requested registers and see how they would represent in the touchscreen. One particular register always matches the value the touchscreen represented as "Inverter plate AC voltage", this is the value represented in the chart in the first post (I added the PCB back and the chart images a bit later). 

Also, the fault code reported in the touchscreen interface is "Inv. Input Overvolt." which is why I'm assuming it's a measure of the input AC voltage but I agree that there does not seem to be a need for such a measure.

 

Do you have means to operate the inverter board at somewhat lower AC input voltage? Some kind of autotransformer or variac to reduce the mains voltage by some 10 Volts?
One can use an ordinary e.g. mains input and 12V output transformer by some simple but tricky wiring to reduce the mains voltage by 12V - the 12V output winding just has to be able to supply the nominal current of the load, not the nominal power.

Are you suggesting running the inverter at a lower voltage permanently as a workaround or as part of diagnosing the problem?  Could probably just drop the AC voltage by running about 50m of 2.5mm2 extension cable if it's just a test, wouldn't want to run it permanently like that though.

[capt bullshot]

Yes, 100:1 probes could do the job as the AD has differential inputs anyway, and your setup should be safe. Just ensure to properly adjust and match them.

"Inverter plate AC voltage" is as meaningless to me as "AC plate voltage", and the almighty internet doesn't give meaningful results, too.
I guess it's some literal but not meaningful japanese/chinese/korean to English translation.

Reducing the input voltage: My intent is for testing, to see whether the error message disappears or the measured value changes according to the change in input voltage. The extension cord won't drop the voltage in standby, so the latter one wouldn't work.

[Eamon]

Well I tried to connect the motor in place of the compressor and while nothing blew up, the motor barely turns at a very slow speed with a bit of a high pitch for a second or two before the inverter faults out  with a "PFC fault".  Time to find a plan B for a load to replace the compressor.

Can you give me some insight as to the 12v transformer trick to reduce the voltage?

Also just occurred to me that I have an old stick welder that is essentially just a giant iron core variable step down transformer, turn a handle to move the secondary winding further into or out of the primary.  I'll investigate what the voltage is on the lowest current setting....

[Eamon]

Well I don't think the welder will be useful directly as the maximum output voltage is ~45v with input of 236v,  but it may be useful in place of a 12v transformer that you mentioned previously, it will certainly handle the current of the inverter and the load even.  If you could provide some more details of the "some simple but tricky wiring" that would be helpful, in the meantime I'll google a bit and see what I can find.

[florentbr]

This reminds me of an SMPS driver that was randomly shutting down.
The waveform of the input rectified DC had voltage spikes, which I assume was causing an over-voltage shutdown (OVP).
I removed the PFC switching transistor and the issue never reappeared.
I didn't check the capacitors since the rectified DC under load was steady without the PFC.
I suspect that the ESR of the capacitors went out of spec.

In your case, the reported voltage seems to be correct when on standby, so I don't think it's a sensing issue.
Since it takes an inductor to raise a voltage, I would pay attention to the PFC stage.
Among those TO-264 you'll find a switching transistor and two diodes for the PFC.
For more information, search for "PFC boost converter design guide".

[capt bullshot]

See the attached sketch for the basic idea of using a "standard mains step down transformer" to add / subtract its secondary voltage from mains. Load current shall be lower than the ampere rating of the secondary.

Previous posters thought about the PFC stage: I have to agree, this is plausible. There are components that look just like a PFC stage, and said "plate voltage" might have something to do with the PFC stage. Disabling the PFC stage could work - e.g. by removing the transistor. In general the inverter is able to run without a PFC stage, it is required for compliance, not for function. Depending on how it is done, the inverter might throw other errors if one disables the PFC.

Once you've started investigating, scoping the waveforms at the DC link capacitors and at the switching node of the PFC might give some insights.

[Eamon]

I've spent a fair bit of time trying to reverse engineer the schematic and have what is attached so far.  I have a little more around the gate drivers for the two PFC MOSFETs but not sure its too important here; it's essentially a ucc27524 with some resistors and an unidentified SOT-23 transistor.

The three 150k resistors in series head off on a trace that jumps back and forth through vias and ends up in a sea of smd passives and a few SOT-23 diodes near the MPC.  I can't identify the pin arrangement on the diodes as the only marking on D5 is "HA7" which is not much use for searching, but the "idle" state voltage at the end of the 3 series resistors is 2.08v

The three 200k resistors that come out after the first choke also head off to a bunch of smd parts and becomes hard to follow but given that I think this is the only path the MPC could potentially measure an AC signal, I'll try to follow it further.

If there is a problem with the PFC circuit, what is the most likely culprit?

No progress finding a compressor substitute so I'll probably just have to put the module back into the heat pump and put the scope on it in place.  Are there particular points in the circuit I should focus on?

[capt bullshot]

The R3 ... R5 string might be the startup supply for the PFC controller IC, or the other auxiliary SMPS.
Doesn't look like a typical AC sense circuit at a first glance.
The R54 ... R99 string most probably is an input signal to the PFC controller, as these need the actual half bridge rectified voltage for their control loop.
If you have a stable DC voltage at E1 ... E3 that is higher then the peak value of the input AC voltage, the PFC should be considered OK. I'd recon the controller turning on/off the PFC stage depending on the compressor motor stage (PFC off while motor is off to save energy). So the voltage across these capacitors might actually rise when the motor starts.

The UCC27524 is a simple gate driver, not a PFC controller. My guess the PFC is done by the MCU in software, as the inverter control loop and modulator is done there too. You could measure the voltage across the electrolytics (with a multimeter) to see if it matches one of the reported voltages, or if this voltage rises unexpectedly high (whatever a "normal" voltage would be, one can rely on guesses only - I'd consider 350V for a PFC  DC-Link "normal").

[capt bullshot]

Another thought:
TI publishes quite a few reference designs using their chips. You might want to search for motor control reference designs using the TMS320F28027 (or similar) maybe even including a PFC stage. There are designers out there, that use these reference designs as a guide or copy parts of the schematic. This might ease the reverse engineering and help understanding the circuit.

[florentbr]

The AC input voltage in question could be measured at R94 even though it's rectified.

What's strange is that the inverterDCvoltage is supposed to rise well over 320v when the PFC is turned on, but your measure doesn't show that.

If the reported AC voltage jumps from ~220v to ~320v, it could indicate that the output DC is finding it's way back to the output of the bridge rectifier. I might be wrong, but I don't think it's supposed to.

I would remove and test C53A, Q1, Q2, D1, D2 and DB2.
DB2 is not used as a bridge rectifier but as a single diode.

The PFC from your schematic seems to match this one: 
https://www.digikey.be/de/articles/hvac-dual-ac-motor-control-with-active-pfc-implementation-using-piccolo-mcus

[Eamon]

The AC input voltage in question could be measured at R94 even though it's rectified.

Firstly, thanks for that link, it does look strikingly similar to my board's circuit apart from specific component values (though DB2 does not seem to have a reciprocal part in the diagram). And reading the article was a bit enlightening too, I think I have identified the phase current measurement blocks as a result (3 x MV7221 OP-Amps connected to the low side shunt resistors from the IBGT driver)

Just double checked with the multimeter in DC mode and I don't see more than about 3V variation at the start of the resistor chain at R94 when the compressor starts and the inverter reports the big jump in AC voltage, unfortunately every time I try to measure the voltage at the end of the resistor chain after R99 as the compressor starts, the inverter reports "PFC Failure" code.  Hopefully the oscilloscope will be more of a passive observer when I get the chance to set it up and try.

What's strange is that the inverterDCvoltage is supposed to rise well over 320v when the PFC is turned on, but your measure doesn't show that.

How high is "well over"? I double checked the +vDC plane serveral times with the multimeter in DC mode and the highest I have seen is about 325V.  Is there something I can do with a multimeter to check if the PFC circuit is turning on?  My assumption from reading the article you provided is that it will be running at a frequency too high for a multimeter to be of much use?

If the reported AC voltage jumps from ~220v to ~320v, it could indicate that the output DC is finding it's way back to the output of the bridge rectifier. I might be wrong, but I don't think it's supposed to.

My thoughts are similar; the reported AC voltage when it jumps is almost exactly the same as the reported DC voltage, i'm just not sure where to measure to test that theory.  Hopefully when I can get a scope onto it I'll learn more.

I would remove and test C53A, Q1, Q2, D1, D2 and DB2.
DB2 is not used as a bridge rectifier but as a single diode.

De-soldering is not a strength of mine trying to avoid if possible. though if the problem is the +DC leaking back to the output of the bridge rectifier, it would seem that C53A could cause that?

[WattsThat]

The inverter supply of 320vdc should not rise above that value, unless the compressor pressure/piston is pushing back against the motor, causing it to be become a generator rather than a motor. This can lead to overvoltage trips and this may be a sign of a bypass valve not opening for startup to reduce the pressure to allow the use of smaller motors. Don’t know what the refrigeration system schematic looks like but it should be easy to locate anything in the refrigerant lines with wires on it. It might be a pressure switch but you should be able figure out if this is going to an input or output on the control board.

Your approach of trying to run the inverter with a different load is good idea but even better, the inverter may run without a motor. If not, a suitable load can many times be made with three 220v incandescent light bulbs wired in delta.

[Eamon]

The inverter supply of 320vdc should not rise above that value, unless the compressor pressure/piston is pushing back against the motor, causing it to be become a generator rather than a motor. This can lead to overvoltage trips and this may be a sign of a bypass valve not opening for startup to reduce the pressure to allow the use of smaller motors. Don’t know what the refrigeration system schematic looks like but it should be easy to locate anything in the refrigerant lines with wires on it. It might be a pressure switch but you should be able figure out if this is going to an input or output on the control board.

Your approach of trying to run the inverter with a different load is good idea but even better, the inverter may run without a motor. If not, a suitable load can many times be made with three 220v incandescent light bulbs wired in delta.

I've disconnected the heat pump controller board and I'm turning the inverter on an off directly with a small application I wrote using a USB to rs485 converter. I've reverse engineered enough of the communication protocol between the heat pump controller and the inverter module and that I can disconnect the heat pump controller and drive the inverter and compressor directly from a laptop so none of the various other sensors related to the refrigeration system should matter. The application essentially just polls the inverter for its state and sends the compressor on instruction in response to me pressing a toggle button on the laptop.  I'm pretty confident that the compressor is not faulty.

If I try to run the Inverter without anything connected to the U, V and W outputs of the inverter module it faults with a "Comp. Driver Failure", I've tried the only 3 phase motor I have at hand and it caused the inverter to report "PFC Fault" instead.  I'll see if I can try 3 resistive loads in delta, but I suspect it will not work as I believe the inverter would be using the phase differences on the three inputs to measure frequency or rotor position or something like that?

[WattsThat]

Is there any feedback from the motor back to the inverter? Some motors use three Hall effect sensors to tell the inverter where the rotor is, needed for proper switching. If there are no sensors/feedback, it should run with a resistive load so long as the current is close to that of the compressor load.

[Eamon]

Is there any feedback from the motor back to the inverter? Some motors use three Hall effect sensors to tell the inverter where the rotor is, needed for proper switching. If there are no sensors/feedback, it should run with a resistive load so long as the current is close to that of the compressor load.

No, the motor has only three wires, the three power phases from the inverter.