Why a power resistor in parallel with relay ? Treadmills power board [Fixed]

[valere]

Hi,

A friend of me, a neighbor, retired for a long time, actually the one who initiated me to electronic back when I was younger... is fixing a lot of electronic stuff for fun. Now that the student ;-) is an electronic engineer, the master asks for help when things get too complicated.

We are digging (well when I've a bit spare time) on why Treadmill is not working anymore (someone gives him this for repair).
The power switch actually burned, the protection varicap also... and well this is a piece of bullshit there are two places for fuse and... one is just not populated (main fuse there is just a standard isolated wire) and the other one (on board) is a solid wire ( )... He fixed that ! He also changed the DC capacitor (400V 470uF). It's not working, the "user interface," said error xx which means "engine default".

If he bypass the (on/off) relay the engine spin at full speed, this made me think the control MOSFET let some of its magic smoke away...

But my question is why there is a power resistor (5K, 5W) R (red on the schematic), in parallel with the relay which 230V AC between its plots... I don't know why this is used. My main guess, it's to channel a small amount of current in the power part of the board to allow automatic health checking and to prevent turning on  (by switching the relay and thus bypassing the resistor)  the main power if something is wrong.

Anyone have any idea ? About the resistor and why this isn't working anymore...

EDIT : Figure / Schema update, Freewheel diodes misplaced (what was I thinking seriously...)

EDIT2 (22/01/2017) : Mosfet is not a Mosfet as advertised on the silk, an IGBT is installed (G4PC40U)

EDIT3 (23/01/2017) : Treadmill power board fixed, It was the power Transistor (and the Capacitor that we already changed before the forum post) see in video there, https://www.youtube.com/embed/bFq2gU_l7fk Thanks you all for your help !

[Andy Watson]

To limit the surge current when the unit is switched on. When you apply power you could be trying to charge-up the 470uF to 320V instantly, the resistor prevents the very high current that might occur. After a second or two (no longer) the 470uF should be charged and the relay will pull-in and short-out the resistor so that you can take full power from the motor-control circuit.

[Tandy]

It is called an inrush current limiter.

[Armadillo]

Anyone have any idea ? About the resistor and why this isn't working anymore...

I am thinking that the resistor is too hot. The temperature is not normal. The contact of the relay could be worn out. Need to replace the relay. But I think the mosfet is shorted. Please check the mosfet and replace the relay.

[mariush]

It's probably the relay dead (no longer closing) or the npn transistor is dead (i don't think it's mosfet), the transistor being  the one connecting one side of the winding to ground through it.

Check the relay , check the npn transistor.

If you want, you can probably replace the microcontroller control with a 555 one time shot circuit. Choose the components so that the 555 waits about 5-10 seconds and then triggers closing the relay (not sure if it has enough power to drive the relay directly, maybe use a npn transistor) and stays on until you turn off the device.

[valere]

Thanks for your help, eevblog forum and people here are really all amazing !

To limit the surge current when the unit is switched on. [...]

It is called an inrush current limiter.

Thanks guys, I feel stupid not to have thought about that*, on the other hand, this is why repairing things is a good idea and my feelings about Eevblog forum was right it's the best place to discuss electronic 
 

I am thinking that the resistor is too hot. The temperature is not normal. The contact of the relay could be worn out. Need to replace the relay. But I think the mosfet is shorted. Please check the mosfet and replace the relay.

The resistor was checked with a thermal camera (Flir one G2 + extension micro USB wire), in fact, I first came on the forum to exchange about it (not only of course). Despite all the possible measurement error, I'm sure it's out of the camera "standard range" meaning more than 125°C (probably between 150 and 180°C), especially considering that a white resistor is emitting less than my standard tuning for black body radiation. And double proof, after disconnecting AC for safety, it's relay hot on the other side of the PCB... Unhopefully, I've not taken any picture with the thermal camera.

As far as I know the relay has not failed, it's the main controller which decided not to switch it on because there is an engine failure. However, we will investigate that if our other idea proves to be false.

Like you, I really think the mosfet is shorted, I asked him to check V_GS (will keep you updated), if it's null since the resistor is too hot and the engine rotate full speed when the relay is bypass, the Mosfet is fried.... (which make sense in my experience Mosfet like to be conductive when they are fried and the previous capacitor was dead, meaning that the Mosfet really had some bad days with an awkward rectified only sinusoid....

On the other hand, there is really no reason for the control system to command the engine to rotate full speed (relay bypassed), except maybe if the optical encoder sensor is in trouble but... I do really find the died Mosfet being more probable.

It's probably the relay dead (no longer closing) or the npn transistor is dead (i don't think it's mosfet), the transistor being  the one connecting one side of the winding to ground through it. Check the relay , check the npn transistor.

I don't fully remember if we have checked the relay (there is 230V AC and 200V DC and not such a good accessibility, we think twice before doing something), however, as far as I've seen, the control, say "engine error" and decide not to turn the relay "on", not a relay failure, but we will double check that in case the relay had a difficult day when thing got smoking bad... I really think the Mosfet is dead, and yes it is a Mosfet as far as I remember (with some enhanced protection integrated maybe), I would have used a Mosfet to do this, also I've seen (judging by the number of pins and the wiring, I've not double checked the reference) something looking like a Mosfet driver IC (makes sense for a PWM, speed controlled engine).

But I understand why a bipolar transistor makes sense, for power system and I had some insane day with an ultra-low voltage crazy super capacitor project, and the bipolar was well really life saving... Try to know what happened to the Mosfet when you alternatively have power, on the one hand, or, on the other hand, or both (surcharge protection) is brain damaging and the magic smoke escaped (we were three friends, all electronic engineers trying to solve that but... the unusual bipolar structure proves to be way easier and skater...). I like Mosfet a lot but they have also some annoying properties.

If you want, you can probably replace the microcontroller control with a 555 one time shot circuit. Choose the components so that the 555 waits about 5-10 seconds and then triggers closing the relay (not sure if it has enough power to drive the relay directly, maybe use a npn transistor) and stays on until you turn off the device.

I like the 555 way of thinking, it made me happy  . However I could do that using many other technic. I probably have some 555 in my stuff, which I did not have with me there. There is a NPN transistor to command the relay on the board, since it's linked to the integrated controller I'm not sure we will force its command or bypass it yet). Also I could probably take any uC small development board and recreate all the engine control** :-P but I do not have that amount of time available...



* [another field of application/robot] In fact, the inrush current limitation is something I wanted to apply on the electronic board I design for autonomous robots without begging satisfied but the solutions I was finding. What I've seen here and what you said will probably made me way more creative... I do have an impressive capacitive load in our robot power systems (well those capacitor are actually one of the main reasons we have such a stable and virtually noise-free power supply all over the system so...

** [another field of application/robot] I will probably end up wanting a millimeter precision control on the treadmills  , I'm never satisfied when an engine controller with a feedback loop is... not perfectly tuned and precise (we end up getting an impressive trajectory control with a sub +/-1mm +/- 0.57° error even during acceleration and deceleration in a 0.9m/s^2 1m/s 8rad/s 20rad/s^2 12kg robot.

[Armadillo]

I am thinking that the resistor is too hot. The temperature is not normal. The contact of the relay could be worn out. Need to replace the relay. But I think the mosfet is shorted. Please check the mosfet and replace the relay.

The resistor was checked with a thermal camera (Flir one G2 + extension micro USB wire), in fact, I first came on the forum to exchange about it (not only of course). Despite all the possible measurement error, I'm sure it's out of the camera "standard range" meaning more than 125°C (probably between 150 and 180°C), especially considering that a white resistor is emitting less than my standard tuning for black body radiation. And double proof, after disconnecting AC for safety, it's relay hot on the other side of the PCB... Unhopefully, I've not taken any picture with the thermal camera.

As far as I know the relay has not failed, it's the main controller which decided not to switch it on because there is an engine failure. However, we will investigate that if our other idea proves to be false.

Like you, I really think the mosfet is shorted, I asked him to check V_GS (will keep you updated), if it's null since the resistor is too hot and the engine rotate full speed when the relay is bypass, the Mosfet is fried.... (which make sense in my experience Mosfet like to be conductive when they are fried and the previous capacitor was dead, meaning that the Mosfet really had some bad days with an awkward rectified only sinusoid....

On the other hand, there is really no reason for the control system to command the engine to rotate full speed (relay bypassed), except maybe if the optical encoder sensor is in trouble but... I do really find the died Mosfet being more probable.

I imagine the events here;

The user press the start or On button, the microcontroller closes the relay and ramp up the PWM signal to the gate of the mosfet so that the speed matches with the user RPM settings, however the motor didn't turn, the microcontroller times-up and flag an error signal and turn off the relay for safety. [here when you measure the relay, the relay is off]
When you jumper the relay, the motor spin at full speed instead of controlled speed, that could mean that the mosfet is full ON or shorted. The resistor should not be hot, the only reason I can think of is, the mosfet is full ON or shorted but the relay is not closed, thus the current at full DC voltage is applied across the resistor, heating it up. It should not be so hot.

And

If the incoming AC voltage is 230vac, and the bridge rectifier is a full bridge, you should be getting more than 300Vdc instead of 200Vdc.. please check the full bridge rectifier.. don't look right. [Remember to discharge the capacitor when working with such voltage].

[Theobald]

Hi
the relay is out. You can't get more than 200V because the current is limited by the 5k resistor. If you're lucky, everything is fine on the motor side, you need only 30mA to drop 150V on 5k resistor. At this time, the power dissipated is 150x150/1500=15Watts, 3 times the rating, it goes hot.


Do you ear the relay click ?

Theo

[Tandy]

Indeed the relay is the most likely culprit being mechanical and all. No need to mess with the mains voltage just put 12V across the relay coil and test for continuity across the relay contacts. If it is working then the next most likely would be that the back emf diode has failed and the transistor switching the coil has been killed out by the coil.

[valere]

The user press the start or On button, the micro controller closes the relay and ramp up the PWM signal to the gate of the mosfet so that the speed matches with the user RPM settings, however the motor didn't turn, the micro controller times-up and flag an error signal and turn off the relay for safety. [here when you measure the relay, the relay is off]

Well, the system is turned on, then diagnostic is done (possibly even without switching on the relay), then if everything is okay, it waits for the user to choose his "walking/running" program ;-) . I do really know that the relay is currently off, of course, but I got interested by the fact that the resistor was too hot for normal operations (thermal camera pinpoint obvious things like that, and more than 100°C is usually not good so I dig this path).

When you jumper the relay, the motor spin at full speed instead of controlled speed, that could mean that the mosfet is full ON or shorted. The resistor should not be hot, the only reason I can think of is, the mosfet is full ON or shorted but the relay is not closed, thus the current at full DC voltage is applied across the resistor, heating it up. It should not be so hot.

It's even worse in fact, I asked for a few more tests, the Mosfet is not controlled (VGS~0V), meaning that neither the engine should spin full speed, neither the resistor should heat that much : I ordered two mosfet, he will change it, then we will see.

In fact, in those tests everything is online (engine disconnected), the capacitor charge, then connect the engine, it spin a bit (capacitor discharge) + VGS kept equal to 0V (not commanded).

The mosfet is an IRFP460 (meaning I was right not a bipolar, anyway I wouldn't understand why putting something else than a Mosfet ;-) in this specific case, well maybe an IGBT if it was a giant treadmill for an elephant  ).

EDIT (22/01/2017) : IRFP460 is advertised on the silk, real part is an IGBT G4PC40U (IR)

If the incoming AC voltage is 230vac, and the bridge rectifier is a full bridge, you should be getting more than 300Vdc instead of 200Vdc.. please check the full bridge rectifier.. don't look right. [Remember to discharge the capacitor when working with such voltage].

You are totally correct, I do know that I've forgotten this part of "high voltage" courses... the V*sqrt(2) or Vmax/sqrt(2)... Well, a good time to fix that in my brain after all (I usually never work with electronic related to the main voltage. Moreover my non DC signals are never sinusoids so I either calculate or measure the real Ueff or Umax).

In fact I said 200Vdc for, I don't remember why, I didn't measure this one, so ok for ~320 VDC, thus I asked him to measure it again (engine unplug, meaning even if the mosfet is fried... this will be ok).

the relay is out. You can't get more than 200V because the current is limited by the 5k resistor. If you're lucky, everything is fine on the motor side, you need only 30mA to drop 150V on 5k resistor. At this time, the power dissipated is 150x150/1500=15Watts, 3 times the rating, it goes hot.

Do you ear the relay click ?

I think it didn't click anymore (first thing we checked - without forcing it yet), but if I had designed this I would have used the inrush resistor to diagnostic if there is any short or mosfet default with the micro controller without even turning on the relay (before doing that in fact). Obviously there is at least engine current measurement (clearly) and I think I've seen something looking like 300VDC measurements for the uC. I think the relay is not turned on because the default is detected, but we will check it (harder to supply than the mosfet by the way).

The relay is a SLA 12 VDC A - 30A

By the way there is 230V AC across the inrush resistor, so if I'm correct, it's 230Veff so we more or less have 230*230/5000 = 10.58W dissipated, so twice the resistor rate which was what I calculated when I've seen that and the reason why it's that hot... indeed.

Indeed the relay is the most likely culprit being mechanical and all. No need to mess with the mains voltage just put 12V across the relay coil and test for continuity across the relay contacts. If it is working then the next most likely would be that the back emf diode has failed and the transistor switching the coil has been killed out by the coil.

We are thinking of testing the relay since the beginning of the battle, but, it proves a bit tricky to be done safely... we may manage to do that but the time the new mosfet arrived. However basically, either we do that when everything is online (AC connected) and it's a bit dangerous, either we do that offline with a 12V DC source but if I'm pretty sure the NPN transistor don't care about it, I'm not exactly sure if the 7812 will like the 12V on is output when it's not powered ;-) , and it will get the chained 7805 online, which means everything else in the logic part will start, mmm not perfectly happy with all of that. 

I am really too tired this evening to recheck how exactly a 7812 is made* and what will happen if we put 12V on its output and nothing on its input 

*by the way I should check for this specific flavor of 7812... which I don't know, since we didn't remove it from the board and it's difficult to read

Next steps
- Check V_GS : done V_GS=0, but the mosfet is conducting > change the MOSFET
- Order two replacement mosfet : done
- Change the mosfet : waiting
- Check the relay (if needed - depending of the mosfet results) : waiting if needed
- Check the NPN + diode command of the relay : waiting if needed

[Tandy]

Putting 12v across the relay for testing is unlikely to damage the regulator.  However if you are concerned the easiest way to protect it is to short the input and output terminals on the 7812, no need to remove it,. Can usually easily done with a crocodile/alligator clip as the input and output are on the outer pins 1 & 3.

[Theobald]

By the way there is 230V AC across the inrush resistor, so if I'm correct, it's 230Veff so we more or less have 230*230/5000 = 10.58W dissipated, so twice the resistor rate which was what I calculated when I've seen that and the reason why it's that hot... indeed.
Only when the relay is off and the current is not null....
Step 1: the relay is off and the load is off.
Step 2: when the 230V is switched on, the 5k resistor limits the current of the charging capacitor. When the capacitor is charged, the current is null, and the power in the resistor is 0.
Step 3: the relay is closed.
Step 4: the motor runs, the current is no more limited by the resistor, the power in the resistor is null.
Theo

[valere]

By the way there is 230V AC across the inrush resistor, so if I'm correct, it's 230Veff so we more or less have 230*230/5000 = 10.58W dissipated, so twice the resistor rate which was what I calculated when I've seen that and the reason why it's that hot... indeed.
Only when the relay is off and the current is not null....

I know... By the way, I already knew that if the MOS wasn't fried, even with the relay off, it wouldn't even maintain 10.58W for more than a few seconds, so no significant heating.

Step 1: the relay is off and the load is off.
Step 2: when the 230V is switched on, the 5k resistor limits the current of the charging capacitor. When the capacitor is charged, the current is null, and the power in the resistor is 0.
Step 3: the relay is closed.
Step 4: the motor runs, the current is no more limited by the resistor, the power in the resistor is null.

I'm well aware of this sequence, I was only missing the ""anti" inrush current" function of the power resistor  . It was harder to identify all of this because we had to figured out what was going were (and we weren't so sure about the tracks under the resistor on the bottom of the board).

[Vtile]

I like the paint schema makes true engineer vibe.   

[valere]

I like the paint schema makes true engineer vibe.   

Take anything like paint, a Wacom Tablet and suddenly remote engineering is easier.  This is my online version of the DaveCAD(TM).
Some other schematics were done by the friend, using pencils, paper, and picturing the PCB haha... I redraw a simplified version by memory a few days after for the purpose of the forum post. I could have used Kicad, but well, Paint is better 

I already explain someone PWM resolution (or something like that to improve their PWM settings on an AVR) at 2 or 3 AM at the French Robotic Cup using a Moleskin Notebook and a rollerball... that was epic, but efficient hahaha same sort of fancy pansy engineering technology that makes true engineer vibe 

[valere]

Hi everyone,

I got some free time to go where the Treadmill is.

We received the MOSFET, however it's the reference written on the board (IRFP460) which like the freewheel diode isn't similar to the real part.  In fact, for the diode the written reference is indeed false and clearly false (not the good polarity for one of the dual diodes), the installed one is the right polarity.

I previously said that I would use a Mosfet or maybe an IGBT (if voltage was sufficient).... and guess what ? The real part was an IGBT.... I wasn't perfectly aware that Gen4 IGBT got to voltage as low as that (~300V).

That was just to keep you updated about the repair, I will get back here as soon as we have more feedback on our repair.

I attached a picture of the board to let you see what it was about.

EDIT : The IGBT is a G4PC40U (IR)

[valere]

A small piece of complementary information : MOSFET vs IGBT. In this particular situation we are in a "gray zone" where one technology isn't fundamentally better than the other one. For the fun fact, the MOSFET advertised on the PCB silk (IRFP460) is compared to the IGBT really installed on the board (G4PC40U) in a publication from International Rectifier / Infineon (http://www.infineon.com/dgdl/choosewisely.pdf?fileId=5546d462533600a40153574048b73edc)

[valere]

Hi everyone,

I want to thank you all for your help ! As I guessed, It was the transistor not the relay 

Like we always do by helping fix this Treadmill power board I learn new stuff, and you help me lots with the "Inrush limiting function of the resistor”.

We changed the IGBT by a MOSFET (the one advertised on the PCB), I would have preferred the IGBT but we already ordered & received the MOSFET when we discover that it was an IGBT and the specifications seems similar enough and I've done my homework by checking them, reading a nice Infineon IGBT vs MOSFET comparison in which I learn much, and checking if the MOSFET would normally be suitable for this purpose (as far as I can tell except that it's a bit less resilient to some issues, it's OK).

Enjoy the Treadmill working at low speed, low load (for now, then the load will be gradually increased) video there : https://www.youtube.com/embed/bFq2gU_l7fk