GoldenMotor.com Forum
General Category => Magic Pie & Smart Pie Discussions => Topic started by: Speedsterharry on November 14, 2016, 03:23:54 PM
-
Hi All,
I was a happy MP4 user up to now, for around 1 year. My kit was used primarily for going to work. One day, I went to work without any worries and at the end of the day, I noticed that the MP4 wouldn't react to thumb throttle (strictly no reaction), although the battery had more than enough energy to move a bit (48V 10Ah LifePO4). I had to go back home on foot and use the train that day (very frustrating time for me). The motor showed resistance when forced to rotate.
The few tests I've done until now: Battery works great as far as I know (plugged a 60W light bulb and the filament would glow). Charger would charge the battery if asked to. The brakes have variable resistance depending on lever position. The 3 LED gauge shows 3 lit LEDs. I don't know what else to do.
I stumbled upon a topic on endless sphere showing exactly the same symptoms, so I suspect this is a problem with the controller MOSFETs, although I didn't experience any loss of power during my 600 meter uphill trip (just before failure). The failure was so unexpected I thought someone vandalised my setup (could a MOSFET fail at rest, is this possible ??? Even after some kind of effort ?).
I'm ready to change the MOSFETs myself, but first: Are the IRF3077 mosfet the best to have in the MP4 these days, for a sub 60V battery ? If there's anything better, I'm ready to try that instead.
If the MOSFETs are not guilty, what else to try ? Is there anywhere to buy a MP4 controller ? I read that this version of the MP was discontinued ....
Thanks in advance
Jerome
EDIT: I can still access the controller settings thru USB-cable and Windows XP machine, via the Vector software
-
Hi Jerome and(https://i.imgur.com/evDSMvT.png)to the forum.
If the wheel can be rotated relatively easily at a very slow speed, but becomes very difficult to turn at normal speeds (almost impossible to pedal normally) then it does sound like one or more of the MOSFETs will have failed in a closed circuit state.
To check this you will need to remove the controller and disconnect the three phase wires, and make sure the connectors cannot touch each other, and you should find that the wheel will spin more easily again if it was failed MOSFETs.
I haven't actually tried, but I imagine it could be difficult to gain access to the FETS, due to the way the controller's PCBs are completely sealed into the cast aluminium heatsink/controller housing.
If the MP4 controller is no longer available, the standard MP5 controller (http://www.goldenmotor.ca/products/Magic-Pie-5-Vector-Internal-Controller-with-Vector-plug.html) or Bluetooth upgrade kit (http://www.goldenmotor.ca/products/Bluetooth-upgrade-kit..html) can be used instead.
However, you may also need to change the connectors on the Hall sensors and phase wires if they are not the same.
Also, if you were using a BAC-601 Smart Display unit with the MP4, I don't know whether it will still work with the MP5 controller. :-\
Alan
-
Hi Alan,
I'm glad this forum is still active, thanks Alan for your help !
I can indeed rotate the wheel very slowly, but not really pedal, including when going downhill.
I understand I can barely do anything from the outside of the wheel, and I'm ready to open the latter. Meanwhile, I ordered a set of 6 IR4030 Mosfets instead of the (apparently) usual 3077s. The allowed voltage is higher but the RDSon too (meaning more heat will be dissipated I fear :/ ).
What if I parallel the new IR4030s over the still-good 3077s, since they are pin-to-pin compatible, ie gate to Gate, Drain to Drain, Source to Source ? Is there a risk for the motor, the controller, the mosfets or me ??? (I'm no electronician so bear with me) ?
I didn't know one could upgrade the controller to that of the MP5, glad to know that. I will try this if all else fails.
I have the BAC-601 display at home but removed it because I feel it's not useful enough (battery levels are for Li-Ion, I'm on Lifepo4) and there's no display for cumulative energy dissipated by the motor (which would be extremely useful to know the real battery capacity - too bad GM didn't think about this option !).
Thanks for those PRECIOUS infos, Alan, they will be put to good use. I'll report back here when I have something new from my side.
-
OK, I opened the motor wheel. The controller is accessible from the side opposite of the freewheel.
I had a hard time opening the flask on the controller side until I followed Alan's advice (http://goldenmotor.com/SMF/index.php?topic=2103.msg10472#msg10472)
However, the controller components are made completely closed to any replacement (a very sad thing). I tried cancelling my purchase of TO220 mosfets because there's very little hope I'll be able to install those in the MP4 controller (that and the Mosfet format is probably not TO220 anyway). Checked the official Golden Motor web site for a stand alone MP5 controller to buy.... nothing. Only available on GM Canada, damn it.
From now on, I don't know if I should order a new controller from GM Canada (which is going to take ages) or try something with my fried controller...
Meanwhile, I'll have to use the car to go to work :'(
Open to ideas :(
-
Another view from the freewheel side. The trap was used to hide a few cables, and not the controller ...
-
A view from the controller. The screws and most components are covered with a transparent silicon thing, probably to prevent moisture damaging components or copper tracks. I couldn't inspect the mosfets since I found the screws are freely rotating >:(
-
aren't the mosfets on the opposite side of the controlers board of the last picture?
-
@asterix: probably so, but like I said in my post, I can't unscrew the board from the heatsink, so couldn't access the mosfets. I think I need to drill the screws from the outside if I want to go further.
The IR4030s order couldn't be cancelled, unfortunately for me ....
Also, I tried disconnecting the phase wires and one of them had the quick disconnect "parting" the wire. Seeing that, I started the soldering iron and prepared a new round quick disconnect, which I planned to solder to the wire for better mechanical stability. I held great hope that this could be the root of my problem ... but no actually :/
-
did you unscrew the big green blue and yellow wires and test them for continuity?
if you remove the battery and slowly move the wheel, is there power produced at motor power supply?
-
@asterix: I'm unsure where I should test for continuity (between which points exactly ?). As for your other suggestion (hadn't think about it, good idea), the motor wheel is still not reinstalled on the bike because I changed my mind: I ordered the MP5 controller on Golden Motor Canada and will resume repair when I get the part. Thanks !
-
there are 3 big wires: green , blue yellow. Those wires are maintained through black screws that can be seen at your picture of the controler.
I do not know exactly what should be measured (I haven't open my Magic Pie, but your post is at least interesting for this purpose, thanks for sharing that info), but I could imagine that those 3 wires are connected to the copper wires of the wheel border (near to the magnets). If those wires are big, it means that big current is supposed to flow through them, so this means that their resistance must be low. So you should measure a low resistance between green and blue, but also blue and yellow. May be if you do not measure a low resistance, it means that a wire is broken somewhere. So this could help traking where to search about broken wires.
-
I'm about to get the new MP5 controller from GM Canada, probably tomorrow (having something delivered by UPS is such a pain though - you cannot really choose the rendez-vous date/time). Also, the customs tax will be around 40 EUR. I have no other choice though....
On the troubleshooting front, the wheel definitely exhibits a resistive behaviour (switching between smooth and resistive), so I presume some mosfets are really dead. But since I cannot access them, why bother :-\ I contacted the seller of my MP kit, and he said GM could do a part exchange for the controller. Will take ages though, but better than nothing.
I have another problem when trying to fit the wheel in the bike frame: the axle is now out of shape after my various attempts at opening the flasks. What are my options on this issue ? Thanks
@asterix: I will make the measurements when the controller is out from the wheel again and report back.
-
Small update:
I got the thread of my axle rectified by a professionnal (because it got damage when I used the hammer to open the flasks). I tried replacing the controller, however the new controller isn't directly compatible with the new MP5: the new connector needs 4.8 mm quick disconnects instead of hollow, round disconnects than get screwed on the controller. Plus, and this is more annoying, the hall sensors connector is not the same between the MP4 and MP5 (see picture)
I changed the phase wires to quick disconnects, assuming the colors were the same with the new controller and let the hall sensors connector loose inside the wheel (I think the GM controllers work sensorless if the hall sensors are shot). When the system is under voltage and I push the thumb throttle, the wheel produces a slight hum but no movement occurs. I tried giving a little nudge to the wheel but still no movement.... :( (the wheel rotates with difficulty, even without controller and battery attached, I hope I didn't damage the mechanics of the wheel)
EDIT: Tried another time to remove the dead MP4 controller from housing to access MOSFET for testing/replacing purpose but no luck. I managed to removed the screws that I thought were maintaining the PCB to the housing but it's not enough. Damn it, I *hate* when companies make such decisions >:(
I measured the resistance between all phase wires: about 30k whatever the colors. Strange ...
-
( (the wheel rotates with difficulty, even without controller and battery attached, I hope I didn't damage the mechanics of the wheel)
does the wheel also rotate with difficulty if it is not screwed on the bike?
I measured the resistance between all phase wires: about 30k whatever the colors. Strange ...
30KOhms ? Is it stable value and identical value if you change the colors of the wires? Are you sure of your measurement instrument?
Normally those wires go directly to the big copper wires, so how is this possible to put some amperes inside 30k ?
I would first try to follow those color wires and make a measurement with a needle on the copper wires (afterwards you can always put some glue on it).
-
one other measurement to be done: the resistance between each color wire and the wheel itself (mass). I suppose it should be infinite (wires isolated from mass).
What happens if you rotate the wheel slowly without any battery: is there some voltage produced at power supply wires black and red?
-
I changed the phase wires to quick disconnects, assuming the colors were the same with the new controller and let the hall sensors connector loose inside the wheel (I think the GM controllers work sensorless if the hall sensors are shot).
Unfortunately, the Vector controllers do not have a sensorless mode, therefore the Hall sensors must be connected for the controller to work.
If the MP4 controller is no longer available, the standard MP5 controller (http://www.goldenmotor.ca/products/Magic-Pie-5-Vector-Internal-Controller-with-Vector-plug.html) or Bluetooth upgrade kit (http://www.goldenmotor.ca/products/Bluetooth-upgrade-kit..html) can be used instead.
However, you may also need to change the connectors on the Hall sensors and phase wires if they are not the same.
The Hall sensor plug will need to be replaced with the correct connector assembly to suit the new controller:
(http://www.goldenmotor.ca/product_images/y/110/MP5_replacement_controller_what_do_I_need__10092_zoom.jpg) (http://www.goldenmotor.ca/products/Magic-Pie-5-Vector-Internal-Controller-with-Vector-plug.html)
I measured the resistance between all phase wires: about 30k whatever the colors. Strange ...
The resistance between each pair of phase wires should only be a fraction of an Ohm, definitely not 30,000 Ohms! :o
Alan
-
the failing resistance test could be in fact a "good news": may be the old controller is still working perfectly well and the problem is "just" related to some wiring defect.
-
I installed the new controller in the wheel, while taking care of the hall sensors connector (hopefully in the right order): same symptoms as before ...
So I think there's something else not working. Since the Vector controller has no sensorless mode, I'm tempted to think one of the hall sensors is dead (would that mean the wheel will not rotate ?). This could be good news because in that case, the old controller might be completely functional. I haven't had the time to test the halls so far. Will do so on tuesday.
As for the 30k resistance, I was talking about two phase wires on the controller side, not from the wheel side (where I get a sub ohm resistance between any 2 phase wires). Sorry for not being clear in my description :P
Thanks for the suggestions, Asterix and Alan, they give me some things to test.
-
I detected a failing Mosfet on the Yellow phase (on the old controller). All the FETs are fine on the new one of course.
I still couldn't check the halls, the test is pretty difficult to convey even with 2 people. I can't access the hall sensors now because the screws holding the motor flask went out of shape when trying to open them, now they are stuck. I Used a drill and drill bit to get rid of the screw heads but there must be some metal left over that holds the lid closed. The motor isn't pretty anymore, I have no clue how I will get the flask closed with watertightness.
Asterix:Are you french by any chance ? I can give you the old MP4 controller if you wish to have a look for a fix, or take photos of the MOSFETs if you can access them.
-
yep Speesterharry, Asterix is "un Gaulois".
Best thanks for your offer.
Let's first send me some high resolution pictures of the MOSFETs, I will try to see what can be done.
See your private messages.
-
@Asterix: My offer was to send you the broken controller (for the price of postage) and YOU take some photos of the mosfets (I can't access them). If you can devise a way to fix the mosfets and let these forums how to do it (with photos), it would benefit the community. If that can help others ....
-
ok, no problemo and many thanks for your offer. I will try to access the mosfets and will publish the results of the investigations here so that it can benefit to the community.
See private message for contact via email.
-
I finally found the time to test properly (I think) the hall sensors, with 2 Li-Ion accus in series (around 8V in total)
HS#1: around 7.9V (steady)
HS#2: around 7.9V (steady)
HS#3: 0 V (never changes)
Done while slowly turning the wheel around the axle
Those readings would mean all 3 sensors are shot ?!? It doesn't matter, I bought 4 replacement hall sensors (SS41).
The only problem is to open the flask. I had 2 screws that were stuck in their place, I tried drilling their heads, thinking that this would free the flask, I was wrong. What could I do to now ? See photos for current state of the flask ...
Asterix sent me a set of photos in private of my dead MP4 controller and managed to get access to the mosfets, I hope he can find the time to post the photos and some instructions to replicate what he did.
-
yes, thanks for the reminder Speedsterharry.
This was the controler at beginning:
-
So as you can see there is some gel, glue that maintains the components inside the aluminium housing.
The trick for opening this stuff and get acces to the mosfets, is to use some screwdriver and to slowly push the board outside the housing:
-
then you put the gel outside as much as possible
-
the problem is that you also destroy some copper wires by doing this...see first picture below.
Also I made a mistake that should not be done: before using the screwdriver, I have unscrewed the screws...bad idea: those screws do not fix the board to the housing.
They are just soldered to the boards, so by unscrewing them...you destroy the copper wiring of the boards.
Hope those advices can help the next person that tries to open such a controler.
-
The only problem is to open the flask. I had 2 screws that were stuck in their place, I tried drilling their heads, thinking that this would free the flask, I was wrong. What could I do to now ? See photos for current state of the flask ...
If you are not using disc brakes, then simply refit the cooling fan/disc mount with the four good screws, as this will be more than adequate to hold the cooling fan in place.
However, if you have a brake disc rotor mounted to the cooling fan, four screws are not really strong enough to cope with the tremendous side loads exerted upon the screws, and there is a chance that the remaining four screws could sheer off completely under heavy braking. :o
If you are intent on using a disc brake with only four of the six screws, I would suggest permanently bonding the cooling fan/disc mount onto the side cover flange with some suitable metal adhesive in addition to the four remaining good screws.
Ideally, the side cover and the cooling fan/brake disc mount should be replaced, unless you are able to have new holes drilled and tapped with the required thread adjacent to the damaged ones, which would then allow all six screws to be used.
Alan
-
If you are not using disc brakes, then simply refit the cooling fan/disc mount with the four good screws, as this will be more than adequate to hold the cooling fan in place.
The main issue is not how to tighten the cooling fan afterwards but how to open the flask to access the hall sensors and change them.
-
To remove the side cover, undo the nine allen bolts holding the side cover in place and place a piece of tape on the cover in line with the valve to make sure it goes back in the same position.
Place a block of wood against the end of the axle on the sprocket side and give it a nice firm tap with a large hammer or mallet to separate the cover from the hub. The joint is usually sealed with a silicone sealant, and it may require a fair bit of force (several large taps) to break the bond between the cover and the hub assembly.
Once the bond has been broken, the side cover should slide off the axle with a bit of persistent pulling and wiggling on the cover.
Check out this thread (http://goldenmotor.com/SMF/index.php?topic=6134.msg34313#msg34313) for more information.
Alan
-
OKay, some news from me:
I managed to open the flask with a set of pure carbide router bits. Not pretty but it works. (pic #1)
The proper tools to remove a stuck screw was found several weeks after (pic #2). I wish I had heard about these before...
And last but not least, I built a small hall effect sensor tester (pic #3) and noticed that the behaviour of my replacement halls was not the one expected: Only detects when magnet is nearby (so not latched) and only sensitive to one face of the magnet.
After inspecting my ebay bills, I saw I bought SS441 sensors which are not exactly the same beast as the SS41 (unipolar vs bipolar and not latched vs latched). The correct replacement parts are now ordered....
Meanwhile, I bought some nine continent Direct Drive wheel (RH205), along with some 12 Fet controller from Lyen and accessories. I'd really like to install them soon but that would mean dropping the Golden Motor system altogether, which is not something I'd like to do right now since I HATE wastage. I hope my next attempts will yield success.
-
I made no progress on my problem due to lack of time.
Today, I noticed the controller (MP5) flashes red 5 times continuously as soon as the battery is activated with the key lock.
Does anyone know what this means and what I can do about it ?
When I push the throttle, there's still the low hum but no wheel spinning (a bit like a wrong phase wires/hall sensors combination)
However, I used the pinout indicated on Goldenmotor.ca
(http://www.goldenmotor.ca/product_images/y/110/MP5_replacement_controller_what_do_I_need__10092_zoom.jpg)
Any suggestion welcome :(
-
Five flashes usually indicates a problem with the Hall sensors (or the Hall sensor connector is disconnected).
If the connector is connected properly, you will need to check the +5V supply to the hall sensors and the output signal voltage from each of the three Hall sensors using a voltmeter.
Take a look at this post (http://goldenmotor.com/SMF/index.php?topic=2409.msg14247#msg14247) and this video (http://www.youtube.com/watch?v=7m8DA6mmo84) for more details on testing the operation of the hall sensors.
Alan
-
Small update:
I changed a failed hall sensor this morning. Now all three hall sensors are tested OK.
However, the controller insists on flashing 5 times in a row. I don't know what to do. There's a small chance I damaged the controller during my tests.
I think I will order an external controller (not Golden Motor), thread the halls and phase wires to the outside of the wheel and see how it goes. If the wheel still wouldn't rotate, I'll get rid of it and install some no-nonsense motor like a RH205. AND this time I know it will work !
Thanks for all the help to those who participated in this thread.
-
Changed a failed hall sensor this morning. Now all three hall sensors are tested OK.
However, the controller insists on flashing 5 times in a row. I don't know what to do. There's a small chance I damaged the controller during my tests.
Was the stator still removed from the motor ring (the ring of magnets inside the hub) when you tested it?
If so, this could also cause the five flashes from the controller LED due to the Hall sensor output signal voltages all being the same, which should never occur when the stator is surrounded by the motor ring.
Alan
-
Oh My God, Alan, you are an incredible GENIUS !!! How can you possibly give such a good advice and have such a strong intuition ??? I don't have time to make further tests right now, maybe tomorrow afternoon. It's sad because I just ordered a NC RH205, controller, etc yesterday. I guess I will make good use of it at one point in time.
There's a slim chance I can salvage my MP4 motor and MP5 controller after all ...
Many many thanks, it's so nice to see the light at the end of the tunnel. I hope I can pull this off thanks to your help :) :)
-
OK I guess I was too optimistic 2 days ago.
When the stator is in the ring of magnets, the controller still sends the series of 5 flashes.
Also, I cannot possibly rotate by hand the stator inside the rotor (completely stuck, whether the controller is connected to the stator or not). I watched a vid explaining about how BLDC work and I don't think this is normal....
-
Also, I cannot possibly rotate by hand the stator inside the rotor (completely stuck, whether the controller is connected to the stator or not).
If the stator is properly centred in the motor ring (and both side covers are fitted correctly) you should be able to rotate the axle (and stator assembly) using a 10mm open ended spanner without too much difficulty, but don't expect to be able to turn the axle just by twisting it with your finger and thumb.
If the covers are not seated properly, it may be possible for the stator to touch against the magnets, making it very stiff to turn.
Also, if any metal particles have stuck themselves to the magnets while the stator was removed, they could easily wedge themselves into the small gap between the stator and the magnets when the stator is refitted, even with the stator correctly centred inside the magnets. ;)
If the axle is very tight to turn with a 10mm spanner, or makes a horrible grinding sound, it is likely to be caused by some form of physical resistance rather than the normal magnetic drag, and you will have to remove the stator and look carefully to locate the cause.
Unfortunately, if you are still experiencing the continual 5 flash fault code on the controller LED with the hall sensor plug fitted correctly, the controller could be damaged. And if the controller has failed, the motor is not likely to run, even if the axle does turn freely with the spanner. ::)
Alan
-
OK Alan. After following your advice, I could rotate the wheel with little resistance, by making several turns with a spanner and refitting the flasks. Probably the little metal particles that created friction, and the fact the stator was not properly centered without the flasks, like you suggested. This part is now sorted.
Alas, things have not changed at all with the controller. Still the 5 flashes. I also randomly interverted phase wires to see if anything happened. In some configurations, the wheel stay stuck, in others, the wheel makes very little "hops" and then stays still. In any case, the red LEDs still flash 5 times.
Im going to try an external sinewave controller with the wheel and see how it goes. Probably get rid of the GM MP5 controller, I have no clue if it still has value :( (my god, more than 100 EUR thrown out the window)