How To tune up the PWM frequency

[JonEQuest]

You most likely blew one or more of the main FETs and now it's shorted.

If you disconnect the controller, use a meter to measure resistance between the battery neg. wire and the motor neg. wire. If it's good, it will measure like a diode (reverse the meter probes). If it's shorted, well, it will measure like a piece of wire.

You can replace the FETs. If you do, I suggest using better (higher rated) ones.

Thanks for the advice. I started to de-solder and check them out of circuit one at a time. Then half way through I thought maybe it was the throttle for some reason. I did notice that of the 4 pairs of transistors the first pair are different than the next three pair of them that are all identical to each other. Other photos of this controller I have found online show all 4 pairs as identical chips. That has me baffled a bit.

 

[fechter]

Try to read the number off them. Some are FETs and some are diodes (in the same TO220 package).
If the diodes short, it's real hard to spin the motor. Alternately, use an ohmmeter to measure both ways across the motor wires. It should look like a diode, not a short.

 

[JonEQuest]

I will do that and post what I find. Thanks.

 

[JonEQuest]

It seems to have shorted transistors as you suggested. I removed all of them and will see what the better replacements cost.

 

[JonEQuest]

I found one shorted TK80E08K3 MOSFET out of the 8. The spec sheet says drain source voltage up to 75v and drain current rating of 80A at or below 175c. I am running six 12V10AH batteries for 72v probably up to 83v unloaded and fully charged, so I am pushing the limits to be sure. I have been out of tech school for many years so I am not sure what the current limits are for these and the higher voltage would effect that too I guess. The avalanche current is 40A. I know that the stock fuse for this scooter was 40 amps and it blew the second I accelerated the first time. For now I am running no fuse but I estimate that my current is somewhere in the 70amp+ range. The motor is rated as 1,000 watt by Currie at 36volts (stock voltage). I am hunting around trying to see what other transistors I may be able to drop in with slightly higher ratings but I am not having much luck randomly guessing what specs to search for.

 

[fechter]

That one is rated for 75V, 240A pulse, RDS (ON) = 7.5 mΩ

There are many alternates that have much better ratings. You should use a 100V rated part and get the lowest RDS you can find.

I upgraded the FETs on my blown Crystalyte controller with IRFB4110s and have run peaks over 80A on the battery side. 50A for minutes at a time. My limitation is no longer the controller.

 

[JonEQuest]

OK, I am looking at Mouser since they seem to have a great elimination filter that you can use to sort through options. The spec sheet of the stock one says Vdss instead of Vds but I guess that would be the same, Mouser refers to the value as "Drain Source Breakdown Voltage" where the Toshiba spec sheet just says "Drain Source Voltage". I was trying to search for anything above 80v with highest amps but the I guess the higher voltage ones won 't break down as easy. Do you think that the IRFB4110 will be compatible for this application? Do you have the same or similar controller? I was starting to get concerned about the resistance and capacitance values being within the specs of the circuit I have. Maybe I am over complicating it a bit. Funny thing is these chips will probably cost me more than the whole controller and this one on e-Bay is starting to look appealing but it won't fit in my compartment. I even had to grind the mounting tabs off the one I am using to fit.

http://www.ebay.com/itm/271603850898?_trksid=p2060778.m1438.l2649&var=570437543903&ssPageName=STRK%3AMEBIDX%3AIT#ht_2620wt_949

 

[JonEQuest]

I think I should have looked at the specs for the IRFB4110 before asking. That looks perfect for my application and they seem to be dirt cheap from China on e-Bay. I may just have to wait a couple weeks to get them.

Another thing I may need to upgrade is the pair of Schottky barrier diodes that are in line with the MOSFETS. The pair I have are C87-009 (ESAC87M-009) rated 90v at 16Amps average output current. One has a slightly different impeadance than the other one so maybe one was already breaking down.

 

[fechter]

IRFB4110s are pretty cheap now. I've had very good luck with them.
The amp rating is not so important. The legs will melt off at 70 amps but the silicon can take more.
The RDS is much more important. VDS and drain-source ratings are the same thing.

The Schottky diodes could possibly use an upgrade too, as they take a beating as much as the FETs. Amp rating counts here.

I have several similar controllers. The power stage is the same but the input side is different. With upgraded parts, the motor becomes the weak link.

 

[JonEQuest]

IRFB4110s are pretty cheap now. I've had very good luck with them.
The amp rating is not so important. The legs will melt off at 70 amps but the silicon can take more.
The RDS is much more important. VDS and drain-source ratings are the same thing.

The Schottky diodes could possibly use an upgrade too, as they take a beating as much as the FETs. Amp rating counts here.

I have several similar controllers. The power stage is the same but the input side is different. With upgraded parts, the motor becomes the weak link.

I decided to order 10 of the MOSFETS. I need 8 but the sale was for 10 and 2 extras can't hurt, especially when it was like $10 total from China. I also ordered upgrades for the Schottky diodes as well. The old diodes were rated for 90v and 16amps (per diode I guess?) The ones I have coming are rated for 100v and 40/80 amps (40 per diode, 80 total. The old ones never went but I figured the same thing. They must take a beating when the throttle is suddenly cranked up or probably worse when it is cut off at top speed and the DC motor suddenly surges back a few amps.

 

[fechter]

I think that should work great.

The most stress on parts happens during high load and partial duty cycle. This would be like a full throttle startup from a dead stop or climbing up a hill that is steep enough for the motor to slow down. At around 30% duty cycle, the current going to the motor can be as much as 3 times higher than the battery current! At least that's what I measured on mine.

 

[JonEQuest]

I think that should work great.

The most stress on parts happens during high load and partial duty cycle. This would be like a full throttle startup from a dead stop or climbing up a hill that is steep enough for the motor to slow down. At around 30% duty cycle, the current going to the motor can be as much as 3 times higher than the battery current! At least that's what I measured on mine.

That is a LOT of current. I just run this thing around my neighborhood on concrete streets. Of course It is a little scooter with 8" rims and 12" tires. The rear tire I did upgrade to a moped tire that is larger but I wouldn't take this thing off road or on hills. I may build a bike next since that would be a lot faster and more useful.

 

[JonEQuest]

Update on the controller repair. All seemed to be going well and I even replaced my throttle with one that has a nice battery gauge and a key to turn on and off the scooter so no one runs off with it.

The problem is when I finally got it back together the first time I hooked it up it just got hot and the motor instantly ran at full speed. Then I pulled it out several times looking for shorts, etc. Somewhere in there I actually put it back and had the battery wires swapped. Yes, very dumb but this thing had me working all day. After 20 reinstalls I was getting sloppy so every time I hooked it up it burned some traces right off the board on either side of the positive lead that runs to the two Schottky diodes. I began fixing the gaps with thin pieces of wire as sort of a fuse. Then I pulled the board again and again and looked for soldering shorts, reinstalled components, etc. etc. I could find no shorts at all and high impedances across the three power terminals but they still blew right away. I finally walked away for a while, then when I came back I realized I had swapped power wires on install time #10 or so. I swapped them back and soldered 2 more little "fuse traces". Hooked everything back up and powered on, no blown traces, no instant motor on full power, then I eased the throttle and the motor started off slow like I should and started to turn, but of course my thin traces blew with a couple of amp draw as they should since they were temporary anyway. I pulled it out again and remade the positive to Schottky diode traces nice and thick. Put everything back together and when I touched the throttle, the controller sparked and popped the first second. I disconnected everything quickly and realized the motor engagement got everything hot REAL fast. So, for the first time, I tested the motor itself. It is ALMOST A DEAD SHORT. LIKE .5 OHMS. it is a gear reduction drive motor, as I turned it by hand my ohm meter did flash higher impedances for a moment but you couldn't land on anything but almost dead short.

Now I know what happened a few weeks ago. The motor has a short, burned one of my FETs and I have been chasing my tail this whole time. The motor does still spin if power is applied but at such a low impedance it take a LOT of amps. I am sure that I could take the motor apart and find some brush pieces causing the short or maybe a partially melted stator. I have been running 72 volts on a 36 volt motor for a while now. I never noticed the motor get really hot though. It already has a heat shield and aluminum fins to dissipate heat so when I felt it get kind of warm after a long run I figured that was normal. The motor even has "Caution Hot" warnings on it. In my experience you can usually over volt DC motors quite a bit. My golf cart has a stock 36 volt motor with some modified windings (less winds for better ratio) for more speed and I have been running it at 48 volts with 2 adults and 6 kids on it sometimes for the last 2 years with no breakdowns. Maybe these Currie motors are already run at the limits when stock. This may be the end of this project for me. I am not buying an expensive motor. Hopefully the brushes just wore down and the wired ends are shorting out on the stator.

 

[fechter]

It sounds like one of the windings is shorted. This can be very hard to track down since the normal DC resistance of the windings are very low to start with. If the motor has a shorted winding, it will take an abnormal amount of torque to make it spin fast. If you just turn the shaft quickly by hand, you can feel the resistance.

If only a part of a winding (say a few turns) is shorted, it can be very hard to track down. Spin the motor with a drill or external source and see which winding gets hot. The real scientific way to find it would be with an inductance or LCR meter.

If you measure resistance from the motor wires to the shaft, there should be no connection. Most shorts I've seen are between the windings and the iron laminations of the armature. If you get a low resistance reading, that's a sure indication of failed insulation.

If the current limiting circuit in the controller works well, it should keep the controller from blowing.

If a winding is shorted, you either need a new motor or you need to re-wind the existing one (not an easy chore). I've done it a few times.

 

[JonEQuest]

I agree, I am hoping it is just that the brushes are worn down and the braided wires have contacted the armature or something that can be fixed with new brushes. If it gets more involved than that I am probably going to scrap or cannibalize this project. I have many hours and a few hundred $$ into this "Free" scooter already. Shame is that I have already modified and welded the swing arm to accept a larger moped tire (I found one to fit the 8" rim) and I have replaced the bearings, tires, brake pads etc. I guess I would just convert it to a good old chain motor if I cannot fix this gear driven one. Then I could get 30-35 mph out of only 48-60 volts instead of the 72v it takes me to get to 32mph now.

John

 

[JonEQuest]

I have taken the motor apart and I am surprised to say it looks real good inside. The brushes still have plenty of life, there is no heat marks on the brush holders and the armature coils look OK too. I tested impedances between individual segments on the armature looking for a dramatic pattern like several ohms across many then suddenly dead short on a few but they all seem very consistent at about 1-2 ohms per. I called Currie and spoke to tech support that basically knows nothing. The lady on the phone said they don't have access to any technical info like motor winding impedance, etc. Finding what ohm reading is typical seem all but impossible. I think the motor may just have a low impedance and the controller is probably still bad in some way or maybe just can't handle the low impedance now with the modified mosfets. I have two testing possibilities one is to wire up the original controller and see if it can operate normally which will likely smoke it. My other choice is to just hook up another small motor (that I don't have) and see if it works fine with the controller. What I would rather do is just make sure the motor I have is good or bad but I am not finding any troubleshooting or testing tips for this motor anywhere on the web.

 

[fechter]

The gear reduction part makes it harder to test.

One simple test is to attach the motor directly to the battery and see what happens. A good motor should spin up instantly and not get too hot after running for 30 seconds or so. You might want a fuse wire in series in case the motor is really bad.

Another test is to lift the wheel and try spinning it by hand with the motor wires disconnected. There should be some resistance but not too much. If the motor has shorted windings, it will be like having brakes on.

 

[JonEQuest]

I think the motor is totally fine. When power is applied it starts right away and spins fast. I also found a nice website that goes through testing windings of brush armature motors and mine tests fine. I plan to drill 4 tiny holes to insert paper clips and hold the brushes so I can reassemble the motor. This seems to be the only way to hold the brushes back that I have found on a few other motors I have taken apart as well.

Here is that site: http://www.groschopp.com/how-to-check-a-motor-armature/

Sooo.... this leads me back to the controller. I replaced the components with upgraded ones. I cannot see what else could be wrong. I cannot find any shorts that would cause this sudden failure when I try to turn the throttle so I am baffled. I am ALMOST tempted to order another new YK43 and just upgrade it with these parts BUT what if the different mosfets ARE the problem. Part of me wants to reinstall the old ones (except the one bad one of course) and see if it blows the traces or not then.

 

[fechter]

It's hard to measure the resistance of the windings unless you have a special meter that measures milliohms. A short between two turns of a winding will be almost impossible to detect even with a good meter. The real test is to apply voltage and measure the current draw of the motor with no load. A shorted winding will cause the no-load current to be much higher than normal. You can test at a reduced voltage, like 12V.

It sounds like the gate driver circuit in your controller is probably damaged. This frequently happens when the FETs short. You really need an oscilloscope to measure the gate drive, though just measuring the DC voltage on the gates at full throttle might tell you something (this can be done with the motor disconnected). You should see 10-12v on the gates measured against battery negative.

A new controller is pretty cheap.

 

[JonEQuest]

Since I have modified this controller to be tougher if I can just fix one thing that may be a better thing for me. I am attaching some photos of it. Do you which component the gate driver is? I guess it would be the last transistor in the middle since the 4 smaller ones on each side are the mosfets and the 2 big ones on each side are the barrier diodes. The only two "chip" components left are the 16pin DIP and the last transistor in the middle. I have not had time to run the numbers on them yet.

You can see the small wires to reconnect the traces that I had soldered. I would rather they burn in a know spot each time like a fuse until I get this resolved and make them full traces again. P.S. they are thin but the last test was with full traces and they were still burning up. I can try with no motor and see what the output is.

 

[fechter]

See below. There are two gate driver transistors. If you measure between all the legs with in diode check mode, look for shorts. Measure each pair of legs in both directions (reverse the meter probes). This makes 6 possible combinations. One pair of legs should measure like a diode. None should be shorted.

The big chip drives the transistors. That could be blown too.

What's the number on the chip? I can't read it in the picture.

Another thing to check is the voltage regulator. U12 big thing in the middle. Actually if you were over volting the controller, that would be a likely victim. One side is fed from the battery positive through a big resistor (the pair next to it). One pin is ground. The output should be 5 or 12V depending on the flavor. Check the number on that too. a 7812 is a 12V regulator. 7805 is a 5V.

 

[JonEQuest]

Very helpful! I will check those things, Thanks!!

 

[JonEQuest]

[Quote "fetchter"]See below. There are two gate driver transistors. If you measure between all the legs with in diode check mode, look for shorts. Measure each pair of legs in both directions (reverse the meter probes). This makes 6 possible combinations. One pair of legs should measure like a diode. None should be shorted.

The big chip drives the transistors. That could be blown too.

What's the number on the chip? I can't read it in the picture.

Another thing to check is the voltage regulator. U12 big thing in the middle. Actually if you were over volting the controller, that would be a likely victim. One side is fed from the battery positive through a big resistor (the pair next to it). One pin is ground. The output should be 5 or 12V depending on the flavor. Check the number on that too. a 7812 is a 12V regulator. 7805 is a 5V.

[/quote]
The big one in the middle is V12. It is a NP power transistor rated for 100v collector to emitter voltage and 140v collector base voltage. The two small ones are a S8550 and S8050. I'm trying to look at what they are rated at and test them. My testing has been at stock 36 or usually just 24 volts so voltage won't be the issue right now.

 

[fechter]

OK.

You might try applying power and measuring the voltage going to the throttle. That should give some clues.
Voltages on the two small transistors would tell us something about the gate driver supply.

One of the small ones switches the gates high, the other one switches the gates to ground. Those are probably pretty generic switching transistors. I'd guess one is NPN and the other is PNP.

 

[JonEQuest]

That is what I figured about the similar numbers. I actually stopped after realizing none of those components were dead shorted or open. They tested like any transistor device would 3-5kohms here, 10mohms there, basically no leg to any leg below 3-4 kohms or totally open. So I put my meter between negative and motor negative and read 0 ohms. So there was a dead short somewhere. I left the meter connected as I de-soldered each new component one by one and sure enough one of the new MOSFETS was bad. As soon as it was removed the impedance went up to something real high as it should. I replaced that one bad MOSFET, reinstalled the others I pulled and the 2 barrier diodes and then it was down to about 20kohms or so. SHORT GONE!!! Then I remade the burned board with de-soldering braid and extended wires from the throttle to the test bench and ran the controller on the bench with my test load of an old fog light and my cordless drill hardwired with the trigger on. I tested with 24volts and it worked perfect. The light slowly lit and drill started spinning with that annoying wine we all know and hate. I then put lots of heat sink paste (they used heat tape originally) and reinstalled the board into the heat sink body. Then hooked up the motor to the controller and sure enough, the controller works fine and the motor spins just like it used to. I was chasing my tail thinking the new components were fine and testing everything else. The only thing I can figure is that either one of the MOSFETs was defective or maybe I overheated it while soldering but I really think it was just defective and shorted the first time it was energized. Thank GOD I ordered 10 of these so I had 2 spares. I am going to put the scooter back together and test it at 60volts for a while. Then maybe go back to 72 volts. The motor actually looked fine on the inside so running double voltage has not seemed to hurt it too bad. It has monstrous torque (dangerous actually) so I may just weld a one way sprocket on the motor to run a chain for higher speed instead of the fixed direct drive gearbox that probably won't last too long at double speed anyway. I will post results of my test ride.