blown 36-48V controller upgraded to 150V

[Antonio Zanardo]

Yes, I upgraded a 36-48V 20A controller in order to stand 115 Volts delivered by my Lythium battery pack.
The upgrading consisted in replacing 6 stock mosfets and three capacitors with six FDP2532 rated 150V 79A and the capacitors with three 100uF 200V.
The bike got an enormous gain in acceleration and speed. However the throttle became extremely sensitive. A very little fraction of a turn could produce a sudden and strong rev up, or, to the contrary, a current drop to near zero. It was very hard to keep a steady speed, as the motor changed continuously its revolutions.
I spoke to the past, as right now the controller is dead. There is a bad short inside. As per Richard suggestions, a test shows that a bank of mosfets blew, and possibly some more components blew with them.
The problem is I am not an electronic expert, therefore it's hard to me to fix the fault without any qualified help. However the excellent power increase I got for (alas!) few days urges me to try every possibility to fix the controller AND to overcome, if possible, the throttle problem.
As a matter of fact, the extreme difficulty to control speed motor could be the culprit of all this mess.
Any idea?

Antonio Zanardo.

P.S. when the short occurred the bms protected very efficently the lithium batteries which did not suffer in any way for the short.

 

[dirty_d]

maybe the voltage regulator for the controller cant handle that much voltage and is putting more than it is supposed to and messing everything up.

 

[xyster]

No idea exactly what happened. But if the main FETs fried, next time maybe try a lower current limit than you were running by adding or subtracting solder to the current-sensing shunt, or changing a resistor in that part of the circuit.

 

[Doctorbass]

I already repaired two controller from now, maybe I can help you.

First, you upgraded your 20A controller that use only 6 fets to 150V and I think this is bad. When you increase the voltage, the current increase too. The 150V fets are very similar to thestock IRF4710 of the 72V 35A controller about the current rating but it have 12 fets to be able to support the current of it. In your case you have 115V, but only 79A of fets per phase. The stock 72V/35A controller have around 160A of fets per phase.

Also, theFDP2532 have a higher mohm than the 4710 and will dissipate more heat.

About the regulator, you should know that the controller have a 14V rail to supply some low voltage chip. the difference between the input voltage like 115V to drop to 14V is directly dissipated in heat that's 101V !
in the stock 48V controller, it's 34V (48V - 14V).. so you have 3 times the voltage applied to the regulator! the heat is very high.

I know that the casing of the 48V 20A controller is smaller than a 72V 35A, so it may have less surface to dissipate the heat.


My opinion is that an upgrade to 150V on a controller that have only 6 fets is very hazardous. You have not enough of fets current to hold the power that is produced when supplying your hub motor at 115V. My suggestion would be to replace your controller to a 12 fets if you still want to keep a 150V upgrade. and yo add dissipating surface on the controller at that voltage to facilitate the regulator work.



Doc

 

[Antonio Zanardo]

Hi Doc,

I wish to point out that I tried the 150V modification for the following considerations:

1) My lithium pack delivers 115V 10Ah with elements connected in series.
2) Given the fact that I can't draw much current from the pack, the only way to get more power is to increase voltage, as W=V*A. In example, if you draw from the pack max 15A for security reasons, you will get 15Ax48V=720W from a stock 48V controller. With the modified controller you get 15A*115V=1725W. A big increase of power, but none of amps.
If what I said is right (please correct me if I am wrong), I am not sure the controller failure was caused by inadequate mosfets.
After all, the mosfets should have blown instantly while I made the first harsh tests at full power for several minutes.
Thus, excluding mosfets and capacitors which can manage well over 20A and 115V, I presume the failure was due to some other components which cannot tolerate the extra voltage.
You mentioned the low voltage regulator. As I said in my previous message, I faced since the beginning a serious throttle malfunction.
Maybe the 1-4V throttle voltage was not constant. Actually, I was unable to control finely the bike speed as the controller was behaving almost like an on-off device.
Was this the sign that the low voltage rail was already malfunctioning?
If so, can it cause the mosfets failure in the long run?
How could I check if this regulator is working properly?

Thank you for your appreciated comments.

Antonio Zanardo

 

[fechter]

At a higher voltage, the controller can develop more current on the motor side. With a 20A limit, you might be well over 100A on the motor side, so the FETs can fail from too much current.


If you measure the voltage on the 3 throttle wires, one will be zero, one should be a steady 5V (+/-), and the signal should vary from about 1v to 4v as you advance the throttle.

There are two levels of low voltage regulator. One is around 12-14v and the other is 5v. You can measure the 5v on the throttle. The 12-14v can be measured on the hall sensor wires to the motor. If the 12v regulator shorts, it will pretty much fry all the control circuitry.

 

[Doctorbass]

Hi Doc,

I wish to point out that I tried the 150V modification for the following considerations:

1) My lithium pack delivers 115V 10Ah with elements connected in series.
2) Given the fact that I can't draw much current from the pack, the only way to get more power is to increase voltage, as W=V*A. In example, if you draw from the pack max 15A for security reasons, you will get 15Ax48V=720W from a stock 48V controller. With the modified controller you get 15A*115V=1725W. A big increase of power, but none of amps.
If what I said is right (please correct me if I am wrong), I am not sure the controller failure was caused by inadequate mosfets.
After all, the mosfets should have blown instantly while I made the first harsh tests at full power for several minutes.
Thus, excluding mosfets and capacitors which can manage well over 20A and 115V, I presume the failure was due to some other components which cannot tolerate the extra voltage.
You mentioned the low voltage regulator. As I said in my previous message, I faced since the beginning a serious throttle malfunction.
Maybe the 1-4V throttle voltage was not constant. Actually, I was unable to control finely the bike speed as the controller was behaving almost like an on-off device.
Was this the sign that the low voltage rail was already malfunctioning?
If so, can it cause the mosfets failure in the long run?
How could I check if this regulator is working properly?

Thank you for your appreciated comments.

Antonio Zanardo

Any motor have an impedence du to the coil insode. the current depend on that impedense. supose you have a motor (like mine) that draw 35A at 33V. in this case, the mean impedense will be around one ohm 33V/35A (R=V/I). If we forget the heat effect on that impedence, it is very constant. when you increase the voltage, the current will increase too. I=V/R. The current limit of your motor will try to keep the max current at 20A by adjusting the pulse width. when too high current is wanted, the pulse width will decrease. Fechter, corect me if i'm wrong, but on the that 20A controller the only gain of adding more voltage will be possible in the case where using 48V if the current can't increase to reach 20A due to the impedense even if the pulse width would be saturated at the max width.. in other words, the impedense of the motor could be too high to permit enough current at 48V...

ex:
impedense of the motor(Z).. or R =4ohm
volt = 48V
in this case, I =12A
P would be 576W

case using 60V
I=15A
P=900W

case using 90V POSSIBLE WITH A 35a CONTROLLER
I=22.5
p=225w

Case using 90V with a 20A controller
I=20 (limited by the current protection)
P=1800W


Like i said, the current increase with the voltage increase.
the way that limit that is the current limit of the controller
the impedense of the motor need to be take in account.

Doc

 

[Antonio Zanardo]

Hi Fechter and Doc,

of course you are right both of you and I was wrong. In my reasoning over watts, I did not consider that the motor coil impedance cannot be changed, therefore by increasing voltage, current also increases as per Ohm law I=V/R, while R stay constant.

Sorry for the stupid mistake, and thanks to all of you for your explanations.

Antonio Zanardo

 

[fechter]

You might consider replacing all the FETs to 100v rated IRFB4110's or similar and reduce the pack voltage to around 100v.

Otherwise, if you stay with the higher voltage FETs, you should reduce the current limit a bit. I'm not sure what a good limit would be, but whatever you started out with was too high.

Fixing the throttle problem will help too.

 

[Antonio Zanardo]

The way this unhappy outcome happened is rather odd.

As soon as the controller transformation was finished, I tested the bike at full throttle for several minutes at top 60kph speed. For what Fechter and Doc said, I suppose the current draw was very very high on that occasion, therefore the fets should have failed at that time. But the fets did not.

Afterwards, I used the bike less severely for about ten days, and everything was ok except for the throttle malfunction.

One morning I was picking my bike to go to work as usual, but I was unable to get the motor running. I suppose the controller "died" either at the moment I put off the power switch in the previous evening, or during the night. I don't know.

How is that possible? What may have happened?

Now I changed the two damaged mosfets by following Fetcher's instructions.
However the resistance reading between power line and motor cable is not uniform, showing there is still a short in some point of the board which I am unable to locate.

Fetcher's suggestion to limit the current is a good one and could overcome the excess of current draw, but I should do the same job on a new 48V20A controller, as the old one is badly damaged. And, what if the throttle problems rise again? Or what if the low voltage regulator can't handle the hig voltage?

I am not sure if that hard modification work is worthwhile.

What would you suggest, considering on the other hand the modified bike performed like a real missile? Is it worth a new try?

Antonio Zanardo

 

[fechter]

The voltage regulator has a 400v rated transistor, so I wouldn't worry about that. The resistance readings on the phase wires are normally unsteady, but as long as none of them are near zero (over 1k) it should be good. You may have blown the gate driver transistors, the two TO-92 devices on the high side driver. You can check them for shorts in circuit.

The throttle problem is interesting, as I haven't heard quite that behavior before. You could test the throttle with the motor disconnected. Just measure the signal wire while advancing the throttle.

 

[Antonio Zanardo]

After various attempts, I was able to fix the dead 150V controller. During a thorough check with the aid of a microscope, I found an apparent uncertain soldering of a mosfet leg. In my opinion the initial failure was probably due to this uncertain joint.
The same bad joint could have caused also the throttle problems.

Now the bike is running regularly with the fixed controller at a 57.5
voltage. No throttle problems at present.
However I plan to raise voltage step by step, first to 86 V, then again to
full 115V while watching what happens during the voltage increase. I will
also install the current regulator described by Fechter, just to be able to prevent overcurrent.

I wish to thank all of you who helped me with your valuable advices.

Antonio Zanardo

 

[fechter]

Good report. Glad to hear you're running again.
We'll be interested to see how it works at higher voltage. I don't think many people have tried going that high.

 

[Antonio Zanardo]

As I said, I tested today the bike with 57V supply and in a short run.

However I know already how the bike performs at 115V: 60kph and a very strong acceleration, although my weight is 92 Kilos. I remind that I am using a Crystalyte 408 motor and a 48V-20A controller, upgraded to stand 150V. The wheel is 26".

I don't know how long the modified controller will last. I hope the previous malfunction depended on a bad connection, wich right now is apparently corrected. Its lifespan will depend also on how well I'll be able to control high current.
I'll keep you informed on evolutions.

One thing let me ask you experts: is there any risk to demagnetize the motor by supplying high voltage and amps?

Antonio Zanardo

 

[Doctorbass]

One thing let me ask you experts: is there any risk to demagnetize the motor by supplying high voltage and amps?

Antonio Zanardo

Simply impossible with rare hearth magnet like these!

Don't border about that :wink:

Doc

 

[xyster]

One thing let me ask you experts: is there any risk to demagnetize the motor by supplying high voltage and amps?

Antonio Zanardo

Simply impossible with rare hearth magnet like these!

Don't border about that :wink:

Doc

Jozzer fried his 408 at 84 volts 35 amps.
http://endless-sphere.com/forums/viewtopic.php?t=1618&highlight=
Somebody else on the PA forum fried the windings of their 408 at the same power.
It sounds like 115V 20Amps is really pushing this motor's power and heat dissipation limits.

 

[Jozzer]

It is indeed, they will take it for a short while (few mins), but they heat up quick, especially at lower RPM's, and as they heat they lose efficiency and heat faster. That said, I did use it like this for some time before the unhappy day with no problems (like 1500 miles or so), and I was being very silly when I killed it.
The controller on the other hand doesn't seem particularly stressed. remarkable bit of kit!

 

[solarbbq2003]

anyone using these, they have 6 lots 4310fets, I told kenny that they forgot to put another 6 fets then stephen pointed out 4310 can replace two old fets?? whoops
seems there is still places for puting another 6 fets might make a good reliable controller adding another 6 lots of 4310's, not sure if drivers are wired up to empty fet holes will have to take a closer look.

 

[Antonio Zanardo]

Jozzer fried his 408 at 84 volts 35 amps.
http://endless-sphere.com/forums/viewtopic.php?t=1618&highlight=
Somebody else on the PA forum fried the windings of their 408 at the same power.
It sounds like 115V 20Amps is really pushing this motor's power and heat dissipation limits.

It's clear that the 408 motor limits have been unwillingly exceeded.
The real question is: which are its real limits? I do not know.
But aren't we all striving to get the best performances out of our bikes?
If the answer is "yes", then we have to take some risks.
I am calculating that in my case 115Vx20A=2300W, which is about 20%less power than Jozzer's (84V*35A=2940W).
Is that enough to not destroy the motor? I hope so. Of course it depends much on the way the bike will be used. I am sure that at a continuous full power usage, the motor will not survive a long time.
But if I'll use the extra power just in emergency, or for very brief test runs, then probably the motor will have a longer lifespan.

Anyway I am convinced that, in the absence of sure experimental data, "one test is worth a thousand opinions", as Fechter says.

Antonio Zanardol

 

[The7]

I am calculating that in my case 115Vx20A=2300W, which is about 20%less power than Jozzer's (84V*35A=2940W).

Antonio Zanardol

Just my speculation if the motor could stand such V and A:

1) 115V 20A will give a higher top speed. Could be 25% faster because it does not require max torgue at top speed. Asssme 84V 35A tops at 50 mph. Then 115 V20A will tops at 62 mph.

2) 84V35A will have a higher torgue for the initial range (say 0- 40 mph). Could be 20% higher. The max power is limited by the ampere rating of the controller.

3) From 40 mph onward, the motor will not be limtied by the ampere rating. Then 115V 25A will have a higher torque and reach a higher top speed.

 

[Jozzer]

It was 30A...so you are not far off! It is plenty of power to burn it...unless you are carefull. It is also possible that the insulation on the copper will leak at more than 100v, I dont think anyone has tried at those kind of voltages. Why do you want to run such high voltage though? Even a 4011 will do over 35MPH at 100v, a 408 at full throttle no load is over 50mph even at 84v. You will spend most of your time accellerating, at inneficient revs, unduly heating the motor IMHO, and never actually see anywhere near the top speed.(much past 35 is pushing it with only 2KWish of power)
In fact, looking at the simulator (http://www.ebikes.ca/simulator/) , at 15mph, running at just 100v, efficiency at 15mph is only 50%...meaning half your power will be going towards turning the motor into a large radiator
How about dropping your voltage to match your desired top speed, and upping the amps to match your desired power usage/output?

 

[xyster]

In fact, looking at the simulator (http://www.ebikes.ca/simulator/) , at 15mph, running at just 100v, efficiency at 15mph is only 50%...meaning half your power will be going towards turning the motor into a large radiator

I think this reasoning is in error, and that I've made the same error in some of my charts and posts.

At full throttle, efficiency briefly passes through that zone of inefficiency as the ebike accelerates, or gets stuck in that zone when climbing a hill full throttle, if the incline prevents the motor speed from rising out of the zone.

Cruising at partial throttle with 100 volts, efficiency ranges from 70% at 15% throttle, to 90% at 100% throttle. The sim also shows that increasing the battery pack voltage increases partial-throttle efficiency.

 

[Jozzer]

So you dont think having max efficiency at a speed that the bike will never see will cause it to run more of the time at lower efficiency? theres a big difference between 80% and 90% in terms of heat generation...especially when your motor can only take a certain amount of waste heat. (my guess is around 350w or so continuosly....meaning 3.5KW at 90% eff, or only 750w at 50% eff!)

 

[Jozzer]

BTW, those graphs only help your point if you accellerate using partial throttle, opening it slowly as you accellerate, or cruise at a lower speed. Doing this, using ONLY partial throttle, the 408 at high volts is very efficient. But as soon as you have to accellerate/stop often, or face a headwind, you can load the motor with 20A and still be stuck at 15-20 MPH (in my experience)....a few mins too many and thermal runaway gets ya!

 

[xyster]

So you dont think having max efficiency at a speed that the bike will never see will cause it to run more of the time at lower efficiency?

The battery pack voltage is chopped by the PWM controller such that at X% throttle (25%...50%...75%), the motor sees X% of the pack's voltage. At 1/4 throttle, his motor sees 29 volts, even though his pack is 115 volts. Higher pack voltages tend to be more efficient because the battery and battery side of the controller requires proportionately fewer amps for the same power output. That's why big EV's, like electric cars, use a much higher voltage then they require (like 300+ volts, often). The hub sim also shows a little better efficiency at higher voltages.