Information on BLDC controller

[Shenta]

Hello All,

I have been reading your post, and some information have been very helpful, so I would like to point out some do and don't as a little contribution for this forum.

1. Mosfet- 4110 vs 4310 -
I have done some testing with both Mosfet, 4310 is used in some higher power controller from China (eg:1500W, 3000W...etc) for E-Bike usage, but it has greater resistance, so the heat from 4310 is higher than 4110, there for 4110 is more recommended. During high load (eg 40A ~ 50A) 4310 tend to have more problem with thermal run away, when it is "DAMAGED", it will first start loosing it's switching characteristic little by little, ( if you measure Rgs, the resistance will not drop slowly as it should be), than after a little bit while, the board will be fried alone with the transistor and diode or Mosfet driver (IRS2110 or IRS2010...etc). Another note: the price for 4310 is around USD0.75/pcs sold in China(local shop price), the fake one is around USD0.60/pcs, 4110 is cheaper than 4310... ^__^

2. Conversion form 36 or 48V to 72V+ -
There are 2 type of Mosfet driver within the controller from China, Transistor drive and IR(2110 or so) drive. For Transistor drive, the power rating is usually 40V only (to save cost), so if you hook up 72V battery, when load comes into the circuit, the whole board will be fried. Some manufacture use 80V transistor, but again, 48V lead acid when fully charged, it is measured app 55~56V, and it will be 84V+ when you hook 6 of the 12V lead acid batteries together. Problem will come when the load is high too. As for IR drive, the problem is little bit less, because by datasheet, it can sustain 100V, but however, if you use scope to see the waveform, you will be surprised how much noise you get out of it. Despite the noise, during switching, there will be a spike as high as 100V+, usually manufacture will try to use fast diode or suppressor to limit it, but again, with this kind of high voltage and speed, not many diode can be fast enough and strong enough to withstand. *the above is measured from batter to float ground or high side driver to Mosfet.

3. Modification of Soft Start -
Hub motor have tendency to draw large amount of current during start, and as for controller from China, when push throttle to max, their PWM is switching at 100%, and you can measure it from scope that with 48V, under no load, the Mosfet is switching at app. 350hz. By SOA (safty Operating Area)of 4310 or 4110, minimum switching frequency with 50V+ starts at 10Khz, so when during start, if you are completely stop, and you reduce the soft start to very little, you are sending 30A directly with no switching into Mosfet, guess what will happen?

4. Regen Break-
the regen breaking is completely total con, please be noted the max speed of BLDC motor is the max voltage we supply into it, we can not get the wheel to move faster is due to the BEMF force is as strong as the batter power we can supply, by this logic, unless in the controller there are some sort of Boost converter, than you will be only trying to charge 48V battery with less voltage supply.

I have been working on the BLDC controller for a little bit while now, things was quite confusing for me in the beginning, after blow up 500pcs+ of Mosfet and 100pcs+ of IGBT, finally started to get some down to earth knowledge. Now I am final testing 5000W BLDC controller, which use 4110 Mosfet, with Sinusoidal drive, the torque is app 35% more than conventional BLDC driver from China with same rating. And with much better thermal protection. But since it is commercial product, I could not provide any information on this, however, if there are any question when you modified your board, I am more than happy to help out.


Cheers
Shenta

 

[wrobinson0413]

.

 

[ZapPat]

Welcome to ES, Shenta.

1. Mosfet- 4110 vs 4310 -
Another note: the price for 4310 is around USD0.75/pcs sold in China(local shop price), the fake one is around USD0.60/pcs, 4110 is cheaper than 4310... ^__^

Another reason why products from china are so cheap I guess... They are two to three times more expensive here.

4. Regen Break-
the regen breaking is completely total con, please be noted the max speed of BLDC motor is the max voltage we supply into it, we can not get the wheel to move faster is due to the BEMF force is as strong as the batter power we can supply, by this logic, unless in the controller there are some sort of Boost converter, than you will be only trying to charge 48V battery with less voltage supply.

You are going to have to read some more theory about DC-DC converter circuits, because you have obviously missed some basic but very important principals about switching power conversion. Read up about synchronous rectifiers and you should find some info about how regen works by effectively raising the motor's voltage above the batteries so that current flows backwards.

I have been working on the BLDC controller for a little bit while now, things will quite confuse for me from the beginning, after blow up 500pcs+ of Mosfet and 100pcs+ of IGBT, finally start to getting some down to earth knowledge.

More than 500 FETs!! Good thing you get them for under a dollar a piece. :wink: I guess some people prefer to get into the hardcore hardware testing stage of their controllers right off the bat!

 

[Shenta]

wrobinson0413:

TKS for your welcome, I am base in Thailand, but as electronic parts are difficult to find here in Thailand, so I source almost all parts from China. Regarding on Fake IR Mosfet, YES, there are many fake one, and sometimes difficult to tell which one is real or fake. So you kind of have to know the people you are buying from, if they are honestly enough to tell you / teach you which one is real and fake, than you will know you should buy from them or not. I have noted in my post the price for real one and fake Mosfet, as for IR 4110, you have to look for those which is "Made in Morocco", for those MOSFET 4310 and 4110 made in Mexico is cheaper but less quality.

Cheers
Shenta

 

[billvon]

4. Regen Break-
the regen breaking is completely total con . . .

I (and several other people) have designed regen controllers that work, so I'd suggest it's not a total con.

please be noted the max speed of BLDC motor is the max voltage we supply into it, we can not get the wheel to move faster is due to the BEMF force is as strong as the batter power we can supply

Actually, there's a technique called phase advance that lets you run the motor faster than its base speed (i.e. faster than it will spin when its back EMF is equal to the battery voltage.)

unless in the controller there are some sort of Boost converter, than you will be only trying to charge 48V battery with less voltage supply.

That's correct. As a hint, notice that a boost converter has an inductor, a switch, a diode and a storage capacitor. Notice also that you have all these pieces available in your controller already! (The motor is, in some ways, a big inductor.) From there it's just programming.

 

[Shenta]

ZapPat,

TKS for your welcome too, the Mosfet are cheaper from China with another reason: Quantity. China government have made a new law forbids gasoline or LPG motorcycle to run in their major city, so the electric motorcycle are the only choice. There are millions of e-bike in china, and 4110 / 4310 are widely used.

Regarding on Regen breaking, please correct me if I am wrong, motor = generator, so for the power we could put into motor, what we put in - loss = we could get out of it. With less voltage coming out of the motor, even after the free wheel diode in the MOSFET, you still need to have a DC-DC Boost converter to make it higher voltage right? So we can charge the power back into the battery. Most of EVcar manufacture use low voltage power pack and Boost when driving their motor, so they can get regen back directly into the battery. Another note is when regen (on a e-bike, app 80Kg without passenger), you need to stop from 50km/h down to 0 within less than 5 sec, the power sending back to battery is at Kilowatt range, instead of charging the SLA battery, it will be dead before you know it.

Things never really go as we all planned, in the beginning I have tried to do the layout follow IR's application note with their 2110s mosfet driver, everything follows exactly as IR suggested, but Mosfet exploited again and again on IR4310 and IR250N with 48V battery and 30A current limit. After almost 1 month of waiting, they reply me it was a dv/dt failure and suggest me to use better rating Mosfet, and they no reply what so ever on the high voltage spike. Here goes app 300 Mosfet... Ouch

Cheers
Shenta

 

[Shenta]

Billvon

Regarding on advance phasing, another word for it is weak field control. Have you ever dis-assemble your motor? If you do, you will find out that your motor will actually run faster than before.

The trade off for weak field control is torque, when you lose torque, under load, the speed will actually drop. Proper weak field control will MAYBE give you very very very little bit more speed, but the real advantage is it actually can reduce upto 8~12% of current needed during at motor's max speed.

Cheers
Shenta

 

[Doctorbass]

ZapPat,

TKS for your welcome too, the Mosfet are cheaper from China with another reason: Quantity. China government have made a new law forbids gasoline or LPG motorcycle to run in their major city, so the electric motorcycle are the only choice. There are millions of e-bike in china, and 4110 / 4310 are widely used.

Regarding on Regen breaking, please correct me if I am wrong, motor = generator, so for the power we could put into motor, what we put in - loss = we could get out of it. With less voltage coming out of the motor, even after the free wheel diode in the MOSFET, you still need to have a DC-DC Boost converter to make it higher voltage right? So we can charge the power back into the battery. Most of EVcar manufacture use low voltage power pack and Boost when driving their motor, so they can get regen back directly into the battery. Another note is when regen (on a e-bike, app 80Kg without passenger), you need to stop from 50km/h down to 0 within less than 5 sec, the power sending back to battery is at Kilowatt range, instead of charging the SLA battery, it will be dead before you know it.

Things never really go as we all planned, in the beginning I have tried to do the layout follow IR's application note with their 2110s mosfet driver, everything follows exactly as IR suggested, but Mosfet exploited again and again on IR4310 and IR250N with 48V battery and 30A current limit. After almost 1 month of waiting, they reply me it was a dv/dt failure and suggest me to use better rating Mosfet, and they no reply what so ever on the high voltage spike. Here goes app 300 Mosfet... Ouch

Cheers
Shenta

Hello and welcome Shenta,

I agree with ZapPat for the regen that is possible even if your speed is lower and the corresponding backEMF voltage is lower than the battery voltage...

Otherwise How would you explain that regen BRAKING exist... this is a boost converter.

the principle is very simple... during regen, for a little part of the PWM, the motor winding is shorted by the mosfet... that energize the motor winding.. then this energy stored into it is released and this spike is alot higher than the battery voltahe and the diodes of the mosfet conduct and current flow back to the battery.

the same principle as a dynamo that you short the output while it is runing..... what happen...? the coil will generate a large spike with voltage higher than the Back EMF...

Or if you short a coil with a 9V battery... you will see that when you brake the contact, there will be a large spike that is higher voltage than 9V and that can give you a shock..... this is the same principle with braking regen...

There is alot of great thread that explain that here.... and also alot of people like me that run 12 x 4110 at 100.8V or higher.. i've seen 112V) with 70A+ without any problem since one years...


Doc

 

[PhoenixOSU]

Shenta, may I ask what FET driver you have been using? I think the LTC4444 does the best job I've ever seen of driving high voltage FETs quickly and reliably with short through protection (4446 is the no short thru protection version). Plus LT has the best **** documentation out there for their parts. Very often people blow hundreds of FETs before realizing some sort of hidden obvious misunderstanding about FETs. Mine was to forget that VGS applied when turning off FETs also, and I tried to shut off an N-channel FET by grounding its gate, when the source was at 24V (20VGS max is the most common), and pop! So there you have it, just stick with LTC parts if you are at all uncomfortable with the technology.

Shaun

 

[liveforphysics]

The price on a lot of 50,000pcs factory direct for IRF4110 from IR's Mexico factory was over $1 per FET.

If you have a source for <$0.75, Im going to take a wild guess that you are buying fakes, and that could explain mass failure issues.

 

[Shenta]

Shaun,

I am using transistor to drive the mosfet, after blow up more than another 500pcs, now we finally got it working as the way we hope it would, Mosfet drive can be expensive and not useful (egIR2110), very often they try to use charge pump to produce a very short high voltage (200V+) to turn on Mosfet quickly, but it also killed the mosfet in the same time. Using transistor can achieve 450ns raise time and 560ns of fall time on the driver side, 180ns and 220ns on the mosfet, which is good enough for me.

I bought my mosfet from IR authorized distributor from China, and now our controller can do upto 84V (90V when battery fully charged) 120A of output current with 12 Fet config.

Cheers
Shenta