Max rated and phase current for 9,12,18 fet controllers

Hey guys,
I was wondering what peoples experiences are with the programmable infineon controllers in terms of how much rated and phase current you can get away with. I guess the theoretical limits of the components are a good place to start!

Obviously the boards need to have the traces physically built up enough to handle high current, but what will the fets allow ?
Anyone got any data or anecdotal evidence ?

I'm particuarly interested in what can be squeezed from the little 9 and 12 fet models - with either 4110s, 4310s or even the commonly used 75a NF75s ? I have to admit the technical intracicies of high side and low side fets escapes me, but for example with a 12 fet controller running 100 amp rated fets, could I assume it will take up to 200 amps ? If the rated current is normally 30 amps and the phase current is 70 amps, if you double the rated current to 60 amps is it going to handle a phase current of 140a ?

The reason I'm curious is I've been experimenting with different controller programming lately and I'm wondering what sort of phase current is safe to use without risking popping the fets. Ideally I want the most punch from the smallest controllers but with reliablility. 40 amps from a 9 fet and 60 amps from a 12 fet is my goal, which I think is reasonable...

The battery wires, shunt, and caps are the only parts that feel the battery current. (seldom an issue for reliability)

The FETs feel only phase current. (almost always the failure point in high performance setups)

Phase current can easily be 4x battery current during times when the controller is current limiting.

This might mean a given controller setup could handle 150battery amps with no issues, and another running a higher battery voltage might melt down from over-current through the FETs at only 100amps battery current, because it's phase current could be ~2x higher.


If we had a way to compare controllers based on phase current (anyone have any ideas?), then it would be a great tool to relate them.

Hyena said:

Hey guys,
I was wondering what peoples experiences are with the programmable infineon controllers in terms of how much rated and phase current you can get away with. I guess the theoretical limits of the components are a good place to start!

Click to expand...

Howdy brother!

In theory a stock (most are P75N75& FETs) in a 6FET configuration could handle somewhere about 80A primary current in burst or perhaps 75A Primary Current and the Maximum Phase Current (will be a multiple between 2 and 2.5x Primary Current for Direct Drive at least).

I've run them on 18S and (darn just remembered from another thread I was trying to remember when I ran a 9FET stock fets using 18S before) doubled the shunts in parallel programmed for 30A / 60A but pulling 60A and likely 120A worked fine and pulled like a raped ape in my 20" geared steel hub. I did first upgrade the caps to 100v, since I was going with Low ESR type... I went with a single 1000uf in place of the two 470uf and then stuck with a 470uf on the FET side - also 100v, I think I replaced one 100uf in the front with another 100uf but I could be confusing units here... This controller ran great on 15 - 18S but I goofed and tested it on 20S and poof... I think I used a pair of 800 ohm resistors in parallel (R1A/R1B with trace severed) to get a 400 ohm 4w power resistor... about 475 is needed for 20S if R6 is 680 ohm (imho).

I caution you - the application above was a geared motor with absolute WOT and a peak of 79A but only for 2-3 seconds then it would taper down to perhaps 15-20A, quite in the normal range (once at speed, even up hill it would only see perhaps 35-40A... I also beefed up the traces (power traces) with a solid 12G on top of what was already there. On a direct drive motor, I think you would see higher sustained currents and it may not be a great idea.

Hyena said:

Obviously the boards need to have the traces physically built up enough to handle high current, but what will the fets allow ?
Anyone got any data or anecdotal evidence ?

I'm particuarly interested in what can be squeezed from the little 9 and 12 fet models - with either 4110s, 4310s or even the commonly used 75a NF75s ? I have to admit the technical intracicies of high side and low side fets escapes me, but for example with a 12 fet controller running 100 amp rated fets, could I assume it will take up to 200 amps ? If the rated current is normally 30 amps and the phase current is 70 amps, if you double the rated current to 60 amps is it going to handle a phase current of 140a ?

Click to expand...

The 4110s an 4310s have posted datasheets... low rds(ON) and a very high continuous current drain rating. Assuming proper cooling, these could handle tons of current!!!

Someone please correct me but the FETs amplify the phase current by dropping the drain voltage correct? Assuming 100% efficiency it would be basically be to get a given current multiplication the subsequent voltage would have to reduce by the same factor (can't remember which law this is)??

If I am correct then looking at some of the drain burst current ratings of the 4110s and 4310 (300A?) then we could assume they would output that current but at a much lower than VCC (2.X factor lower voltage roughly). Also the output is Pulse Width Modulated so they are bursts not 100% on... which brings me to your previous question (sorry it's late)

To try and clarify the High and Low side FET banks... think in terms of a brushed motor... as it spins the brushes make contact (physical) and complete a circuit - this in turn energizes the proper coils and causes the motor to turn faster... On a solid state controller... the FETs act as the brushes being timed by the input pulses (leading or trailing I don't recall) of the Hall Sensors to determine what Phase needs to be fired and in what polarity... then the proper High and Low phase are powered energizing the coils (two at the same time on a BLDC controller I think) momentarally (really for a split second) then the next set is energized... the point of having High and Low is basically like having 6 SPST switches... 3 connected to +VCC or Positive and 3 Connected to negative... these thrown in the proper order will create the sinusodial commutation our BLDC motors require.

Hyena said:

The reason I'm curious is I've been experimenting with different controller programming lately and I'm wondering what sort of phase current is safe to use without risking popping the fets. Ideally I want the most punch from the smallest controllers but with reliablility. 40 amps from a 9 fet and 60 amps from a 12 fet is my goal, which I think is reasonable...

Click to expand...

What FETs are you running, how many and what type of motor - DD or Geared?

-Mike

liveforphysics said:

If we had a way to compare controllers based on phase current (anyone have any ideas?), then it would be a great tool to relate them.

Click to expand...

Yeah that's what I was kinda hoping for in this thread
Basing capability on the specs of X mosfet multiplied by the number in the controller.

Mike, thanks for that explanation!
I'm running direct drive hubs - GM/9C types.

On my Norco I'm running a 12 fet 4110 controller that I beefed right up and saw peaks of over 100 amps battery current. I backed it back off to run with peaks of 80 amps and 30-40 amps constantly (18S) and it was fine for a few rides then suddenly popped one day when I turned it on. (motor felt rough as I went to ride off like a phase or hall wire was loose, I check everything was ok, tried again and it didnt go at all)

I guess the next question is does a soldered / modified shunt affect both the phase and battery current ?
With my modified shunt I dialed the software back to 40 amps battery and 90 amps phase current. This read 80 amps battery current on my watt meter, so do I assume that was 180 amps of phase current ?
And if so, is 180 amps of phase current a killer to a 4110 12 fet controller ? Or was it a fluke of bad luck that it popped when turning on. I'm hoping for the later, and it'd be nice if it actually blew while pulling huge current at warp speed rather than just powering up the caps...

Ideally for my high(ish) powered rides I'd like a rated current of 50 amps with say block time of 5 seconds allowing bursts up to 80 amps. This would hopefully protect the motor somewhat from being cooked by noobs.

Just for statistics...

We already measured burst current of 251 Amps battery current on a high power modded 18 fets 4110 controller... calibrated FLUKE 337 clamp meter measurement and it was repetable constant results...

And measured a sustained current during acceleration ( 30 to 60 % of the powerband) of 160A (15kW)

But that was with massive copper flat bar heatsink link and two kapton layer insulation between msfet case and heat sink to maximize heat transfer.

Doc

liveforphysics said:

If we had a way to compare controllers based on phase current (anyone have any ideas?), then it would be a great tool to relate them.

Click to expand...

Perhaps something for swbluto too look into on his next marathon session in front of the pc...
Could it be done similar to the simulator he put together? Never seen it or used it so no real idea if its
possible just suggestion.

KiM

I'm afraid its going to be a bit more complicated than that. A lot will depend on what throttle setting the phase current is generated at.

For instance, suppose the phase current is 20 A, generated at WOT from a suitable battery voltage. Then the FETs are not in PWM mode, and the battery current is the phase current. The FETs are only switching to do the commutation.

But if that 20 A phase current is generated from twice the battery voltage, then the FETs are in PWM mode and the phase current into the motor is switching between flowing through the top FET and flowing through the body diode of the bottom FET. The phase current out is going through another bottom FET as normal. The top FET is getting half the RDSon dissipation but way more switching losses, and the bottom FET is getting a real thermal punishment.

Strapping Schottky diodes across the bottom FETs might help, but for some reason that doesn't seem to be done.

Nick

Tiberius said:

Strapping Schottky diodes across the bottom FETs might help, but for some reason that doesn't seem to be done.

Nick

Click to expand...

The controller BigMoose is helping me with uses monster schottky diodes to clamp freewheeling voltages. Keeping this flyback energy away from being clamped by the intrensic diodes in the FET bodies works out to keep quite a bit of heat out of the FET packages. Save that thermal overhead for Rds-on losses.

liveforphysics said:

Tiberius said:

Strapping Schottky diodes across the bottom FETs might help, but for some reason that doesn't seem to be done.

Nick

Click to expand...

The controller BigMoose is helping me with uses monster schottky diodes to clamp freewheeling voltages. Keeping this flyback energy away from being clamped by the intrensic diodes in the FET bodies works out to keep quite a bit of heat out of the FET packages. Save that thermal overhead for Rds-on losses.

Click to expand...

It ought also to be possible to turn the bottom FETs on at the right part of the cycle. It sounds risky, but the timing is known and its the sort of thing that is done on some DC-DC converters to reduce diode losses.

Nick

Tiberius said:

liveforphysics said:

Tiberius said:

Strapping Schottky diodes across the bottom FETs might help, but for some reason that doesn't seem to be done.

Nick

Click to expand...

The controller BigMoose is helping me with uses monster schottky diodes to clamp freewheeling voltages. Keeping this flyback energy away from being clamped by the intrensic diodes in the FET bodies works out to keep quite a bit of heat out of the FET packages. Save that thermal overhead for Rds-on losses.

Click to expand...

It ought also to be possible to turn the bottom FETs on at the right part of the cycle. It sounds risky, but the timing is known and its the sort of thing that is done on some DC-DC converters to reduce diode losses.

Nick

Click to expand...

That's synchronous rectification you guys are talking about, and using this switching scheme is much better than adding any type of adding additionnal diodes to the low side. However, you also loose the ability to coast when the throttle setting is lower than motor BEMF. This is because you now have a bi-directionnal energy transfer circuit when actively switching both bottom and top side FETs. So instead of coasting you would get throttle based regen when using synchronous rectification on the switching phase. To get the coasting possibility back while still using sync. rect., you will need to actively regulate current flow using some form of feedback loop. Your throttle could then just set the desired phase current (torque control), and the MCU would be the one adjusting the controller's output to aim for the desired current.

Hyena - It is important to use a very good and thin thermal transfer material between FETs and the heat spreader. This has a big effect on how much current your FETs can handle. Some controllers use than thick grey silicone stuff - it's absolutely terrible at keeping your FETs cool. Many have kapton which is good and can be very thin, although I think many controllers use kapton with adhesive which performs quite a bit worse than some good thin "kapton MT".

ZapPat said:

That's synchronous rectification you guys are talking about, and using this switching scheme is much better than adding any type of adding additionnal diodes to the low side. However, you also loose the ability to coast when the throttle setting is lower than motor BEMF. .....

Click to expand...

Good points, ZapPat, I'd forgotten about the automatic regen.

Here's another point. I was talking above in terms of the top FETs doing the PWM switching and the bottom FETs taking the circulating currents through their body diodes. It works just the same the other way round, with the bottom FETs doing the PWM and the top FETs taking the punishment.

In fact it might make more sense to do it that way, because driving the bottom FETs is easier, and switching them might be more efficient than switching the top ones.

Nick

Tiberius said:

Here's another point. I was talking above in terms of the top FETs doing the PWM switching and the bottom FETs taking the circulating currents through their body diodes. It works just the same the other way round, with the bottom FETs doing the PWM and the top FETs taking the punishment.

Click to expand...

True, and I guess your are talking about driving your FETs in boost mode (Instead of buck), which puts your controller in regen only mode. However if you invert the polarity of the second driven phase which is usually held low (hold it high instead), then I guess we would end up with regular drive mode once again. There might be some problems though, which brings us to your next point...

Tiberius said:

In fact it might make more sense to do it that way, because driving the bottom FETs is easier, and switching them might be more efficient than switching the top ones.

Click to expand...

The problem with inverting the polarities of both the PWM'd phase and the held phase as I think you are suggesting is that most FET drivers use charge pumps for high side FET driving. This means your bootstrap cap has to provide enough energy to maintain an acceptable voltage to keep the high side FET gate(s) charged during the whole commutation period. At very low RPMs a commutation period could be quite long, maybe up to half a second at startup speeds? There are ways to get around this, like adding a small independant charge pump to each high side FET supply, but it adds more complexity. I did this on an old prototype once using some low power 555-like timers (pics posted somewhere here on ES a long time ago), but reverted back to the regular FET driver based charge pump for the last few prototypes I've made since.

Pat

The voltage drop across freewheeling diodes is often the limiting factor in terms of heat generation. The FET body diodes aren't extremely good, their voltage drop is frequently around 1.3V near max rated current. If you're running at 100A phase current, that's 130W during freewheel! This is why low duty cycles are so much harder on a controller than full throttle.

If you do synchronous rectification, you get the usual I^2*R loss during freewheel. For a 4110 at 5 mohm, that would be more like 50W. You do get some switching loss, but that might be another 5W. 55W vs. 130W is a huge drop in heat. Adding external schottky diodes helps as well, a power schottky might have a drop of 0.6V or 0.7V, so 60-70W loss. The downside, besides an extra component, is that heat is all concentrated in one diode package. Using the internal FET diodes (or synch. rect.), the loss is usually spread over 2 or 3 FETs.

In most cases, the FETs will hit thermal limit before they reach the rated current, except in short bursts. Note on the datasheet that the rated current assumes a case temp of 25C, which is not going to happen in the real world.

rhitee05 said:

The voltage drop across freewheeling diodes is often the limiting factor in terms of heat generation. The FET body diodes aren't extremely good, their voltage drop is frequently around 1.3V near max rated current. If you're running at 100A phase current, that's 130W during freewheel! This is why low duty cycles are so much harder on a controller than full throttle.

If you do synchronous rectification, you get the usual I^2*R loss during freewheel. For a 4110 at 5 mohm, that would be more like 50W. You do get some switching loss, but that might be another 5W. 55W vs. 130W is a huge drop in heat. Adding external schottky diodes helps as well, a power schottky might have a drop of 0.6V or 0.7V, so 60-70W loss. The downside, besides an extra component, is that heat is all concentrated in one diode package. Using the internal FET diodes (or synch. rect.), the loss is usually spread over 2 or 3 FETs.

In most cases, the FETs will hit thermal limit before they reach the rated current, except in short bursts. Note on the datasheet that the rated current assumes a case temp of 25C, which is not going to happen in the real world.

Click to expand...

I might add that your example of the sync FET vs the diode assumes only one FET per leg (6 FET controller). A 12 FET controller would have half the loss yet (22W), and an 18 FET would only see a 18W loss, versus an almost constant intrinsic diode loss of 130W (a bit less for each paralleled device, but not much).

This also means that actively switching all FETs (synchronous rectification) is still 3-4 times better than using additional schottky diodes, not to mention less power components to add where layout is critical and already quite tricky. But of course it also implies custom firmware(/hardware) being made, since a regular ebike controller logic section doesn't do this. :| Happily though, many common MCUs are capable of doing this using special hardware on chip, including those used in ebike controllers (infinion, XC116,...). It's all a matter of different programming.

Talking about controller limits without discussing the motor it's to be used with is inappropriate. Those controller limits need to be tuned specifically for a motor, because the low turn count motors are much harder for controllers to drive due to increased current limiting.

John in CR said:

Talking about controller limits without discussing the motor it's to be used with is inappropriate. Those controller limits need to be tuned specifically for a motor, because the low turn count motors are much harder for controllers to drive due to increased current limiting.

Click to expand...

Its good that you said this. Im about to run a 5303 on a 12 fet controller at 72v 45A. I have the choice of either running a 4310 or the 4110. I know what the obvious choice is.

I noticed the senario you just suggested. I was running a 4310 mosfet with the new 9C 10x6 and it ran hot as hell. Im expecting the same senario with this 5303 on the 4310... I dont know how the 4110 would perform.

Are you talking about the motor or controller running hot as hell? On my 2 turn motors, all controllers have run hotter than I expected. Running a 6 turn 9c should be much easier load on a controller than a 3 turn X5.

ZapPat said:

I might add that your example of the sync FET vs the diode assumes only one FET per leg (6 FET controller). A 12 FET controller would have half the loss yet (22W), and an 18 FET would only see a 18W loss, versus an almost constant intrinsic diode loss of 130W (a bit less for each paralleled device, but not much).

Click to expand...

Entirely correct, thanks for adding that. Ignoring switching losses, the total I^2*R losses for 2 FETs in parallel would be half that of a single FET, and each device sees half of that, so the per-FET loss is 1/4 what it was before. Switching losses increase, but not as fast so it's a net gain (up to a point). AFAIK, the net loss won't decrease much using 2 FET body diodes in place of one, but the losses should at least divide about equally between them giving a per-FET reduction as well.

I don't know anything about how the Infineons are programmed, but in theory applying sychronous rectification is dead-simple. Whichever high-side FET is being PWM'ed, you drive it's low-side companion with the inverse PWM signal. The low one is on when the high is off, vice versa. Ideally one or the other would be on at all times, but in the real world some dead-time has to be allowed. This causes the body diode to start conducting, then the FET channel takes over once the low-side is turned on, then back to the body diode as the low-side turns off.

That would be a pretty low-level change, though, if the firmware isn't designed for it.

John in CR said:

Are you talking about the motor or controller running hot as hell? On my 2 turn motors, all controllers have run hotter than I expected. Running a 6 turn 9c should be much easier load on a controller than a 3 turn X5.

Click to expand...

It seemed like my 9x5 GM ran alot cooler than I expected than but when I put the 10x6 9C1606 on the controller and phase wires were pretty hot on the 9C. Both had stock harnesses terminated with andersons outside the axel. Ive ran the GM without any issues with the phase wires getting super warm but the 9C and my usual riding style would have defintely cause the harness or motor controller to fail first day out the box.

Looking at the power curves in the simulator explains alot i think. The GM wattage peaks around 15-20mph and falls from there when the 9C peaks around 30-35mph an falls. Since Im a high speed person then it makes since it would be running hot since it would hover in the mid 30s on speed. The 5303 power peak is even higher between 45mph to 50mph before it falls. Looks like I need a bigger controller.

Well, hasn't this been a can of worms!

I was rather wishfully hoping for a reply like "a stock 9 fet with 75a fets will take 70a, a 12 fet with 4110s will take 120a, an 18 fet with 4110s will take 200a" :lol:

As has been pointed out, there's too many variables to apply a blanket answer.

How about I narrow it down to a garden variety 9C 9x7 and the standard 9 fet and 12 fet 4110 controller.
What would be reasonable/conservative limits for these NOT to blow up during regular use ? (allowing for PWM extremes at part or low throttle and at variable loads)

Rather than start a new thread, I will drag up this old one to (hopefully ) get an answer to my question
5304 motor on 26 in MTB wheel
Lyen 18 FET 4110 65 amp controller

I previously cooked a mosfet upping the block time when running at 84 volts..so i want to avoid cooking anything again

Now running 24 series..at hot off charge of 99.3 volts

battery current set at 65 amps and phase current 2.5 times that at 162 amps

block time 0.5 seconds

8 or 10 gauge battery wires...10 gauge to the hub

I am sure i read 2.5 times somewhere but cant find it now...have scanned through the previous posts here and my eyes started to glaze over...so simply put for me

Are my current setting safe? for the controller ..not for my limbs maybe ...I can only assume they are as I have not cooked anything yet..but am not doing long period WOT or lots of stop start full current WOT acceleration

Can I safely up the limtis any more? or am I pushin g it as it is...wold liek to squeeze a bit more out of it if I can without risk of killing the controller again.

Looks very safe to me should be perfect for that setup.

Yes I know...old thread. But there is a lot of good info . IMO...unless you define a bunch of the variables like battery voltage, battery capacity, wire conductivity, type of FETs, wheel diameter, type of motor, weight of the bike and rider, etc., etc. It is very difficult to compare data.

I had the same basic question as the OP (Hyena) and I was not able to find a definitive answer so I bought an extra controller and started increasing the battery and phase amperage. Figured I'd keep going until something broke. Well I chickened out sort of...I actually reached the maximum battery amperage recommended by EM3ev for my BMS...50A Continuous, 55A Burst, with a little over 60A circuit breaker...not really a circuit breaker but that is how the BMS is programmed.

I am running a 12T MAC (geared hub motor) in 26" wheel with a 723mm OD tire. Total weight for me and the bike is 275 lbs, I am running a 14s5p battery with HG2 cells, and an Infineon 12 FET controller with IRFB3077 FETs. The controller is programmed for 50A Battery and 140A Phase current with a one second "Block" time (block time is the time the controller allows the programmed limits to be exceeded). When starting from a dead standstill, I have repeatedly wrapped the throttle wide open and I get 58A max on my Cycle Analyst. Initially had 60A Battery current programmed and I would flip the "Circuit Breaker/BMS" and have to reset it by turning my battery off and back on again.

All that to say my BMS is the limiting factor at this point and not my controller...no problems whatsoever with the controller. I used the Grin Tech Motor Simulator to estimate the max phase amperage with 50A Battery current and it was just under 140A so I plan to continue to run 50A Battery and 140A Phase settings with the "1" Block setting...since it is the max amperage I can run with my set up due to the BMS limits.

Hope this data helps someone .

This is a very interesting topic.
I wonder how voltage influences current capabilities as I am in the process on standardizing all my batteries to 52v/14s. Does this mean I can safely push more amps than I did on 72v as long as I keep the watts the same? Or is it a case of amps is amps no matter the voltage?

The short answer on increasing the amperage with a lower voltage is probably NOT.

The long answer is "It Depends".

The amperage squared times the resistance determines the amount of heat produced...assuming you have enough Voltage to push that much amperage through the circuit. So for overheating concerns, voltage does not enter the equation. There are other failure modes besides overheating so that is why the long answer is "It Depends".

I like your idea of standardizing on a 14s/52v battery. There are some rules/laws in parts of the world that limit the voltage to no more than 60v or all kinds of additional safety and handling requirements kick in...that is probably the main reason 52v is becoming so popular. Just for those that don't know, a 14s/52v battery when fully charged will read 58.8 volts.