18 Mosfet Infineon Controller

[Doctorbass]

So, if a 6 x 4110 FET controller can handle 50A OK, then a 6 x 3077 FET controller should handle around 68A with the same temperature rise as the 50A 4110 FET version. Scaling this up to an 18 x 3077 FET controller would give a max current capability of around 204A. Pretty impressive, but the real world limit might be how well the controller can drive the gate charge. Still, as these Infineon controllers use discrete component drivers, it might well be possible to do some simple changes to increase the peak gate drive current.

Jeremy

Jeremy, maybe we could work together to discover if the discrete component driver can be ameliorated!

I plan to rebuild the busbar of my first 18 fet controller with thick cooper sheet. I dont like the 18 gauge style wire soldered on the trace to cary... 65A...

I hesitate between cooper sheet and 8 AWG car ausio bulti braid wire... i'll see wich one is the easyest to work with..

Doc

 

[steveo]

So, if a 6 x 4110 FET controller can handle 50A OK, then a 6 x 3077 FET controller should handle around 68A with the same temperature rise as the 50A 4110 FET version. Scaling this up to an 18 x 3077 FET controller would give a max current capability of around 204A. Pretty impressive, but the real world limit might be how well the controller can drive the gate charge. Still, as these Infineon controllers use discrete component drivers, it might well be possible to do some simple changes to increase the peak gate drive current.

Jeremy

Jeremy, maybe we could work together to discover if the discrete component driver can be ameliorated!

I plan to rebuild the busbar of my first 18 fet controller with thick cooper sheet. I dont like the 18 gauge style wire soldered on the trace to cary... 65A...

I hesitate between cooper sheet and 8 AWG car ausio bulti braid wire... i'll see wich one is the easyest to work with..

Doc

Hey Doc,

Is that just a bare board? .. no parts on it? .. diy ?

Go with the 8AWG .. size wise its better... will solder quite easily also ..

When putting this wire accross the bus bar i think its called.. what is the advantages again i think you spoke about it somewhere else ...

p.s.

ran the 4115 with my limited 32 amp shunt on low batteries .. pulling easily in the 70-75km/h range .. It is halarious keeping up with cars .. i rode in a coffee shop parking lot with my dirt bike helment .. and i heard people gigleing .... then i gave it some throttle...heheh

Oh and garys bms board is working perfectly with my controller .. cutting the power when the a123 are drained..

still waiting on my parts to fix one of my broken china psu's to complete my 150v charger.

-steveo

 

[Jeremy Harris]

Doc,

I like the idea of soldering solid copper bar to the board. I've been looking at getting some 5mm square copper bar to try (if I manage to get a couple of bare controllers to play with).

I want to be able to drive a tandem array of HXT 80-100 motors (mounted on a common shaft, synchronised and fitted with Hall sensors) with one of these controllers, so will need around 12kW or so of drive capability at around 60V or so.

I'm more than happy to try reverse engineering the FET driver to see how it can be beefed up, or even see if it needs to be beefed up, but first I need to get my hands on a board to play with. I have a 6 FET controller coming from e-crazyman, so I can make a start on looking at how it drives the FETS and go from there.

Jeremy

 

[georgeycc]

hi
hopefully everyone is still on this thread.
i have something to discuss about the mosfet rating.

when people mention that a controller could handle 50 amps ok, do they really mean battery current or armature current?
the current is different on these two except at full throttle where there's no PWM modulation! the motor inductance and
the controller switching circuit is actually a buck DCDC converter that amps up the current. so the armature current could well be
100+A at low speed while the battery current stays at 10A.

as for the mosfet rating, the datasheet says IRF4110 could handle 120A, at Rdson = 4mOhm, so that is 120x120x0.004=57.6W to the package.
if you have exotic cooling, then 18-fet controller would give you 360A peak armature current. i dont understand why are people down grading
the current rating so much to a very cautious 100-A? as Jeremy mentioned, 25W per TO220 package should be reasonable, so each 4110 should be
able to handle (25/0.004)^1/2 = 80A, then 18-fet should still give 240A armature current no problem? the 18-mosfet controller is, to me, simply
too overkilling. am i missing something here?

====update====
heatsink datasheet for TO220, passive cooling.
http://www.angliac.com/NewBookPDF/Heatsinks/273_277.pdf
can reasonably assume temp coef to be 5degC/W, altho the juction temp can be as high as 150deg C, 100 deg C sounds safer. that would translate to
15W per TO220 device, so the current limit might be (15/0.004)^1/2 = 61A

 

[Doctorbass]

hi
hopefully everyone is still on this thread.
i have something to discuss about the mosfet rating.

when people mention that a controller could handle 50 amps ok, do they really mean battery current or armature current?
the current is different on these two except at full throttle where there's no PWM modulation! the motor inductance and
the controller switching circuit is actually a buck DCDC converter that amps up the current. so the armature current could well be
100+A at low speed while the battery current stays at 10A.

as for the mosfet rating, the datasheet says IRF4110 could handle 120A, at Rdson = 4mOhm, so that is 120x120x0.004=57.6W to the package.
if you have exotic cooling, then 18-fet controller would give you 360A peak armature current. i dont understand why are people down grading
the current rating so much to a very cautious 100-A? as Jeremy mentioned, 25W per TO220 package should be reasonable, so each 4110 should be
able to handle (25/0.004)^1/2 = 80A, then 18-fet should still give 240A armature current no problem? the 18-mosfet controller is, to me, simply
too overkilling. am i missing something here?

====update====
heatsink datasheet for TO220, passive cooling.
http://www.angliac.com/NewBookPDF/Heatsinks/273_277.pdf
can reasonably assume temp coef to be 5degC/W, altho the juction temp can be as high as 150deg C, 100 deg C sounds safer. that would translate to
15W per TO220 device, so the current limit might be (15/0.004)^1/2 = 61A

it depend on the use of these controlelr!

The calculations are made in a situation where the controlelr is continuously used... but these max power calculation make sense only with burst high power use!

Maybe there is just my drag ebike setup that can continuously take that max power calculatoin... when the back emf increase..to a given point.. around 60km, and that the current start to reduce eventhou i am WOT... i'll switch the winding into delta and the kV of the motor will change and make the motor to draw more current for continuing to accelerate. .. and that should last.. for 20sec...

I also decided to replac ethe actual aluminium bar for a full copper bar with 28% more surface and volume than the original aluminium bar.

I just wonder if when i'll start the 20sec race with that bar to let say 25oC... the temp increase willstill be far to the point where the temp stabilise (equilibrium point) where the the heat produced is the same sinked to the exterior

In other words.. starting with a cooper bar that is at ambiant temp for let say 20sec.. does it will reach the max temp limit at 15kW?

Doc

 

[liveforphysics]

Jeremy H- I just had my 72v E-crazyman controllers arive today. I haven't even opened the box yet, but I will take explict and perfect macro photos of the top and bottom of the board for you, with labels of the part numbers and any components that aren't clearly visible.

I VERY much appreciate that you volenteer your skill and time to review the controller circuits for us, and let us know the limits. My controler is a 12fet 72v setup, and I would like to configure it to drive 12 of the best possible T-247 package FETs with it. I would like each controller to be able to supply 250amp bursts at 72v to each motor on my E-bike. I eventually will need help getting something setup to run a 50Kw motor for my bicycle. Yes, I do have the battery for it

 

[BatteryMooch]

Hi guys, I just found this thread. Thought the other one was the only 18FET thread.

A few things to consider....
- The Rds-on of a MOSFET can easily double as it gets hotter, often 2.5 times the 25C temperature value.

- The heat generated is also a result of the time the MOSFET spends in its linear region (as it is turned on/off), where it generates a LOT of heat. All this depends on the driver and how fast it can turn the MOSFET(s) on and off. Once fully on, then you can look at the Rds-on and current specs to determine the power that needs to be dissipated.

- That MOSFET mounting/heat spreader bar's thermal resistance will easily exceed 30C/W. It will only be able to dissipate a few watts of heat to the ambient air and the thermal equilibrium point will be WAY above the allowable max temperature for the MOSFETs. Even with fan cooling. I wouldn't trust it do do any cooling, just spreading of the heat to allow you the longest running time (or highest duty cycle) before the MOSFETs overheat. Your biggest bottleneck might be the removal of heat in this design, not the MOSFETs, etc.

- If measuring the temperature of the MOSFETs mounted to that bar, drill a small hole through the bar and measure directly against the middle of the back metal mounting plate of the MOSFET using a thermocouple with some thermal transfer compound on it. Measuring the temperature anywhere else will give you a very, very low (false) number.

- The temperature of the inner MOSFETs will be higher than the outer ones.

- I recommend running the MOSFETs at 80% or less of their max junction temperature rating for good reliability, about 140C max for the IRFB4110. So, after taking the case temp, calculate the junction temperature with the following equation: Junction Temp = (Case Temp) + ((Power) x 0.61). The 0.61 is the junction-to-case thermal resistance for the IRFB4110and multiplying it by the power being dissipated by that one FET gives you the temperature rise above the case temp that the junction is operating at.

 

[georgeycc]

- The heat generated is also a result of the time the MOSFET spends in its linear region (as it is turned on/off), where it generates a LOT of heat. All this depends on the driver and how fast it can turn the MOSFET(s) on and off. Once fully on, then you can look at the Rds-on and current specs to determine the power that needs to be dissipated.

hi Camlight,
thanks for your very insightful comment
however you could also factor in the fact that each fet could only turn on for 1/6 of the total time. the worst case would be stall during when one fet is in continuous-on state. i can't agree with you more on this topic that thermal removal would be the real bottleneck rather than the mosfet itself.

 

[BatteryMooch]

Good points!
I think actual temperature measurement is about the only way to know how well the 18 FETs are doing.

 

[Jeremy Harris]

Remember that the spreader bar is bolted tight to the case all along it's length, and that the case is fairly well finned and made from a reasonably thick extrusion.

I did a few quick (and rough) sums and came up with a junction to air thermal resistance in the 6 FET controller of around 6 deg C per watt, of which the case was around 3.5 deg C per watt (the other 2.5 deg C per watt was the junction to case, case to thermal spreader bar, thermal spreader bar to case, losses). Assuming that the 18 FET case is three times the length, but the same width and height, then I'd guess that it would be somewhere around 1.2 - 1.5 deg C per watt, or a total from junction to air of around 3.7 to 4 deg C per watt.

I agree that it would be a very good idea to measure the FET case temperatures, this would give a great real-world insight into what's happening and how accurate some of our guesses are. I also agree about the switching losses, but as the switching speeds in these things are modest, I'm not sure how high they will be. The ratio of switching time to on/off time is pretty large; I think typical PWM frequencies (the highest switching frequency) are around 12 to 15kHz (assuming that this is similar to an RC controller). My guess is that we may be able to beef up the drivers to get faster turn on/off times, which will reduce the switching losses. It may be that the existing drivers have plenty of spare current drive though, and can adequately deal with FETs with a greater gate charge requirement without slowing down.

If someone really wanted to push things, then it would be easy enough to discard the case and the alloy spreader bar and just fit a hefty, long heatsink directly to the FETs. Looking around at what's available, it should be possible to get the overall thermal resistance down to around 3 deg C per watt from junction to air by doing this, which would make a massive improvement. Adding a fan would perhaps get this down to around 2 deg C per watt, which would allow a practical device dissipation of over 50 watts (allowing for temperature derating), which should be enough for most purposes.

One thing worth noting is that, although the IRFB4110 is rated for 120A, the TO220 package it's in is only rated for 75A. This limits max sustained current, as the bond wires will fuse if it's run at 120A for more than short bursts. The lower voltage IRFB3077 has enhanced bond wires, so can run at 120A continuously. It also has a lower Rdson of 3.3mOhm max. The snag is that it's only rated at 75V, but it's definitely a better choice than the 4110 for those of us who don't need to go over about 60 to 70 volts or so.

Jeremy

 

[liveforphysics]

Do you think the IRFB3077 will work ok for my 72V controller application? My peak charged voltage is 84v, and nominal is 74v. I know they leave some overhead in the max voltage. Do you think it would be ok to choose this FET for my application?, or should I look for something 100v capable?

I'm planning to re-do the mounting/cooling setup, so swapping to a larger package isn't a big deal. Do you know of a TO-247 package that would be a solid performer for my pack that is 84v fully charged?

 

[BatteryMooch]

The IRFB3077 might, maybe, work for a little while but I don't think it's a good idea to use a 75V maximum rated MOSFET in an application where you know that the voltages will be going up to 84V. For me personally, I wouldn't exceed 60V with the IRFB3077. And that's only after confirming that I didn't have any spikes going over 60V.

IMHO, you want a 100V part, at least. 150V would be better but it becomes hard to find a low Rds-on value. But, paralleling more MOSFETs takes care of that and you have the bulletproof-ness (is that a word?) of the 150V rating.

Perhaps the IRFPS3810? It's 100V, 9mOhm Rds-on, 0.5 junction-sink thermal resistance (nice!). I've used it for a couple of projects and it's a nice MOSFET. You'll need clips to mount it though. They're better than screw mounting anyway. At 150V, you have the IRFPS3815. Both are about $8 each. Not bad when you consider the very low thermal resistance which can allow the use of fewer MOSFETs for a given heat sink.

A search on the IRF, IXYS and Infineon web sites will turn up a decent selection of parts. You can then use http://www.findchips.com to find the ones that are actually affordable. Hopefully others have some suggestions too.

 

[Jeremy Harris]

I agree, the 3077 (or any 75V FET) is really only good for a 60V nominal system - I'd not trust it to have any appreciable ability to withstand being run at more than it's rated voltage. The chances are that a lot of devices will take more than the rated voltage, due to manufacturing tolerances, but you can't rely on getting a batch of "all good" ones.

For me, the 3077s look to be a good choice, because I want to drive a big RC outrunner motor fitted with Hall sensors. 60V is really the limit for these, I think, at least at the moment.

Jeremy

 

[liveforphysics]

Jeremy- I believe you are planning to run the same pair of 130KV HXT motors that my bike uses. I know some giant scale airplane guy runs them at 16s, and they run fantastic. I was planning to run them at 20s, but I guess I could re-configure the pack to run at 15s, which would be only 63v right off the charger. I was thinking it might give me a little extra kick with the 20s though. The motors run my bike well at 10s and 12s, but I set them up with excellent heatsinking, and they never get above luke-warm at those low voltages. I would like to try to push them to the limits.

How many cells do you plan to run Jeremy? Maybe we can work together with our hall mounting solution and controller modding.

 

[Jeremy Harris]

You're right, I am looking to run a pair of HXT 80-100s, just like you. I reckon 16S (with LiPo) is probably OK with the 3077s, as it's about 60V nominal, perhaps 68V hot off charge. I may run either 15S or 16S, it depends on what packs come up at the right price when I'm ready to buy. I have a 48V nominal (in reality it sits at around 53 - 54V), 40Ah, Headway pack that's on my motorcycle conversion, so may use that for testing.

A pack of small neo magnets arrived in the post this morning, which I'm going to play with as a possible solution to shaft position sensing. These are 3mm diameter (just under 1/8th inch) and 6mm long. The plan at the moment is to machine up an alloy disc and drill 14 holes spaced evenly to mimic a set of rotor magnets. I can then bond in these small magnets, facing alternate directions and bolt the disc to the end of one motor bell. I'll make another disc, with grooves machined in to accept Hall sensors. This will be statically mounted, perhaps with some adjustment to allow timing changes. I think this approach might be simpler than trying to fit Hall sensors inside the motor, although I may yet change my mind.

Jeremy

 

[georgeycc]

http://www.fairchildsemi.com/ds/FG/FGPF90N30.pdf
for high voltage people, you might be interested in this IGBT rated at 300V,90A at $1.97 retail.
looks like a drop-in replacement for mosfets but it lacks the built-in diode. i bought a set of the IGBTs and haven't had time to try them on, should be interesting although the switching-loss might be bigger.

there's always spikes spanning the full voltage range when the mosfet are turning on or off, it's the inductance of the motor. these spikes are supposed to be clamped by the diodes but still it'll go slightly over
DC bus since the diode has finite response time. just my 2 cents.


-george

 

[Deepkimchi]

I still think Doc should drill little holes in the front & rear plate of the controller, point it forwards and mount a funnel on it for drag races (not in the rain).

RAM Induction Cooling.. :lol:

 

[Deepkimchi]

Or maybe use this for cooling..

http://www.youtube.com/watch?v=zUc6znC848o&feature=related

 

[Doctorbass]

Or maybe use this for cooling..

http://www.youtube.com/watch?v=zUc6znC8 ... re=related

Ah ha! :lol:

I had easy acces to this cooling 2 years ago when i worked at Sherbrooke Univercity's department of physics!!!

Liquid Nitrogen... -196 oC and liquid helium.. -269 oC !!.. but it can stay at liquid stateenough time at ambiant temp...


Now i only have acces to liquid nitrogen :wink:

Doc

 

[solarbbq2003]

xc864 chip info.......go nuts
http://www.infineon.com/cms/cn/product/channel.html?channel=db3a30431b0626df011b315ce2eb6a27

 

[Jeremy Harris]

Thanks, Brett, but that's not the chip used in the Infineon. The chip in these controllers is the XC846, not the XC864 (easy mistake to make - I made it a few days ago!).

As far as I can tell, the pin out of the XC846 seems to be the same (or very similar) to that for the XC866. I suspect that the two devices may be externally the same, but perhaps have differing internal specs (like RAM etc).

Jeremy

 

[methods]

I'm still looking at ways to get high power (around 200A, 60V+) from an affordable controller. I've been trying to buy the 18 FET board from Keywin since last September, but although he quoted me a price for a couple, I've not yet managed to get a PayPal invoice from him (I've just emailed him again to ask for one).
Jeremy

If you cant get an 18 fet version let me know.
I have these (see picture) in CA right now.



No fets, no wires.
Complete with programming cable, insulators, fasteners, case, heatsink, etc.
100V caps

I am going to build two of these up with 4110's and run them at 150A each.
I am also going to send off two to Luke.

I will send one or two to you at cost, free shipping.
(but only because you look like a nice guy :wink: )

-methods

 

[Emoto]

Hope this is a good place to ask.
Im about to buy a infineon eb3 18 fet but cant decide on the model ill be running 18s 10ah lipo so a 4110 fet should be fine but if i rarely want to add a couple of cells 24s will i be pushing it.
So should i mod the 4110 for extra voltage or do the multi voltage mod with a jumper cable to switch between higher and lower voltages on the 4115 controler.

 

[NeilP]

A 4110 controller will just about handle 24s as long as you keep the hot off charge voltage down to below 100 volt, it is taking a risk though, but do check the caps will take it.

Also as I recently found out, it also depends on the power input stages and the R12 value..the part that regulates what voltage for LVC HVC that the voltage sense part of the controller sees.
With an EB3xx controller I think the front end will be OK for variable input voltage..unlike the earlier ones...where / who are you buying form ? they will tell you what if any mods are needed for 100 volts for the 4110 controller.

best to stick with 4110 if you are not always running at above 100 volt, as running the 4115's at too low voltage leads to heating and losses in the FETS. Not sure the exact reason, but I do know it is not good. best to run them at their design voltages

 

[Emoto]

A 4110 controller will just about handle 24s as long as you keep the hot off charge voltage down to below 100 volt, it is taking a risk though, but do check the caps will take it.

Also as I recently found out, it also depends on the power input stages and the R12 value..the part that regulates what voltage for LVC HVC that the voltage sense part of the controller sees.
With an EB3xx controller I think the front end will be OK for variable input voltage..unlike the earlier ones...where / who are you buying form ? they will tell you what if any mods are needed for 100 volts for the 4110 controller.

best to stick with 4110 if you are not always running at above 100 volt, as running the 4115's at too low voltage leads to heating and losses in the FETS. Not sure the exact reason, but I do know it is not good. best to run them at their design voltages

Couldnt find any info on EB3 controlers, i hoped there operating voltages were a larger varation on EB3's ,@ 100v on 4110's is pushing it, as far as the r12 bit im still learning, but must make a dicision soon,
But as you say the 4115's on lower voltage would be smart to aviod those probs, Which would affect me most of the time.
So if i increased the caps would that increase my voltage range/ do you know of a thread that has done this on a EB3.
Thanks for the help