Modding my EB236 150V controller the same as my EB318 12kW

zombiess

10 MW
Joined
Jan 31, 2011
Messages
3,048
Location
Oklahoma City
I have a factory stock EB236 controller board and also a spare board for it. I decided to build the spare board into a 36 FET IRFB4115 controller so to play around with. I also wanted to do the same buss bar mod to it that I did on my EB318 controller I got from Lyen which now takes up to 12kW and regularly handles 10kW of input when I ride on 30S2P or 30S3P LiPo charged to 125V.

So here are the pictures of my progress.

First up is the size comparison so you can see exactly how big this is.
On the top is a normal 18 FET controller
The next one down is a 24 FET controller
The last one is the 36 FET controller which uses the same case style as the 24 FET, but just longer to accommodate the extra FETs
Very bottom is the internals of the 36 FET controller with parts setup for 60V max. I have the heat sinks removed. The coil of wire is the silicone multi stranded 6 gauge wire that will be connecting this controller to the batteries. All phase outputs will be 8 gauge.
controllers.jpg


Next up is the construction of the buss bars for each bank of FETs. The construction of all 6 of these took about 4 hours. Had to take the #6 eyelets, cut off the plastic, then pull them apart so they were flat instead of round and tin them with solder and make sure I got the spacing as close as possible. Buss bar is sold 8 gauge copper grounding wire picked up from the hardware store for 93¢ per foot so I nabbed 10 ft of it along with 4 ft of 6 gauge... just in case I need to beef up a trace somewhere and go over board. I also have the FETs separated into groups of 6 matched by Miller Plateau resistance. I'm trying to keep them all as close as possible in each back so that they will current share better.
eb236-buss1.jpg


This is the buss bars heat shrinked to insulate them against shorts to the heat sink / spreader bar. FETs still groups waiting to be installed.
eb236-buss2.jpg


One side completed. Decided to use the nylon screws again because I've had such great luck with them on my 18 FET which is being abused regularly and they are holding up great. They have a continuous operating temperature of 188C. Only downside is that you can't crank down on them with as much force as metal screws, but that doesn't seem to be much of an issue in my EB318 using the same mod since I'm always monitoring the temperature.
For the insulators I decided to scrap the stock ones (not sure if they are Kapton or not, but they seem rather ragged, have thermal paste on them and I don't trust them) try some Wakefield 173-7-220P insulators which are rated at 0.33°C/W 12.6 W/m-K. Not sure how well they will work but wanted to give them a shot.
The fasteners are a #6 3/4" nylon screw with a #6 brass washer, the buss bar made from #6 eyelets, the FET, the Wakefield insulator, heat sink spreader bar, bronze lock washer and finally a #6 brass nut holding it all together. I went with brass and bronze fasteners to hopefully help increase the thermal mass on each FET's tab and allow it some more time before it heats up too much. Past experience with my 18 FET shows me my buss bars really help soak up some of the extra heat and they aren't even as good as this 8 gauge solid copper wire on this setup.
eb236-buss3.jpg


That's it for now. I should have the other side mounted up tomorrow and start construction on the board by adding the caps/transistors/etc to it. If this one works out well I'll cut down the factory board in the first pic, rebuild it for 150V use and put it into a custom black Hammond case with shorter alum buss bars mounted to the case much like the stock setup is. All assuming I have time and will to do it. Wanted to get this build going first for when I get my 13kV motor, next up is most likely one of my water cooled TO-247 power sections. Got plenty of controllers to play around with in the future :D .
 
Updated my build.
Installed all the top side components with some pretty decent low ESR 560uF 160V caps. Used 250V Polypropylene caps all throughout, going to add some more once I get closer to finished.
eb236-top01.jpg

eb236-top02.jpg


Beefing up the traces on the bottom side of the board to assist with the extra current. Combination of 12 gauge and 8 gauge solid copper.
eb236-trace01.jpg

eb236-trace02.jpg


Plan is 6 gauge battery feeds, 8 gauge phase outputs, all tied together on the tabs.

Hope to have this finished this weekend.

When I build the second one I hope to be able to cut down the board and fit it in a nice black Hammond heat sink case. Already bought the aluminum buss bars, need to drill them and make sure the caps aren't to tall or else I'll have to change a few parts up. I'll see how this one goes first. Sure is a time consuming project and not the cheapest either, but I want custom and built to my own specs. Next up, finishing the water cooled TO-247 projects now that I finally have some practice building a controller from scratch.
 
Doctorbass said:
Nice project Zombiesss :wink:

Why did you choosed the 150V fets?.. we debated that many times and the 4110 100V can give 33% more power due to their twice higher current.

Maybe you want ... pure SPEED 8)

That remind me to finish my 36 fets projects!
DOC EXTREME 36 fets project
http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=21722&hilit=36+fets+extreme

Doc

Why IRFB4115's because... well as you know, you and I are receiving a certain motor ;) That and I'm always running 100V anyways and I hate running components at 100% of their rated values, it just begs to fail. Besides, 6 paralleled miller plateau matched IRFB4115s connected through buss bars for B+ and phase out are a seriously bad ass setup (if it works right). Based on what I've been doing with my 18 FET controller taking 12KW many times always taking 10KW, this 36 FET one should be able to handle 20KW or more, depending on how fast the FETs switch. But I noticed these use 0 ohm gate resistors so I worry that these might switch on the slow side. If so I'll have to intercept the digital logic output and feed it into a buffered driver. Good thing the stock case has tons of room. I'm hoping it all works out just fine stock. I think it will since the IRFB4115's have half the Qg (gate charge nC of the IRFB4110s) so they should switch faster.

Thanks for that tip on the Hammond case. I grabbed one of them and it should work great for the other 36 FET board (stock need to desolder all the junk components) I'll build at some point.

After I get this working it's back to the grind stone on my water cooling projects. My 12 FET and 36 FET TO-247 power stages are just sitting here waiting to be finished, but I needed to get this one done first. It's like there isn't anywhere near enough time to do all this stuff. Building a controller like I'm showing pictures of is many many hours of work if you pay attention to the details and do it nicely as I'm sure you know.
 
I'm officially sick of soldering. This project is taking too long, need to wrap it up this weekend hopefully. So many hours of work!
 
I found this article about gate driver setups and circuits and stuff. I read through it
and agree with what it's saying. I would observe the gate signals and would not add
any gate resistors if you son't see any ringing. Another thing to check for, maybe the resistors
are not in the gate but in the source line. Also look at the drain-source signals for
overshoot, add snubbers if necessary (see passive snubbers article, this one also
makes a lot of sense to me).
 

Attachments

  • gate_drive.pdf
    345.3 KB · Views: 144
  • passive_snubber.pdf
    1.9 MB · Views: 115
Lebowski said:
I found this article about gate driver setups and circuits and stuff. I read through it
and agree with what it's saying. I would observe the gate signals and would not add
any gate resistors if you son't see any ringing. Another thing to check for, maybe the resistors
are not in the gate but in the source line. Also look at the drain-source signals for
overshoot, add snubbers if necessary (see passive snubbers article, this one also
makes a lot of sense to me).

I've seen both of those before, but this build is just a very basic Xie Chang 36 FET IRFB4115 controller, nothing special, uses individual 0 ohm gate resistors and I matched all the FETs in each bank at 4V Vgs so it shouldn't be an issue. It just so happens I'm modding it out to handle lots of current (at least 20KW) and hopefully improve it's reliability. After this is done and installed on my race bike I'll move onto my water cooled power stages powered most likely by Xie Change EB312-AS-2 boards, then your controller tied to a 12 FET water cooled power stage or something else that just came to the top of my priority list I thought of today that I and several others need pretty bad. So many projects so little time. My to do list is getting longer and my house is a disaster. I'm losing at least one day this weekend just to preparing all the paper work I'm sending to my Fiance for her Visa interview (did I mention I'm getting married too?).

Just a few months ago I was the king of boredom, nothing to do, sat on the couch, watched TV, rode my ebikes, vegged out. Now.... LOL. I'll get it all done in the next few months, just doing one electronics project at a time for the moment. Major props to Lyen an anyone else who builds their own controllers.
 
zombiess said:
So many projects so little time. My to do list is getting longer

Same here. I switched to working only 4 days a week last january just to have more time for all my projects, cycling, GF etc etc.....

Just watch out with the high current and over-volting the drain / source of the FET's.
More current means more energy stored in the PCB track parasitic inductances, which
translates to higher voltages spikes accross the drain / source on FET turn off.
With 150V FET's on 100V supply, additional snubbing may not be necessary but I'm not
taking the chance with mine...
 
Lebowski said:
zombiess said:
So many projects so little time. My to do list is getting longer

Same here. I switched to working only 4 days a week last january just to have more time for all my projects, cycling, GF etc etc.....

Just watch out with the high current and over-volting the drain / source of the FET's.
More current means more energy stored in the PCB track parasitic inductances, which
translates to higher voltages spikes accross the drain / source on FET turn off.
With 150V FET's on 100V supply, additional snubbing may not be necessary but I'm not
taking the chance with mine...

This one is pretty similar to my 18 FET I modded and run at 125V max on stock capacitors and no extra PCB trace beefing except what Lyen does. I'm going to have a lot more capacitance on this one with some additional low ESR 1uF 250VAC caps to hopefully help. Not a really a snubber setup, but they help absorb the spikes and save the FETs. Even without these mods my 18 FET IRFB4115 is working great at 10KW output. I'm removing a lot of the issues with PCB track parasitic inductance by connecting directly to the tabs of the FET and also adding copper to the PCB traces.
 
zombiess said:
with some additional low ESR 1uF 250VAC caps to hopefully help. Not a really a snubber setup, but they help absorb the spikes and save the FETs.

This because the 1uF cap is kinda doing the same as what you can much better achieve with snubbers.

What you need to do is to protect the ('off') FET from is the inductance starting from where the drain wire leaves the FET housing
all the way to where the source wire enters back into the FET housing. This inductance includes all of your PCB traces (not the motor
as this is shorted by the 'on' FET) and the wires all the way to the battery and back.

What you're doing by placing the 1uF is that you're effectively shorting the current loop, it no longer goes from the FET all
the way to the battery but only to where the 1uF is placed. Less inductance -> less energy in the ringing -> FET lives.

A snubber does not lessen the inductance, what it does is it effectively absorbs the energy stored in the inductance.

An analogy: lets take the suspension of a car. Assume you have a 1970's American car with the super bouncy suspension. What
you've done by placing the 1uF cap is you've swapped out the springs in the suspension for stiffer ones. But what you really
want to improve your car is some decent dampers together with the springs. This is what the snubbers are, they are the
dampers.
From Top Gear UK I know for a proper sportscar you need both stiffer springs and good dampers (I like bouncy myself :D )
 
Lebowski, I know a proper snubber would be best but I'm going to stick with the KISS principal on this since these controllers have been built with IRFB4110 FETs, often times not even matched for miller plateau and been run ragged at their Abs Max voltage rating of 100V (FETs and Electrolytics) and over 200A of battery current on 24 FET versions. They don't even have very much in the way of capacitance let alone a proper snubber design or even that great of a driver, but they work and live at some pretty high power levels. I'm hoping mine will do the same.

I am interested in learning more about snubber design and what it would take to add, but I'm lacking the time to experiment and learn right now. I'd love some help on working through the math just to see what it would take in theory to improve it if you have the time.

auraslip said:
Out of curiosity, how much do you think a stock lyen 18 fet can handle compared to your 18 fet?
The case of my 18 fet gets warm running it a 4kw.

I have no idea since I haven't tested a stock one beyond about 5.5KW myself on level ground, but that was at the Grange race last Oct. I can tell you trying to climb 20-30% grades with my modded 18 FET at 40% throttle probably isn't the wisest of ideas since I saw my FET temps get to 88C but the case was barely warm since I stopped before it had much time to transfer, you can hear me talking about the temps in my hill climb video. Had I stayed on the throttle and kept climbing I don't know how much hotter the FETs would have gone or if they were leveling off since I couldn't pay attention and I forgot to hit the log button before I started the climb like a dummy.

I believe Lyen suggests not exceeding 40A for a 18 FET 4115 controller when run at 30S LiPo. My guess is I've abused some of my previous controllers pretty hard and never knew exactly how bad I was abusing them because I've had the cases on 12 FET (IRFB4110) and a 9 FET (IRFB4310) pretty warm before running 24S LiPo at 100V.
 
Some more update pics.

Phase output and FET tabs tied together with 8 gauge
eb236-buss4.jpg


Top down view. High side is not soldered into place yet, just placed to check spacing was OK. Tight fit but it's all good.
eb236-buss5.jpg


Low side soldered into place. Need to get a hotter iron to smooth out and reflow the solder better. This is just one massive heat sink now. Really hard to reflow even small areas even with a 180W iron. I have one that also uses a torch heating a tip, but it's still hard to get this to flow. Will probably have to use a combination of the two to get it all nice and smooth where the FET legs wrap over the 12 gauge wire. It still needs some fill under the legs on some of the FETs.
eb236-trace03.jpg


Finally the case this is all going into. Modded it a bit with some nice strain reliefs. Middle one is for all the small wiring such as halls, ebrake, cycle analyst, etc to run into. Right side is same size as the middle and fits the three 8 gauge phase wires which will exit through it. Left side shows two small strain reliefs which will each have a single run of 6 gauge battery power wire running into it.
eb236-case01.jpg


The whole setup is probably a bit over kill, but as long as I built everything right it should be once mean 125V 36 FET controller.
 
me and stevo did 150v controller with 4115 thay whould blow out under load and did it on a 32fet did same we changed fets to bigger batch fets and problem was gone it may have somthing to do with switching because only one whould blow ramdomly threw out the controller we wired the 12fet with the bigger batch and workes like a charm :p now we need to make an adapter to do an 18 fet
 
Pick up one of these guys. Even tho it's only 80w the tips are so massive, they will flow 60/40 solder from 12v battery terminals into 2awg wire. Just search this guy, Sears won't let me view anything but their shoddy mobile site.

Cooper Tools 185-SP80L Marksman 80 Watt
 
drunkencat129 said:
me and stevo did 150v controller with 4115 thay whould blow out under load and did it on a 32fet did same we changed fets to bigger batch fets and problem was gone it may have somthing to do with switching because only one whould blow ramdomly threw out the controller we wired the 12fet with the bigger batch and workes like a charm :p now we need to make an adapter to do an 18 fet

I had over 100 IRFB4115 FETs and I matched each banks resistance at 4V gate voltage. Some of the banks are using the Halogen free version because I had some of those as well and they all had really low resistance at 4V Vgs. Out of all my FETs I chose to match the banks with the same version (didn't mix halogen free and lead free versions) and also the ones which had the lowest resistance at 4 volts of Vgs in the hopes that if the gate drivers were a little on the weak side that this would aide in switching at the same time for equal current sharing. I'm also only running 126V max through the controller so I have head room. I thought I read that Stevo did 144V max voltage which is only leaving 6V before the absolute max voltage rating.

Wish me luck, changing blown FETs is a PITA. I hope what I'm doing will minimize problems. Like I said though, so far so good on my 18 FET 4115 controller for about at least 150 miles of use at 60-105A and it's FETs aren't matched.

It is good to hear that once you guys start using a large batch everything was working OK. What voltage did you guys run and did you match the FETs for their Miller Plateau voltage like I did? Also, what was the most watts you pulled out of these. Max battery amps @ voltage and max phase amps you tried and had it all live?
 
zombiess said:
drunkencat129 said:
me and stevo did 150v controller with 4115 thay whould blow out under load and did it on a 32fet did same we changed fets to bigger batch fets and problem was gone it may have somthing to do with switching because only one whould blow ramdomly threw out the controller we wired the 12fet with the bigger batch and workes like a charm :p now we need to make an adapter to do an 18 fet

I had over 100 IRFB4115 FETs and I matched each banks resistance at 4V gate voltage. Some of the banks are using the Halogen free version because I had some of those as well and they all had really low resistance at 4V Vgs. Out of all my FETs I chose to match the banks with the same version (didn't mix halogen free and lead free versions) and also the ones which had the lowest resistance at 4 volts of Vgs in the hopes that if the gate drivers were a little on the weak side that this would aide in switching at the same time for equal current sharing. I'm also only running 126V max through the controller so I have head room. I thought I read that Stevo did 144V max voltage which is only leaving 6V before the absolute max voltage rating.

Wish me luck, changing blown FETs is a PITA. I hope what I'm doing will minimize problems. Like I said though, so far so good on my 18 FET 4115 controller for about at least 150 miles of use at 60-105A and it's FETs aren't matched.

It is good to hear that once you guys start using a large batch everything was working OK. What voltage did you guys run and did you match the FETs for their Miller Plateau voltage like I did? Also, what was the most watts you pulled out of these. Max battery amps @ voltage and max phase amps you tried and had it all live?


Personally i never took the time to match them and play with the mosfet data. I could but maybe i'm lazy to do that from now... :lol:

What i know is that the controller behavior was way difefrent with lower turn mwinding motor than with higher turn winding!!!

If you want to see how tough is your controlelr build, use a low turn winding motor that compare with the 5303 or 5302 :twisted: If they dont fail, they are perfect!

I never had any problem with all my 18 fets controller using 4110 of 100V max rating at 24s LiMn and 250A phase current when i was using them with 5304 or 5305 or 2807 motor. I thoiught they was really tough too!.. until i connected thjem with the 5303 and the 5302.. Ouch... the high phase current make them suffer at low rpm!

That's the ultimate test, low turn winding motor will show their limit !

Doc
 
Finally got an iron that is hot enough to do the work I need. 400W/150W. Works great. Project is still a total pain, but if I build another one I know a bunch of short cuts now, such as install the copper AFTER soldering in the FETs. Would have saved me many many hours. All that copper and solder just love to act as a giant heat sink preventing the solder from melting the way I want it to without lots of time spent trying to get it hot. Not to mention fighting the occasional solder bridge.

No pic update yet, but 6 gauge power wires are installed. FET heat sink with high side is now installed with all three banks connected by the buss bar tabs using 8 gauge wire and then connecting to the far side of the board as well as the power feed side (these boards have two holes for feeding B+ power, one on each end and mine are now shorted together through beefed up traced and 8 gauge wire connecting the tabs.

Will hopefully test the controller out tomorrow or Monday to see if it actually works or if I made a goof somewhere and need to find it. No shorts anywhere, checked multiple times due to causing several accidental solder bridges because I'm so use to working with low power soldering irons. As long as I didn't fry any semi conductors from heat, I'm just a few hours of work away from testing it to see if it spins a motor before finally installing it into the case.
 
auraslip said:
Sounds awesome. You're doing a lot of really cool things lately!

I have a lot more to come, I'm just getting started. It's just about lack of time at the moment. I'm busy as all heck because I've had a few things come up that have caused delays like everyone does in their life, but it's nothing major. I'm hoping to be back on track by next weekend at the latest.
 
Cool man.
 
Arlo1 said:
Cool man.

I just wanna see if this actually works after all the hours and hours of work I've put into this thing. I did learn a ton about building a controller though. Such as it's probably wisest to add the copper to the PCB traces last if you don't want it to take forever soldering stuff because the entire controller has become one gigantic heat sink! Even my 400W iron struggles now, but it's also really awesome to see how well the heat sinking works. I was worried I was dumping a lot of heat into the legs of a FET I was struggling with that had a solder bridge (giant soldering irons and small parts = not so fun) and was using the 400W setting and probably had it on there for a good 2 mins when I realized what I was doing and thought "Crap!", that's too hot for the FET. Flipped over the board. FET case was just barely touchable (not even close to damage levels, especially with my overly sensitive girly finger tips). Alum heat sink was a bit warm. Every thing was cool again in less than a minute, except the soldering iron tip which still was able to melt solder for a good 3 mins after I let off the trigger :D

As long as I didn't goof anything up, I think this controller is going to be a serious monster at 125V. Already have the 6 gauge power wires installed and 8 gauge phase wires. FET tabs are connected to phases through the 8 gauge wires. High side has 8 gauge connected to all tabs then attached again on the other side of the board. Other side is tied directly to the 6 gauge power B+
 
Had some time to make more progress. The board is done, just need to solder on a few connectors and then test this bad boy out and see if it actually works and spins a motor before I assemble it into it's case and mount it to a bike for real testing.

Bottom shot of the completed traces. I was able to just barely squeeze in two 1uF 250 VAC low ESR polypropylene caps at each end on the bottom side. Since I couldn't get them installed the middle I decided to use the left over 0.1uF caps and parallel two of them on the bottom to go along with the two on the top side. In all this board has 4uF of low ESR polypropylene caps to hopefully help absorb any spikes. It has 3,900uF of low ESR 160V aluminum electrolytic capacitors.
eb236-trace04.jpg


Top view showing the buss wires and 6 gauge power feeds
eb236-top03.jpg


All connectors soldered in and run through strain reliefs
http://dynamic.opticalanarchy.com:8080/zombie/bike/eb236mod/eb236-top05.jpg

Just about done, just need to add on the spades for the phases and figure out how to connect the power to it. Got the one phase wire installed with a spade, just need to change the color of the heat shrink so I know exactly what it is. Accidentally installed a red one... oops.
eb236-allwired01.jpg.jpg


I'm hoping to test it out tomorrow, still need to do the regen mod to allow it to work on high voltage, but that will only take a few mins. Everything just barely fits in the case. That 6 gauge wire has some really thick insulating on it and just barely fits because of two electrolytic capacitors at the end.
 
Got to test out the controller today but only no load. Worked great at 46V running the bike at no load. Tested out 3 speed switch, cycle analyst, throttle etc. All seems to work great. Time to put it into all back into the case and try it out on the bike.

Only bad news is I put a scope on the gate of an FET and did some measurements to see the driver performance... really poor. See my other tech thread for details on that
 
Back
Top