Infineon 9FET Controller and 9C 36v FH from e-bikekit.com

[mwkeefer]

Hello All,

Although far from my intended final design (using HV110 & MS eDrive 2Stage + Tower Pro 5330-9T), because Matt has been so busy with the eDrives lately (I didn't want to add one more order to his load) I had to either back burner testing and development of my pre-controller (combination systems monitor, step up/down buck converter, current limiter, cruise control, variable - cell level LVC, etc) for a month or two or purchase a hub motor and start testing, logging and debugging my code / design.

I suppose the subject makes this overly obvious but I obtained a 26" 36v 500W front hub kit from Jason @ http://www.e-bikekit.com just so I could get my eBike up and running (for above reason and a few others).

What I have learned about these kits, the hub motors and the 9FET infineon controllers could fill a small book (or with my posting style perhaps a large book) but I wanted to give anyone thinking of using this kit as a starting point or even a final eBike the benefit of my testing and experimentation.

1.) Forget 500W rating, far under-rated! I have personally acheived > 2000 W sustained output by opening up the programming on the Infineon controller, raising the Rated Current to 38.5A and the Phase Current to 75A. I have tested using 36S3P 9.9AH nimh pack and a 12S 5AH Turnigy Lipo Pack (Build of 4x3S in Series) with the lipo getting better results for all out accelleration from stop or low speed. The nimh pack clearly had better range but at near 10AH rated capacity, that is no wonder.

2.) Forget 36v rating. Even the Lifepo4 batts offered by Jason at http://www.e-bikekits.com are 12S config or 46.2v off the charger, so a 12S lipo pack @ 50.4 off the charger (@ 4.20v per cell) is fine. I haven't tested with 15S yet as I am working on a soft powerup circuit to prevent arcing when batteries are first connected but 63v off charger should be the MAX V (based on 63v FETs). The 15S will likely blow the FETS but so far the only Voltage I am sure will toast the 9 FET ESC is 72V > (tested and confirmed to smoke controller).

3a.) When using your own battery option with this controller and hub, you will want to use Parameter Designer (http://endless-sphere.com/forums/viewtopic.php?f=2&t=5713&start=180#p163592) to configure a safe Low Voltage Cutout, Rated Current and Phase Current. I have not gone above 40/75A which gave me > 2000W peak without even warming the motor to the touch under load (or the batteries for that matter).

3b.) You should also know, the small red or orange wire from the controller to the batteries is actually a on/off lead and as such can be run through a normal SPST switch to turn the controller active/inactive. This is important because the current draw in off mode (wire disconnected) is 10ma while the draw with wire to power source is 700-800ma minimum. When your not riding (parked, at work, whatever) turn the power switch off to avoid draining more power from your batteries than you need. This reduces the draw to a bare minimum without needing to disconnect / reconnect the main power leads and deal with an arc (working on this)

4.) These controllers need airflow!!! - Even without modifications or changes to current limiting this controller needs airflow or it will burn up. I am lucky to have an extensive logging / monitoring system in place so I quickly noticed the ESC approaching 150F and remounted it under my front mounted handle bar bag. I used hobby pine to make a reinforcement panel that goes inside the bag (with a hold cut out to pass wiring through) then I just screw the controller upsiide down to the bottom of my bag and run the harness through the hole I pre-drilled in the wood panel. Bottom mounting the controller puts it in the airflow and the heatsink does it's job (mount with sink screws facing forward) max temp with this mounting position was steady 128F while running 1/2 throttle @ 16-18mph up a 20+ grade just under a mile long.Controller returned to < 118F within 10 seconds of completing the climb and reaching level ground. Within 15 seconds controller was < 100 degrees stable. I also ran the same test but at WOT without pedaling and as expected the temp stayed lower (presumably becaue at WOT the FETs were 100% on and switching only to limit max current which seems to generate less heat than partial throttle settings. Maybe a fan fould keep the temps lower when under load with partial throttle, I will test when I have a chance.

There are many more "tips and tricks" things to be revealed about thhese e-bikekit.com hubs and 9fet controllers which I will post later as time permits, up next: How I managed to install front hub on same Specialized Hard Rock Pro (Disc) that Jason tried and broke the drop outs. Including my el-cheapo torque arm solutions using off the shelf (ACE hardware) parts or 10MM bicycle brake wrenches. They are not pretty yet but they do save my front forks (recently upgraded to full lockout capable) and they have stopped all spin on the axle within the dropout even under the heavy torque of my modified controller programming (and a few other tweaks).

Hope this helps someone!

- Mike

 

[fechter]

Thanks for the report.

You can see some photos of the controller insides here: http://endless-sphere.com/forums/viewtopic.php?f=2&t=11177

The 9 FET configuration makes for a nice compact package and I love the black finish.
With an IRFB4110 upgrade I think it would perform nicely at higher power levels.

 

[dnmun]

i agree, these are powerful little controllers and the 9C motor is like a little 5xxx series C'lyte.

i haven't modified the software settings, but did solder up the shunt on mine and used knuckles resistor bypass mod for resetting the LVC.

i have actually run the controller and the 9C i got from jason up to 90.6V in stock form without blowing up the FETs or the caps, but only freewheeling in test mode. i think the FETs can handle the 80V or so of the nominal 72V lifepo4 once some power draws down the voltage. breakdown voltage climbs with temperature also.

i have upgraded 5 of the controllers to 72V lifepo4 for jason and he got one of them up to 45mph with 2 of his EBK lifepo4 packs in series. he did blow up one of the last ones, and hope to see the inside of it eventually too. in order to understand what happened.

http://endless-sphere.com/forums/viewtopic.php?f=2&t=10222

where did you put the temperature probe in the controller and in the motor? what did you use for the probe? a thermistor bulb? and did you open up the hub to place the probe inside? any pictures?

i can sell you some of the 100V1000uF caps for 30 cents each too, and just ordered some 4110s from methods to see about upgrading one or two of these 9 FET controllers to 100V.

make sure to replenish the silicon thermal joint compound too when you reassemble so it will handle the heat better.

 

[mwkeefer]

fechter,

thanks for all the great work youve done with the different components like the throttle->servo with current limiting circuit you did. big inspiration ! (along with Matt and many others)

dnmun

the temp sensors are a mix of std eagle tree temp thermistors and motor loop style (high temp). I mount the temp sensor for the ESC inside the case by placing it against the middle FET and applying a small amount of zinc heat transfer compound then I use some high temp (rated at 100C) epoxy to case it in. I previously tried embedding into the FET side of the ESC sink and it does give a better overall ESC temp reading on the outside (sensor is still there just not in use right now) but I wanted to monitor the FET temps directly (hoping that by watching 1 FET so close I can determine approx temp of other FETs in circuit faster).

The motor temp sensor began with (I know ghetto) me just duct taping the sensor on after a hard run to get an accurate temp reading but not monitoring real time. Currently I have embedded it within the motor housing and have run the wire out through the std pass out, I am reading internal temps but obviously since I am not in a stator directly my measurements will be slower to react and perhaps not true internal peaks but its close enough for govt work = )

Thermal compund - I have found the compound for PC heatsinks works much better to couple the FETs to the sink case, better than the zinc. I haven't touched on the thermal paste here yet because I hadn't yet encouraged anyone to open their 9FET ESCs - that will be coming since I will cover water proofing, beef up, additional options and connecting / wiring those options, soft start up circuit mod to remove arc when connecting power, etc. I just dont have time to do the write up until early next week and figured any novice to mid experience builder would spend that time looking at the specs / docs on Parameter Designer. I plan to follow up with what I have deciphered with regard to the programming options (and how to hack Parameter Designer to set non-standard variables to tweak the LVC, etc).

Pics... I have to pull the hub apart again soon when I do I will take pics of sensor mounts. I am working on shooting the other pics now (while balancing my real life) and should have them up by weekend.

I would be very interested in obtaining a few (10-12) of the 100v1000uf! I may have some parts (components) you cuold use laying in my parts bins... send me a PM or email = )

with regards to Jasons blown 72v controller, I am reasonably certain it occoured under WOT and was a result of heat dissappation combined with the off charger voltage. I haven't seen it but have spoken with him and knowing the design of thsese ESCs I would bet 1.00 = )

that said, I would be curious if you tested the RPM at your voltage and perhaps the no-load current. I promise if you load it will toast the controller. I am curious in the RPM and load to determine kV for the motor and ensure what I have tested is consistent at that high a voltage.

-Mike

 

[dnmun]

i agree, a load would have toasted the controller. it was actually inadvertant because i had added one more SLA to the stack than i thought. just spinning it freewheel when i first hooked it up.

this was curious too as to how it happened, because i was spinning the motor up as i added batteries, and thinking i was around 80V i spin it up and it dies out as soon as it spins up so i shut down and then discovered the 90.6V. but i think the reason it was shutting down was because the throttle line was pulled so high with the 5V bus from overvolting that the throttle inhibit above 4.7V cut in.

so it was abused. i estimated the 12V rail was at least 22V and 62V to the LM317T. so they will take some abuse from overvolting, but now i am more sensible and not too stupid at one time.

jason told me it blew up almost immediately. i had not tested it before sending it to him so i coulda put too much solder on the shunt, actually i esentially bypassed the shunt on that controller. wish i could program them yet.

i can't imagine how powerful the 9FET will be with 4110s since methods was almost doing wheelies with the 6FET and 4110s.

 

[CamLight]

the temp sensors are a mix of std eagle tree temp thermistors and motor loop style (high temp). I mount the temp sensor for the ESC inside the case by placing it against the middle FET and applying a small amount of zinc heat transfer compound then I use some high temp (rated at 100C) epoxy to case it in. I previously tried embedding into the FET side of the ESC sink and it does give a better overall ESC temp reading on the outside (sensor is still there just not in use right now) but I wanted to monitor the FET temps directly (hoping that by watching 1 FET so close I can determine approx temp of other FETs in circuit faster).

Mike,
Is the temp sensor mounted against the black epoxy case of the FET or against the metal tab of the FET? If it's the metal tab, the junction temperature (the important one ) is 45C-80C higher (depending on the FET). If the sensor is on the black epoxy case, the temp reading is only valid once the system comes to thermal equilibrium and then the value is still 45C-80C higher.

Can you get one directly against the back of one of the FETS (thru the sink)? That's the best way to get an accurate reading. Then all you need is the junction-case thermal resistance spec for the FET and calc the temp difference between the case and the junction.

 

[mwkeefer]

John,

Thanks for the info!

My initial placement was through the heatsink butting up against the fet tab behind (the metal) but the reaction time here was far too slow (I think you would call it thermal equilibrium) so I moved it inside and yes it's on the metal portion of the sink in the center FET so my readings are high (I calibrated using the max difference between the pre-heated outside case and the temp reading of the metal junction of the FET) so my readings are fairly close to accurate (+- 5% by my math) but the react time is greatly improved which was a design goal as I need an ultra fast temp sample from the actual fets not the dissappation by the sink or the fet case.

Thanks for explaining how to calibrate the right way (datasheets are great aren't they?) the solution I have now can be temporary (and really is) but it was a 3-4am engineering session / rebuild and I had to finish by 8am so... I did the best I could, with what I knew in the alloted time.

Your way is clearly better! Thank you for the heads up, not just for my purposes but for everyone else reading this thread also.

-Mike

 

[CamLight]

Excellent!
Good luck with the rest of the project. I look forward to reading about it.

 

[dnmun]

if you can develop the relationship between power through your 9C and the hub temperature, then that should be applicable to the other 9C mtors too, so then people could have a reasonable estimate of how hot their motor is for the amount of power consumed, and maybe we have to consider the rate at which it is consumed too maybe. but that would help people without a temp probe even more than you. so this could be really useful.

 

[woobins]

I just bought this kit for my first e-bike. This thread couldn't have come at a better time. Thanks for the info you've posted so far and please don't be stingy with the updates. I'll be watching this like a hawk.

 

[Takemehome]

Well, I reprogrammed my controller at 29 amps. instead of 22.5 and boosted phases at 70 ( my BMS is fused at 30 amps )
WOW Freewheel no load 54kmh ( 120% = my speed 2%L)
On the road, it feels like a kickdown when you go WOT passed 30kmh.
It takes me up to 42kmh in a flash, NO PEDALING. ( me at 260 lbs + bike, motor, battery and all my bags >300 lbs )
When on the bike path, I use LS speed ( 60% ) that's 20-23kmh. ( got to control myself to stay civilized )
Added cruise control, that's great on a long trip.
Regen is between 4 and 9% ( depending road or town )
Got to love this programmable controller.

FWIW:
To change speed, I use a little dipswitch hot glued to the throttle with a tiny push button for cruise control , connected to the controller with a small gauge ribbon wire. There is no current involved here, so you don't need big wires and switches.


Here's my asv file.
2:EB809
70
29.0
31.0
1.0
60
0:Switch Mode
30
120
60
1.0
3.0
1:Only fake Indicate
1:Comm GND
2
20

If there's anything I can change, like improve regen for example, please do tell.

 

[mwkeefer]

Wow... I only spent a day traveling to e-bikekit.com and already I have a pile of unanswered but answerable questions...

Lets begin with dnmun since I know a bit of your situation.

First excellent work on the "accidental" engineering. I'm not poking fun, we all make mistakes and if nothing blows up then we learn from those mistakes (if somthing blows up we learn not to try it the same way again) in your particular case we have learned a good bit and you didn't destroy the hub or controller in the process! Kudos.

And so we are on even ground, I once accidentaly charged a 3S lipo on a PB (Sealed Lead Acid) setting with my old Triton charger. Obviously this overcharged the pack, and did cause a cell to rupture however... there was no heat, just exhaust gas. At this time I was still charging my lipos in my oven (you don't get much more blast proof in a residental setting) but always had them in an aluminum baking pan with paper between them and the pan to insulate. No burn marks of any kind on the paper (it even bent up on the sides with the baking pan). I screwed up but I learned that overcharging turnigy 3S lipo packs will not cause life threatening explosion or heat... the cells just puff some gas (we probably shouldn't breathe) but otherwise no danger (in my one test case, I haven't tried to duplicate it yet as I need all the packs I have).

Onto your questions:
" i esentially bypassed the shunt on that controller. wish i could program them yet." - Why can't you?
The program you need is Parameter Designer and is available for download on ES under Infineon Controller Technical. The adapter is any TTL voltage level (3.3v-5v) to RS-232 (5-10v) signal level adapter. Since the signal goes only in one direction (to the controller) it make it that much easier. You can buy these adapters premade from sparkfun which will work with a variety of devices including but not limited to the controller, the bms inside the pack (I think, sw is not available at this time... not sure it ever will be but if it is this adapter would be configurable to program it.)
If your handy with a soldering iron, I can point you to half a dozen single transistor (NPN I think, normal switching type) designs available freely on the web. Add a few resistors and a cap and you have your adapter to program the controllers (its how I did it at first).

One thing you should know, if your using the e-bikekit.com lifepo4 batteries then you should never exceed 25A Primary Curent, you can set the phase to whatever you wish... I suggest beginning with 2x the primary (50A) and working up from there... testing performance.

I am working on a manual for parameter designer now and will be a few days still (alot of things have come up) but it will explain all the parameters available to the controllers and what tehy really do. Some of this like Block Time are so conflicted when you do research that I have just done tests at each setting to determine what it is doing, same thing with the low voltage cutout and the tolerance setting. I assumed it used LVC + Tolerance to reach the final cutout v but really it works in reverse and its LVC - Tolerance = Real Cutout Voltage.

Give me a few more days and all will be revealed, I am working on regen with the e-bikekit.com lifepo4 batteries, ensuring it will be safe at any speed (ie it doesn't cause overcharge at high speed braking) and still need to do testing. I will be testing with lipo, sla and nimh chemistries also.

methods has come up with some really interesting ideas and tests, the guy is smart (real world and engineering) the kind of smart we need more of around here. I thik he may be in the right direction with the 6 FET version beefed up with the 4110s, it's the controller is smaller (making mounting easier) and well it's just impressive work. The problem is getting china to use the much more expensive 4110s. I doubt it will happen in reality, many are ordering the blank PCBs and building them up on their own... I think fetcher and a few others are doing this right now. The issue is cost of mass production vs requirements for kit as it is intended. As intended (500w rated, more like 800+ at full charge but...) the 9FET infineon is even overkill, max rating on the battery (not burst max but continuous max) is 25A (look at your manual).

I can tell you that without hardware mods, and only using software reprogramming I have configured my controller:

Board Type: EB809 (that's our 9FET board)
Phase Current: 100 (Amps)
Rated Current: 40 (Amps)
Limit Voltage: 32.5 (Volts + Tolerance for actual cutout value)
Tolerance: 4 (actual cutout = 36.5v but not sure of variances between controllers yet)
Block Time: 1 (default setting everyone seems to use, I have tried others but have not finished compiling the
logged data to determine measured effects)

Minimum cutout voltage for pack: 36.5v which is cutting it close so I usually self stop at 38.5v without load (3.2v per cell - very safe for lipo)

Your cutout voltage and rated amps need to reflect the ability of the power source (batteries)... if you want I will help you pick optimal configuration.

And just FYI: My controller is also setup for EBS regen (only off ebrake inputs, not based on throttle), has a brake light, has small dial controller mounted to handlebars to select speed mode (Legal, Economy, Performance, Turbo, Reverse) yes, reverse and it works = )... makes backing up fun again - lol. As I said before until I finish testing the EBS and regen braking to determine its max voltage, current supply vs speed is safe for use with the LiFePo4 packs I will post information and I believe Jason will incorporate it into his kits from here forward (the regen and brake light maybe, not the multi speed and reverse... at least not as far as I know). He has a tough job, trying to keep costs low and quality high... I've been in his shoes and it's never an easy place to be... any changes could have to be applied to hundreds or thousands of units already in stock and so changes after the fact are usually frowned upon. That said, I do know for certain that he works very hard to ensure the quality of his motors, controllers and kits and in my humble opinion does a good job of it. I can also tell you he is very interested in including ebrake and if possible regen into the kits (plus for everyone).

Hmmm... what am I forgetting.

Oh yea, for the guy with the working regen setup already... your config looks good. I assume you don't want your throttle to control the ebrakeing. In other words you want reduced throttle to result in freewheel coasting unless you pull your brake cables? If you want to test that throttle mode, change the Slip Current Charge Mode to 0 Up 15 Mark. This means when your throttle is lower than 15% it will freewheel otherwise as you reduce throttle from 100% the controller will ebrake.

If you want more brakeing power, you could increase the load seen by the controller but I think 30-40A will be the limit, it may be matched to Rated Current value. A simple way to add a load would be to parallel a suitable light bulb (automotive trailer lamps are available in 12-48v and would likely survive over volting a bit, also a load resistor could be used but that's a waste of energy as heat) and then a 5v relay to switch it in and out of circuit (hint the red wire going to the ebrake is 5v + and the black is 5v ground = )) this would increase the load the controller sees and may increase braking effect to some extent, I have not tested it yet but will.

One other idea I am toying with (especially if the LiFePo4 pack BMS turns out not to support regen brakeing or limits the input amperage severely (recommended charge rate is 2.5A remember, that's not much load for braking use.) is to inline a 1 Farad "Car Audio Stiffining Capacitor". If the ESR is low enough (even the cheap ones seem to be now) then an inline cap would do several things at once (all positive):

1.) The capacitor would be capable of discharging much faster than the LiFePo4s at much higher current rates basically absorbing the abuse which the battery would suffer under constant WOT launches without peddling. Caps are designed for this, think of it like a 2AH battery capable of 100-120A peaks (more really) for 5-10 seconds.

2.) Durring regen braking, the cap would eat up as much current as the motor can output - 20, 30, 60 amps and it would charge the capacitor storing this electricity for the next launch or controller load (ie throttle).

3.) Since the capacitor would charge first at a higher rate the BMS of the LiFePo4 would not be bothered because the short term recoup storage would be in the capacitor. The same applies with all other battery technologies.

Since this could be used with SLAs or LiPos... and since in our type of use the highest load is presented to the batteries (of any type) while starting from a dead stop. This is where caps of this size would buy us extra power by reducing the current demand on the batteries. Since the cap would charge from either direction (battery to cap or controller to cap) this might be the ideal recovery and heavy duty ebraking system. Don't expect this to be part of a kit but I will pick up one of the low ESR 1 farad caps tommorow (might need 3 in parallel to handle voltage) and do the testing. I know some hybrid cars use this method for short term burst ability while maintaining steady draw rates from the main batteries.

Geez... I can't believe I almost forgot, I finally saw the new disc brake adapters (thread on) which will become part of the e-bikekit.com standard kit (I think, they will be available in either case) ... they are nice! True as possible (ie flat) and once threaded into place tortional force will keep threads seated so no spin offs!!! - Very nice work.

I also got my eyes on some of the new hub / front wheel kits and in a word... black. Beautiful looking, ran a quick test at his warehouse test station and they really are nice, I won't get a chance to really test them for a while since they aren't available yet (I don't think) but they seem quieter and I for one think they look better in black and will blend with more bikes (drawing less attention to the ebike part of the ebike) which is what we all want right?

Finally someone asked about taking my temperature ratings at different RPM levels and using them to approximate motor temps based on what I call a flat mathmatical equation using my total watts @ rpm @ volts = temp readings to create a graduated scale. I don't know if it will work or not yet, there are several factors:
1.) Is the wheel in motion? At what speed?
This will determine how quickly the hub can dissappate heat, cooling itself under a given condition.
2.) How tight are the assembly tolerances on these motors
At this point I have no idea, I have spoken with Jason asking to test a random batch of his motors to see how consistent they are across the spectrum and determine a variance number and Jason is all for it but my schedule will take this type of testing a few weeks out (2-3).

Once we have a variance or tolerance level between random hub motors and we add in airspeed (assuming groundspeed is airspeed which doesn't account for head or tailwinds but should suffice) then yes, we will be able to model the temp, top speed ... basically we can model anything we want once I have collected the required data to base our modeling on.

I will even write a small C# or vb.net front end GUI app for users to calculate everything in a nice and intuitive manor (this is what I do for a real living after all, programming).

Ok... it's 1:00 AM and with the exception of some issues with one of my other partner businesses I have spent the past 15 hours dealing with ebikes, hubs, motors, controllers, computer problems, a broken down mercedes (not mine) and about 300 miles of driving total (I was working on the above mentioned in the passengers seat, while my associate was driving) and to be bluntly honest... I am too freaking tired to go out for a ride on my own eBike and yea it is fully charged and ready to go, I'm not (too bad they don't make people chargers, I know... I mean legal ones).

With that I bid everyone a goodnite, more info tommorow (maybe even some pics now that I've fixed my servers routing tables).

-Mike

 

[bikeelectric]

I wired in a direct plug for the Cycle Analyst on the Ebikekit controller last night. I put the throttle/control wire on the ebrake minus pad and it seems to control it somewhat.
I am using this with 48V 10Ah Headway pack. I had to put the max amps setting way up from the Cycle Analyst default since it kept cutting the motor in and out. After that it runs now. I think I put it up at 150 Amps from the 50 that was in there. Not sure if anything is correct since I haven't figured out the shunt calibration. Anyone have an idea where I should start at with this ?
Later today I will try to determine how close it is by using another amp meter to check what the Cycle Analyst is showing.

 

[mwkeefer]

bikeelectric,

Calibrating your shunt has been covered in the 12 FET Infineon Controller thread, knuckles Regen and More thread and the Infineon Controller Technical Information thread. Calibrating specifically with Cycle Analyst is covered in the manual you would have received with your cycle analyst from e-bikekit.com (or whomever it came from).

Calibrating the shunt worked best for me when I used the method of passing a known current and voltage into the controller (it's odd, you have to drive positive voltage into the controller negative input and then negative into one of the motor phase wires) but once your done ... your done.

If your handy with a soldering iron, you may want to consider lowering the impedeance of your shunt using Methods method of solder wick and solder. This could be very important for you since @ 48v and 100A you will be putting somewhere in the neighborhood of 10 Watts of power dissappating through the stock (shipped) shunt. It will get HOT HOT HOT and all that heat is nothing more than waste.

I am still working on the documentation for the Infineon 9FET controllers and will include references to the above listed threads for more information on the topic as it really is well documented already by Geoff, Methods, Knuckles and the rest of the Infineon elite.

If you need more assistance... u know = )

-Mike

 

[bikeelectric]

Thanks for the info Mike.
I had done some reading and found the calibration method you referred to. I used an alternative method for now which was to simply put an amp meter in series with the battery and then compare readings with what cycle analyst and make the percentage adustment to the Rshunt value. This got me in the ballpark.

I didn't realize for a long time that this was an Infineon Controller but since I discovered that , I found so much information on here .

The tip about lowering the impedance of the shunt with solder is something I will read up on and then most likely do. It makes sense now that you explain it that it would get hot. When I first got this ebikekit I had the battery and controller in pack and it got pretty warm. But since I moved the controller outside I don't ever notice the case warm.

I will look forward to your how to guide for these controllers .

Larry

bikeelectric,

Calibrating your shunt has been covered in the 12 FET Infineon Controller thread, knuckles Regen and More thread and the Infineon Controller Technical Information thread. Calibrating specifically with Cycle Analyst is covered in the manual you would have received with your cycle analyst from e-bikekit.com (or whomever it came from).

Calibrating the shunt worked best for me when I used the method of passing a known current and voltage into the controller (it's odd, you have to drive positive voltage into the controller negative input and then negative into one of the motor phase wires) but once your done ... your done.

If your handy with a soldering iron, you may want to consider lowering the impedeance of your shunt using Methods method of solder wick and solder. This could be very important for you since @ 48v and 100A you will be putting somewhere in the neighborhood of 10 Watts of power dissappating through the stock (shipped) shunt. It will get HOT HOT HOT and all that heat is nothing more than waste.

I am still working on the documentation for the Infineon 9FET controllers and will include references to the above listed threads for more information on the topic as it really is well documented already by Geoff, Methods, Knuckles and the rest of the Infineon elite.

If you need more assistance... u know = )

-Mike

 

[mwkeefer]

Larry,

I had my logger, infineon and customized pre-controller in a handle bar sack. It got hot (logging > 100%).

My solution is simple... I cut a small piece of hobby wood (pine) to fit the bottom of the inside of the handle bar bag. Once I placed it in and ensured it provided adequate surface area for the controller, I removed it and used epoxy to bond it to the proper location (super glue or even duct tape would work in a pinch). With this new 1/4" wooden base, it was simple to drill a 1" diameter hole for the wiring loom to pass through. Then I did the obvious, screwed the controller upside down below the bag. Face the transisters (FETS) screws forward for maximum cooling effect.

The result... controller temps are sweet even climing massive grades at partial throttle (when producing the most heat). This seems to be a reasonable solution to the issue of heat dissapation.

To further make things a bit more... bullet proof... the bag was sprayed with water proofing for nylon weaves (the outside construction) and a boot was placed to cover the wireing harness and provide weather resistance.

Riding in the rain with this setup has not caused me any issues yet (I emphasize, me and yet).

Hopefully this will help you find a good final resting place that is both stealth and convenient for you.

The bag still contains all my wiring (totally replaced from the original kit, additional leads wired into the controller to allow for ebrake, cruise control (automatic and manual), various leds and feedback lines) which is drasticly lighter and easier to manage nicely inside a small front tool pouch style handlebar bag. I am a believer that the bag provides a nice bit of vibration dampening to the ESC which should also help extend it's life. I also have my eLogger v3 and a loom of sensors (wired overcat 5 cables) in this bag.

- Regards,
Mike

PS: I am in the process of refitting a new control unit into the case and an additional LCD display to view more detail at once. but I should complete that mon / tue and will post those pics. If I can find pics of the prototype I will post them sooner. but I think the whole thing is fairly obvious. (Then why did it take me a month to come up with it? Lazyness - if it's not broke don't fix it)

 

[Takemehome]

Mike, you are right about the cruise control.

If you hook a momentary between CR and GND you get a "manual" cruise control.
You get to your desired speed, push the button and release the throttle.

If you put a switch between CR and GND then you can enable-disable "automatic" cruise control.
Automatic is set by holding the throttle steady for the time selected with the Parameter Designer ( Auto Cruising Time )
I strongly recommend to disable it if you ride in town.

Denis

 

[mwkeefer]

Denis,

That is exactly how I have it wired (more or less) with a enable/disable cruise control and the timing set to 15 seconds (if I ever hold the throttle other than WOT for more than 15 seconds, call an ambulance and the TV crew cause I'm dead and still riding... now that's A S S I S T A N C E!)

Right now I am testing to see if activating manual cruise control has any effect on the Speed % variables used by the controller because I have a suspicion it does.

My pre-troller will bypass this entire feature at the controller level so that it can be based on Speed, Inclination and Current Thrust (Gs) forward. No more current based cruise control (can't really call it cruise control).

I noticed somthing the other day with this infineon: No matter if you have "Fake Indicate" or "Up to 15%" selected, if you have EBS at anything other than 0, you will find that your wheel puts out voltage even when the batteries are disconnected if you are moving. At roughly 8mph I was seeing 17.?v, I have the log. I promptly reprogrammed the EBS value to 0 and the voltage went away, I didn't test for current or anything but without using the ebrake cable cutout, just enabling EBS was causing volts to run backward into the controller. This could be really bad when going downhill and picking up far too much speed (25-40mph) the voltage could far exceed your packs safe range of input/charging voltage. Again I did't test just disabled, since right now I don't want regen (testing other things) and without regen the controller should never feed electricity back to the battery or power source.

I would follow your "strong" recommendation but I put the cruise control activate momentary in a very awkward location that I won't ever accidentally hit. I did want the ability to disable the feature with pushon/pushoff button to mimic a car operation but since I am now writing the routines to provide Cruise Control at speed basis (minimum) for the infineon... I doubt I will ever use the Infineon cruise control. Not sure why they aren't picking up RPM off a single HALL sensor and computing speed when Cruise Control is engaged then maintaining the speed (so long as AMPS MAX aren't exceeded).

-Mike

 

[bikeelectric]

How do you disable the cruise control once it is activated ? Push the momentary button again ?

Mike, you are right about the cruise control.

If you hook a momentary between CR and GND you get a "manual" cruise control.
You get to your desired speed, push the button and release the throttle.

If you put a switch between CR and GND then you can enable-disable "automatic" cruise control.
Automatic is set by holding the throttle steady for the time selected with the Parameter Designer ( Auto Cruising Time )
I strongly recommend to disable it if you ride in town.

Denis

 

[mwkeefer]

The cruise control (manual activated with push button) can be stopped by any of the 4 following methods:

1.) Sending the eBrake line to Ground (happens when you pull your brake levers and a NO switch makes contact driving the eBrake line to ground (5v)

2.) Change the throttle, just start using the throttle again and it will disable the cruise control.

3.) Press the manual momentary switch again while cruise control is on and it should turn off (I haven't tested this actually but its supposed to work)

4.) Disable cruise control all together using a switch wired from the controller between the enable cruise control pad and 5v ground. You can use the black wire (5v gnd) of one of your ebrake lines for ground and simply run the single wire up from the contoroller to put a SPST push on / push off so you can turn the cruise control on / off like a car.

If you are going to run the wire to the front handlebars to some special controler box, I would suggest taking a while... examine the other features available (maybe even wait for my users manual for the 9fet infineon + the config software) and instead of a single wire... you may find you want 3 more for electronic speed control, 1 to disable all eBrake / regen brakeing, 3 for led lights to display status, 1 for a security interface, 1 for a speedometer, etc).

All I am saying is plan it first. If you use the ebrake gnd line (tap it wherever is close) then just run wires from the pads on the controller up to the appropriate places on the handle bar mounted control box. A favorite of mine for interconnecting my "black box(s)" to the infineon is simply embedding a CAT5 jack in the infineon and in my interface / controller box. Now I use flat, white cat5 cable (hides so much neater) and since you can get weather resistant boots for cat5 cable and silicon is always available you end up with a modular, easy to repair and maintain weather resistant solution that looks uber clean!

Hope it helps

-Mike

 

[woobins]

Mike,

I went ahead and built an adapter cable and am having a blast playing with the programmable settings on this 9 fet infineon board. Regen is especially neat considering my beach-cruiser only has a rear coaster brake and I can now supplement that by using the front motor to slow the bike down. If you don't mind, I have a few questions based on what I've observed so far.

The regen feature seems to be a bit unpredictable as to how much stopping force is applied at different speeds -- at low speeds it's barely perceptible. You made a reference earlier to the regen being based on the current that's passing through the controller. IE, the higher the amperage, the more noticeable the regen effect, and vice versa. Is this correct?

When regen is activated via the ebrake, the 9C motor exhibits a pronounced shudder/'growl' effect, similar to how it sounds when accelerating at low speeds under load. I'm assuming this is normal, if not a bit unsettling. Correct?

Currently, I'm running at 48v using 4 12ah SLA's. What values would be appropriate for this setup? Right now I'm using these:
PhaseCurrent: 70
RatedCurrent: 35
LimitVoltage: 42.5
Tolerance: 1.0

I'm tempted to bump the rated current up to 40 and phase current to 100 but I don't know what kind of strain this would put on the SLA's. I also wonder if that LVC is overly conservative.

Do these controllers have a thermal protection mechanism? I have mine mounted inside a fairing where it gets limited airflow. I'm planning on drilling some ventilation holes but I'm still curious if it will shut itself off if it overheats.

Given my current battery pack, would it be worth my while to upgrade the fets to 4110s? Or should I just wait until I eventually purchase a better pack, buy a 6fet board, and upgrade that one?

I'm very much looking forward to your upcoming documentation about these controllers and appreciate the current information you've provided so far. Keep up the good work! It's appreciated.

 

[mwkeefer]

woobins,
Sorry to take so long to respond but your question(s) are actually quite a bit more complicate to explain and took some thought as to how to approach each one (because a few are related) but here goes nothing:
Congratz on the adapter cable - makes life so much more fun (and easier), still haven't found a way to flash the controller while it's powered up by battery but am looking for a way to reflash the eeprom memory while in transit. If I can figure it out then I can build a small "profile programmer" where you store your ASC profiles from your PC and then use this device to select the profile number and then reprogram on the fly. I am afraid this may not be possible without serious mods as the infineon appears to require only 5v power input to accept programming over serial link.
Regen-
Unpredictible, actually it's quite predictible but I won't go into the math here... a better way to say it might be un-reliable or inconsistent.

Good news first:
The growl is normal and to be expected, shudder... not so normal (I don't think, Ive never had any) so please, PLEASE check your axle bolts and make sure they aren't loose.

Bad News:
Regen is never going to perform the way we would hope (not with these controllers, not with any in so far as I know and perhaps not in the near future) unless someone comes up with a better and more predictible braking force combined with actual recouperation durring the ebraking process. I have some ideas but testing is slow going as I am using LiPo right now and as you will soon see, regen with lipo could be DANGEROUS!!! (SLA should be okay, not dangerous in any case).
Here's the breakdown:
Each time the motor (hub) receives 1v of input from the ESC it goes a certain RPM. All brushless (and brushed I think) DC motors have this parameter known as Kv (the number of RPM per volt of input). These 9C hubs have kV (under load) somewhere in the area of: 9 RPM per Volt
So using the above information… if you were to run using a pack of 43.5v the expected wheel motor (wheel) RPM would be: 43.5 * 9 = 391.5 RPM @ 43.5v.
Just a little more math:
Lets figure out how fast (under load but not accounting for grades or head/tailwinds) we would be going if our 26” wheel was turning at 391.5 RPM @ 43.5v input (assuming we have sufficient current available to get up to top speed) we could calculate MPH using the following formula:
RPM / 336 * Wheel Diameter
*** Please remember your wheel diameter will not be exactly 26”, the tire height and tread will effect it’s diameter by increasing it slightly. The best bet to perform this calculation is to use string or tape to surround the wheel + tire outer diameter and then measure it. My personal rollout is 27.75” but for the purpose of simplicity we will use 26” as in our example.
First: 391.5 / 336 = 1.165
Second: 1.165 * 26 = 30.29 MPH
Now we know our top speed with a 9kV motor (Its close, I have to go through the logs and calculate it better) is 30.29 MPH while providing it with 43.5v
What you have been experiencing is the exact inverse of the above example, what I mean is:
When using a motor as a Generator (IE: Regen Breaking) each kV RPM produces 1v output. So lets use our same 30.29mph speed from before. This means, yep you got it… the motor is pushing 43.5v back out down the ESC and to the batteries.
I can tell you that at approx 8mph your regen is putting out a maximum of 17-18v
To wrap up:
1.) If your pack voltage isn’t below the regen voltage then there will be no regen taking place, effect of braking or recharging of battery.
2.) If your speed is crazy (35-50mph) your regen is still going to follow the kV and will peak far, far above your nominal voltage. This could blow up lipo batteries and possibly damage SLA.
3.) When your pack voltage is below the regen kV generated (not too far below) and when the regen voltage is not above your pack high voltage cutout, you will get the best braking and regen effect. The lower the battery, the faster you are moving the better the regen will be.

I know that’s not all the details but I hope it explains the behavior you observed and were so kind to bring to light here. I had written a bit about this in the manual but, you helped me to think of better ways to explain a rather complicated interworking of mechanics and electronics.
4x12AH SLAs (what are they wheel chair batteries), how much do those weigh??? WOW (Seriously they are fine so long as you can handle the weight and live with the range limitations)
Without specifics on your SLAs (AGM, Make, Model, DataSheet) then I have to assume normal SLA operating parameters. That said your pack is really:
55.2 v fully charged (13.8v per battery or 2.3v per cell)
48.0v nominal (12v per battery or 2.0v per cell)
40.4v low voltage cutout (10.1v per battery or 1.683v per cell)

With those numbers you could lower your LVC a bit for sure, I would go from 42.5 to 40.5 with a .5 tolerance. This should cutout between 40.5 and 41. Probably won’t amount to a hill of beans but, that would be optimal (if the setting isn’t there, hex edit the parameter designer or PM me and I will build one for you).
In regards to the current settings –
This again depends on the SLAs your using and their discharge rate at what C? Most SLAs of 7-12AH are rated at ½ C or lower. The more you ask of SLAs the more the voltage drops and the faster they drain, you shouldn’t expect the rated 12AH of capacity drawing 35 Amps… more likely 7-8 AH if your lucky.
I wouldn’t push the SLAs further than the 35A you have them at now that is 5-6C already.
You could raise your Phase Current a bit, I find 2 times + 20% to be right in the ball park for perfect (I am using lipo but I don’t think this effects the Phase Current… this is the current the motor sees). I would go to 85-90 (don’t expect much performance difference but try it).
Thermal Protection – Yea they blow up, burn out FETS (The power transistors) and could short out your power pack. No seriously, there is 0 thermal protection (as it seems with most controllers, power packs and motors) although we are addressing this in our new pre-controller unit also.
Forget upgrading the FETs with your power pack… I am running 12S2P lipo at 45/100 (Rated Current) with an LVC of 38.5 and a tolerance of .5. My batteries will give me about 20-24mi without pedaling in my hilly, high traffic area and will average speed between 22 and 30mph. Best part is they don’t even warm up (maybe on a horrible hill climb) and they weigh about 5 lbs total. All this I run on the STOCK Infineon with only calibration of the shunt (search for methods posts on calibrating the shunt in these controllers), replacement of the existing wiring with higher temp, silicon jacketed, oil and gasoline resistant 700c 10G AWG for the Battery and Motor Phase Wires, Added loom for electronic 3 speed and a few other simple mods.
In other words, you probably don’t need the upgraded FETs… what are your goals, targets, budget, etc… maybe I can point you in the right direction?
I will get the documentation finished as I can… logging data requires riding and since my days (and nights) are spent working or taking care of my children, sometimes things take a while.
I have written software to import the eagletree logger data into SQL Server 2005 and several stored procedures to model the data and extrapolate important and relevant data. This will speed the documentation along nicely as I can refer to my logs using SQL to find averages and specifics needed to get the technical details accurate.
PM me if you need help figuring out brand / model specific LVC and C ratings
-Mike

 

[mwkeefer]

Hello all,

I thought it was time to post a status update and some further information regarding these controllers.

First, I have obtained from Jason at e-bikekit.com several controllers for testing and modification.. currently I am testing them in sequence without modification while logging using eagletree. This data will provide us all with baseline differential of the performance and setttings of the "stock" infineon controller from e-bikekit.com.

Once I have worked through the testing phase I will post my results and all the data (full logs, route maps, etc)... and the extrapolated constants (assuming there are constants - JK).

In the meantime... I figured I should update some real world info:

I have upped my lipo pack to 15s which I normally charge to 62.55v @ pack... this gets my cells to 4.18v. My testing is still done with a single 20/30C (100/150A) 5AH Turnigy Pack (leaving me a 5ah pack for the return ride). My maximum speed has gone through the roof at this voltage level... I have clocked (GPS verified) over 45mph using a fresh pack.

The more important piece of data here... this is a STOCK, uncalibrated infineon I am using now. I have upped the amps to 32.5 and changed the block time to 5 and the speed %2 = 120% but otherwise no hardware mods, no fet upgrades, no caps... no enhancements!!!!

This is good news... and we now know that the stock infineon will hold up to a 15s lipo voltage with NO ISSUES WHATSOEVER!!! I have logged 200-300 miles in this configuration (different current limits) at up to 30a nominal and have found this combination not only uber stable but F A S T!

Off the line launching will stil be best usng a speed % of 100% but, the 120 is a WOT top speed boost in phase triggering within the motor, essentially it skips phase triggers thereby causing an advance of timing simliar to old fashioned distributor on a gas engine would advance the timing. This was very difficult to log (very fast switching) but I got it = )

I have not given up the manual (I am still working on it) and I am still around for those who need assistance or whom need questions of specification answerted... if I can, I will = )

-Mike

 

[dnmun]

why not just put both the lipo in parallel, rather than having to swap out?

your 15S will keep you under the 63V of the caps, but several people have used 72V directly without replacing the caps and they did not blow immediately. so you have alot of space. i ran my 9FET controller at 90.6V stock freewheeling unloaded for a few seconds and nothing blew up. 80V for a longer time with no problem, but not under load.

but of course if you depend on the cap to not blow, then it will, the inverse murphy's law.

you are doing good here, that is fast.

 

[mwkeefer]

dnmun,

I realize I can parallel my packs (and will eventually) but for now I conduct tests and loads which very may well kill my controller... or atleast smoke a cell. That said, I keep them out of parallel right now for Testing Purposes and to ensure I always have a good pack for a home trip.

Since I limit my distance using the first pack and my single pack far out performs the LiFePo4 10AH pack that comes with this kit (lipo = little to no voltage sag under 22-28a loads on single 5AH pack) I know I will always have sufficient power to get home.

Really it is just the engineer in my mind telling me to keep some headway (no pun intended) for the return voyage = )

Besides... you may not realize but I am pulling 2-4 test runs per day of 5ah (about 12-15mi normally) under different configurations, etc. Add to that the controller array I am trying to test for a baseline of real measurement on these and the 4 5AH 15S packs I have at my disposal go fairly quick.

I am also doing test runs with one of Jason's batteries at each suitable configuration testing voltage sag, top speeds, accelleration, 0-10, 0-20, 0-30 speeds, etc.

I guess it is more than just a reserve for the trip home, I don't have time to test a full 10AH usually... I am too busy testing other things or doing real work of engineering or writing software = )

With regards to the 72v nominal pack that Jason runs or the 90v you have run for a short while... Long term exposure to > 63v will kill this controller Under Load. Underload is the key, for freewheel testing anything may work for a few moments but ? Why push it when all you must do to handle 72v nominal is really replace the caps to 100v (or 75v but if your replaceing them anyway... go 100v). Jasons controller survives his 72v chopper because of these mods you did for him and also because his LiFePo4 packs wired in Series have such extreme voltage drop under his constant full loading (his CA said 24-26A but I am guessing it's not caibrated because I rode it and the launch isn't heavy enough for 25A at these voltages...). That said his chopper is C O O L and fun to ride, his new one should be PHAT!

In either case, please don't let me stop anyone from testing (when you can take the $$ hit) as I would be trying crazy voltages like 80-90 (I have a pair of 5.2AH 73.5v nominal liions in my segway... would love to velcro those on my ebike in parallel) but I can't afford to blow up the controller(s) as only 2 are actually my own, the rest are on loan or I have to build for people and I think we know just based on the component that at 63v +-10% (I think thats the caps in these) then really you could pop one at 57v in theory and if you add murphys law, well I see you noted that in your post = )

I must admit this is fun, going near 50mph (dual GPS + rpm confirmed 49mph - not a burn out!)
was crazy until I bought new glasses... these are sun glasses, night glasses and rain/fog glasses from my local Performance Bicycle with pop-out lenses. Before these my eyes were tearing so badly at speeds that I had started wearing my motorcycle helmet - full face. Now with the new glasses 40-45 is comfortable and the wind stays out of my eyes.

** I do not encourage others to go this fast!!! I have been on motorcycles (fast) since I was 10 years old and freestyle bmx, serious offroad trail riding mtn bikes and I have a solid understanding of the equipment / mechanics of speed + the ability to react fast enough for this... not everyone does have these reflexes and so... please don't try this at home!! I really am an untrained professional = )

-Mike