72V upgrade for 36V 9 FET Infineon controller

[dnmun]

i finally got around to this after jason at e-bikekits.com called and asked me if i would upgrade one for him. he sent me one which has a different LVC set to 31.5V versus the 30.5V of the others i have. so i upgraded 2 instead.

the older 30.5V LVC controllers have a 330 ohm power resistor in the input to the voltage regulator and this new one has a 200 ohm resistor. other than that there is no apparent difference. even the surface mount resistors in the LVC bridge are the same.

however i did find that the controller labeled with the 31.5 LVC ended up having a 55V LVC after i added the 1.6k ohm bypass resistor to the bridge, and the older model with the 30.5V LVC preset ended up with a 51V LVC after i added the same 1k6R bypass resistor. no idea why.

i used 2 270 ohm 2W power resistors on the newer model controller and the 330R and 200R resistors from the original 2 controllers on the older version. so 540R and 530R input power resistors was really the only difference between them. about 2W total when running under load.

i am fairly certain of my chart from knuckles data since i was able to correctly estimate i would have a 37V LVC when i added the 5k1R bypass resistor on my other controller i modified for the 48V nicads. it was exact, right at 37V! i had expected these 2 new ones would end up at 53V each, so not sure what caused the discrepancy.

i have now determined absolutely that the idle controller current is only 53.3mA and this climbs to about 61.7mA under full load. if i had known this initially, i would have increased the input power resistors from 530-540 ohm to 600ohm because the 12V rail was pulled very high by the 86V input voltage setup. in that case where the idle voltage was 86V, the input to the LM317T was about 54V, with the 12V rail at 19V and the 5V regulated output at 6V. but when it was running under load, the input voltage to the regulator was around 51V, with the 12V rail is around 17V and the 5V output regulated to about 6V but the test points were reading 5.7V, maybe there is a zener clamping the voltage output from the regulator. i decided to leave it since the pack will likely rapidly drop into the 76-77V range where the input to the LM317T is about 48V, and the regulated output drops to 12V and 5V.

at the low end, around the LVC, the voltage regulator still has about 24V input. 12V and 5V on the busses. with 600 ohm in place instead, at 51V for the LVC, the voltage regulator would see 14V input under load.

i soldered up the shunts when i had them exposed and was soldering the bypass resistor in place. i had a bad moment when the LVC bridge capacitor lifted with my soldering iron and i had to reposition it and resolder it in place along with that end of the bypass resistor.

i think the large 100V1000uF caps will make these 75V FETs work at the high voltages of the 72V lifepo4 packs so i am hopeful this is a simple solution, otherwise i will upgrade to 4110s to keep the heat down if these pop.

total cost to upgrade 2 controllers = $4.30 for 6 big caps and 4 resistors.

 

[rkosiorek]

now i am totaly confused by that post. too many resistors boards and parts mentioned to follow.

just where exactly is that LVC bridge. what are the board labels for the parts (R??)? and what should be the values for which resistor for let us say a 38 -39V cutout?

or is it better to change the LVC in software? which is easier for a novice to do?

rick

 

[dnmun]

the LVC hack adds a resistor in parallel to ground in the resistor divider that takes the signal level to the 846. i agree with jeremy that the software modification of the LVC is the proper way to do it. i can't program the 846 yet, that's next.

if you have a BMS on your lifepo4pack, there is no need to worry about LVC in any case.

i just ordered some 300 ohm 2W resistors because i think 600ohm is a better choice for the input power resistor. 63mA at 600R is 2.4W so 2 resistors to split the heat is better. imo

i can take a picture of my chart and you can try to read off it the resistor value for a specific LVC. all taken from knuckles thread, i think for dummies thread. each horizontal line is 39,49,59V from knuckles. obviously exponential and the tail is limitied by the 30.5V of the infineon programming.

the main reason for doing it software is the risk of lifting the cap on the bridge, i did it already, major anxiety putting it back down with my shaky hands. it stuck to the tip of the soldering iron because it got too hot.

 

[rkosiorek]

a new member was asking about setting the LVC lower for his 36V pack. he wanted to drop it to 30.5V

so i guess the correct answer is the best way is to set it in software.

i'll pass that on to him.

rick

 

[rkosiorek]

never mind, you guys are already taking care of him n a different thread.

rick

 

[dnmun]

you can't lower the LVC below the factory setting using this hack with the added resistor.

but i have a 48V nicad pack so for the controller i rebuilt to use on the 48V nicad i reset the LVC to 37V so i could use all the pack, with a little sag at the end. i think that was 5k or 5.1k ohm. on the chart. but the LVC on that infineon was set to 30.5 at the factory.

 

[dnmun]

upgraded 3 more of the 9 FET infineon controllers i got from jason at E-Bikekit.com and this time i built it more solid and used a bigger cap on the S/D buses. i did not modify the LVC this time because he can use the BMS on his 2 36V lifepo4 E-Bikekit battery packs to shut off the controller if the battery gets drawn down to where the LVC on the BMS cuts in and can shut off the controller by sinking the ebrake line.

i soldered up the shunt even more this time and actually pulled the 14G stranded ground power cable over the top of the shunt and soldered the negative side of the input caps right on top of the shunt, which essentially is bypassed now. i used a 100V2200uF cap on the S/D buses and the same pair of 100V1000uF caps on the input. built up the bus and phase traces with solder and added a little short piece of bus wire to build up the jumper from the big 2200uF cap to the S/D buses. this is like a big coulomb accumulator soaking up the voltage spikes and softening the blow each cycle on the FETs to extend the reliability and prevent breakdown at the limit of the p75n75 FET's specs above 75V, while pushing 40A.

i changed the power resistor to 2 300R 2W resistors in series on the underside along with the bypass shunt for the voltage regulator. soldered all 3 in place and then clipped the legs and mounted the caps on top after. i used just a dab of silicon rubber this time, enuff to keep the caps from moving but so that i could cut it out later if needed and bend it up out of the way to program the infineon if needed in the future. or do other repairs.

thanks so much to knuckles and richard for documenting the board so i could attempt to do something like this and feel confident it will run for a long time at 72V lifepo4 without burning up. lotsa lotion helps too, you can see my liberal goop of silicon heat sink compound too, real insurance against cooked FETs, imo. along with fat caps.

 

[malcav]

So after looking at what you have done to the board i am wondering why you left the shunts on? Why not just drill a bigger hole in the board and put the neg and capp leads strait to the board and replace the traces with copper wire? And if i might ask one more question what would you say the max amps you can get out of this board?

 

[dnmun]

i made these for jason, don't know what the max amps would be but assume around 40A. i still do not have 72V battery yet.

now i would not solder up the shunt but instead reprogram the controller to allow higher current rating without soldering up the shunt.

also if i do this again, i would go with just one big cap on the input and not two.

i am about to get a chance to do this for an 18S lipo, and will use just one of the 100V1000uF caps on the input. and adjust the input power resistor to reflect the 75V peak and 37V low.

 

[texaspyro]

also if i do this again, i would go with just one big cap on the input and not two.

Two smaller caps in parallel are almost certainly better than one. They should have at least half the ESR of the single device.

 

[malcav]

would you mind pointing me in the direction of the required info to install the whips to program the micro contrlers