Golden Regen controller problems

[biohazardman]

I have 36 and 48V GM regen controllers. Both have the nasty habit of toasting the 1.5KE62A diodes on the back of the board. They are in series with a resister red, black, gold should be 20 ohms or is that .2 @ 5%? Meter reads them as 0 - .2. Any ideas for an upgrade to stop the insanity?

 

[dnmun]

what are all the other wires?

i am guessing that the diodes withstand some of the flyback voltage and then at a certain point the regen voltage on the highside source is so much the current gets dumped through those resistors? a little load 27R 5W wirewound resistor to soak up the current spike?

show the other side too, close enuff to see where the traces run. they look identical.

want caps? i got caps coming, big caps, just what you need. hehe.

 

[biohazardman]

what are all the other wires?

i am guessing that the diodes withstand some of the flyback voltage and then at a certain point the regen voltage on the highside source is so much the current gets dumped through those resistors? a little load 27R 5W wirewound resistor to soak up the current spike?

show the other side too, close enuff to see where the traces run. they look identical.

want caps? i got caps coming, big caps, just what you need. hehe.

Other side as requested.

 

[dnmun]

i got this board from duane today, already pulled the zener diodes and the resistors off. it was .2 ohm, except a bunch were burned open, some of the zeners were burned out too. confirmed the zener is 62V, and found this for the output FETs.

the digikey entry http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=IRF2807-ND

and the data sheet http://www.irf.com/product-info/datasheets/data/irf2807.pdf

i would recommend increasing the zener to 68V and raise the resistor to 5W from 3W, and increase the resistance to .3 ohm.

you run this at 48V lithium or max 60V then the 63V 470uF caps i have on my 9FET infineon board would help too. when i pull them off my board, i can add them to your board.

maybe if the voltage spikes can be partially absorbed in the extra caps right off the S/D busses, and then if we add 6V to the peak before the zener avalanches, maybe that will reduce the severity of the breakdown, and add a little resistance just to reduce the current about 30%, maybe only 50% (when you consider the current at the higher zener voltage) of what it normally it would be at the 62V setting. but just guessing about that. but it could help reduce the rate at which it burns up, and i don't think it will take out your FETs because they are 75V FETs. 13mohm, so they will get hot at high current.

i also wanna remove that stupid relay they use for the LVC, just hardwire it, and use the LVC on the BMS to protect itself and the controller, no need to waste even a tiny little bit of current on the coil of that relay. thats why the GM controller has so much voltage regulators, overkill compared to the infineon. wasted current in the controller is just more heat.

this controller will work at 72V too i bet, just raise the input resistor another 200 ohms, use the 100V caps, and i think those FETs will still work, but don't know until you try it. then you can swap to the 4110s like you were talking about.

about $30 for parts of 20 sets of diodes and resistors, enuff to rebuild the 3 controllers you have, and 2 sets extra for spares. so that would give you 3 regen 48V controllers. mod the 36V to 48V at the same time.

i recommend spares, because it turns that out if you prepare in advance and save the spare parts for an eventual failure, then your hack will never fail. the inverse murphy. lose the spares, it fails in no time.

 

[biohazardman]

But But But that is way to much werq and I have a Shenzen so instead you reset the LVC for me and I will soon be on the road again. Near exact replacements, for the diodes and resistors will werq on the old GM controller for to have as a back up unit less werq and cost that way. Thanks much Dennis

Duane

 

[dnmun]

i replaced one of the power resistors on the shenzen duane had, and changed the LVC from 59V to 40V.

i used 13kR for the parallel value, but his controller had different values for R3 and R73 from what richard had in the initial shenzen thread, ended up using a 13kR in parallel to get to 40V LVC.

duane's shenzen had 2 - 510R power resistors and i replaced one with a 15R 1W to reduce the voltage drop since i am converting his 72V shenzen controller to run at 48V lifepo4.

View attachment 5

anyway, i am convinced that the reason that goldnmotor used the 62V zener as the transient voltage suppressor was to protect the 2 - 63V input caps when the large flyback voltage spikes were generated during the regen and they chose to limit the spikes to 62V because the boards they had to use to do the regen on were built with the 63V caps, 1000uF.

i am certain that they decided it would be cheaper to use the old boards and add the zeners to protect the caps, rather than replace the caps, plus the 100V caps won't fit standing up. these 100V1000uF caps are 18mm diam, 41mm long. pins are 7.5mm apart too, versus 5 for the 63V caps.

i replaced the input caps with 100V 1000uF caps and i actually wonder if the controller might actually run again now. but i doubt it. i think the voltage spikes were high enuff they can still damage the FETs whether the hi side flyback FETs or the driver FETs i don't know which. but since the FETs are rated to 75V, i think it is possible to raise the cutoff as high as the 72V zener in 5W, but i think it would be easier to live with myself if i blow up his controller if we only raise the zener voltage to 68V, change the power resistor which has to sink the load to .27 ohm, 5W.

does anyone have an idea if this will work? or can we raise the zener right up to 72V? or will it run now without the zeners replaced?

View attachment 4
View attachment DSC00238.JPG
View attachment 2
View attachment DSC00241.JPG

 

[dnmun]

i guess this is called a clamping circuit, where the zener breaks down at the 62V to clamp the voltage between the source and drain to 62V but the specs lead me to believe it doesn't fully clamp until up to 85V, so with that in mind, it went back together with the 62V zener and a .22ohm ceramic 3W resistor. no shrink since it is not touching anything, except a few pieces of shrink where it was close to the phase wire legs. so i am putting it back stock, almost.

 

[317537]

Sorry to dig this thread up but I have one of these controller without the resistors.

What I had in mind is I built a cap bank with 5x 1000uf caps off board but 5 inches down the power lines, close to the controller. Put a few TVS's and a good wad of 1ohm 10 watt wire wound resistors that = .2 ohms.

would it give me a chance to back off the accelerator and stop the bike before it took out the fets?

If is was to add 6 more tVs's on the cap bank would the resistors even be needed?

Is the bank I plan to add close enough to work?

And would the caps pull all the regen away from the tvs's over the fets?

BTW These controllers blow when the power is disconnected while bike is in transit. I want to make it rock solid. any other suggestions would be awesome.

 

[dnmun]

les, the further away the caps are from the di/dt source, the less it can soak up the spike. that's why they put the big caps right on the S/D busses to absorb the current that can't get through the inductance(ie. the S/D bus).

i think duane quit using the regen just to be on the safe side, and i wonder if it would have been ok to clamp it at a higher volatage and just swapped out the caps from 63V470uF to 100V1000uF nichicons. i know the infineon controllers don't use that clamp on the S/D busses either so i think it was all done by GM to reuse the stock boards with the 63V caps installed when they did the regen and started selling the controllers as regen. bigger caps don't fit unless you do something extreme like what i did. just barely fits too.

i coulda sent you some of them too, i bot 2 boxes on ebay. i charge my cost, 30 cents each plus postage. now that's a deal.

 

[317537]

Ok I will put the caps in the box. :wink:

My cap bank that = 5000uf is solderd up on a board it fits snug inside the box, just means i have to deconstruct my controller box holder. Including the 2000uf already in there, I will have 7000uf. I can find a tap really close to the fets even closer than the caps already in there..

My added 6 extra TVS can bridge across the back of the cap bank circuit board Ive made up

Will these added TVS's and caps protect the fets when the power is cut? Meaning if there is no battery to take the regen, the caps soak up the spikes and the TVS's get hot. Will this be a better controller?

 

[dnmun]

no you absolutely have to have a battery or a resistive load to soak up the current generated by the motor, just from turning. the TVS diodes whould have resistance in series also to absorb the energy of the part of the spike that exceeds the clamp, lower resistance means it absorbs more of the spike but .22ohm is pretty low, and using a 5W instead of 3W will help it handle the load better, but 5W wirewound is expensive and hard to fit underneath the board, use a peice of plastic underneath to prevent the wirewounds from rubbing on the case and shorting out. fat heatshrink will work too just like they had originally. 2k uF is enuff caps too, not much to gain from adding another 5k, imo. the curve is very steep with distance when you look at voltage vs length of the inductor. i did add 3,000uF with the 3 caps, and i had some 2200uF caps too which i used on 3 of the 72V controller upgrades and those had a total of 4,200uF spread between input and S/D buss.

 

[317537]

Yea I understand the zener clamps and resitors and I wonder why they added the tvs's in there without the resistors in the first place.

Do the TVS's have to be right over the fets behind the boards or can I tap to them with the resistor and run them off my cap bank board hovering over top of them?

 

[dnmun]

if you don't have sink for the current, nothing matters. it will continue to burn up the FETs and the TVSs with or without the resistor, the resistor is just there to absorb the little bit of spike left over. but if the battery is full and won't take a charge, then you will blow the TVS and resistors anyway, and maybe the FETs too.

 

[317537]

I was thinking if I had 6 extra TVS's and 60 watts of resistors and used 5000uf of caps and use them like this.

View attachment idea.JPG

Picture above. The red lines are the planned tvs's tap off the out terminals from fets of the and the yellow lines tap through the resistor to the negative, or off the bus via the darlinton, those transitors burn a ship load of current in their own right.

A power cut is a short event. Like about 10 seconds maximum. I think I might be able to work this.


Wouldnt the TVS's zener breakdown and use up a lot of current as they can take a lot of heat the caps would charge up and kick the resitors in. I know its only 60 watts but what if I used a power transistor like a 100 amp darlington to off load to a power resistor bank.

Your saying if the battery cuts out then light out with these contollers.

Edit:

possibly put some bleeder resistors across the cap banks as well to make sure you dont get a series accumulation.

 

[317537]

The power shouldnt be cut but I have had fuses go on me in the past. Bumped the swtich once with a shopping bag and batt terminals shake lose over time ect. Its a safe guard but

I would like to solve this issue for the educational reasons or something to achieve in life or what ever lol.