izeman
1 GW
btw: HOW THE F*CK do those aluminium bars work that kelly uses? i have never seen them before. where is current allowed to flow? i can't see any connections or pcb traces. how can they transfer that much current?
New Kelly pseudo FOC controller series
- Thread starter crossbreak
- Start date
izeman
1 GW
don't know the exact type, but they are TO-220 cases. so any clamp designed to TO-220 should work i guess. thanks for googling it.Kodin said:
Um... I can see an issue here. Imagined it differently.
When you drilled, you have possibly shorted copper layer to aluminum plate. If not shorted, there is very small gap now, that could short in future. Not sure what to suggest now.
When you drilled, you have possibly shorted copper layer to aluminum plate. If not shorted, there is very small gap now, that could short in future. Not sure what to suggest now.
izeman
1 GW
yes. after i was finished i thought the same. it's not shorted now but the air gap must be minimal now.
i may remove that one FET and see what i can do. that's why i asked how those aluminium copper thing works.
i may remove that one FET and see what i can do. that's why i asked how those aluminium copper thing works.
izeman
1 GW
i guess there is a layer of copper on aluminium. but how is it isolated?
to me i looks like there is NO gap between the copper layer and the aluminium bar. it looks as if it was the same material.
to me i looks like there is NO gap between the copper layer and the aluminium bar. it looks as if it was the same material.
izeman
1 GW
so any idea what i could do? i feel really stupid now.
Probably what you could do is desolder that fet and then somehow remove copper around the hole, without damaging the isolation layer. And then enlarge the hole in FET. This would move phase voltages further from aluminum plate.
Just not sure how the copper can be removed safely, maybe just cut it with sharp knife half way heep, around perimeter, and then peel it off. It will peel like a normal trace from PCB.
One more hard to do thing actually desoldering the FET, as you will have to heat all that chunk up. Maybe with gas torch...
Just not sure how the copper can be removed safely, maybe just cut it with sharp knife half way heep, around perimeter, and then peel it off. It will peel like a normal trace from PCB.
One more hard to do thing actually desoldering the FET, as you will have to heat all that chunk up. Maybe with gas torch...
SplinterOz
1 kW
circuit said:Nice case you have got there.
Am I seeing this right, are those aluminum-PCBs? Nice and expensive move by Kelly.
I would buy it, but no support for proportional brake is making me go for sabvoton.
By the way, can someone already describe the smoothness and amount of noise and torque in comparison to Sabvoton?
Sorry I have not got around to it. I did try to power up my 2nd hand Sabvoton (haven't even done that yet) and it wouldn't turn on. gotta check wiring etc... but in all honesty with the rest of the stuff I have on it may take me a week or two before I get it done.
izeman
1 GW
i guess i got it working properly again. scraping away the copper took about an hour, but it was doable. removing the FETs from the heatsink was a real PITA. i needed to use a torch to heat it up as my 80W soldering iron with a 5mm tip was not NEARLY up to the job. no idea how they do it when they build them?!
so this is how it came out. i scraped away a bit, and drilled a chamfer in the connecting side of the FET. so there is NO WAY there can be a short. i hope the whole thing survived the massive heat input. i will NOT repeat that for the other side
View attachment 1
so this is how it came out. i scraped away a bit, and drilled a chamfer in the connecting side of the FET. so there is NO WAY there can be a short. i hope the whole thing survived the massive heat input. i will NOT repeat that for the other side
View attachment 1
izeman
1 GW
the FETs used are MAGNACHIP MDP 1932izeman said:
http://www.magnachip.com/product/down.php?filename=MDP1932.pdf
izeman
1 GW
coming along nicely. i decided to skip the additional screws all together. it was a bit too risky for me. now i at least can be sure that there will be no shorts. and i tested for shorts off course before adding power to it.
this is what it should have looked like from the very beginning. kelly are you listening?
this is what it should have looked like from the very beginning. kelly are you listening?
this power stage looks really STRANGE
good to see you have no short. hopefully there also will not be one under voltage.
for instance on my lyen 18fet i measured no short, but under load it had and blew. i have tightened the screws for the fets to much so there was very small gap left to the heatsink. multimeter did not beep.
i only want to bring this up just because i had this problem. i do not want to make you frightened
good to see you have no short. hopefully there also will not be one under voltage.
for instance on my lyen 18fet i measured no short, but under load it had and blew. i have tightened the screws for the fets to much so there was very small gap left to the heatsink. multimeter did not beep.
i only want to bring this up just because i had this problem. i do not want to make you frightened
izeman
1 GW
that's why i didn't install the screws. i should have left it as was. but i don't think that there can be a short now as i scraped away a lot of copper and with the chamfer in the backside of the FETs heat sink those parts can not touch.
i already connected it to main power and it was working ok.
i already connected it to main power and it was working ok.
izeman
1 GW
this is not necessary for my needs. the controller has a max phase current of 240A for 10s, which is WAY more than my motor can handle. so i probably will have to reduce it. i hope to be ready for another test ride tomorrow.
izeman said:coming along nicely. i decided to skip the additional screws all together. it was a bit too risky for me. now i at least can be sure that there will be no shorts. and i tested for shorts off course before adding power to it.
this is what it should have looked like from the very beginning. kelly are you listening?
View attachment 1
Yes,I am listening.But this is really not what Kelly expected.The controller is opened and modified.
The big KLS controller KLS-8080I included the shunt,fuse and hall device for measuring.riba2233 said:Now you must only find a way to bump up the current limit. There's no classic shunt or hall, right? It reads Vdrop from mosfets? Maybe change resistance in that circuit
These devices are not included in min case type KLS-S
izeman
1 GW
you sound upset.fany said:
there are no real modifications. i only transfer the pcb to another case, as the case it came in, is way too bulky to mount it on a bike. i was suggesting that you could use a similar case and showed a way how to do it. i also used other wires as the ones installed are hard to route in the bikes frame. so no real big deal. i shouldn't have drilled the heat sink so. that was stupid.
if i order the next controller from you i'll do the very same thing again. and i'm definately not the first one to do it
izeman
1 GW
i did some test ride. it's really super smooth and quite powerful.
it just seems to me that it doesn't run up to full speed. could be an almost empty battery as well.
so i put it on a bike stand and rerun calibration. calibration for the MAC seems somehow different as well. it really doesn't matter how phase and halls are connected. the controller finds the correct combination. but it took 5 tries to get a proper calibration. the first times it just stopped with a "hall error" problem. i tried again and again and finally it went through.
if i run the bike on the bike stand with wheel lifted from the ground there is another funny phaenomenum: i go WOT very fast and it goes up to eg 68 km/h and then drops to 62 km/h and stay there. when i release the throttle is jumps up to 68 km/h again and then drops to zero. i can monitor that with the kelly pc program as well. i guess this is because of the PID controller trying to prevent rpm overshoot, but why is it not reaching top rpm possible?
i will now charge the battery and make some comparison video.
it just seems to me that it doesn't run up to full speed. could be an almost empty battery as well.
so i put it on a bike stand and rerun calibration. calibration for the MAC seems somehow different as well. it really doesn't matter how phase and halls are connected. the controller finds the correct combination. but it took 5 tries to get a proper calibration. the first times it just stopped with a "hall error" problem. i tried again and again and finally it went through.
if i run the bike on the bike stand with wheel lifted from the ground there is another funny phaenomenum: i go WOT very fast and it goes up to eg 68 km/h and then drops to 62 km/h and stay there. when i release the throttle is jumps up to 68 km/h again and then drops to zero. i can monitor that with the kelly pc program as well. i guess this is because of the PID controller trying to prevent rpm overshoot, but why is it not reaching top rpm possible?
i will now charge the battery and make some comparison video.
izeman
1 GW
you asked for it, you got it
i made 4 videos, 2 with the old KBS 48101 controller, and two with the new KLS-S 6018S controller
both starting from the same point, going WOT (as good as possible, going uphill with an iphone in one hand). the #2 videos also start from the same point.
you can hear the noise difference VERY clearly. it's even more obvious when you hear it "live". all you hear now is gear noise of the motor and the chain and tires. the motor is almost totally silent.
with the old controller you can hear this high pitched whine which is totally gone.
old controller part#1
[youtube]JWwqtk_0Fj0[/youtube]
new controller part#1
[youtube]mE6A-7sjBS0[/youtube]
old controller part#2
[youtube]qVI9FdKCDT4[/youtube]
new controller part#2
[youtube]qVI9FdKCDT4[/youtube]
i made 4 videos, 2 with the old KBS 48101 controller, and two with the new KLS-S 6018S controller
both starting from the same point, going WOT (as good as possible, going uphill with an iphone in one hand). the #2 videos also start from the same point.
you can hear the noise difference VERY clearly. it's even more obvious when you hear it "live". all you hear now is gear noise of the motor and the chain and tires. the motor is almost totally silent.
with the old controller you can hear this high pitched whine which is totally gone.
old controller part#1
[youtube]JWwqtk_0Fj0[/youtube]
new controller part#1
[youtube]mE6A-7sjBS0[/youtube]
old controller part#2
[youtube]qVI9FdKCDT4[/youtube]
new controller part#2
[youtube]qVI9FdKCDT4[/youtube]
izeman
1 GW
and now for the bad news. if you watch the videos you may notice that the NEW controller takes more watts to maintain a slower speed. so something must be wrong.
in video #1 the hill is not as steep as in video #2. so the motor reaches it's maximum rpm.
with the OLD controller it reaches 40km/h and uses 2.220W (more or less)
with the NEW controller it reaches 30km/h and uses 3.000W.
so two things to think about the new sine-wave controller:
.) why does it limit it's RPM?
.) why does it new SO many WATTS to maintain such a slow speed?
EDIT: i found setup screenshots from the first identification process, and they look different.
NEW one (bad one) is left, OLD one is right. i can't really remember how i ran before as i didn't really test it.
in video #1 the hill is not as steep as in video #2. so the motor reaches it's maximum rpm.
with the OLD controller it reaches 40km/h and uses 2.220W (more or less)
with the NEW controller it reaches 30km/h and uses 3.000W.
so two things to think about the new sine-wave controller:
.) why does it limit it's RPM?
.) why does it new SO many WATTS to maintain such a slow speed?
EDIT: i found setup screenshots from the first identification process, and they look different.
NEW one (bad one) is left, OLD one is right. i can't really remember how i ran before as i didn't really test it.
izeman
1 GW
no. i haven't done a no load test to compare. but i can make a video to show the strange behaviour of limited rpm that raises when you release the throttle.
i guess auto detect is a bit off, but i don't know of any parameters to manually changed. they are greyed out and i have no description what they mean.
i guess auto detect is a bit off, but i don't know of any parameters to manually changed. they are greyed out and i have no description what they mean.
izeman
1 GW
@fany: this is mainly asking you, as you are the professional who knows how those controllers work
i got hope again
i tried again doing "auto config" but it always failed. so i swapped two phase wires and it instantly found a working combination. i tried it again and it worked again. strange.
but the problem i had before now is somehow gone. it now uses less power to climb hills, and the strange effect of reving up when releasing the throttle is gone.
but there is still some issues: it seems to me that the motor could run faster, but something is stopping it from doing so. when i drive in first gear on the flat the power is pulsing very hard. hmmm, i hope this is not my CA. i will repeat the test without the CA, and a throttle directly connected.
TPS is also a big question mark. WTF does TPS stand for? this term is used all the time and not explained - well at least i didn't find it.
if i want to set throttle dead zones: TPS deal low and dead high are the values to adjust. it's a percentage value. but what's the base value? 5V? so 20% = 1V??
@fany: you may please adjust the manual, as it is not described clearly i guess.
also what does "(6) TPS Low Err" stand for? is it the low error throttle voltage when the controller is turned on, or is this monitored all the time? is the percentage value the same as for the dead zones? reference value = 5V?
the decription "Hall active pedal, if lower than the value, report the fault of TPS Type" says NOTHING to me. what PEDAL? do you mean throttle?
i set max speed to 4000 which is default. in the monitoring tab it shows speed of about 2000. is this the speed it is referring to? so if i set max speed to 1000, the rpm in monitoring tab should not exceed 1000. correct?
if you say "(3)Accel Time: Accel Time, the time of TPS Torque from 0 to max, accuracy 0.1s, 5 is equal to 0.5s. Range: 1~250" you mean that with the default value of 5 it takes 0.5s to ramp up torque (which is equivalent to phase amps) from zero amps to max amps. correct?
even if you say that the PID parameters in "(8)Torque Speed" for KP and KI will be removed in the future, i still would like to ask you to explain what those really do. i think that (as far as i'm familiar with PID tuning) a too high P value will lead to oscillation of the rpm and the I value is for damping the over shoot. but this could be totally wrong and you implemented it differently, or i just understood it wrong.
those are the only values to tune the controller, so it may help to explain them. what does "speed err limit" do?
ok. thanks for following my post so far and for reading it. i hope you can answer and explain most of my questions.
thanks!
i got hope again
i tried again doing "auto config" but it always failed. so i swapped two phase wires and it instantly found a working combination. i tried it again and it worked again. strange.
but the problem i had before now is somehow gone. it now uses less power to climb hills, and the strange effect of reving up when releasing the throttle is gone.
but there is still some issues: it seems to me that the motor could run faster, but something is stopping it from doing so. when i drive in first gear on the flat the power is pulsing very hard. hmmm, i hope this is not my CA. i will repeat the test without the CA, and a throttle directly connected.
TPS is also a big question mark. WTF does TPS stand for? this term is used all the time and not explained - well at least i didn't find it.
if i want to set throttle dead zones: TPS deal low and dead high are the values to adjust. it's a percentage value. but what's the base value? 5V? so 20% = 1V??
@fany: you may please adjust the manual, as it is not described clearly i guess.
also what does "(6) TPS Low Err" stand for? is it the low error throttle voltage when the controller is turned on, or is this monitored all the time? is the percentage value the same as for the dead zones? reference value = 5V?
the decription "Hall active pedal, if lower than the value, report the fault of TPS Type" says NOTHING to me. what PEDAL? do you mean throttle?
i set max speed to 4000 which is default. in the monitoring tab it shows speed of about 2000. is this the speed it is referring to? so if i set max speed to 1000, the rpm in monitoring tab should not exceed 1000. correct?
if you say "(3)Accel Time: Accel Time, the time of TPS Torque from 0 to max, accuracy 0.1s, 5 is equal to 0.5s. Range: 1~250" you mean that with the default value of 5 it takes 0.5s to ramp up torque (which is equivalent to phase amps) from zero amps to max amps. correct?
even if you say that the PID parameters in "(8)Torque Speed" for KP and KI will be removed in the future, i still would like to ask you to explain what those really do. i think that (as far as i'm familiar with PID tuning) a too high P value will lead to oscillation of the rpm and the I value is for damping the over shoot. but this could be totally wrong and you implemented it differently, or i just understood it wrong.
those are the only values to tune the controller, so it may help to explain them. what does "speed err limit" do?
ok. thanks for following my post so far and for reading it. i hope you can answer and explain most of my questions.
thanks!
