PROJECT no 4 Doc's Giant DH comp 2003

[John in CR]

AJ, Mr. I have 10 fans in my computer, but I don't use air cooling, it's quite simple. Doc doesn't hassle and mock others, so others don't hassle and mock him.

 

[liveforphysics]

Hey Doc,
Since you already have your windings split in half, what about putting a controller on each one? Wouldn't 2 18fet controllers be better than 1 36fet?

I think you're on to something here John. As I've been learning (the hard way), current handling does not scale-up well. On paper, if 1 fet is good for 50, 2 fets is good for 100, 3 fets is good for 150 etc. In practice, the distance in the legs and board traces between them, along with various other factors makes for a plague of paralleling switching issues and unbalanced inductive path issues etc etc. Better to have more groups of FETs each switching less current, rather than having 1 larger single group switching all the current. It would be a pain to run 6 wires out of the hub, but I think it may be the best solution for milking every last bit of power potential from off the shelf controllers.

 

[AussieJester]

And AJ.. Doc and Methods ..LFP.. and a few more don't qualify to be among us " normal " power hungry bastards.. they are in a class of their own lol..

oooooh i see forgive my ignorance, i forgot they are "special" people that can do no wrong...and YPedal Man, how you can read my worklog and miss so much info startles me sometimes I guess i need to start using BIG BOLD TEXT and ridiculous devil smiley faces everytime i post something that is even slightly out of the ordinary But i guess seeing i use only use a 6kw 'toy motor' and push it to 10kv its meant to not blow up and should expect people to continually question the reliabilty of my bike. frock you all....

As for you John, i am not the only one i see getting a hard time, if you dont have frock motor and blow something up according to you and a handful of others its unreliable what a crock of shit. I might give people the occasional stir John but i always offer words of encouragement too, your just arrogant wanker who constantly stirs to get a reaction from me and never offers a word of encouragemen or advice, as i said yesterday, done with you John Methods had you pegged months ago i should of listened, a know it all that is always right and can't be told a thing...

KiM

p.s still waiting for proof of that data on hair driers John, more facts you pulled from you usual reference ..i.e you ass.

 

[Doctorbass]

Hey Doc,
Since you already have your windings split in half, what about putting a controller on each one? Wouldn't 2 18fet controllers be better than 1 36fet?

I think you're on to something here John. As I've been learning (the hard way), current handling does not scale-up well. On paper, if 1 fet is good for 50, 2 fets is good for 100, 3 fets is good for 150 etc. In practice, the distance in the legs and board traces between them, along with various other factors makes for a plague of paralleling switching issues and unbalanced inductive path issues etc etc. Better to have more groups of FETs each switching less current, rather than having 1 larger single group switching all the current. It would be a pain to run 6 wires out of the hub, but I think it may be the best solution for milking every last bit of power potential from off the shelf controllers.

Yes I agree about that inductive and current path and other gate capacitance..problem

I might have an idea about the 5 times blow i had... Could it depend a bit on the phase wire lenght or wire gauge??

Because on that DH ebike I first used the spare controller i had but i assembled it with .. less beefy traces and also have 12 gauge silicon phase wire instead of 10 gauge like other controller...

So you believe that a 36 fets controlller could be a desing problem due to bigger inductive problem?( let say the 36 fets controller PCB is very similar than the 24 fets.. but just wider...)

If the stars are perfectly aligned, one day.. i could have acces to a 36 fets controller 8) and test it for you

My DH setup is ... PERFECT for that!

 

[liveforphysics]

I think a 36-fet should be better than the 18fet, but it certianly couldn't perform close to as well as 6 x 6fet controllers if you know what I mean. Though it doesn't show it on paper, in practice it seems to be diminishing returns as huge groups get paralleled up IMO.

Bigger FET packages in smaller numbers is the key to building a reliable monster controller IMHO.

 

[Doctorbass]

Monday ( tomorrow) it will be raining.. one rare time this month 8)

I have many thinks on the to do list this week:

That thread is like becoming a blog to me about that DH ebike... :lol:

I'll bring with me that blasted controller and will clean all the mess and black stuff with isopropylic alcohol and will take a closer look to the PCb traces and smt to see if i can still recover it a 6th time !! :wink:

And when i'll finish at work, i'll dissassemble the 5303 and will investigate on that serie mode that causing vibration problem. if the relay are ok.. then i'll just solve the problem and leave it with 2 electric speed.. otherwise i'll remove that PCB and relay prototype and will wire the motor normally in serie or parallel mode...

note: ( i'm still not decided on wich mode i'll chosse.. 5306 or 5303..

From what i see after more than 200km of use:

I used the parallel mode for around 90% time.

5303 PROS:
-I really love the acceleration still ALOT present at high speed!!
-at 80km/h i still feel i could do a whelee!!
-the powerbans seem linear
-Top speed very interesting !! :wink:

CONS:
-CONTROLLER BLOW EASY !!! 5 time inn a weel !!! with 18 fets set at 150A
-My average Wh/km increased by 30% for the same speed when rising around 50kmh !!
-Motor heat too easy at low speed ( cross country trail..) can reach 180 celsius in a minute!!!

5306 PROS:
-Torque still impressive without any worrie for the controller fets !!
-Phase wire dont heat up
-less volt lost in the phase wire from the controller to the motor
-Motor cooler at low speed.
-better Wh/km at most used speed(50-60kmh)
-easy to use

CONS:
-lack of top speed.. I still would continu to accelerate at 70kmh but BEMF = V batt
-Acelleration from 50 to 70 is less impressive


About the battery pack ,
-i'll finish that important setup before my garage go in flammes!
now all 3 parallel 6spack ( 15Ah 6s) will be shrinked together to limit the movement inside the battery case.
-i'll put some thin layer of FR4 between each of the 4 6s groups like i was supposed to do...
-I'll install a permanent DB36 pins for the battery balancing that will go to the 4x6s charger to end the last % of charge and ocasionally fully balance the pack to 4.200V (but i still charge to 4.15)

i'll make a box that include the 4x celllog8 as alarm and monitor.

I want also to have a way to completly disconnect the lipo to the celllog8 when i dont charge or when i dont use the ebike. for that i have to clarify if i can or not just disconnect the pin 1 of each of the 4 celllog8.. or if i better need to disconnect all pins 2 to 9.
( i bought 8x small 4pst relay for that. ( i want to disconnect them to avoid unbalancing the pack or draining the cells... I know its just 8mA per celllog8 but I want these mAh for riding.. not for supplying these things!

Switch /breaker install

Since each time i charge i disconnect and reconnect the battery ( serie100V use and parallel 50Vcharge), i need to ass a switch with a precharge resistor ( from now the switch is the fuse that i remove and re install after each use and i connect a 100ohm resistor before re installing the fuses to abot that little blast..

I have in stock 3x 40A siemens industrial breaker rated 380-415V AC.. and i wonder if i could use them in DC ?? for 100V... according to the normal 32V DC and 125Vac rating on many components.. If i follow the rule.. I could possibly use these 415v rated for 100Vdc. They use a spring loaded contact wich help for reducing DC arcing...

the thing i'll try tomorrow is to test if the current share between all 3 paralel breaker dont pop before reaching 120A.. because i feel thata little unbalance in all 3 breaker could make one of those to trip and open.. before reaching 120A total..

A friend of mine here brang me his 7.5Vdc 200A psu for testing that.. i'll see tomorrow of all 3 breaker can hold 100A 120A.. 140A.. and how many time they hold that ..etc


I swaped the controller from my mongoose to the DH bike

I temporarely installed the controlelr of my mongoose on the DH ebike.. BUT I REDUCED THE CURRENT TO 30A rated current and 65A phase current from now.. I dont want to blow something again!!

What is strange is that on EVERY infineon controller that i have without the shunt modified, the max continuous current on the CA is always twice the max rated current( battery current) of the infineon!!!! icecube57 is also experiencing this...

It's like the calibration on the infineon controller are off by 200% ??? do we are the only one to experience that?



... and... as welll so many other things to finish !!

Oh.. i'll save that text,, last time i wrote a post like this.. IE crashed and i lost everything!

Doc

 

[Doctorbass]

Bigger FET packages in smaller numbers is the key to building a reliable monster controller IMHO.

Yes! for sure.. it's interesting.. but :| .. does they exist at the same price of a 36 fets controller?

I need IGBT!!. lol...

following this.. i bet i'll melt that X5 before the end of 2010!

Doc

 

[jsplifer]

Wow! High Current Carnage. Gnarly.

I'll say it again, real need for modular, scalable controllers.

Hope your downtime is minimal Doc.

 

[liveforphysics]

Bigger FET packages in smaller numbers is the key to building a reliable monster controller IMHO.

Yes! for sure.. it's interesting.. but :| .. does they exist at the same price of a 36 fets controller?

I need IGBT!!. lol...

following this.. i bet i'll melt that X5 before the end of 2010!

Doc

The bigger FET stuff generally costs more in a big way. The single reason TO220 is so popular is price. It's extremely cheap, if a manufacture wants to do a super cheapy low power version, they can source super cheapo TO220 fets for under a dime each, and have it work fine for the typical 15-20amp controller setup that 99% of the market is happy with.

 

[Doctorbass]

Now i'm testing industrial Siemens AC breaker to use them as DC breaker.. they are 380-415Vac rated so i just hope that at 100V Dc they will not have arcing problem if they open... i'm testing their rating by applying different current to tehm and measuring the time they take to pop. i'm using a 0-5v 0-200A EMI-Lambda DC power supply to acheive that. ( got for 80$ on ebay!)

From now with a G40A breaker it take:

5 minute to open at 60A
1 minutes to open at 70A
20 sec to open at 80A
and 2sec at 200A
and.. probably 0.01sec at 10 000A short circuit like it occur with our controller blow.. hey.. that's 2000 joules.. oopss.. that might be a little under 10000A finally.. :lol:

20 sec at 80A is perfect cause i'll use two of those in parallel so i'll have 160A for 20 sec on the drag racing lane.

I tested that they have 2 miliohm each ( same as 2x 4110 mosfet in parallel...)

Power dissipation inside it at 200A for that 40A rated breaker is 28Watts during these 2 sec.

 

[John in CR]

Doc,

Something Fechter suggested for my low turn count motors is to put some inductance on the phase wires to help protect the controller. Do you have any ideas for cheap toroidal cores that will work with our 10ga phase wires?

 

[liveforphysics]

Doc,

Something Fechter suggested for my low turn count motors is to put some inductance on the phase wires to help protect the controller. Do you have any ideas for cheap toroidal cores that will work with our 10ga phase wires?

http://cgi.ebay.com/Toroid-Core-Powdered-Iron-1-8-47mm-inductor-x1-/180515863525?pt=LH_DefaultDomain_0

These should do OK for filtering the PWM if I'm understanding inductor core function at a basic level. They will have losses increase with commutation frequency though, so at higher speeds, they might start getting hot, but they are pretty big, so they likely can take it. I'm not inductor expert or anything, but I do think these would help smooth the PWM spikes into something much flatter. Inductor core selection is actually pretty damn complex. lol. I've been trying to get a firm handle on them for over a year now, and it still feels like a weak understanding.

http://powerelectronics.com/mag/power_inductor_core_material/

http://tritium.fis.unb.br/Fis3Exp/fcim.csdc.com/fcimis/compid/ind/toroid.html#irontoroid

 

[Mark_A_W]

Doc

My 18 FET Infineon "Crystalyte" controller also draws double the current that you program with the PC.

But my 18 FET Infineon "soft start" controller was accurate.


I don't know why.


And umm...I think your controller is dead now


Good to see the DH frame is working out for you, I love mine, it's rock solid.


Mark

 

[Doctorbass]

Doc,

Something Fechter suggested for my low turn count motors is to put some inductance on the phase wires to help protect the controller. Do you have any ideas for cheap toroidal cores that will work with our 10ga phase wires?

In my lab i have alot of inductors and alot of stuff to build some ( past car audio 160dB+ and SQ sound system stock) but what is strange is that on DC brushed controller they specify tio have the shortest wire lengh between the motor and the controller to avoid inductance problem!!! :|

now with our BLDC controller we would add inductance between motor and controller???

that seem to decrease efficiency ALOT !!

so why using a 5303 if we need to add external inductance... would it be better to use 5304 with higher voltage?.. you know what i mean?...

that is becoming really wierd!

 

[Doctorbass]

Doc

My 18 FET Infineon "Crystalyte" controller also draws double the current that you program with the PC.

But my 18 FET Infineon "soft start" controller was accurate.


I don't know why.


And umm...I think your controller is dead now


Good to see the DH frame is working out for you, I love mine, it's rock solid.


Mark

Now 3 of us can now CONFIRM that the real measured max current and the max current programmmed of the instant start 18 fets controller is twice the supposed value??? ( without modifying the original shunt)

 

[John in CR]

I thought it was just on the battery side wires that short cabling is beneficial, because the inductance essentially makes it harder for the controller to draw the current it needs by slowing the flow down a bit. On the motor side I thought the goal of short as possible phase wires was only to reduce the wire losses. My understanding of the increasing the inductance on the phase wires is that it's a relatively low loss way for the controller to see what looks like a higher turn count motor and effectively take some of the current limiting burden off of the controller's back. If this view is incorrect hopefully someone will step in to explain.

I have a handle on why a toroidal inductor is better for this purpose, but I'm unclear whether we'd want a ferrite core or if an iron powder core like the one Luke linked on Ebay is fine. I have a container of ferrite that we collected at the beach, so mixed with some epoxy I could make some toroidal cores. I was also wondering if I could take a speaker magnet and demagnetize it with temperature to make a good core. What about just using that ring shaped magnet as a core?...I'm really unclear about what would happen running current through a wire wrapped toroidally around a ring magnet.

Yes, for Doc just going with a 5304 is an easy solution, but my motors only come in one winding, so I don't have that luxury. Plus for mine, going from 2 turns to 3 with a rewind would be too big a jump.

John

 

[bigmoose]

Luke I hope the good Dr does not think I am hijacking his thread for a little talk on magnetics... but here is a starter link with some design equations and free software http://www.smps.us/magnetics.html You are on the right path with your previous links. You are trying to get a little inductance, but with less resistance than would be in the motor. As a guideline I would want to try to get say 80% of the phase inductance of a motor that I know does not fry the controller with the new (low phase inductance) motor phase inductance + the external inductor. You want to keep the external inductor out of saturation at max current. You can take it close to saturation, like 85% of the way there. You may find that your inductor wants to be wound with copper tape for the currents and inductance you are after. The pwm frequency of the controller will determine whether to use iron powder or ferrite to keep hysteresis and core heating losses in check.

 

[Seriousknot]

Electronic Noob here, just had a thought. What if Back EMF is causing 'crosstalk' in the phase wires and thereby overloading controller? Solution would be to shield individual phase wires. Or possibly braiding the three phase wires to attenuate signal. Either way is cheap and simple to implement. Couldn't hurt anything.

 

[icecube57]

The more and more I look at my 5303 the more I just want to change it to a 5306 and breathe a sigh of relief. Minimum cost alternative and my controller will like me for it. And I can run harder with higher voltages without getting any backtalk from the controller. Controller and phase wires would run cooler. Good torque off the line. Safer Top Speed. Lower wh Consumption. Anything is better than the 80wh/mi I was getting. Running a lower voltage would help the controller out alot but a 5303 at 44v is ok but you desire more also... in the torque department. A higher voltage makes a difference in pulling power at the top end of the power band. But you take a hit on consumption and exceeding the rating of the fets. As Doc puts it you need a larger controller than the 18fet. But if you do get a controller that can deliver you have to deal with the heat saturation of the windings and how quickly you would loose performance. I noticed it alot running mine at 66v. Is still fast but after the 1st minute or two you feel your performance start to suffer. I think the 5303 and 5302 in a "normal" wheel size is just going to be a whore. Its going to realize you have a bigger controller and its gonna say give me more until it burns itself out. The inductance is that high. Its a messed up cycle the motor gets saturated you stay on the throttle long enough to get the same performance. But yet if you back off and go partial throttle you risk blowing the fets. But you know the fets cant keep that high current up forever. It makes you wonder if the 4110 are rated for 120A in an ideal situation. You have an 18fet controller you supposedly have 360A phase current capeability yeah i know there are variable blah blah blah. But we are exceeding the capeabilty of the fet packages then what is the motor really pulling. Especially in that first few seconds of acceleration and what amperage is it trying sustain. Battery current tells us nothing we need some actual phase amperage numbers to better understand whats going on. Adding inductors. Wouldnt that increase winding length and resistance to kinda simulate a higher winding motor but has alot of losses.

 

[John in CR]

It seems like there's something about those 5303's that make them harder to drive than my 2 turn motors with a lot more stator slots and magnet poles. Yes I blew an 18fet 4110 controller with a 24" wheel, but Luke and I abused it for almost a week, and it didn't fail until after I really heated it up racing a small moto for over a mile up a slight grade, and then shortly afterward did it in with partial throttle up an 8% grade. Keep in mind that all up, me and bike, was about 100lbs heavier than the good Doctor Bass and his rig. Also, I've never gotten only 80wh/mile. I don't know if it's the angled laminations or overlapping windings, but there's something different, because I'm pretty sure a 170lb person couldn't have blown the controller on mine without bogging it down on a pretty extreme hill, and they would enjoy some extreme performance even at the 20s LiMn we were running. Luke and I were running a sealed motor, and sure it got toasty but the windings are still pristine despite running currents that made the long 10ga phase wires pretty hot. The one detriment I know of is that mine cogs enough to require someone with tree stumps for legs to pedal only, an d it looks like Icecube didn't have much trouble pedaling the X5.

 

[icecube57]

One thing that I notice about my 5303 after a hard ride and the motor is hot the drag/cogging increased. Kinda opposite of what you expect. So at first when its hot its hard to pedal but as it cools down it gets easier and easier. When im babying the motor it seems like i can coast and pedal with little effort. I lay into it really hard a mile with 4-5kw you feel a drag and you slow down alot quicker.

 

[icecube57]

Doc

My 18 FET Infineon "Crystalyte" controller also draws double the current that you program with the PC.

But my 18 FET Infineon "soft start" controller was accurate.


I don't know why.


And umm...I think your controller is dead now


Good to see the DH frame is working out for you, I love mine, it's rock solid.


Mark

Now 3 of us can now CONFIRM that the real measured max current and the max current programmmed of the instant start 18 fets controller is twice the supposed value??? ( without modifying the original shunt)

This really makes me wonder. If the settings in the controller are about half of the actual. Meaning if I set 45A battery current and its peaking at 80-90A. Which ive seen on my stand alone. Same applies with a lower setting of 30A battery current setting yeilds about 50A-60A battery current. It makes me wonder if we set the phase to 120. Is it putting out 240A phase current peak. Even if its not double but close to it then i can easily see how the controller and the fets are being is being exceeded in the case of a 12fet. If one 4110 mosfet is good for 120A "on paper" with proper heat disappation then in a 12fet controller you are meeting and exceeding the thermal capeabilities of the fet packing by using the 120A phase current setting. It can only do that much current with proper cooling. I also understand the load is spread amongst the # of fets you have in you controller per phase. So a 12fet has a theoretical 240A phase current capacity in a perfect world. Lyen was telling me the stock values for the 72v 45A Infenion with stock shunt is 70A phase and 28A battery. This means the controller is running at 60% of what what its designed to do with phase current theoretically being 140A peak. This is a good safety margin. I think pushing it to 80% capacity would be pushing the reliability margin. 96/35 would be the new 80% capacity.
My old agressive settings were 115P/45B.
With the 2:1 programming ratio the old agressive settings on my controller makes 95% capacity being used. To close to the edge with the theoretical 230 peak phase and 80-90A battery.
The adjusted 80% makes it 192P and 70B Which would give you 96P 35B programmed software values
The adjusted 70% makes it 168P and 63B which would give you 84P 31.5B programmed software values

These values were created using the max current for the 4110 mosfet
2,3 or 4 depending on # mosfets = "X"= capeability of mosfets in phase bank times "X"= Mosfet Current Capacity you want to use divided by 2 = (suggested program values)

 

[Seriousknot]

Hey Doc,
How's the prognosis for the controller? Do you think she can be resurrected? Here's something you might find interesting....
http://jnaudin.free.fr/steorn/indexen.htm#EMF a way to cancel Back EMF!!

 

[John in CR]

Hey Doc,
How's the prognosis for the controller? Do you think she can be resurrected? Here's something you might find interesting....
http://jnaudin.free.fr/steorn/indexen.htm#EMF a way to cancel Back EMF!!

LOL! Steorn LOL!

 

[bosnimf]

i'm using 6 x zippy 6s 5000mAh in 24s 3p ( 15Ah)

Doc[/quote]

could you give a link of the site you bought your batteries from?