Infineon controller REVIEW+tests with 18x IRFB4110

[Doctorbass]

Hello Robin,

Here is what you asked for :wink: .. 2 minutes spent..

(i also include the 846 MCU pinout i found on the french http://cyclurba.frforum.. it seems they observe us 8) ..

Silicium is on both forum and also transfer infos..

Doc

 

[liveforphysics]

This is the first time I saw this thread. Awesome work Doc! I love it! What a great forum!

 

[geoff57]

hi
Doc great picture of how the controller is connected up.
One or two things though
1: On the brake section EBS+ that is NOT designed for the battery to be connected to it it takes 12v+ it was designed for a vehical with a 12v brake light system and so you could tape into that insted an ebrake lever.

2: The SM pad you have marked Digital Speed Indicator is under review at the moment it is what you say but not the way we expected it to be a how the output format has yet to be worked out since this was intended for a cycle analyst i may never be useable for that as a direct connection.

3: This is not one you have wrong this is an extra feature we have finaly worked out how it works, thanks to Mark at Team Hybrid wanting a pedelec controller (which is why we know about the pedelec sensor connection on other threads) Keywin sent Mark a controller with a pedlec sensor lead itted and a throttle with a green button built in, the button was a momentary make type (he also sells one with a red button that is identical but the button locks on then off) on reciving the controller Mark found it was set to 72 v not the 36v he had asked for so it was sent to me for convertion with the throttle which Keywin told Mark was to change the speeds but as normal did not say how .After I recived the controller and opened it up I saw where the pedelec sensor was connected and duplicated it on a spare controller I had not wantin to touch the controller sent untill I had been in touch with Keywin. The first thing I found out is that the pedlec sensor is a latchin hall sensor rotation sensitive, it will not work in reverseand the speed the motor go's at is the max speed of the three speed settings dependent on how fast you pedal, pedelec and throttle can be fitted to a bike at the same time the pedlec overides the throttle though this may be an advantage but have a feeling this may cause problems i am setting it up on a bike for road testing soon. The second was a bit mor dificult this green button throttle and how could it posabley change the speed setting we had been using a 3 posion switch (ON-OFF-ON) to change speed settings so far, I cannot remember if I worked it out on my own or if Keywin helped but either way if the cntroller is programed to cycle mode not switch and a mometary make switch is connected to X1 then on closing the switch the controller wil change up a speed, from speed 1 to speed 2to speed 3 then loop back to speed 1. Of course the switch that keywin intended we use is the green button on the throttle.

sorry the last one was a bit long just an explination of how I got there as well as what it does.

Geoff

 

[Doctorbass]

I forgot to mention that the PCB pic with descriptions of the connections was sent by Keywin 4 days ago. so i dont know if it's more revelant about my controller pcb or the overall 18 fets pcb... just to clarify :wink:


Geoff.. I see that i'll need to read alot of tread about that controller before to put power on it !... there is so many (under review) ( still R&D desing)... (experimantal options... etc).. I just want to be sure how to use it !.. It's really imnportant to try step by step.. and not to understand all at the same time!!.. so i'll take time to try it.

I just received my nine continent motor... and i'll try it on it soon.

I think i'll finish the copper plate on monday... ( drilling every holes for mosfet screwS, polishing it crystal clear, cleaning it... buying mica instead of using the grey poor insulator..etc)

Anyway.. thanks for your clarifications on the PCB :wink:

Doc

 

[wanders]

(i also include the 846 MCU pinout i found on the french http://cyclurba.frforum.. it seems they observe us 8) ..

Wow, I've been looking for a datasheet / manual for the 846 for the longest time, and all I've seen is that Chinese powerpoint presentation, which unfortunately does me no good. Does this mean that the 846 is similar or a minor variant of the XC866 (which is the label that is on your pinout)? I want to explore the possibility of doing regen braking a bit differently than the Shenzhen Sucteam design. Data on the XC866 is relatively easy to find.

 

[wrobinson0413]

.

 

[geoff57]

Hi
I have a bit of bad news I opened up the 18fet I was sent today and noticed no speed "blue wire " coming out of the controller just before I opened it so I searched the board for the SM circuitry, it is missing the tracks are there just not the components, I used a 12fet board as a comparison. I then looked at a picture of your board and found the same compnents missing they are named the same on both boards just have not been fitted.
componets are
R4
R5
R14
R35
Q1
some of them are surface mount but if they are realy wanted they could be fitted.

Geoff

 

[Doctorbass]

Hi
I have a bit of bad news I opened up the 18fet I was sent today and noticed no speed "blue wire " coming out of the controller just before I opened it so I searched the board for the SM circuitry, it is missing the tracks are there just not the components, I used a 12fet board as a comparison. I then looked at a picture of your board and found the same compnents missing they are named the same on both boards just have not been fitted.
componets are
R4
R5
R14
R35
Q1
some of them are surface mount but if they are realy wanted they could be fitted.

Geoff

Oh... so what these missing component usually do?

Could it work without them?.. I mean.. what is their use.. ?

i'll see for that tonight..

Some pcb pics are from Keywin and some are from me.

Anyway i still have noty installed any mosfet and no power on it

Doc

 

[geoff57]

Oh... so what these missing component usually do?

Could it work without them?.. I mean.. what is their use.. ?

i'll see for that tonight..

Some pcb pics are from Keywin and some are from me.

Anyway i still have noty installed any mosfet and no power on it

Doc

you only have to worry if you are going to connect the cycle analyst hall line to the SM pad and at the moment we have doubts on how well that works philf brought them up, if you are not useing the SM pad dont worry about the missing components.

BTW who fitted those caps the 1000 uf 100v ones the 18 fet controller I was sent had only 470uf 100v caps in it the fets are 4310's, also what values are the three power resistors on your boards,mine came fitted with 3x 1K5 ohm resistors.

Geoff

 

[wrobinson0413]

.

 

[Doctorbass]

Excellent work WRobinson!!!!.. that's a real support for tthat controller!!

I still have some IR2101 that Bob sent me with the fets.... (the drivers that goes into the crystalyte).. I wonder if replacing the actual discret component driver with them (soi 8) placed on a pcb would be helpfull...?

Geoff57, i'll check for thge resistor value tonight

Doc

 

[wrobinson0413]

.

 

[Doctorbass]

Thanks Wrobinson!.. it was designed for the 4310 ( higher rds on)

Geoff.. here is the value: 680ohm :wink:

Doc

 

[ZapPat]

I also added the mosfet gates of the IRFB4110's to load down the driver, and noted that the simulation is showing about 3.8usec turn on time and about 1.2usec turn off. Again this is probably ball park, but very informative. What it says to me is that your controller is going to have relatively high switching losses compared to some of the other controllers that might be using higher performance drivers and layouts.
Wayne

Wow! 3.8uS ON + 1.2uS OFF times sound really terrible to me! I aim for about 1/10th of this transition time to keep switching losses down (like ~200nS). With 3 paralleled 4110's your resistive (conduction) losses should be fairly low, so the switching losses would be quite high at low-medium current use. At really high loads though I guess the conduction losses would catch up, since they go up with the square of the current, whereas the switching losses have a fairly linear relationship to current.

Wayne - Very good work on figuring out that discreate high side driver circuit! Personnaly I think I would not use such a circuit, specially if all it can do is in the micro-second range when switching such large FETs. I have a question for you about scoping out switching waveforms: You mention using "an isolated scope or high voltage differential scope probes, and a current probe" for checking waveforms. Right now I use regular scope probes rated at over 100MHz, and set them at "1X" attenuation to avoid picking up too much magnetic junk in the probe (I also use the small ground spring to minimize probe loop area). I can use this at low voltages easily, and use offsetting in my scope to check out higher voltage signals (like the overshoot caused by the ringing). I am aware of the scope saturation danger of doing this, but I keep my settings low enough so that it doesn't seem to be a problem. Anyways... I was wondering if you could suggest what kind of probe(s) I could find and wouldn't cost too much (used or whatever is OK) so that I would be sure that the ringing I'm seing is really the FET signal and not just induced junk in the probe? I know that setting my present probes at "10X" makes for terrible readings and so always use them in the lower impedance "1X" position... but this does also make me wonder how much junk I'm still reading, and what options would be more accurate for such readings? Thanks for any ideas/suggestions about this!

Pat

 

[Jeremy Harris]

I agree, the switching times do look really slow. I think it might be an idea to re-check the circuit, to make sure that it's correct, as there is one bit that looks a bit awry - C2A. This capacitor is fitted across the gate-source connections of the FETs, so is adding to the gate capacitance. We don't know what it's value is, but it's certainly going to slow the switching speeds down.

I will try and take a look inside my small Infineon tonight and see if I can puzzle out the circuit it uses (tricky, because it's tightly packed and there's a lot of wires in the way).

Jeremy

 

[Doctorbass]

Guys... you seems to have better skill than me about analog discret device...

I understand what you explain and do, but my scope is not the best for testing(old heatkit..16Mhz) and RCA probe.. and feel better than me for trying things on that board.

So i would offer to one of you to send you the board + mosfet for testing and try to ameliorate it like the way you think... Than, i could build the final version with the other board and being sure it is ok.

( have two identical board)

Doc

 

[ZapPat]

Guys... you seems to have better skill than me about analog discret device...

I understand what you explain and do, but my scope is not the best for testing(old heatkit..16Mhz) and RCA probe.. and feel better than me for trying things on that board.

So i would offer to one of you to send you the board + mosfet for testing and try to ameliorate it like the way you think... Than, i could build the final version with the other board and being sure it is ok.

( have two identical board)

Doc

I would be tempted to take you up on your offer, but as Wayne mentioned the FET layout is a bit poor for three paralleled FETs, so this will limit your maximum switching speed. There might be room for some improvement, but personnally I wouldn't go so far as to try fitting on an IR FET driver chip.
Anyways, I would rather continue concentrating on my own controller project, and then send it to you so you can compare it with this beefy 18 FET controller this summer. There are soooo many interesting projects to do, it's hard to keep focus on the most important ones sometimes!
Maybe Wayne might be interested? He knows his FET stuff for sure!!

 

[wrobinson0413]

.

 

[wrobinson0413]

.

 

[Jeremy Harris]

OK, I've just spent fifteen minutes pulling apart my controller (a baby 6 FET model, which looks like it has a similar drive circuit) and I've tried to take a close up photo. I'm posting it here before I start digging around trying to cross check the circuit that Wayne has very helpfully already provided.

Here you go:

Hope this is helpful. I'll try and probe around to see where some of the hidden tracks go. I can confirm that the two through-plated holes to the left of the two 2k2 resistors to the right of D2A are connected to the big ground track underneath the board. D1A is marked T4 (it's the diode above and to the left of the orange 100nF capacitor) and Q0A (immediately above it) is marked G1. The double through-plated holes above D0A (next to capacitor C0A) are connected to the controller supply track under the board (+12V - confirmed by looking at the LM317 regulator voltage setting resistors). The high side FET gate is the right hand track under the "Q2A" text, the source is the big track to the left of this (the drain connection is underneath the board).

Hopefully our ideas will converge on a common circuit configuration later!

Jeremy

 

[wrobinson0413]

.

 

[Doctorbass]

Well, just for fun I went to IR's website and downloaded the spice model for the IRFB4310 and ran it in my simulation. It looks like the smaller miller capacitance on the 4110 relative to the 4310 negates the larger input capacitance, so the turn on time is approximately the same. It also appears that the driver has a bit more trouble sinking the additional current from the miller effect on the 4310. So over all, you probably will not see very much difference in switching speed between the two mosfets, and if anything maybe a slight improvement in the 4110 for turn off. Not what I would of expected at first glance, but it does imply that the layout is the limiting factor in how fast they can switch the fets since it was tunned up for a slow switching speed.

I fihure that it's a great news?
since it was built oem with the 4310...?

 

[solarbbq2003]

i posted on other infineon threads
http://www.infineon.com/cms/cn/product/channel.html?channel=db3a30431b0626df011b315ce2eb6a27
lots of info on the chip, most of it goble-de-gook to me
someone should download and put on es forum the files incase they disappear

 

[ZapPat]

Well, just for fun I went to IR's website and downloaded the spice model for the IRFB4310 and ran it in my simulation. It looks like the smaller miller capacitance on the 4110 relative to the 4310 negates the larger input capacitance, so the turn on time is approximately the same. It also appears that the driver has a bit more trouble sinking the additional current from the miller effect on the 4310. So over all, you probably will not see very much difference in switching speed between the two mosfets, and if anything maybe a slight improvement in the 4110 for turn off. Not what I would of expected at first glance, but it does imply that the layout is the limiting factor in how fast they can switch the fets since it was tunned up for a slow switching speed.

I fihure that it's a great news?
since it was built oem with the 4310...?

Yes indeed this is good news for you! After looking at both data sheets yesterday morning I thought this might be the case, but Wayne's simulation is better proof than a hunch. Now to know how accurate the simulation is we would have to scope out the real waveforms.

It's a bit challenging to check out full load FET transition signals on some of these ebike controllers though, since they often have a cut-out mecanism to shut off the outputs when a stall is detected for something like one second (or at least the infinion-based designs seem to do this). A digital storage scope helps with this problem, but it still is an annoyance.

PS: Thanks for the probe docs, Wayne... much good info and examples to give me an idea of how much distortion can occur. I'll consider getting an active probe soon.

 

[wrobinson0413]

.