Infineon controller REVIEW+tests with 18x IRFB4110

[Jeremy Harris]

I'd be interested to see if this has any effect, I very much doubt that it will. The series resistor is only 2k2 and looking at the chip cap it can't be more than about 0.1uF. The time constant of this is just way too small to have any noticeable effect at all, my guess is that the soft start is in software.

The odd thing is that my Infineon (a baby 6 FET one) doesn't seem to have any noticeable lag on my bench test set-up. This makes me wonder as to what the software setting may be in yours. Have you tried flashing it with vanilla settings to see whether that makes any difference? I've just spent half an hour playing with different software settings on mine to try and re-create the sluggish start-up you and Luke are seeing, but for some reason I can't seem to make mine do it.

Jeremy

 

[methods]

IIRC the effect was not that noticeable on the bench
Gotta get the current limit up really high and try it on the road.

The most noticeable delay is when I am cruising at 20mph and I wrap the throttle around to WOT
Even my wife was complaining that it was unsettling.
Just long enough delay to make you think it is broken the -BAP- it hits with 2500W.

Good idea to try some simple vanilla settings.
I did solder my shunt so perhaps it has something to do with the ratio between my battery current setting and my phase current setting
I had to leave the phase current really high while setting the battery current lower as a side effect of my mods.

I am also wondering about some of the other settings. . . I asked Keywin if the delay was software or hardware but have not heard back yet.
If the delay is software it would sure be nice to have settings for it.
The kelly has a great set of controls for the throttle input. . .
Even a check box for "Geared", "Smooth", or "High Performance" would be great on the Infineon.

Curse knuckles for requesting a "soft start"

btw: Why put a .1u cap right there if not for soft start? Doo far away from the chip to do any good as a bypass. Cant think of a reason for putting a cap right there. . . But then I am a total hack when it comes to board design. I usually borrow very heavily from others designs.

-methods

EDIT: I take back what I said. Put the hub in the stand, get it going WOT, let off the throttle, then wrap it back to WOT. There is a clear delay on mine. If there is not a delay on yours then I am very hopeful that it is just an artifact of my software settings. I am going to go play with it right now.

 

[Jeremy Harris]

btw: Why put a .1u cap right there if not for soft start? Doo far away from the chip to do any good as a bypass. Cant think of a reason for putting a cap right there. . .

Just for noise suppression, I suspect. It will take any spikes off the throttle line and give a certain amount of protection to the MCU input. The time constant for that circuit is around 220uS if that capacitor is 0.1uf. For the capacitor to be causing a 1 second delay it'd have to be something like a big 470uF electrolytic.

My test set-up is just a motor on a test stand, so I don't have the inertia (and corresponding high current demand) that you'll have. That may explain the difference.

Jeremy

 

[methods]

I pulled the chip and you were right.
No difference. I was hoping that someone figured out how to stuff 475uF into 1sq mm

I will post some results somewhere later if I ever get it figured out.
For now, if the 18 fet boards have the same delay then they are useless. I am just going to sell them off at cost.

I will try a few more of the software settings and start a thread if I get anywhere.

-methods

 

[dnmun]

if you pull the cap and discover you need it later, could be hard to solder back, so a jumper is better option in any case. hope it is software.

 

[methods]

I already pulled it and put it back.
The cap does nothing appreciable.

When I pull surface mount I just "lean" it up on one end
To put it back I heat the bottom and "lean" it back down.

At work I have to work under a microscope to haywire boards all the time.
I rarely get the design right the fist time

-methods

 

[dnmun]

deft hands, i can hardly see them, good work. its gotta be software.

 

[Deepkimchi]

Just got my 2 Infineon 18 FET controllers in the mail. That was fast. At work so I can't unpack them yet. :lol:

DK

 

[fechter]

I'd guess your delay is in software. The other possibility is something in the current limiter circuit is lagging, but that should not be doing anything if you're running no-load.

 

[GGoodrum]

So, is the current thinking that this lag/delay is something that is in all of these Infineon controllers? Has anyone asked Knuckles?

 

[geoff57]

So, is the current thinking that this lag/delay is something that is in all of these Infineon controllers? Has anyone asked Knuckles?

hi
I'm on it, the delay or soft start was asked for by mark of Team Hybrid the problem is that keywin desided to have all controllers soft start from then on, the delay setting is programed in the infineon by the factory and cannot be altered by the programming software we have.
I am contacting keywin to find out if hard start controllers can be bought again, I will let you know as soon as I get a result.

Geoff

 

[Doctorbass]

So, is the current thinking that this lag/delay is something that is in all of these Infineon controllers? Has anyone asked Knuckles?

hi
I'm on it, the delay or soft start was asked for by mark of Team Hybrid the problem is that keywin desided to have all controllers soft start from then on, the delay setting is programed in the infineon by the factory and cannot be altered by the programming software we have.
I am contacting keywin to find out if hard start controllers can be bought again, I will let you know as soon as I get a result.

Geoff

Thanks Geoff.. I think this become VERY IMPORTANT here to us :lol: :wink:
If he could only send us a chip with the modified version.. i guess some of us could discover their talent about micro SMT soldering technique :lol:

Softstart is useless for me ... I hate : SOFT..... it not correspond to my ebike caracteristics

Doc

 

[GGoodrum]

Hmm, I'm surprised this issue hasn't come up before. I'm also puzzled why this had to be hard-coded into design, instead of making it one of the programmable options. What's the point of creating a high power, high performance controller if you're going to cripple the responsiveness?

Hopefully, there will be a quick-fix, or a hack that can be done to fix this, as it would be a shame not to be able to change this on all the existing controllers out there.

-- Gary

 

[Deepkimchi]

Trying to spoil the dreary mood - here's pics of mine..

 

[geoff57]

hi
I have had a long chat with keywin on skype, I advised him to get in hard start controllers and he is going to get in 50 boards.
I did say that the 18 fet must be only in hard start not in soft start as the way this controller is being used soft start is NOT wanted, i do not know how long it will take for keywin to change to hard start on the 18 fet but hopefully he will.
As for the soft start controllers out there any ideas that can cut or kill soft start on an infineon controller it does not matter the fet number will be useable on the 18 fet because of the design of the infineon family of controllers.
Geoff

 

[Deepkimchi]

Aww crap - might as well join in... would have been nice to find this bug out about a week ago BEFORE I ORDERED MY 2 BOARDS..

I don't have the gear or expertise to remove that 846 chip....

Crap Crap Crap Crap crap CRap BJ

 

[GGoodrum]

Geoff --

What is needed, long term, is to make the soft start function programmable, with an option to disable it above a certain speed, if this is possible. Some, like "dirtdad" claim that slow speed performance is really bad without the soft start function, so we need a way to either be able to program a "happy medium", or have a way to disable softstart once under way. Maybe a better way is to leave it off, but have it switch enableable. this would be the easiest way to implement it, I think. Use one of those "3-switch" settings to control this function. That way, you leave it off unless you need to go slow.

I'm going to go see Dennis today or tomorrow, and try his bike myself, just to get a sense of how big a problem this is, and to get a "feel" of whether or not it might be possible to get an acceptable gain level that still allows slow speed performance, without killing throttle response.

-- Gary

 

[liveforphysics]

It's worst at low speeds. You give it throttle, then give it more throttle, then give it more throttle, while it's slowly inching forward, then BAM!! Flips up into a wheelie. I launched myself into the street at intersections a number of times when I just want to creep forward a foot or two.

Maybe if your bike has no power, and you can just pin the throttle anytime you want to take off, and you just wait, and then it gradually starts moving forward it wouldn't be a big deal. On a bike with power, it's a serious saftey hazard, and makes the power train feel like riding a sponge through the mud, rather than a crisp instantly reactive machine.

 

[methods]

Hmm, I'm surprised this issue hasn't come up before.

I only brought it up like 9 times in every single Infineon related thread.
Just like I brought up the subject that the shunts are too high of resistance for 50A let alone 100A+ like many are going to run.
The board needs to be populated with a < 1mohm shunt.

The good news is that I happen to know for a fact that the chips ARE reprogrammable.... Even after they are soldered to the board and shipped out.
One need only get their grubbies on the source. (or at least a precompiled binary)

More than just a few people have been beating down Keywins door about this. No doubt he will have a solution soon as Geoff says.

-methods

 

[methods]

What is needed, long term, is to make the soft start function .......
-- Gary

Until they do a firmware update you can solve the problem with hardware ($0.50 cents worth).
Buy the "Instant Start" version when they are available
Put a soft start circuit consisting of a resistor and cap between the throttle and the controller. (same as found in Xlyte controller)
Use a switch to route the signal either directly to the controller (good throttle response) or through the RC circuit (soft start as found on the Xlyte)
Switch back and forth as desired.
There are about a thousand variations that could give you whatever behavior you want

The instant response should not be an issue anyhow. . .
What is needed is throttle resolution and throttle control.
I could take off with my 100V 200A monster just as smooth as butter... but put a nOOb on the same bike and he would smash his face with the handlebars.
But then the kelly is totally programmable as far as throttle resolution and throttle response.
I nagged Keywin to copy the programmable features found on Kelly.

-methods

 

[methods]

I am dumping mine.

18 fet Infineion boards w/programmer - $72 shipped

-methods

 

[GGoodrum]

It could actually be a resolution issue, but more people than just Dennis had a problem with this. Also, his bike is powerful enough, a 5304 on a 72V LiFePO4 setup with the 18 FET controller set for about 65-70A, I think, which is well over 5kW. Anyway, a new programmable function would be optimal.

I know I could get a new "hard start" controller, but what about those who already have them? It would be great if Keywin would send out an executable that could be flashed to existing controllers, but frankly, I'd be more than a little surprised if that happenend. Too easy for a competitor to clone it. I'm sure that's why he's shipping complete controller chips.

-- Gary

 

[methods]

If anything comes out I will crack it.

We just interviewed a kid the other day who *Majored* in "Firmware reverse engineering".... i.e. microcontroller hacking.
He had an awesome presentation.

Tedious work. Will rot your brain.

-methods

 

[Jeremy Harris]

I'm pretty sure that the core of this chip is the old 8051, so it's odds on that there are already reverse compilers about that may handle the object code and convert back to something vaguely resembling the original source code. If we knew what the original compiler was, then we could probably get back to a near exact copy of the source code.

The first step is to find a way to get the chip to do a dump of the embedded code, so we can get to work on it. It's 20 years since I've done any work in assembler, so I'm very rusty and it'd likely take me a long, long time to suss it all out. My guess is that the source code was originally written in a variant of C, as there are loads of C to 8051 machine code compilers about. Ideally we want one of you young guys with loads of energy to have a go at it!

Jeremy

[edited to add:

There's some interesting info on the Infineon XC800 series controllers here: http://tinyurl.com/cxpy3c that may be of interest. I've not read it all yet, but wouldn't be at all surprised to find that some of the stuff there is what's been used in these controllers.

One interesting find is that the XC846 is the ROM version of the XC866 - they are the same chip internally, but the XC866 has flash programme storage. This may be bad news when it comes to trying to re-programme these chips........... ]

 

[wanders]

There's some interesting info on the Infineon XC800 series controllers here: http://tinyurl.com/cxpy3c that may be of interest. I've not read it all yet, but wouldn't be at all surprised to find that some of the stuff there is what's been used in these controllers.

One interesting find is that the XC846 is the ROM version of the XC866 - they are the same chip internally, but the XC866 has flash programme storage. This may be bad news when it comes to trying to re-programme these chips........... ]

This is good and bad news, of course. If you can find the exact corresponding XC866 version of the '846 that's in these controllers, you should be able to flash in the ROMmed code and then change it. But the best solution is access to source, using the Infineon tools to build the image, etc.

Jeremy, can you recheck your tinyurl link? When I click, it goes nowhere...