Infineon has REGEN BRAKING (and more)

[geoff57]

I've been playing with the programming again today. Anyone know what the Speed % settings do? I set all three to 120%, no load speed is about 40 rpm faster but motor makes a lot of noise at full throttle.

GET IT OFF THAT SETTING UNDER NO LOAD

 

[Mike1]

I assume from the CAPS that this is a bad thing. Wont be using it like this, just ran it long enough to get an rpm reading, couple of seconds.

 

[geoff57]

Hi
something like that just don't use over 100% in no load now read below.

speed % over 100 may work on the road but it DOES NOT WORK UNDER NO LOAD i have done a lot of testing on this all no load, I keep a motor on the bench in its torque arm that is in a vice. Tis is what I do all my tests on and get all of my data from.

my ride was stolen and I want to build it back again right so I have no way to do field tests over 100% that will change soon with me or someone else soon I hope.
Geoff

 

[Mike1]

I guess its not just throttle calibration then.
Some sort of timing advance maybe?

 

[geoff57]

I guess its not just throttle calibration then.
Some sort of timing advance maybe?

hi
No it's a throttle calibration setting I've just come to the opinon that 110%, 115% and 120% is close to timing advance, exactly Im not sure on the over 100% settings.
The best desription I came up with for the speed% settings is if you think of it in terms of it as a ratio between the throttle and motor rpm I have some very acurate data working with an infineon and a puma motor at 48v, this uses the fact that the throttle sends out a signal between 0.8v and 3.4v, I was able to mesure rpm in 0.1v incriments so I had about 20 steps to work with opposite our normal twist grip throttle where you only have about 80 degrees of movement to work with getting accurate result just broard results, the best results on the road are observation.

 

[Knuckles]

I've been playing with the programming again today. Anyone know what the Speed % settings do? I set all three to 120%, no load speed is about 40 rpm faster but motor makes a lot of noise at full throttle.

See ... http://endless-sphere.com/forums/viewtopic.php?f=16&t=7361#p110931

 

[Mike1]

Thanks Knuckles,
Could I use DX3 at the same time as SL to further reduce the reversing speed below the current 15% minimum?

 

[geoff57]

Thanks Knuckles,
Could I use DX3 at the same time as SL to further reduce the reversing speed below the current 15% minimum?

mike not likely SL minimum setting 50% and it gives full torque.
the only thing you could do is to use the SLA as a means to further reduce speed, personally I would not bother just be careful.
Geoff

 

[Mike1]

What does SLA do? I must have missed that one.

 

[geoff57]

Hi
SLA I think is designed to allow the SensorLess module to work with the infineon when powering a geared motor, I don't know about DD motors I have not got one to test with yet.

The sensorless module has problems at about 80% throttle, after that things are crazy not good, while testing I had fitted a switch between SLA and GND.
As an experiment I set the switch to on shorting the SLA pad to GND I then tried the throttle again, I observed that I had to turn the throttle at least 20% before the motor started, the throttle could now go to full throw without problems BUT max rpm was not as high.
The circuit oard showed me that the track going from the throttle to the infineon chip now has a branch on it going through a resistor down to if used GND this reduses the voltage the infineon recives from the hall sensor so a sensorless module can be used.
Now all we have to do is redesign the system so as to not require a rpm drop!

Geoff

 

[ZapPat]

Hi Phil, thanks for the reply.
Yes there are 15 P75NF75 mosfets. I thought there would be some multiple of 6 so a little confused by that.
Mike.

This controller appears to use 2 high side FETs and 3 low side FETs per phase. Seems odd at first, but actually this can be good specially if the controller is being used often at partial duty output (meaning at partial throttle and/or while the contoller is doing current limiting). Not too usefull for someone spending most of their time zipping along at full speed and full throttle, but good for going up hills at partial speeds. The Low side FETs in this situation work harder than the high side FETs, making such a design choice logical.

BTW, has anyone else spotted what looks to possibly be a switching power supply on this controller? Mike, could we have a close up picture of the area of the board around where the smaller red wire goes? With a bit of an angle so we can see the chip number of the heatsinked part, or maybe just tell us what it is? Also, can you make out what number is on that surface mount part, numbered "U1"? Thanks!

Hi Zappat, I finally got those component numbers.
The heatsinked part is a LM317T
U1 has the following on it.

CY8C24423A-
24PVX1 0719
B 02 CYP 62569.

Thanks for that, Mike.

Humm... that's funny. I checked the number and it seems to be more suited to the chip marked as U2 in your picture. It would be the microcontroller (brains of the controller), and has 28 pins, all which fit with U2, not U1 which has only 8 pins. Don't sweat it though, it was more for my curiosity than anything! We can see that this infinion-like board uses a cypress processor instead of an infinion, so is most likely not running the same program either.

 

[Mike1]

Hi Zappat,
Yes that probably is the wrong chip info I gave you. This is someone else's controller so info getting lost along the way. I should have one of these soon when my replacement part arrives. Hopefully I'll be able to do bit more with it than just modify the shunt.

Mike.

 

[methods]

So I have been thinking about the pitfalls of firmware programability.
First one that comes to mind is overpowering the shunt.
I read previously that the standard shunt value is 3.8mOhm
I dont know if this is for 1 "M" or 2 "M's" in parallel

Seeing as that the controller comes in 250W, 350W, 600W, ... 1500W versions
I could see some poor chap buying a 350W, beefing up the fets to 4110's, then letting the current rip after a firmware update

If one were to up the current limit in software but not decrease the shunt resistance this would be the result:

<Shunt resistance> <current limit> <power across the shunt>

R I W
0.0038 15 0.855
0.0038 20 1.52
0.0038 25 2.375
0.0038 30 3.42
0.0038 35 4.655
0.0038 40 6.08
0.0038 45 7.695
0.0038 50 9.5

Now I dont know about you but 10W is Wha-wha-whaaaaY too much to drop across that little bad boy.

On the little 350W baby controller I am modding out I have soldered the shunt up to allow it to handle more current.
I have not calibrated it in yet but I plan to get it down to at least 1/2 that relatively high shunt value.

On a second programming note I sure would like to find a way to change the ramp up speed
This controllers throttle response is way under damped.
Maybe there is an RC circuit on the board someplace that will allow me to tune the response.
I doubt it... If it were me doing the programming I would do it in the firmware.

-methods

 

[methods]

Size VS. number of fets:

Here are the different versions available on ebay:

<voltage/power rating> <size> <current limit>

24V/250W - 180W_____125*70*35mm___18A
36V/350W____________140*38*37mm___25A
48V/350W____________125*70*35mm___15A
48V/600W - 800W_____180*85*45mm___30A
72V/1500W___________180*85*45mm___62A

Would I be safe to assume that the 250W & 350W units use 6 fets while the 600W uses 12 and the 1500W uses 18?

I am fascinated by the manufacturing decision to run separate designs for separate power levels
I wonder if they ever tried using really high powered fets in the smaller 250W/250W units?
I am going on the assumption that the larger units are fairly identical except for parallel fets, maybe some bigger caps.

We will find out shortly as I am going to draw at least a KW out of this teeny tiny 350W box.

-methods

 

[geoff57]

hi methods
I don't know quite where you’re coming from but I have an idea.
First the "M" shunt is now the standard shunt used on all controllers each "M" is 3.8 ohm's.
The number of shunts is determined by the board type not the power rating; the program knows how many shunts there are on each type of board.
as long as the old trick of adding solder to the shunt is not done then reprogramming the controller will work fine two 12 fet boards sent out with different fets and caps on it programmed to meet the needs of the components installed both would have 2 "M" shunts installed as is normal for a 12 fet board 6 fet use 1 and 18 fet use 3.

This should clarify that firmware programming will be no problem.

What site did you get those details from?
as for higher power fets in smaller controllers I presume you mean the 4110s as far as that is concerned the words camel and eye of a needle comes to mind before you can persuade the china connection to fit the 4110 fets we have been trying for months.

Geoff

 

[methods]

The number of shunts is determined by the board type not the power rating; ....

The number of shunts is determined by how much current the board is designed to source, i.e. the power rating.
The reason they add shunts in parallel on the larger boards is to handle the additional current.
3.8mOhms is way too high of a shunt resistance for a +50A board. 1mOhm or less would be more appropriate.
The 6 fet boards can get away with 1 shunt since they are running 30A or less.

What site did you get those details from?

I calculated them with Ohms law: V = I * R
R = shunt resistance = 3.8mOhms
I = rated current of the controller, i.e. 30A
Power = I^2 * R = rated current squared * shunt resistance

The power across the shunt must be minimized while taking care to still develop enough voltage across the shunt to provide good resolution to the A/D's in the uC.

This is how you tell if your shunt is big enough.
This is how they decided to parallel up the shunts on the 12 fet and 18 fet boards.

I was just pointing out that if someone uses the firmware to up the current on a 6 fet board they need to solder the shunt to handle the additional current.
(well, they dont have to, but it would be advisable)

-methods

 

[geoff57]

The number of shunts is determined by the board type not the power rating; ....

The number of shunts is determined by how much current the board is designed to source, i.e. the power rating.
The reason they add shunts in parallel on the larger boards is to handle the additional current.
3.8mOhms is way too high of a shunt resistance for a +50A board. 1mOhm or less would be more appropriate.
The 6 fet boards can get away with 1 shunt since they are running 30A or less.

What site did you get those details from?

I calculated them with Ohms law: V = I * R
R = shunt resistance = 3.8mOhms
I = rated current of the controller, i.e. 30A
Power = I^2 * R = rated current squared * shunt resistance

The power across the shunt must be minimized while taking care to still develop enough voltage across the shunt to provide good resolution to the A/D's in the uC.

This is how you tell if your shunt is big enough.
This is how they decided to parallel up the shunts on the 12 fet and 18 fet boards.

I was just pointing out that if someone uses the firmware to up the current on a 6 fet board they need to solder the shunt to handle the additional current.
(well, they dont have to, but it would be advisable)

-methods

methods I wanted to know what ebay site you got the controller info from.

if you add solder to the shunt forget the programming the numbers are incorrect the shunt will work but the figure set in the computer will not be the true value.

Geoff

 

[methods]

methods I wanted to know what ebay site you got the controller info from.

Ecrazymans.
Isnt that where everybody gets there controllers form?
Do you know somewhere else can I get one?

if you add solder to the shunt forget the programming the numbers are incorrect the shunt will work but the figure set in the computer will not be the true value.
Geoff

You could also desolder the 3.8mohm shunt and solder in a new known value shunt then scale the software values.

After you add the solder you calibrate the shunt.
For instance the old value was 3.8mOhms and you add solder
The new value may be 1.9mOhms.
Then you just scale the software settings by the difference, i.e. 2:1

Set the software for 20A if you want 40A
Set the software for 40A if you want 80A
etc.

I think we are talking past each other because you are interested in creating a plug and play system and I am interested in geeking the controller out to be electrically ideal.

-methods

 

[geoff57]

methods I wanted to know what ebay site you got the controller info from.

Ecrazymans.
Isnt that where everybody gets there controllers form?
Do you know somewhere else can I get one?

if you add solder to the shunt forget the programming the numbers are incorrect the shunt will work but the figure set in the computer will not be the true value.
Geoff

You could also desolder the 3.8mohm shunt and solder in a new known value shunt then scale the software values.

After you add the solder you calibrate the shunt.
For instance the old value was 3.8mOhms and you add solder
The new value may be 1.9mOhms.
Then you just scale the software settings by the difference, i.e. 2:1

Set the software for 20A if you want 40A
Set the software for 40A if you want 80A
etc.

I think we are talking past each other because you are interested in creating a plug and play system and I am interested in geeking the controller out to be electrically ideal.

-methods

hi methods
I think we are both going at the same thing only we have already ben there done that and discarded it as not viable for production.
let me explain the early controllers were sent out from china with diffrent shunts on that had no uniform value.When the programming became availble to us the shunt value for the program was fixed, so we had to work out a way to mesure the shunt then we could use a multiplyer to get the correct value set in the controller even if the programmed value is diffrent. much as you did above, once the "M" shunts were in use then calibrating shunts was no longer needed and just ment more time.

if I was doing things for myself i might do it diffrently.
the controllers you found on ecrazymans web site are not what you thought they are as below:
dimentions wattage amps volts fets
180*85*45mm 600~800W 30A 48V 12
180*85*45mm 1500W 45A 72V 12
140*38*37mm 350W 25A 36V 9 (i think)
125*70*35mm 180~250W 18A 24V 6
125*70*35mm 350W 15A 48V 6
these are correct to the best of what I know of the controllers ecrazyman does not sell the 18 fet model yet only to order.

Geoff

 

[methods]

Right.
You are talking about production of plug and play units
I am talking about modding a 6 fet unit to run more than 30A. (not really the same at all )

I am not making suggestions for production, rather, I was pointing out the pitfall of someone replacing the fets and turning up the current limit via software without regard to the relatively high resistance of the shunt.

So I would argue that regardless of what is viable for production the shunt should be soldered or replaced if one is to significantly increase the drive current of a given board.
(i.e. if someone is hacking a board)

-methods

 

[philf]

Size VS. number of fets:

Here are the different versions available on ebay:

<voltage/power rating> <size> <current limit>

24V/250W - 180W_____125*70*35mm___18A
36V/350W____________140*38*37mm___25A
48V/350W____________125*70*35mm___15A
48V/600W - 800W_____180*85*45mm___30A
72V/1500W___________180*85*45mm___62A

The size of the 36V/350W unit is incorrect in the postings... It's a 6 FET unit, and the same size as the other small units. If you look at the numbers above, the thing would be a 1 1/2 inch nearly square tube that was 5 1/2 inches long. Obviously not right. Meanwhile, if you look at the pictures Keywin has posted, and use the laser label as a size reference (the same paper label stock is applied to all of the controllers), you can see there are really only two sizes. I don't think Keywin offeres ANY 9 FET units. Only 6 and 12. Of the 6 FET types, its the 36V/350W that I have. The 48V unit I have is the same size as the 72V, and with 12 FETs.

It's interesting to know the *intended* impedance of the shunt. 3.8mOhm, eh? Most of the shunts I've purchased top out in rating at 3 - 5 watts (more commonly 3), but I've stuck with 1mOhm devices, so can suck a lot more through them to hit that peak. I haven't used one of these in a controller (I use them externally for other current measurement) because I wasn't sure what value the controller was expecting. Haven't gone to the effort of actually measuring the drop to find out for myself, but If 3.8mOhms is a bankable number for the "M" style shunts, then I have something to work with (thanks!).

The thing about ANY shunt is that it's impedance will change if you heat it up. So you definitely don't want to be sucking more through it than it can easily handle or it will start giving you artificially inflated readings as it warms. Shunt material is also very important - you can easily make your own shunts, but the most commonly available wire stock has a comparatively poor temperature co-efficient. Copper being among the worst. I've always wondered how consistent the "solder blobbed" shunts are through their operating range.

Sure would be sweet to get at the firmware to see how it's handling all of this. Ideally, each controller would be calibrated as a final phase of manufacture. I sincerely doubt that *THAT* happens, which means considerable leeway between units of the truly imposed current limiting feature. When you're talking mOhms, there can be considerable differences between one unit and another due to soldering alone - notwithstanding the %accuracy of the part itself...

 

[philf]

Methods snuck in with another post while I was writing that .

I'm also only interested in the hacking aspect, but tend to like to do things with as much information as I can get - just in case I need to repeat a given exercise, which often happens when someone sees something I've done and asks, "Cool! Can you make *ME* one?"

 

[methods]

The reason I take shunt calibration so seriously is because I have been burned

Usually we think of the shunt as being used for the current limit or maybe on the CA for current display

To me, the true value (and danger) of a shunt is using it to keep track of Ah usage (especially when you run an insane Lipo time bomb with no BMS). With my 100V 100A 5305 I needed to have a very good idea of Ah used and Ah left. Often times on rides it would take me 3-4 Ah just to get home.... If my shunt calibration was off by just 5% that means I am walking

It would be good to test the temperature drift of a soldered shunt. Good point Philf.
I have seen the peltier effect after long hard pulls of over 80A but it usually is only a few hundred ma offset.

Where do you buy shunts?
I would agree that 1mA is a good round number to use.

-methods

 

[Mike1]

Hi Guys,

Keywin recently told me that the M shunts were measuring a little low and that a 40 amp settings would yield a real limit of 48 amps.
Anyone else heard this? I haven't tested it yet but will very soon.

Mike.

 

[methods]

Like Philf eluded to...
It is nice that they are "targeting" 3.8mOhms but in reality, a few hundred microOhms (3.8, 3.7, 3.6, 3.9...) is the difference of how the shunt is soldered.
If jack does it you get 3.5mOhms
If Jill does it you get 3.7mOhms

If you want accuracy you need to calibrate the shunt yourself.

-methods