Changing the LVC in a 72v generic controller

[Zenid]

On the back, you can see the two plates that make it clear that the resistors are all connected in parallel, and a multimeter reading does indeed give 500Ω.


So, here's my own mod, with a 1KΩ 7W resistor in parallel


Giving the 333Ω (about) that I want.


So I took my bike to pieces, removed two batteries, and rewired back into the 48V configuration. I hooked up the new controller, and powered it up with the hall sensor disconnected at first, to take a reading, It read 4.7V. So far so good...

But then I plugged in the hall sensor, powered it up and tried the throttle and... nothing Sliding the PCB from its case in situ reveals the blinky red light of doom

Bummer... what are we missing here? I thought this should have fooled it into thinking it was connected up to 72V, but evidently something's missing. Now my suspicion is this: There are two power inputs to the controller - the high current one for the FETS, and a low current one (the thin red wire) that the ignition turns on and which takes care of other stuff (powering the CPU?). I think that maybe modding this resistor only takes care of the high current part, but not the other part of the circuit.

Does anybody here who knows a bit more about controllers have any idea about what else might need changing, and where it would have to happen?

[EDIT] The blinky red light gave the same 8-flash lvc error as auraslip got

[auraslip]

We really need a pin out from the mcu.... with that we could back trace the voltage dividers coming off the lvc input. Until then it's just a guessing game!

[Zenid]

We really need a pin out from the mcu.... with that we could back trace the voltage dividers coming off the lvc input. Until then it's just a guessing game!

Could you explain this? My understanding of these boards is limited... Where's the 'lvc input' and 'voltage dividers'? What do they look like? Is the mcu that CPU that is programmable on better boards?

I thought our reasoning here was basically okay - we're fooling the board into making our 48V appear as a 72V would to the motherboard circuitry, with the lvc being scaled down accordingly. Is there somewhere that gives a good explanation of how these things work?

[auraslip]

Well every board is different.... but basically the resistors we changed were part of the 5v regulator system meant to provide 5v to the hall sensors and other controller parts. The mcu probably has a trace leading towards it that senses a voltage and uses that to determine LVC. Whether this feeds of the 5v regulator system is undecided, but it probably is. We just need to find out which pin to the mcu has the voltage dividers(resistors) attached to it that decides LVC.

[Zenid]

Well here are some close-ups for anyone who might want to try and help decipher this. The threads to the left of the mcu are attached to various bits and bobs that could be of interest, it's a question of figuring out which one. Also these things are tiny and I think you'd need near-microsurgeon precision to do anything with them...
This one - I think is the most promising:
http://zenid10.files.wordpress.com/2011 ... 0x1200.jpg

http://zenid10.files.wordpress.com/2011 ... 0x1200.jpg
http://zenid10.files.wordpress.com/2011 ... 0x1200.jpg

The markings on the board must be a clue. Any idea what they mean. The RXD is interesting, as it's usually part of a TTL interface.

The threads to the right don't seem to be as many, but interesting is that 4-way junction, which looks like it might be for a TTL interface (though the RXD is elsewhere on its own). Even with a USB/TTL interface, getting software to talk to the chip is a different matter entirely though...
http://zenid10.files.wordpress.com/2011 ... 0x1200.jpg
http://zenid10.files.wordpress.com/2011 ... 0x1200.jpg
http://zenid10.files.wordpress.com/2011 ... 0x1200.jpg

[auraslip]

http://zenid10.files.wordpress.com/2011/03/mcu2-1600x1200.jpg

I was thinking that bottom trace running from the MCU and then down to the left might be the lvc one. It seems to feed right off the main voltage regulator, and run through some voltage dividers and then directly into the mcu. It was at like 2.5v with my 48v charger. I was guessing that was the one because it doesn't seem to have any other function.... I'd learn how to replace the smd resistors on that trace, but I've run out of time to work on this controller because I got a new bike to build as well as school work!

[Zenid]

http://zenid10.files.wordpress.com/2011/03/mcu2-1600x1200.jpg

I was thinking that bottom trace running from the MCU and then down to the left might be the lvc one. It seems to feed right off the main voltage regulator, and run through some voltage dividers and then directly into the mcu. It was at like 2.5v with my 48v charger. I was guessing that was the one because it doesn't seem to have any other function.... I'd learn how to replace the smd resistors on that trace, but I've run out of time to work on this controller because I got a new bike to build as well as school work!

I'm afraid you'll have to bear with me, I'm not strong on the jargon. Where is the 'main voltage regulator'? - Do you mean where those big resistors step down the external voltage to the 5V needed by a lot of the components on board? (I don't even understand how you can get 5V from 72V and 500Ω )

Yes I had my eye on that little brown resistor(?) by "M4". I tried to measure it with my multimeter, but the reading fluctuates upwards as soon as I try - I think the meter must be charging up a capacitor somewhere or something...

I have a little time on my hands. If I can have some direction as to how to figure this out (safely!) I'll experiment on my board. The first stage - like you say - is figuring out how to replace those teensy little resistors

[EDIT] I've figured out how to read those teensy resistors and confirm their value. Are those little brown things capacitors? They seem to behave like that when I measure them...

[auraslip]

dude I'm only like one step ahead of you lol... a month ago I didn't know what any of this meant!

Maybe we should rename this thread and put it in the ebike tech section, or perhaps we can make a new thread there. It'd likely garner more attention.

[amberwolf]

I'll move it and rename it, just tell me what you want it called. Also, all of us will need to re-subscribe to the thread once moved, as that breaks subscriptions for some reason.

[auraslip]

thx

Just.... rename it something like changing the lvc in a 72v generic controller.

[amberwolf]

Done.

[Zenid]

Thanks Amberwolf!

So... Can anyone out there help us figure out what might need changing in these generic 72V 'McControllers' in additon to the resistor mod detailed above? A hyperlink to existing, relevant discussion would be nice as even with search it's not always easy to locate relevant, existing discussion...

[Zenid]

On some documentation for the infineon controller, modifying the lvc from the board is detailed as follows:

"The Low Voltage Cut-off (LVC) circuit consists of tiny surface mount transistors and one capacitor.
This circuit connects to an MCU pin-out so the MCU can detect the battery voltage.
The R12 resistor is about 1.2 kohm and connects from the MCU pin-out to ground"


Some info from a website on how controllers work ("4QD-TEC") also says this about one of their own circuits:

"There is also an on-board regulator: power input is to pin 15 (pin 8 is 0v), pin 16 is a regulated reference voltage output of 5v. This is used to drive the speed control pot (the 3 pin connector). It is also used to drive under-voltage shutdown into pin 10"
"If you discharge a lead-acid battery too much you permanently shorten its life. So this circuit measures the supply and compares it with the 5v reference supply. If the battery voltage falls too low the transistor inhibits the pwm circuit by feeding a signal to pin 10."

So what we need is to identify the corresponding resistor on the "X6M06" board, that might correspond to Infineon's R12:
Looking at the tracks from the mcu, you can see that many of these don't seem to serve any purpose and simply terminate nearby, so they can be ruled out.


Panning back a bit, you can see how the tracks to the left and bottom of the mcu are organised.


The infineon documentation mentions surface mounted transistors being involved. You can see some of these below right of the mcu (panned right from the previous photo)


So where would we begin when it comes to testing voltages or continuity to try and identify the correct pin and resistor for the lvc?

[amberwolf]

I'd forgotten about the R12 thing on the XieChang ("Infineon") controllers! That also deals with the regen voltage, IIRC.

If R12 goes from MCU to ground, then presumably R11 goes from MCU to either B+ or the output of those big power resistors.

Whichever way the XCs do it, the XMs probably do, too. Somewhere I have a scrap XC board of the type for that R12 info (other than my working Lyen 6FET on DGA), but cant' find it at the moment.

The other ones I have are older and don't have the same layout/etc. However, the one I checked (the blown one from Auraslip) has an R10/R11 voltage divider that goes from the output of one of the big power resistors to ground, with it's center tap heading off toward the MCU area. I can't get an in-focus pic of it zoomed in enough to see anything, though.


I took a gander at the 12FET XM I have here, and it has a few things that could be the voltage divider, the most suspiciously set up of which is the string of four 2.2Kohm resistors here:

But those resistors are the same value in Auraslip's 72V controller, which nixes that idea, if it is indeed comparable in function/design to the XCs:

Buuut: I think a comparison of the resistor values between our controllers ought to tell us which ones are good candidates. If we know which ones are different, we can then trace out just those and see where they go.

[Zenid]

{...} Somewhere I have a scrap XC board of the type for that R12 info (other than my working Lyen 6FET on DGA), but cant' find it at the moment.

The other ones I have are older and don't have the same layout/etc. However, the one I checked (the blown one from Auraslip) has an R10/R11 voltage divider that goes from the output of one of the big power resistors to ground, with it's center tap heading off toward the MCU area. I can't get an in-focus pic of it zoomed in enough to see anything, though.

As luck would have a non-working EB-212 infineon type "XC" board that I got off ebay in the hope of fixing it (dew got onto the board when it got left with the case not-quite-closed and it blew a FET, just to add to its non-workingness)...


But I also have an excellent camera with macro zoom and can take snaps of any parts you want, or try and trace various parts if you can give me pointers...

[EDIT]
The resistors on my XM seem slightly different to those on Auraslip's... hmmm

[amberwolf]

But I also have an excellent camera with macro zoom and can take snaps of any parts you want, or try and trace various parts if you can give me pointers...

I'd just trace the path for the R12/R11 divider, to see where it goes on the XC type. Probably the same as what it appears to be on the old type I checked, above, which uses R10/R11 for that.

[EDIT]The resistors on my XM seem slightly different to those on Auraslip's... hmmm

I can't read the third one from the left, but all the others in the row are 222, which is 2.2Kohm. The bottom one that lines up with the obscured third one is 472, or 4.7Kohm. That's all the same values as mine and Auraslip's, so I think those 5 can all be eliminated as part of the solution to the LVC problem.

[Zenid]

The problem is I can't find R11 or R12 labels actually BY resistors. At the bottom of this picture along to the right of the big resistor on the 'Infineon' XM board, just under the smaller of two capacitors, you can see "R10, R11, R12" printed all squished together but apparently referring to nothing, unless this is some way of referring to those further up the board..
http://zenid10.files.wordpress.com/2011 ... middle.jpg

Here's further up right, towards the tracks entering the MCU
http://zenid10.files.wordpress.com/2011 ... -right.jpg

Here's further to the left... What is that big coppery thing?
http://zenid10.files.wordpress.com/2011 ... 1-left.jpg

Oh and here's way over to the right so you can see all the mcu tracks as they go off elsewhere.
http://zenid10.files.wordpress.com/2011 ... -right.jpg

[amberwolf]

Yeah, they do that because there's no room for readable print there. Fairly common on SMT stuff (when they bother to label at all).

In "p1020029-middle.jpg", the C5 marking above the column is the top capacitor in that column, and C9 in the relocated printing is the one below it. Then the R-markings are for each of the resistors in that column, in the same order as the markings.

So in that pic, R12 appears to be 122, or 1.2Kohm. R11 would then be either 458 or 856, I can't quite tell. Neither one makes sense to me, given that 458 would be 4.5Billion ohms (I think) and evne 856 would be 8.5Megohms. Hopefully I just cant' read the value in that pic. I guess it might also be 85b, but I'm not aware of what the lowercase letter would indicate.

I can't see R10 in the pic, it's obscured by the cap. In the other pic, it looks like it's 01c, which again I'm unaware of the marking standard.


The same goes for the relocated R and C position markings in the other pics.

In the p1020031-left.jpg pic, that large coiled-wire-on-black-doughnut is an inductor (specifically, a toroid inductor). Sometimes simple transformers are made that way, too, but this one only has (AFAICS) two board connections, rather than 4 or more, so it's just an inductor. Probably there to help fight inrush currents into some part of the circuit.

[Zenid]

Yes, the two bottom ones are 856 and 01C (hidden behind the capacitor in "middle"). The problem is, none of these have a track running to the MCU, which is where this LVC cut-off pin should be hiding, so now I'm just as confused...

However looking back at the "middle" picture, and right by where the main 72V low current feed hits the big drop-resistor, there's the promising "Vcc-L" -(Voltage circuit cutout - low??) But what is it talking about? Could it be where there's an option to solder an extra wire to tell the controller to cut off given a certain signal? It probably isn't even relevant
[EDIT fixed link]
http://zenid10.files.wordpress.com/2011 ... middle.jpg

Does this give you any better idea of what pin might be accepting the lvc signal from what resistor? It doesn't look to me like resistor 11 or 12 has anything to do with the lvc, otherwise there'd a track going straight into it, right? And if this doesn't give an example of what one looks like, then it's not much good for deciphering our actual XM boards

[EDIT]
This looks interesting. This is an old pic from where I prised the capacitor all the way up so the mcu and tracks were clearly visible. There are some candidates coming off the tracks on the right...
http://zenid10.files.wordpress.com/2011 ... 0x1200.jpg

[amberwolf]

Yes, the two bottom ones are 856 and 01C (hidden behind the capacitor in "middle"). The problem is, none of these have a track running to the MCU, which is where this LVC cut-off pin should be hiding, so now I'm just as confused...

It might run from there to some other part that buffers the signal first, and hten to the MCU. A transistor, op-amp, etc.

I still have to find out what the 856 and 01C mean, as they don't make sense as resistor values but I can see that's what they are.

However looking back at the "middle" picture, and right by where the main 72V low current feed hits the big drop-resistor, there's the promising "Vcc-L" -(Voltage circuit cutout - low??)

Vcc is usually used to mean some voltage source intended to power things. Vcc-L likely just means it's a Logic voltage source (as opposed to the FET power/etc), or Low voltage source, to power all the 12V-or-less stuff.

But what is it talking about? Could it be where there's an option to solder an extra wire to tell the controller to cut off given a certain signal? It probably isn't even relevant

http://zenid10.files.wordpress.com/2011 ... middle.jpg
The link to your file is broken in the post; it's missing part of the URL, so I can't see what you mean.

And if this doesn't give an example of what one looks like, then it's not much good for deciphering our actual XM boards

That's true enough.

[auraslip]

Don't be alarmed, but I think I've cracked this. I'll test this weekend, and see if it works.

[auraslip]

Well. I got it working with my meanwell 48v chargers
I spent a lot of time repairing and figuring out my infenion 18 fet, and what I learned helped a lot with this controller.

Basically, the voltage detection system feeds directly off the VCC as we thought.


My finger is on what would be the equivalent of the infeinons R12. In fact, their is even a blank spot for an extra SMD resistor. All you need to do is to replace it with one (or two) of a higher resistance. I haven't quite figured out the ratio though. I really need to lower the voltage on my power supply until the LVC kicks in to figure out exactly what I need to run it at 16 lifepo4. I'm thinking of removing the SMD resistors, attaching a a 1kohm resistor to the hole drilled in the trace in series with a 1.5k resistor grounded.

Related: you can see by my finger where I tried to drill a hole through the trace to run a resistor to ground. The drill bit was too big, and ate through all of the trace so I had to make a little jumper to the otherside.

One problem I had was the meanwells kept shutting down if I drew too much power. I don't know what that's about?

One final question: Do I need to switch out the resistors for the 5v supply to run at 16s?

One more question: I rubbed off all the thermal compound on the heatsink. Where can I get more for cheap. That shit is expensive for what it is!

[amberwolf]

Check the local computer shop; they probably have a bunch of the little packets in their cabinet leftover unused from heatsink kits they installed for people. It'll just be the titanium dioxide stuff, probably water-based, but thats probably all that was on there to start with.

You don't need much, just enough to barely fill in all the scratches and stuff on the heatsink and mounting surface of components that bolt to it, so you could evne just scrape it off something else, like another dead piece of equipment with still-"wet" paste.

You can even make your own out of the dried-out white powder on old equipment, by mixing in water with it and spreading it out really thin (basically scraping it into the imperfections), or maybe use mineral oil or similar instead of water, so it dries out slower. Sometimes people have used silicone oil, but that stuff tends to cover every surface for miles after it's applied, and makes things icky and tacky and slimy all over, even when you only put it in one place.

[wesnewell]

I've been using 30 year old bearing grease for years on cpu coolers. Works great. Doesn't run and doesn't dry out much. I'd have no problem using it.

[auraslip]

So I ended up installing a 2.2k resistor rather than the SMD resistors. Some napkin math calculation says it should hit LVC near 37v (or 2.3v per cell for 16s).

I've tried to figure out how to do regen on this controller, but I've got nothing. I've figured out how to activate the brake through the high line. Seems a bit weird to run full battery pack voltage to the handle bars but what ever.

Here is the resistor wrapped in heat shrink.