1500 Watt 48V Speed Controller

[Jason]

This is a china 48v-60v 1500 Watt Speed controller. Ebay link as of December 25th 2018

Main Picture

s-l1600.jpg (94.02 KiB) Viewed 79791 times

It's ebay specs are the following
Rated Voltage: 48 Volts
Rated Power: 1500 Watts
Phase Angle: self-study
Throttle Voltage:1.1-4.2V
Brake level:High/Low
Under Voltage Protection: 42 Volts±0.5v
3 phase 60° or 120° brushless motor
Operating Conditions: -40~80 C
Size: Approx. 24.5x8.7x4.3cm

Connection Table:
Power Cables: big red + big black
Motor Cables: big yellow + big green + big blue
Electric Door Lock Cable: The middle-red cable with 6.3mm plug
Hall Cables: small red, small black, small yellow, small green, small blue
Throttle Cables: Small Red(5v) + Small Blue(signal) + Small Black(Negative)
Speed Cables:Blue(high speed) + Black(normal speed) + Yellow(low speed)
High E-level Brake: Yellow( for 12v braking bike)
Low Elevel Brake: Brown(braking signal) + Black(connecting the 2 cables together will realize the power cutting off braking function)
Anti-theft Alarm Cables: Red + Black(power) + Red(E-start) + Blue(Motor Signal) + Green(Motor Lock)
Self-Study cable:Self-study(Phase Learning)
Function: White cables connecting together
Cruise(Optical):Cruise Function: Gray Cables(Two plugs connecting together will be auto cruise function;if connecting the 2 cables to a switch, it will be switching cruise function)
LCD Display Cables: Blue(speed shift button) + Green(speed calbe to LCD)
Speed Meter Cable: Green( Not suitable for LCD Speed Meter)

The critical wires to get started are the following

Motor Out
Ring terminal Yellow: Motor
Ring terminal Blue: Motor
Ring terminal Green: Motor

Power In
Ring terminal (large) Red: 42v-60v input
Ring terminal (large) Black: Ground
Ring terminal (small) Red: "on/off" Attaching this directly to battery+ will turn the controller on and off. Hooking and estop switch to this wire, and the other end of the estop to the battery positive will give you a simple e-stop circuit.

Halls/Sensors 5 Total Wires in a 6pin connector
Small Red: Sensor 5v+
Small Black: Sensor 5v-
Small Yellow: Sensor Signal
Small Green: Sensor Signal
Small Blue: Sensor Signal

Throttle 3 Total Wires in a 3 pin connector (Works great with hall effect throttle)
Small Red: 5v+
Small Blue: Signal
Small Black: 5v-

Sensor learning
Small white cables. When connected it is learning mode. Normal running is with these connectors disconnected.

1. Lift your rear wheels and insure they are unloaded.
2. Plug the white cables together.
3. Turn on your kart/plug in. Your wheels should begin moving.
4. If the wheels are spinning the correct direction. (If the kart were on the ground, it would move forward) If correct unplug the white cables you are ready to go
5. If your wheels were spinning the wrong direction. Blip the throttle, the wheels should now spin the opposite direction. If it is correct now, unplug the white cables and you are ready to go.

Reverse Two wires in a two pin connector
Black:
Grey:
Connected together will allow reverse, open as stock, it will move forward. Either one will be controlled by the throttle. A simple switch can be installed here to allow for simple forward/reverse capabilities.


Hooking up these cables, should be enough to get your motor moving under throttle control.

[Jason]

Interesting post with lots of detailed information on the internals of a similar (maybe the same) controller.

https://endless-sphere.com/forums/viewt ... 30&t=89366

[Brian]

I got curious about this, so I did some poking at components. Was considering trying to reprogram it, but that looks annoying (details following). Looks like I'm going to be buying one of these in the near future to start checking what connects to what.

The FETs appear to be counterfeit . They say "FB4410Z" and have an International Rectifier logo. However, all IR part numbers start with IR, and pictures online confirm they put the whole thing on parts... Also, IRFB4410Z is an obsolete part, and the date code claims they were made in 2012. All the current actual IR parts have "PbF" at the end because they're lead-free. Anybody got pictures of more of these FETs to see if the date codes change or not? IRBF4410Z has an on-resistance of 8 milliohms typical, which with 3 in parallel would be only ~5.3 millohms per phase which is pretty nice. No idea what those FETs actually are though. Seems like they must be pretty low on-resistance because people say the controller doesn't get hot I guess?

The shunt (the one you glob solder on to make it go more faster) looks like it's across the low side of the input. Control based on that is effectively controlling current through the fuse, which makes sense for being reasonably easy to tune to keep the fuse alive. There are definitely holes for a third metal piece, which makes me curious what else they use the same PCB for.

The pictures in that Endless Sphere thread are pretty nice. However, the poster appears to not be able to read his own picture of the MCU... It is in fact a GPM8F3116A-<undecipherable>, likely GPM8F3116A-QL01<something> from the datasheet. The datasheet is at https://datasheet.lcsc.com/szlcsc/Gener ... 193858.pdf and is surprisingly nice (the sections I read are better than TI's ). The manufacturer seems to be a reasonably large Chinese tech company (from a quick Google search). Datasheet says it's an 8051 with 16K of flash, 512 bytes of "XRAM", and 256 bytes of "IDM SRAM". From a quick poke, those terms seem to be ways an 8051 accesses memory. 768 bytes of RAM isn't much to work with; I'm kind of impressed at what they've done with it. It has a (very) few peripherals, but the 3-phase motor control one looks most interesting. It's got PWM outputs which can be commutated in hardware, fed from either halls or comparators. I'm pretty sure that's how they drive a motor with that little CPU power.

[Jason]

Wow, you really did a lot of research on these buggers. When I get my karts back in, I'll see If I can get more pictures of the 2017 purchased controller and the 2018 version. Specifically fet and microchip pictures.

[Brian]

Ooooh, found something fun: I think it's easy to play with resistors for precision setting of the current limit, and easily varying it at runtime. I've found that the MCU actually has builtin overcurrent comparison. For keeping cost down, I bet that's what they use. It's literally just a comparator which software can set the outputs to be forced on/off based on. If this is actually what they're using, pins 36 and 37 will have two voltage levels to compare (if not, ignore everything in this post). I expect one of them is connected to the FET side of the shunt, and the other is connected to a resistor divider. If you add/replace the resistors in that divider, you can actually set the current limit, and do it in a way that's reproducible across controllers (vs having to tune solder globbing on each one). That also makes it really easy to hook up another resistor in parallel along with a button to vary the current setpoint on command (or even a potentiometer to get really fancy), without having to run full motor current through a button.

EDIT: They've also got a builtin op-amp set up as an amplifier to an ADC input. That's on pin 35, with the gain set by a resistor between pins 33 and 34.

[Jason]

That is awesome. Rick was actually telling me the same thing earlier, perhaps he can pipe in with his findings. My rudementary understanding suggest that you both came to the same conclusion, which to me doubles the chance it will work.

[Brian]

The ones Aren ordered showed up, and they're completely different from that ES thread you linked with pictures of some controller. On the plus side, this one looks way easier to program (labelled JTAG connections). However, that means all the stuff about current setpoint tweaking via resistors is invalid.

MCU is labelled XCM-K NCN95. First Google result (only one that looks like anything other than matching random numbers) is https://www.infineon.com/dgdl/XC864_um_ ... bbbbcc22a4, which is an Infineon XC864 (another 8051). Unfortunately, that's not even available in the relevant package, so it's definitely not correct... At a glance, Infineon's similar parts in the same package don't have power pins in the right places either. Might still be interesting to look into for tracking down a reference manual, but hopefully I can just read the JTAG identifier off of it instead.

The MCU vendor appears to be http://www.xcmmcu.com/. I found an ES thread saying they make clones of other people's MCUs, but no specific data on this part number. Interestingly, poking around their website, it looks like they're the source of these common PCBs too. They've got photos of PCBs with a lot of fairly distinctive features, but only mentions of 18-FET (Google Translate somehow ends up with "18 tubes"??) versions without any pictures. Most of their pictures appear to be edited to (clumsily) obscure the markings on the MCUs too (possibly physical editing via a piece of tape; grainy photos make it hard to tell).

Poking around for info on XCM in general, I found https://endless-sphere.com/forums/viewt ... &start=925 saying this chip even with different markings is another NEC chip. That all seems to be pretty light on details (almost like nobody's crazy enough to try reprogramming these chinatrollers), so I'm not sure I trust it. I'm kind of hoping that's wrong, because if so it's a Renesas special instruction set which means using their C compiler and stuff...

Regardless, should have more details soon once I get set up and turn the thing on to try poking at it over JTAG.

[Brian]

EDIT (again): I think this actually does work. The controller I tried it on has other problems (it appears to fail self-learning hall settings). Haven't actually tried it though, so who knows.

EDIT: None of this actually works. You end up with a speed controller which will spin wheels in the air just fine at 30V, but then cut out as soon as you ask it for torque. I think I have an alternative solution, but I'm going to actually drive with it before posting it this time...

They're pretty easy to modify for running on 8S too. The ones I have should work OK down to around 20V, before needing an electrical redesign (gate drive regulator starts dropping out). These are labeled as 36V-48V, but the PCB has options for low-voltage cutoff up to 80V, so I suspect most controllers designed for 48V will have similar designs. Some higher-voltage ones might not like going down that low though.

The modification itself is just adding a resistor. For mine, 20k makes it start spinning with 19V, and stop spinning around 18V. You just solder the resistor in parallel (ie on top of) the existing 24k resistor. It will have "2402" written on the top. It's near the spots for jumpers to set higher voltage cutoffs. It's the biggest resistor in that area too.

Here's the resistor to attach to (ignore the yellow wires):

before circled.jpg (374.34 KiB) Viewed 79011 times

And here's what mine looks like afterwards:

after circled.jpg (377.92 KiB) Viewed 79011 times

I ended up with the new resistor sideways on top because that's where it was easiest to hold.

I used an 0603 (RMCF0603FT20K0) because that's what I had laying around, but an 0805 (RMCF0805JT20K0) or 1206 (RMCF1206JT20K0) would be easier to solder if you're ordering parts specifically for this. I think the resistor we're soldering onto is a 1206.

To find the equivalent resistor on other controllers, you're looking for a resistor where one side is attached to VCC (the battery input) and the other goes to all of the following:

• Resistors to the higher-voltage jumpers
• A resistor to ground (2k for mine)
• Probably a small capacitor
• A pin on the control IC, probably through another resistor

Replacing the resistor between that trace described above and ground is another option for lowering the voltage cutoff. You have to actually replace that one, because it needs to get a larger value, so it's a bit more annoying.

Calculating the new cutoff voltage for a different controller with either method is straightforwards. You calculate what voltage the control IC reads at the stock cutoff, and then modify the resistor divider so it sees that same voltage with your new, lower input voltage. I found a random calculator for the resistor divider at http://www.ohmslawcalculator.com/voltag ... calculator and one for the parallel resistors at https://www.allaboutcircuits.com/tools/ ... alculator/.

[Craig]

Interesting, I'd love to love to lower the cut off to 36v on a 12s pack. I'll look at the controllers I have.

What does the Speed cable do? Does it limit RPM or just current?

[Scott]

Were you ever able to figure out how to program these?

I’ve been poking around a couple E-S threads and it certainly seems possible. There seems to be a few parameters that you can set through the e-bike display. (I’ve ordered one to try)

https://endless-sphere.com/forums/viewt ... =2&t=96736

And there is a configuration program for some of the controllers, via a programming contector, but is not real clear where that would land on the board.

https://endless-sphere.com/forums/viewt ... =2&t=27927