Firewheel custom battery pack

[esaj]

23 minutes ago, HunkaHunkaBurningLove said:

Ahh I hate that when it happens.  At least it's pretty minor. 

Yeah, not a biggie. At least all the wires are "right way around" (ie. I didn't mix up positive and negative ).

Quote

I've had to disassemble an Xbox 360 multiple times after trying to troubleshoot the DVD reader.  It ended up to be a dying laser head so it was a quite $5 eBay part to swap out.  I spent so much time trying to adjust the alignment distances and play around with it I think I took it apart and reassembled it 10 times.  It's all a learning process though.  What doesn't kill you makes you smarter. 

Yeah, little things can sometimes be really hard. At "best" I had to disassemble and reassemble the Firewheel shells 6 or 7 times before I got all the wiring running correctly and nothing pinched or rubbing the tire... Really NOT looking forward to that again, but it might be my only option, since it looks like KS16 doesn't have big enough battery compartments for my packs.

Soldered the discharge-side connectors in place, the voltages were already pretty much equal, so I just connected them together:

So far, so good, nothing exploded or smells like burning  

EDIT: Still need to do similar "1-to-4" -connector thingamabob for the charging-side, and test discharging & charging... not necessarily in that order, if I can find something to use as load for the discharge testing.

[esaj]

Charging side (pretty much) done.

Charging side doesn't use that much current, so I picked 18AWG wires for the connectors (the packs themselves have 16AWG in both discharge- & charge-sides).

First I connected the bundled 4 positive & negative wires to the XT60, the other end of the XT60 will be connected to the charge-port.

Close up on negative-side wires. The wires are twisted together, then I've rolled a single thread from another wire to keep them together, soldered them together and cut the excess off.

Between the XT60 and the packs, I picked Deans. I must say that I really don't like soldering Deans, though. It's hard to get the wire to keep in place during soldering...

Almost done for the 1-to-4 -connector.

Finished 1-to-4 -connector. Time to solder the charge wires of the packs to the other ends of the Deans-connectors.

Live battery wires being soldered. At one point, I started soldering the other side of the connector without the heat-shrink in place for the already soldered side. Bad mistake, I ended up short-circuiting the connector with the soldering iron tip (the BMSs don't have reverse polarity protection diodes, so the charging side wires also give out voltage!). Luckily, it seems nothing was damaged...

All done and packs connected together (both via discharge- and charge-side 1-to-4 -connectors).

Also took a look (before making the charging-side connectors) at how the packs would fit on the Firewheel shell:

Yeah, it doesn't look THAT bad. Actually the packs sit better in the compartments than before, as the BMSs are smaller. I've already epoxied one of the broken "pillars" that the screws go through, but might have to use some acetone + Lego-block -ABS to fix some others. How does it look with the boards?

Holy fuck, how am I going to fit everything in there? It doesn't look as bad in the picture as in reality. Do note that the front light, light switch, power switch, charging port & charging 1-to-4 connector must still be added, and the motor wires (3 phases + 5 wires for hall-sensors) will also run through the between the battery compartments on the left-side shell...  I'll (again) leave out the battery display to get more room on the front, and use a separate voltage-display. This is going to suck big time...  And probably take more than a couple of full days to get done.

 

[esaj]

I tried to load-test the packs, but don't have much of anything to use as load for such high voltages.

Using a 15W led-lamp with 100ohm/2W front resistor. The lamp has about 45V voltage drop, so I calculated that the resistor needs to drop 11V. Current becomes 11V/100ohm = 0.11A. The resistor needs to dissipate 11V * 0.11A = 1.21W. It becomes hot, but should be able to take it. Which was pretty precise, when you compare to the measurement (112.5mA = 0.1125A). There are, however, two problems.

Problem number one:

CAN'T SEE SHIT, CAPTAIN! Of course I could just turn the light pointing away (and obviously wouldn't hold the wiring in place by hand if using it for a longer time ). So not that biggie. However, at 0.11A, it's going to take a loooong time to discharge four packs at 56V. Assuming that 56V would be about 30% of the total charge, that would mean that the packs still carry about (3.2Ah * 4) * 0.3 = 3.84Ah of charge. Discharging at 0.11A it's going to take about 3.84Ah / 0.11A = 34.9... hours. I don't want to leave it on for about a day and a half. If there's something wrong with the packs, they could blow while I'm asleep...

So, plan B: Tomorrow, I will take about a 1 meter long piece of 3.3xx ohms / feet Kanthal (or maybe it's Nichrome?). Enough for about 10 ohms of resistance anyway. 10 ohms of resistance with around 56V will give me 5.6A, which is enough to suck the batteries dry in less than an hour. In that gauge, the wire should be good enough to take about 8A, (if it's 26AWG, like I suspect, I haven't marked it ), and at 5.6A, that gauge should reach a temperature of around 950 C. Just need to build something that can either hold it, or do it on top some yard bricks/stones/whatever they're called on the concrete floor of the garage...

[esaj]

What I'm about to describe here is something that probably shouldn't be attempted at home   While not overly dangerous (if done carefully), there are certain things that can go wrong and lead to battery damage, fire or explosion. Should anyone still try to replicate this, at the very least, do not attempt it with shunted packs. These are unshunted packs with full protections (overcurrent, overdischarge/undervoltage etc.).

As the led-light would have taken far too long to discharge the packs, not to mention that in "normal" use, the current is much, much higher than 0.1A, I decided to use a heat-resistor wire (so-called "Kanthal"-wire), as that was pretty much the only thing I had that could dissipate large power without burning up.

I took about one meter's worth of 26AWG Kanthal, measuring it's resistance to be somewhere around 11-12 ohms. I then connected that to two separate "nailing plates" (pieces of steel that have holes for nails in them, used for example to connect wooden beams or whatever), wiring it so that there would be smaller resistance (there's a picture later on) near the plate, so large temperature difference wouldn't snap the wire (at least as easily). I then placed the plates between two yard bricks and connected the 12AWG wires coming from an EC5-connector to it with bolts.

Here's the test setup. The batteries on the left lay on top of a piece of rubber mat. The power's not on at this picture, the yellow DMM is measuring the battery voltage from the connector (55.4V) and the red one is measuring the resistance between the two nailing plates (still unconnected! Don't try to measure resistance on live circuit!), which is around 10 ohms total (The meter is showing 10.4ohms, but the probes alone add about 0.2-0.3ohm, and it's a crappy meter anyway accuracy-wise ).

Here's the connection of the Kanthal-wire. Notice that I've added a spring to hold it more tight between the stool and the wire and to prevent it  from touching anything flammable. Depending on the Kanthal-wire gauge, resistance in the circuit and the battery voltage, the wire can go beyond 1000 Celsius-degrees, or above 1800 Fahrenheit (it burns and gets cut somewhere around 1300-1400 Celsius / 2400-2550 Fahrenheit). Certainly enough to set things on fire, that's why I'm doing it on top of the garage concrete floor.

The EC5-connector on the circuit side is connected to the plates via bolts.

Close up on batteries. And then it's time to connect the EC5-connectors. Note that there are no fuses or on/off switches here. I'm relying on the BMSs alone to save my ass should something go wrong   Don't be stupid like me...

And we have lift off! Hard to say what the actual temperature is, maybe around 800-900 degrees Celsius?

The "multipoint" -connection on the plate-end works as planned: There's lower resistance between the bolt and the "really hot" part of the wire, as there's multiple windings of the Kanthal. Earlier experience has taught me that having the hot wire against larger, colder metal plate easily makes it break due to temperature difference.

I was a bit worried if the hot wire could actually make the spring bend enough to release the wire and drop it on the floor, but no problem there.

Measuring the voltage during discharge, 51.8V here. I haven't measured the current separately, but knowing that the circuit resistance is around 10ohms, that's pretty much around 5.18A (a little under 1.3A per pack).

The voltage starts dropping faster, the lower it is. Towards the end, around 48V, it starts to drop really fast, and leads to the BMSs cutting power once it drops below 44V or so (hard to say exactly the point, as it happens pretty fast, and the multimeter updates only a few times a second). After a short recuperation, the voltage had bounced back to around 53.8V or something around that, and then I ran it down to shutdown again.

After running the batteries to shutdown twice, I changed to a thinner gauge Kanthal. Here the circuit has 25.4 ohms of resistance, so around 50V, the total current will be "only" around 2A.

Discharging with the 25-ohm setup.

Here are two videos I shot of the BMS overdischarge shutdown. The first piece at the beginning (up until around 5 seconds or so) I started to record a bit too late (but just caught the point where it shuts down). The latter, slow-motion video is showing better the voltage drop "cliff" before it shuts down (the voltage starts dropping really fast):

 

I ran again the batteries to shutdown twice with the 25-ohm setup, and then decided it's enough. Time to check the charging side.

I hadn't made the GX16-3 connector before, so I needed to do that first. The "cups "of my GX16-3's are really small, so I ended up having to use 20AWG wire to fit them. Well, I'm not going to use large charging currents, GX-connectors can't take much more than 5A anyway   The black & red pieces of insulation tape are there to show me which is positive and which negative pin. I measured them with the connector attached to charger before starting to solder anything.

Close up on the solders, although not that crips image. Nothing special really, pretty easy to solder properly.

While making the connector, the batteries had recuperated a bit, up to 53.4V. I dug up my old Charge Doctor (V1) to measure the charging, but as I'm doing it in another room, I can't use my computer to capture the charge log. Also, I have to pay some attention when the charging goes to the constant voltage -stage, because these packs DON'T have the charging side reverse polarity protection diodes, and the Firewheel charger gives out 67.8V (4.2375V / Cell), presumably because the diode in the original packs' BMS has a forward voltage drop of around 0.6V. But now that there are no diodes, the voltage goes straight to the cells. Might be that the charging-side overvoltage protections cuts the charging at that point, and then I'll have to continue with another (4A) charger, but cannot use the Charge Doctor with that, so then I cannot measure the capacities.

Interestingly, the charger puts out around 2.8A (pretty close to the Charge Doctor V1 limit of 3A, if memory serves). With the original Firewheel-packs, the charging voltage usually started around 56V, and the current was only slightly above 2A.

Now, I'll just have to wait... It should take about 6-7 hours for the packs to completely charge. Will see whether they can already take in the full 768Wh, or if they need to be cycled first. Hopefully the house won't burn down...   And then have to repeat partially discharging them with the Kanthal-wire -setup, as I don't want to leave them laying around with full charge.

 

[esaj]

Time to revive this topic again, as it looks like I'm not going to get the packs to fit into KS16, and Gotway refused to sell me an ACM16 without batteries (although, I'd have to check the dimensions with that too first). As you can see from the above posts (moved from my Off topic-forum Headache-topic), I've (finally ) got the BMSs replaced and am currently working on testing the batteries. After the above initial testing (discharging as 16S4P & charging as 16S4P), I'll still likely test the packs again one-by-one and then start working on putting the whole thing back together. While I love to ride Firewheel, working on the shell is just such an immeasurable pain in the ass...

[esaj]

Looks like my Charge Doctor's current measurement has gone bonkers, need to measure & re-calibrate it at some point:

Yep, 943.9Wh and still not even in constant voltage phase   I've read that new high quality cells can go  5-10% above their rated capacity, but this is not plausible  Also, the CD was showing around 2.6A of charge current throughout the constant current phase, while normally it's been something like 2A.

The voltage measured from the discharge-side of the packs while charging was 0.8V behind the charging voltage throughout the constant current-phase of the charge.

Once entering the constant voltage-phase, I cut the charge after seeing that the voltage on the battery discharge-side raised to 67.3V. After cutting power, it drops to 67.0V, and after letting them sit for about half an hour or so, they settled to around 66.4V. As I can't get reliable Wh-numbers from the Charge Doctor, I'll do a full balancing charge on another charger later on that puts out 67.2V. Just to be on the safe side, I don't want to overload the cells    The packs did get slightly warm (but not hot) towards the end of the charge.

I tested the packs one-by-one running each for 10 minutes with that 10ohm resistance. The wire is pretty much white hot at here, at the start the voltage was over 61V, meaning that there's over 6.1A of current flowing, and the wire is dissipating roughly 61V * 6.1A = around 370W. According to the Kanthal-tables, that should be pretty much around 1000 Celsius.

During that 10 minutes, the output voltage of the pack would drop to around 57.4V (of course bouncing back up once power is cut). The packs did get clearly warm to touch, but not hot.

After testing the packs one-by-one, the voltages still recuperated to above 60V (don't remember the exact value). It was already late (3AM or so? ), but I wanted to push the voltages even lower for storage, so I put the packs together (they were all at the same voltage), and added another Kanthal to the setup, dropping the resistance to around 5.1-5.2ohms and then discharging all the packs through it for a good 20-25 minutes:

That's about 10-11A flowing through the wires (in total) at about 57V, so somewhere around 600W of total output. Afterwards the voltages recuperated to around 58.4V, which I think is just ok for leaving them at (around 3.65V per cell). I'll likely still later on test charging the packs one-by-one, but all in all, I think these are working just fine.

Next up, I need to start working on the shells, likely have to make some acetone-ABS -slurry to fix some stripped threads...

[Hunka Hunka Burning Love]

If you start adding a few more Kanthal wires to form a satanic pentagram, I'm going to start worrying about you esaj.    Stay on the good side!  You're going to open up a portal into the dark realm, and after seeing a bunch of late night movies I know it won't end up to be a good thing!

 

[Hunka Hunka Burning Love]

@esaj How's the custom battery pack mod going?  Any luck getting the shells together with all the packs installed?

[esaj]

11 hours ago, HunkaHunkaBurningLove said:

@esaj How's the custom battery pack mod going?  Any luck getting the shells together with all the packs installed?

I fixed the stripped screw holes with acetone / lego blocks -mix some time back, but haven't had the time to work on it since. Try to get around to it soonish, currently I'm finishing up another project that might take up to couple of days still ("just" wiring/attaching things/usb-port opening, one simple board & some software & testing to do ).

[esaj]

Well, this thing keeps on being disastrous... I got the MT60-connector (I thought I had ordered 2 pairs, as I usually get at least one more than I need, in case of problems). Easily soldered on the mainboard, but of course I then fucked up soldering it on the motor end on the floor (couldn't really see that well what I was doing in an awkward position)... the connector overheated and melted too much to be usable.   And no spare ones.

I put in an order for 10 pairs more (), but it'll likely take another month to arrive. Looks like I won't be riding much this summer...

I could use other connectors, but due to the very limited space, I cannot use any large connectors, don't have any that have 3 pins, and wouldn't want to repeat the entire dismantle & rebuild-process yet again once I get the MT60's. Guess I'll go back to calibrating HX711...

EDIT: Wow, at cheapest I found 2 pairs in a domestic shop (most don't even have them, or the price is like closer to 10€ per pair ) for the same price as 10 pairs from Aliexpress (10€)... So the usual "at least 5 times as expensive" -rule seems to hold, but at least they should arrive this week. We're not such a backwater country after all  (there are many connectors and electronics stuff you can't find domestically at all or the prices are just ludicrous, like 20x more expensive than elsewhere).

[Jrkline "Wheel Whisperer"]

On 6/4/2016 at 0:41 PM, esaj said:

Between the XT60 and the packs, I picked Deans. I must say that I really don't like soldering Deans, though. It's hard to get the wire to keep in place during soldering...

Having added numerous battery packs to my various wheels,I find it much easier to due the soldering if I drill holes in them first.Plus the fact the the connections are much more vibration tolerant.Why the don't come pre drilled is anyone's guess.

[esaj]

Looks like I won't be wheeling this summer... 

After receiving the new connectors, I replaced the connector I f***ed up the first time, and then proceeded to put the thing back together once again.

Yeah, same old problem: too many wires and connectors, fitting everything is a pain. At this point I also short-circuited the mode-control to sports-mode and tested that the mainboard still works, which it did. The welcome message plays ("Welcome to use Firewheel, sports-mode", and then, "please restart the unicycle", as it's laying on it's side and won't try to engage the motor), and the lights can bet turned on and off.

Going to be tight.

It took a good few attempts to get the wires so that I could close everything up. Finally I got everything in and the plate screwed shut. Of course at that point it's impossible to see what's going on in there.

After screwing the plate shut, I tried to power up the wheel again. Nothing, no messages or working lights. I unscrewed the mainboard and checked for any obvious damage, nothing visible. I did some measurements, and while the input voltages at the battery input are ok, the voltage at the regulator (ST LF33, 3.3V very low dropout regulator) input was 1.8V and output was something like 0.8V. So something wrong there. Looking around further, I lifted the large capacitor next to the inductor:

Well that doesn't look right. Spluttered soldering tin and what looks like flux. It could be that the spot was missed in cleaning, as it's covered by the capacitor (the white stuff on top and around the inductor is some sort of glue holding the capacitor in place), but I doubt it. I took a few more shots through a loop.

I then cleaned up the mess with a toothbrush & isopropyl alcohol:

The SMD-resistor values are as follows:

43Y = 273 * 0.01 = 2.73 ohm
03D = 105 * 1000 = 105k ohm
01B = 100 * 10 = 1000 ohm
56B = 374 * 10 = 3740 ohm
23D = 169 * 1000 = 169k ohm

As far as I could measure (my multimeters can't really measure values below a few tens of ohms very precisely, but the 2.73ohm seems to be ok, showing around 3 ohms on my meter), the resistors and the diodes are ok, so not sure what could be the problem. I haven't been able to identify the small chip next to the heatsink, it has markings on two lines:

3CY8  (or maybe 30Y8)
S8 1B  (not sure if there's a space in between)

Probably that's the chip controlling the step-down (that's likely what the inductor, diodes & resistors are for, dropping the battery voltage down for regulation to power the ARM-processor, gyro & such), so it could very well be the culprit.

Anyway, I do still have one working board (that's currently still attached to the laminate-piece that holds it in the custom-frame), but I'm pretty sure I can't get things to fit in there with these packs, so I don't want to fry another board    Maybe if I find the time and enthusiasm, I could try to rebuild the custom-frame, now that I don't need to fear cut-outs during braking...

In the meantime, I also have (yet another) project, that this time doesn't involve electronics or wheels, but "fun times" like moving 2.2 tons of sand uphill over some steep steps with plastic buckets & other stuff:

That's probably going to keep me busy for the rest of the week...

 

 

[Rehab1]

You are extremely impressive when it comes to electronics and computer skills! Way outside my purview but so interesting! 

On the other project, what are you building?

[esaj]

20 minutes ago, Rehab1 said:

You are extremely impressive when it comes to electronics and computer skills! Way outside my purview but so interesting! 

Well, I was always interested in computers ever since I was a kid, learned to write (simple) programs on my own when I was around 8 or 9 with a Commodore 64... now, about 25 years later, I've worked as a programmer (actually currently my title is something like "Server System Architect" or some other nonsense ) for about a decade. But, I must admit, that it has become pretty dull & boring (at least on the work projects), I'm more interested in the low-level stuff and my work is (nowadays) mostly simple business-logic & database design.

On the electronics-side, I'm just a hobbyist, and had someone asked me about a year or a little more ago, if I would ever consider doing electronics, I would have said no  And now I design my own power supplies, tools and other stuff...  Currently working on a LiPo/Nimh-powered linear supply, a milliohm- & milli/microvoltage-meters based on 24-bit HX711-chips and an analog function generator...  But like said, I'm far from a professional in that stuff.

 

Quote

On the other project, what are you building?

Oh right, it's probably not obvious from the picture because that wall's already (more or less) finished. We're paving the sides of the walls with those yard bricks & stones, so pretty much just digging & cleaning up all the soul, filling with fine sand, inserting filtering cloth and then putting the bricks & stones in place. Sounds simple, but it still takes a lot of time. Plus, our yard is a bit "difficult" when it comes to moving stuff up to the house, there's a slight hill and some steps, making it very difficult/almost impossible to use a wheelbarrow, so we had to move all the sand bucket-by-bucket from the side of the road... Took a few hours, and sure made my arms and legs hurt the next day   Maybe half-way through now, although the very last bit (that's actually further away from the house, next to a slope that leads to the road) should be pretty fast and straightforward. Been at it since monday, I'm slow...

 

[esaj]

The device marking "S81B" seems to refer to Texas Instruments LM5007, a high-voltage 80V step-down switching regulator.

http://www.ti.com.cn/cn/lit/ds/symlink/lm5007.pdf

Will try to measure the voltages at the pins tomorrow to see if it seems to match & if there's anything wrong with that.

[dmethvin]

Oh that brings back memories. Not good memories mind you...

 

[Hunka Hunka Burning Love]

Sorry I don't have much to add to this conversation, but tick tick tick cu... cut the RED wire.. no no the blue, the blue one!

and this link:

http://www.cabletiesandmore.ca/cable-management-products.php

[lizardmech]

Do you have any clearer pictures of the board? There's usually a 3.3v linear regulator powering the MCU and MPU. Also on the firewheel board do you know what gate driver they use? Also another thing I been have trying to find out, the allegro current sense, do they run a 5V version and use dividers to go to 3.3v logic or are they running the current sense IC on 3.3v? The firewheel board is more interesting than others because it's based on a higher end motor controller design than the other brands.

[esaj]

6 hours ago, dmethvin said:

Oh that brings back memories. Not good memories mind you...

Yeah, the step-down burning was somewhat common with older boards, thought they fixed it with the newer gens. Of course, something could have shorted the step-down, as the compartment is so full of wiring   In your picture, the first leg of the LM5007 doesn't seem to be connected "upwards", on my picture, you can see there's still a splutter of tin at the 1st leg, which might connect it to a wrong trace/leg of some component just under the heatsink-pad. Will try to remove that later to see if it helps.

5 hours ago, HunkaHunkaBurningLove said:

and this link:

http://www.cabletiesandmore.ca/cable-management-products.php

I've glued some of the wiring to the bottom of the mainboard compartment, but for the rest tying kinda hard, as some of the wires are very rigid, plus they need slack so I can close it in the first place. The original large batteries use very different wiring (4 * 4S4P packs in series), so only one set of power wires needs to go into the mainboard compartment. If you look at the earlier posts with pictures in this thread, I think you'll see what I mean...

3 hours ago, lizardmech said:

Do you have any clearer pictures of the board? There's usually a 3.3v linear regulator powering the MCU and MPU. Also on the firewheel board do you know what gate driver they use? Also another thing I been have trying to find out, the allegro current sense, do they run a 5V version and use dividers to go to 3.3v logic or are they running the current sense IC on 3.3v? The firewheel board is more interesting than others because it's based on a higher end motor controller design than the other brands.

I have some high-res pictures of an older gen board with the heatsink removed somewhere. Don't know about the gate-driver, will try to check and get some pictures later today of this board. All I remember is that the Firewheel uses surface-mount  IRF7759's:  

http://cdn-reichelt.de/documents/datenblatt/A100/DS_IRF7759.pdf

I don't think any other wheel uses surface-mount mosfets, plus none of them have that low Rds(on) at 1.8 milliohm. The tradeoff seems to be higher gate charge (around 200-300nC), so probably the gate drivers also have to put out some serious amperes to switch them on and off fast enough. Also, the Firewheel cooling solution is a bit weird, the aluminum block is relatively small and they use a thick-looking thermal pad between the mosfets and the heatsink.

[Rehab1]

@esaj Your EUC project and programming/electronic skills are beyond  my purview! I took electronics in high school, was a programmer for about 6 months on an IBM 360 using Fortran and Cobol but no longer possess those skills. I did design a system in 2000 called Intelisense that was interactive light and sound used to motivate children during therapy. It was based on MIDI. The system was to complicated for the average therapist to use so it never took off. Unfortunately I still have about 30 units still sitting in my pole barn. My son was an awesome programmer and had his own computer company but unfortunately he died a few years back.

Best wishes on your projects! 

[dmethvin]

13 hours ago, esaj said:

Yeah, same old problem: too many wires and connectors, fitting everything is a pain.

I definitely had the same problem and spent a bunch of time figuring out how to get some of the connectors into the battery compartment and handle area so the control board wouldn't be so crowded.

3 hours ago, esaj said:

Yeah, the step-down burning was somewhat common with older boards, thought they fixed it with the newer gens.

That board looks more like my older board that burned, but who knows how many versions they made! If you look at this album it shows that my older board had two electrolytics including a 330 that covered the step-down. The new board doesn't have that capacitor. 

http://imgur.com/a/8EObi

 

 

[esaj]

7 hours ago, lizardmech said:

Do you have any clearer pictures of the board? There's usually a 3.3v linear regulator powering the MCU and MPU. Also on the firewheel board do you know what gate driver they use? Also another thing I been have trying to find out, the allegro current sense, do they run a 5V version and use dividers to go to 3.3v logic or are they running the current sense IC on 3.3v? The firewheel board is more interesting than others because it's based on a higher end motor controller design than the other brands.

I quickly looked up the relevant chips on the board, I'll get you some pictures later tonight:

Controller:  STM32F103 (T6?)
http://www.st.com/content/st_com/en/products/microcontrollers/stm32-32-bit-arm-cortex-mcus/stm32f1-series/stm32f103.html?querycriteria=productId=LN1565

 

Gyro:  LSM330DC  (Not 100% sure about this, the chip has markings  330DC   2320   AA6EW
http://www.st.com/content/st_com/en/products/mems-and-sensors/inemo-inertial-modules/lsm330dlc.html

 

Gate driver:  HIP4086ABZ  (80V, 500mA 3-phase MOSFET driver)
http://www.intersil.com/content/dam/Intersil/documents/hip4/hip4086-a.pdf

Surprised that it's "only" 0.5A typical, apparently a single chip for all 6 gates.
 

Current sensors:  ACS709T  (3-5.5V supply), at least two pieces, a third might be under the heatsink (if there's one for each phase)
http://www.allegromicro.com/~/media/Files/Datasheets/ACS709-Datasheet.ashx
EDIT: There's a chip near the current sensors that seems like it could be a L78L regulator from ST in SO-8 -package:
http://www.st.com/content/ccc/resource/technical/document/datasheet/15/55/e5/aa/23/5b/43/fd/CD00000446.pdf/files/CD00000446.pdf/jcr:content/translations/en.CD00000446.pdf

The markings are something like "78L058" or "78LO58" or "78L0S8"...

 

Misc:

LCX541 buffer/line driver: https://www.fairchildsemi.com/datasheets/74/74LCX541.pdf
 

 

[lizardmech]

I can never work out what these are for, you can see the resistor for the FET gate and a diode to speed up switching, but those items marked L I have no idea.

 

[esaj]

Here are the hi-res images:

Direct link:   http://i.imgur.com/Yjf7c0I.jpg

Older board with heatsink removed:

Direct link:  http://i.imgur.com/HCi8sZs.jpg

Back-side of the older board:

Direct link: http://i.imgur.com/slnu7NR.jpg

There seem to be some vias, so probably there's at least a ground plane in-between, but don't know if there are even more layers (probably, because there are no visible traces around the MOSFETs).

 

4 hours ago, lizardmech said:

I can never work out what these are for, you can see the resistor for the FET gate and a diode to speed up switching, but those items marked L I have no idea.

"L" usually indicates an inductor, but not sure what for they're there... phase-shift compensation or some form of charge/voltage pumps or buck/boost converters?

[lizardmech]

They must be thin film inductors, I didn't know they had small ones like these. It's probably a 4 layer pcb, ground on the bottom, signals on layer 3, 3.3v plane on layer 2 and high voltage on the top layer. I can't work out why they put the inductors there, they are on the mosfet driver outputs, so it goes through the inductor then onto the mosfet gate. The diodes next to the ACS current sensors is strange as well, you can extend the current sensing range by adding shunts or copper paths in parallel to the sensor, but I have never seen diodes used like that.

I think this is the best of the chinese controllers, it's only really let down by the bad heatsink. With a CPU or GPU heatsink on there it would probably outperform the other EUC boards.