Cheap FOCer 2 (Open-source, Low-cost, VESC 6 based ESC) (v0.9 Release. Beta testing ongoing)

So in the event the battery suddenly doesn’t receive the regen current, then the bus voltage will rise proportional to the amount of current. The large electrolytic caps can help absorb it up to a point but eventually the voltage will exceed the limits of the hardware and cause failure.

5 Likes

Sorry, I wasn’t clear. Appart from the through hole component, heat sinks etc.
I meant the SMD components, that are hard to solder by hand, or the errors made by the manufacturer, that are not easy to debug.
So from what I understood, the drv problem is solved.

Ilias

1 Like

I’m a little confused as what you’re trying to ask… There’s SMD components that need to be soldered but they are large enough that they can be done with a regular iron and patience.

As far as errors go thats kinda luck of the draw from JLCPCB.

I’m not sure if anyone has ordered since the DRV problem has come to light so it’s hard to say for sure if they’re still using ones from the bad batch. I would hope not.

Does anyone having the 5V buck regulation problems have a test to distinguish a poorly soldered DRV from a defective one? On my device:

  • DRV PowerPAD shows continuity to 5V ground with a thin probe through via
  • EN_BUCK switch operating properly
  • D6 is populated with SS110 part
  • DRV from (as surely you can tell garbage) lot 88ASYCTG4

Here is what I see on the 5V and VCC supplies as I slowly bring up the input voltage on a power supply.

Input Voltage Behavior
0-3.7V VCC and 5V at zero
3.7 V VCC jumps to 250mV, 5V to 1V
3.7V-4.5V decaying sawtooth appears on 5V and VCC with 20ms period
4.5V-7.3V sawtooth period and amplitude reduces, humming audible
7.3V-8V ripple disappears, VCC stabilizes ~3V, 5V stays roughly equal to input
~8V VCC still ~3V, D1 pin 1 on 5V smokefails

Anyone that wants to replicate this be careful, the U1 VCC LDO datasheet says a max voltage of 15V is allowed, but the D1 hall diode array will fail before that and probably the CAN chip too.

2 Likes

Yes @doomy has discovered a test by using the diode tester function of a mulitmeter. Its in this thread.

2 Likes

v1.0 Change Checklist

  • gate drive improvements for higher stability/reliability: :white_check_mark:

  • better placement of large caps: :white_check_mark:

  • Comms pinout same as normal VESC :white_check_mark:

  • Bigger pads for the SMT caps: :white_check_mark:

  • Standard servo header for the PWM input: :white_check_mark: I’ve made a footprint that accomodates either the “standard” servo header or the JST one like the normal VESC.

  • JLCPCB-specific component position format: this will happen. It’s just a matter of trial and error to get the orientation the way JLCPCB wants it.

  • Thermal reliefs: :white_check_mark:

  • Hole or pad for vsense wire for remotes, connected to POS::white_check_mark:

  • IMU that JLCPCB can solder. This one is controversial right now. JLCPCB can support the MCU-6050 but that part’s days are numbered. The BMI160 is newer and supports more axis in case the VESC firmware one day supports the unused axis. Maybe it’s worth two variants of the design so that either could be supported. Let me know what you think.

11 Likes

Is there enough room on the board to have pads for both? And leave the option to the end user. If not 2 variants would be nice.

Any idea when the new design will be posted to github? :drooling_face:

2 Likes

idk if I could squeeze both onto the PCB.

I’m looking to get v1.0 out there within the next couple of weeks. I wanna knock it out and then get back to work on my high voltage stuff.

4 Likes

I haven’t done anything with the IMU yet, but I’d opt for the BMI160 over something that will be depreciated. If its not too much effort doing both (two different versions) would be nice for the masses. Edit: to better outline my position here, Id opt for BMI160 in lieu of both if it means that V1 would arrive sooner.

Additions to the Github for v1 (or the current revision) may want to include a PCBWay guide as well, being that JLCPCB/LCSC tend to have some supply/stock issues (namely DRV8301) and having a secondary option could help improve accessibility. That said, I am working on ordering through PCBWay and testing out their ‘turnkey’ options, if it works out ill submit a guide to the Git.

6 Likes

Yeah an alternative to JLCPCB for the guide would be great!

4 Likes

I am not sure if that is really important to others as well, but I think that current button type is a bit awkward, i.e. normally open. We discussed this and perhaps normally closed button would be better approach and those who do not need it, simply can but a jumper on. That is just a suggestion :slight_smile: Still, thanks. I have tested my scooter yesterday and I plan to upload some video or something shortly if boards do not explode by that time :slight_smile:

@shaman So following up on the PCBWay Guide/Pricing.

I still need to push in another quote request for simple SMT assembly rather than Turnkey, but I can say that the Turnkey prices are rather exorbitant.

Note These have not been purchased.

So not counting shipping Turnkey (which means all components soldered) sits at around 73.90USD per CheapFocer, add another 15 dollars or so a piece for some misc parts such as a heatsink, case, etc. This turns out to be just as expensive as any other ROTM v6.

As I said, I’ll check to see where we stand on pricing when the through hole components are removed from the quote. Though I cant imagine that it will drop the price 200 dollars.

5 Likes

They wont drop much. Only plus side with pcbway is that if you get it fully assembled with the imu and stuff and order like…20pieces, its worth it.

2 Likes

Well, I ended up disabling overvoltage protection of batteries… those cheap li-ion batteries of hoverboard are less important and I do not expect long-lasting performance. Killing VESCs seems way worse as that will disable usage of scooter immediately whereas more current to batteries might not make them happy, but that might not result in immediate breakdown. As I do not have mechanical brakes, I really need good regen.

We also tried to make overvoltage protection circuit with FETs and zener diodes to protect batteries and emit surplus energy to body of scooter. But releasing ~1kW of energy would require a lot of FETs and that seemed too pricey.

1 Like

Holy hell. I’m pretty sure I saw this on an earlier batch of TB6 and it was driving me insane. I need to go check some things and I’ll report back.

Have you run this for an assembly quote through Gold Phoenix? I’ve found them to be less expensive than pcbway, but their parts inventory on hand is limited so they import more complex components in from digikey. GP ive worked with for a long time though, solid.

4 Likes

!

I actually use 4 of those hoverboard batteries in my ebike. I made connectors so that I can configure them for all parallel or 2S2P(referenced to the modules themselves). This then gives me whatever voltage is appropriate for testing my controller. So far I haven’t had an issue with regen braking and I have the battery regen current set to about 20A.

7 Likes

Thanks a lot, that is good to hear. We also learned here at our Makerspace, that those hoverboard batteries tend to vary a bit. For example mine do not charge more than 40.7V. They charge up to ~42V then they discharge down to 40.7V. That happens to all I have that I purchased from the same vendor. However, some of folks here have good ones that charge up to 42V.

By the way, one suggestion is to add optional buzzer to next version. When VESCs are enclosed (as it is in my scooter) it is hard to hear when it is fully armed especially when going through pedestrian crossing (as I tend to turn scooter off then).

Up to know, around 30km covered and no issues. DRVs are from the same faulty batch (however, two out of 10 show signs of faulty ones (diode testing method)). Two VESCs for two hubs, 10". Each motor current 16A, regen ~20A. Batteries 36V. The only thing I notice is minor jerking of motors especially at the end of braking, when scooter is already quite slow. It seems like some sort of current limiting, but since speed is very low at that point, it is really weird.

I haven’t, but I will thanks for the suggestion!

1 Like

Well fuck…

2 Likes

“not in vogue”? Curious what that means. Either way, doesn’t sound good for the DRV8301 to remain in stock. Not the best for the CFOC2…but that doesn’t mean I’m stopping :slight_smile:

1 Like