Honda Insight Forum banner
801 - 820 of 830 Posts

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #803 · (Edited)
Best case would be end of November... that's if there aren't any major problems with RevC PCB (which I won't receive for at least another week). @Bull Dog is a beta tester, so I'll come up with a more minimal 'quiz' for that kind of early adopter. It'll probably only have a few questions, but all TBD.

...

Unrelated:
Here's a proof-of-concept video showing the mechanicals that adapt an 18S module to 12S:
For those that haven't been following along, this is useful because LiBCM will only initially support 48S (i.e. QTY2 18S & QTY1 12S modules). There are probably around QTY800 18S modules (enough for QTY400 cars), whereas there are probably only around QTY80 12S modules (enough for QTY80 cars). These are both guesses. Anywho, the 12S availability is no longer an issue.

FYI: Those leftover QTY6 cells will be useful for a companion product I intend to release sometime in 2022. I haven't specifically announced the product yet, but essentially it will replace the OEM lead acid 12 volt battery with a 3S2P lithium equivalent (i.e. QTY6 lithium cells). That will remove around 25 pounds from the vehicle. I don't intend to answer any questions about this product just yet, as it's a bit ethereal... but it's coming.
 

·
Premium Member
Joined
·
110 Posts
Yeh I’m ready when it is. 😎
 
  • Like
Reactions: prototypist

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #805 ·
RevC PCB shipped!
I always love the small glimpse of the PCB that Seeedstudio sends:
Electronic instrument Computer hardware Font Electronic component Electronic engineering


Seeed isn't the cheapest PCB manufacturer, but they do GREAT work! We use them for all our PCBs. If you're ever looking for a PCB manufacturer, I recommend Seeed. And no, they didn't pay me anything to say this.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #807 ·
Can it be assumed that 18S+/18S-/18S+ is the ideal 54S pack configuration?
No idea... haven't done the research yet. It'll either be 18S+/18S-/18S+ or 18S-/18S+/18S-. If I had to guess, I'd say 18S+/18S-/18S+ is the more likely "ideal" configuration.

Note that 54S is not initially supported... only 48S is.
 

·
Premium Member
Joined
·
110 Posts
I finally opened my boxes and I have the 12s+ 18S-x2, 18S+X2. So one ideal set for this, one 18S+ to break up for the future unnamed project, and one potentially for an 'i'm tired' assist on my Greenspeed trike.
 
  • Like
Reactions: retepsnikrep

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #810 ·
Correct:
-An 18S+ battery has the BMS connector on the positive side.
-An 18S- battery has the BMS connector on the negative side.

...

The RevC PCB should arrive tomorrow. However, my Pick and place machine isn't working for some reason, which is going to make assembling the PCB super annoying if I can't fix it tonight. The built-in touch screen stopped displaying anything... I can tap on the screen (where I know certain buttons are) and the machine responds correctly, but I'm not able to build boards without feedback from the machine while it's placing.

Super frustrating.
 

·
Engine-Off-Coast
Joined
·
2,126 Posts
Automotive tire Steering part Motor vehicle Steering wheel Audio equipment


New LiBCM screen. I wanted to match the black and white OBDIIC&C screen I have.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #812 ·
The RevC PCB has arrived, and I am going to place it by hand... pick and place machine is dead in the water. I've never had any problems with it, all the way back to 2015... but now the screen is broken. Manufacturer says the machine is too old to repair, so I'm looking into reusing the hardware with openPnP... that's obviously going to delay things a bit.

Two other options for production boards:
-Buy a new machine ($$$$).
-Send it out to a contract manufacturer... but that's a logistical nightmare and I lose a bunch of agility.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #814 · (Edited)
It's a generic LCD, but is no longer in production. Last night I purchased several items from the same era with similar looking screens... maybe one will be the same. It's a long shot, but if it works it'll save me ~$8000.

Here's the first hand assembled RevC PCB, and a walk through showing the new addition to the manufacturing line (in my garage):

Now it's off to start testing whether the various improvements actually work.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #815 · (Edited)
Apologies on the silence these past couple days... I've been diligently validating all the various changes on the latest PCB. So far all is well, although I've had a heck of a time plugging connectors in properly... let me tell you, if you plug the wrong BMS connector into the wrong port, bad things happen:

In the video I say "I'm not going to replace the resistors that measured correctly." It turns out that was a mistake... in the future, the most headache-free repair method is to remove ALL components that aren't isolated from the LTC6804 that spans both packs. Of course a better solution is to not plug the connectors into the wrong ports.

...

Next I've been hammering out the final battery current sensor circuitry... you'd think I'd be done with this by now, but in RevB there was some instability, which worked well enough, but wasn't perfect. Notably, there was some ringing around 30 kHz (~15 mVpp), due to a poorly configured feedback node. Here's what the original RevB circuit looked like:
Rectangle Slope Font Plot Terrestrial plant

Certainly not the most well behaved closed loop feedback circuit I've ever created, but stable enough to get the job done (with some oversampling and post-processing in firmware).

...

For RevC I thought "I'll just throw a lowpass filter onto the output, and while I'm at it I'll match the impedance on the opamp's inputs." Conceptually these 'fixes' were 'so simple' that I didn't bother to model the small signal stability... my rationale was that if it didn't work, then I could simply depopulate the new parts, which is certainly true...

...but down the rabbit hole I went. After modeling the circuit, it turns out my 'simple fixes' added two nasty poles that made the ringing quite a bit worse ;). After verifying the RevB behavior (by removing said components), I finally fired up LTSPICE and modeled the back-of-the-napkin RevC circuit:
Rectangle Font Slope Urban design Parallel


Holy smokes, Batman, who designed this circuit?
Look at that peaking around 30 kHz... that explains the ringing! Seriously, how is this circuit even stable? It looks like a poorly-tuned oscillator.

It seemed obvious to me (in hindsight) that I needed to add an integrator element (i.e. capacitor across feedback resistor), and also move the 100 nF capacitor (C_R56R57) between the gain resistor (R66) and the offset resistors (R54/R55/R56/R57). This substantially stabilizes the opamp:
Rectangle Font Slope Parallel Terrestrial plant


The peaking is entirely gone, and as such the opamp is now rock solid... there's no longer any reason to oversample the current (in firmware), but I'm going to leave it there anyway because it'll still reduce digitized environmental noise. I'll take 108° phase margin and 60 dB gain margin!

At this point I believe I'm 100% finished with the OEM current sensor circuit... it's been a real learning experience. I'm kicking myself for not modeling the 'simple' changes prior to ordering RevC, but fortunately none of these changes require a board spin... just value changes (and a couple stacked components).

...

In regards to the HVDC isolation between the ribbon cables and the (new) temperature sensor cables, I've decided to wrap the ribbon cables in flat-wrap. This will provide a secondary isolation sleeve between each HVDC BMS cable and its respective temperature sensor cable. That satisfies UL's safety isolation requirements. I believe that was the last lingering mechanical isolation puzzle piece.

...

Otherwise nothing much to report... tomorrow I'll finish up testing and then throw the RevC hardware into the car. Hopefully my time spent thoroughly testing each module on the bench will prevent any new surprises from surfacing in the car. TBD.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #816 ·
Safety Cover Update:
I've finally received a reasonable price and timeline quote for the polycarbonate covers. I should have QTY100 polycarbonate covers in 3 weeks. These will be indestructible, unlike the acrylic prototype covers, which crack if you look at them wrong.
 

·
Premium Member
Joined
·
2,036 Posts
Woohoo! That's awesome. Acrylic is such a PITA to work with.
 

·
Linsight Designer
Joined
·
2,677 Posts
Discussion Starter · #819 · (Edited)
Can you use different connectors for the harness tap to pcb interlinks so it's physically impossible to do what you did?
Not as long as the (identical) 18S OEM battery connectors are still used. Even if I did key the connector (at the LiBCM PCB), if a customer swapped the cables (i.e. bay 2 adapter plugged into bay 3 battery), then they'd experience the same issue.

Note that (accidentally) swapping connectors wasn't possible on the RevB units, as the BMS adapter cables were much shorter. Shorter cables only worked on the RevB units for two reasons:
-the only supported battery module configuration was 18S-/18S-/12S+. With the various configurations, the BMS cables must now be longer (to allow the same cable to be used on all permutations).
-the mating connectors were straight (i.e. not right angle). With right angle connectors, the cable is more difficult to plug in, hence it requires more "pull-away" space for hand access.

As long as LiBCM supports more than the 18S-/18S-/12S+ configuration, this is going to be one of those things that I'm certain some people will (accidentally) do... I'll stress the importance many times over in the (new) installation videos (to be filmed).

Honestly I'd much prefer to remove the OEM BMS harness/connector entirely, and then route a (new) ribbon cable from LiBCM all the way to each cell... but that's more work for the customer. However, if we do take that route, then it's trivially easy to swap any 18S+/18S- pack to 18S-/18S+, respectively. That would allow anyone to use just one standard configuration - regardless of which 18S modules they have - but it would require customers to remove the BMS bolts on each and every module, then flip each cell around (as needed, in order to make QTY1 18S+ & QTY1 18S- module).
FYI: The standard configuration would be 18S+/18S-/12S+.

Thoughts?
A: Does it make sense to remove the OEM harness/connector?
B: Should there be just one standard 48S configuration (18S+/18S-/12S+), which requires customers to flip cells as needed?
 

·
Registered
Joined
·
101 Posts
I'm for one standard 48S configuration (18S+/18S-/12S+), if it only requires flipping cells as needed as I have 2 x 18- so I would probably be doing it anyway to get the best cable lengths.
 
801 - 820 of 830 Posts
Top