Tesla SDU/LDU Update Topic is solved

[johu]

They do!

Just one little glitch, the Tag Connect is mirrored (I'd have to plug it in 180° rotated) so using the pin header for now.

Thanks so much, this is how open source works

[Isaac96]

Ahhhh!!!! So close!

I did have to manually make that footprint so I'm not exactly surprised.
Are the mounting holes reversed in relation to the pads? If so, that's a pretty easy fix, I can just reroute the traces.

-Isaac

[johu]

Yes, the alignment holes are swapped. If the "one" hole swaps place with the "2 holes" it would be perfect. The holes are a good fit, otherwise.

[Isaac96]

Yes, the alignment holes are swapped. If the "one" hole swaps place with the "2 holes" it would be perfect. The holes are a good fit, otherwise.

I'm just now having some time to take a look at this.
I've looked at the design, the datasheets, and the fabricated boards, and the pinout looks completely correct to me.
https://www.tag-connect.com/wp-content/ ... -CTX_1.pdf contains the pinout of the 6-pin TAG connector.

Code: Select all

1 VCC
2 SWDIO / TMS
3 nRESET
4 SWCLK / TCK
5 GND (also connected to GNDDetect)
6 SWO / TDO

Now if you take a look at the TC2030-IDC datasheet https://www.tag-connect.com/wp-content/ ... -Rev-B.pdf it shows the pin numberings. Bottom left is pin 1 (VCC) and top right is pin 6 (SWO, unused IIRC).

Comparing this to the boards as built, you can see connections that look to me to be correct.


Well damn. Looked at the PCB files and it looks like I put GND at pin 6 instead of pin 5. I will fix that and send updated Gerbers for the next runs.
SWO is ignored, correct?

-Isaac

[johu]

Yes SWO is ignored.
This is how it's layed out on V3 mainboard:

[Isaac96]

Excellent. I've just emailed you some Gerbers with the corrected pinout (it was just GND misplaced).

-Isaac

[Isaac96]

Two things to note.
1 - I just sent files without a board outline, that is in the next email.
2 - JLC/LCSC seem to have discontinued the MP2359!!!!! https://lcsc.com/product-detail/DC-DC-C ... 14259.html

Rather rude of them I must say. Not sure how to deal with this -- time for another regulator?
-Isaac

EDIT XRP7659 is pin and function compatible. Not available from JLC though; it's $1.36 at Digikey. I'll poke around and see if there are other easy replacement options.

EDIT2 Might have found the replacement - https://lcsc.com/product-detail/DC-DC-C ... f=jlc-SMT Pin compatible, seems to be functionally identical.

[johu]

Yes in the datasheet they tell you to use MP2331H - which doesn't exist

Replacement in their lib is RT8259GE

[johu]

Would you be up for an LDU update? I figured out a replacement circuit for the discrete current sensor oscillator, see attachment.

The comparator of choice would be TLV3201AIDBVR in a tiny SOT-23-5 package with low propagation delay. V1 and V2 represent the two ends of the 8 MHz crystal of the STM32.

All other changes like on SDU.

[Isaac96]

Absolutely! That looks great, it was the only real concern with doing the LDU. LCSC has 16,000 in stock, nice.

I'll probably have a first draft by Saturday or so (it's exam week for me); will begin with that exciter circuit and send to you for review as I'm not exactly experienced with HF design.

-Isaac

[Isaac96]

Alright I'm getting started -- at least downloading the files for now.
Is v4b the correct version? That is the latest available on Damien's github.

[Isaac96]

I've finished the regulator replacement and TAG6, about 2/3 done with updating footprints (haven't got the capacitors sorted). So now I'm looking at the current sensor clock system.

V1 and V2 represent the two ends of the 8 MHz crystal of the STM32.

If you look at the PCB files, the crystal is a few inches away from the clockgen circuitry.
So there are going to be some long lines somewhere with high frequency signals on them. My gut says that isn't great.
Should I extend the crystal traces or the op-amp traces? It's got to be one or the other.

Or just add another crystal? (though it might need some sort of kickstart... not really sure how those work)

-Isaac

[johu]

Oh, forgot to follow up on this. I think to not disturb the crystal it is probably best to keep the traces to the crystal short and have a longer one running from comparator to line driver. I'm thinking 8 MHz is not that high of a frequency. After all it is being transmitted via a rather lengthy cable to the current sensors.

[Isaac96]

Oh, forgot to follow up on this. I think to not disturb the crystal it is probably best to keep the traces to the crystal short and have a longer one running from comparator to line driver. I'm thinking 8 MHz is not that high of a frequency. After all it is being transmitted via a rather lengthy cable to the current sensors.

Sounds good. I'll have to make some space next to the crystal but it's a SOT23, pretty easy to fit there.

But what is the line driver? The v4a has the LTC1799 right next to the output; the clock goes straight to TP25 and the ICLK_OUT lines on the sensor harness.
Should another chip be added?

-Isaac

[johu]

By line driver I mean IC9 which is currently also being used to buffer the LTC1799

[Isaac96]

By line driver I mean IC9 which is currently also being used to buffer the LTC1799

Heh... Looks like I had literal tunnel vision, thought that was the output connector itself.

So I've placed and routed everything; had to slightly move a ton of RC networks to make good space for that trace, but all looks good now.
I've gone through BOM and updated everything, generated Gerbers and CPL as well.
Running through JLC, the inductor appears to be short of stock, so changed to a different part which is apparently identical.

Okay, good thing I looked at the visualization and BOM again. The schematic specifies NCV8401, an integrated low-side driver/MOSFET package. JLC doesn't have those. They have NCV8403 which is a drop-in, but that's out of stock. ST has the 'Omnifet' line but those are equally unavailable for assembly.

Hmm... What should be done? Looks like Damien used to hand/oven solder those parts. Johannes, what would you prefer? Should I leave the footprint there for your use or figure out a new FET and driver solution?

-Isaac

Bonus pic of the board as JLC imagines it.

[johu]

Thats great news!
I think simple FETs would also be ok. I can browse around.
The SDU uses the larger footprint BTS133TC, would these fit?

[FJ3422]

... The schematic specifies NCV8401, an integrated low-side driver/MOSFET package. JLC doesn't have those. They have NCV8403 which is a drop-in, but that's out of stock. ST has the 'Omnifet' line but those are equally unavailable for assembly.

Parts availability is currently a general issue.. You could have a look at the AUIPS2041LTR. I use it as a self-protected low-side contactor switch. Limits at around 5A. Currently in stock at JLC, but already discontinued at the manufacturer website. Higher powered alternatives may me the BTS133 or BTS3028, however also not largely stocked at JLC.

Also watch out with the polarity of the diodes; last week I corrected the polarity in the placement file before upload (of some BMS-master design), and noticed after a few days that JLC rotated them during their 'DFM-Analysis'. They could not tell my why, silkscreen indications were OK, so they rotated them back after a short chat-conversation.

[Isaac96]

The SDU uses the larger footprint BTS133TC, would these fit?

TO-263 vs DPAK -- should be plenty of room. I'll whip up another revision, should be ready by tomorrow, then hopefully all is ready to go.

Parts availability is currently a general issue.. You could have a look at the AUIPS2041LTR. I use it as a self-protected low-side contactor switch. Limits at around 5A. Currently in stock at JLC, but already discontinued at the manufacturer website. Higher powered alternatives may me the BTS133 or BTS3028, however also not largely stocked at JLC.

Also watch out with the polarity of the diodes; last week I corrected the polarity in the placement file before upload (of some BMS-master design), and noticed after a few days that JLC rotated them during their 'DFM-Analysis'. They could not tell my why, silkscreen indications were OK, so they rotated them back after a short chat-conversation.

Thanks! I'll be sure to double and triple check those markings. We had no trouble with the SDU board but there's always room for mistakes.

-Isaac

[Isaac96]

Switched over to BTS133, copied the footprint from the SDU board.

But now JLC is spitting 'system error' when I give them a BOM or CPL. This happens with old known good files so it's an issue on their side, I'll try again later like they advise.

Almost done...

-Isaac

[Isaac96]

JLC's systems appear to be working again and the v5 SDU looks good in their system.
The usual small rotation issues but all the pin1 markings are correct.
Johannes, I will email you the design files and BOM/CPL immediately.

-Isaac

[FJ3422]

Hi Isaac / Johannes,

Will the V7 design files be made available for us hobbyists at some moment ? I would like to modify a couple of things to my personal taste (different (possibly isolated) CAN-transceiver, ESD-protection on the CAN, terminator on solderjumpers, additional high-side signal outputs for the contactors (to connect to a BMS)). Would prefer taking the V7 files as a base, but if they will not be released (which I would fully understand) I will take the V6 files to start with.

@Johannes, is the tacho-output programmed / configurable in the STM32-application, or isn't it implemented yet ?

Thanks !

[Isaac96]

I don't know what Johannes plans to do, it's his choice.
Generally once the boards are available for sale, the files are published (with or without the JLC fabrication files). That is at least the precedent set by the great Damien.

-Isaac

[johu]

Yes the tacho output is programmed, just not connected and yes I will publish the files in a month or so.

[Jack Bauer]

If you do such modifications please feed them back into the community.