Develop a QCA7000 board?

[catphish]

Assuming it makes sense for this device to handle AC charge negotiation too.
For CP, here's a starting point.

[catphish]

I was thinking of supporting the WS2812 RGB LED (which is 5V, serial controlled, full RGB in one part), but if we want to be LIM-like then we need three separate outputs to control each color with a dedicated pin.

I had a quick look and I couldn't find a nice push button with a WS2812 LED. If such a thing exists then this would be great, but if not, we may have to go back to R+G+B high side switche, which seems to be the standard, and for which there are lots of nice panel mounted illuminated buttons.

[catphish]

Here are some general thoughts on overall design, though I may be getting ahead of myself here:
Connectivity should likely include:
* 8-30V power input
* CAN
* CP/PP/PE for both AC and homeplug
* 12V GPIO including mosfet-buffered 12V inputs, low side drivers, H bridge motor output for locking, high side drivers for LEDs
* Charge port temperature sensing (I don't know much about this one)
Hopefully most other IO can be done by CAN

You will need a programming headers for STM32 and flash.

Consider DTM13-12PA-12PB-R008 (i love this thing) or Modice ME-MX connector / enclosure.

[muehlpower]

Installed in the LIM

the high side driver for the LEDs is a VNQ5E250AJ-E
HV contactor driver negative BSP603S2L
HV contactor driver positiv BSP613P
driver for Locking Actuator L9951XP

[tom91]

driver for Locking Actuator L9951XP

That is a big chip to drive a locking actuator. Having a peak of 3A should be more then plenty.

https://www.phoenixcontact.com/en-gb/pr ... 2v-1624129

https://www.midaevse.com/electromagneti ... x-product/

[muehlpower]

That is a big chip to drive a locking actuator. Having a peak of 3A should be more then plenty.

https://www.phoenixcontact.com/en-gb/pr ... 2v-1624129

https://www.midaevse.com/electromagneti ... x-product/

I only listed what is installed in the LIM.

[uhi22]

Created a further draft version of schematic and board. Plan is not, to have a fully featured product, but a further step to gain experience and learn from the failures.
What is contains: QCA, power supply (one step down, two linear regulators), STM32, 3 analog inputs (for temperature sensors and PP sensing), CP logic, two low-side drivers for the contactors. Lots of test points to be able to solder add-ons to the unused controller pins.
What is not planned for this edition: Highside driver for the external LEDs, bridge for the lock motor, more inputs, automotive connector and housing.

I plan to send the data to JLCPCB end of this week. In case somebody sees major errors, any hint is welcome. As said, this will not be the last version, so we just collect the minor findings and feature requests for the future.
KiCad files are here: https://github.com/uhi22/foccci

[royhen99]

SN65HVD234 CAN transceiver needs EN pin pulling high. If not using enable pin better to use SN65HVD230.

[muehlpower]

I would like to look at the layout, but my kicad won't open these files. Am I doing something wrong?

[royhen99]

They open in Kicad 7.0.

[asavage]

They open in Kicad 7.0.

Yeah, I got caught by this, too:

https://www.kicad.org/download/linux/

. . . once you update to 7.x and start modifying a project, you will no longer be able to open that project in any 6.x version.

Fortunately, for Ubuntu, upgrading Kicad from 6.x to 7 is fairly painless: drop the old PPA, add the new PPA, apt update/apt upgrade, and when opening Kicad 7 for the first time, select default project location (import previous) and symbol files (default, for me) to use.

Then I could open Uwe's files.

[uhi22]

SN65HVD234 CAN transceiver needs EN pin pulling high. If not using enable pin better to use SN65HVD230.

Good point, thanks. I just cheated using the openinverter GS450H_VCU_V3 - Schematic.pdf, took the device name and the connections, but did not read the comment. Maybe these are the small little pitfalls added to the schematic, to avoid too fast and successful copying

Just pushed a fix now.

[catphish]

Comments on the proposed design so far...

Important
=======

1) You've populated the TPS54302 feedback divider with 2 x 10k resistors. These are almost definitely not the correct values for 5v output.
2) The SWD port should have a pin for 3V3_uC, and ideally also NRST. Since you're breaking out SWO, I'd probably put it on that connector too.
3) You've used the same 25MHz crystal for the STM32 and the QCA7000, but with different load capacitors. Was this intentional?

Less important thoughts
===================

4) You've used AMS1117-3.3 for the 3v3 regulators. These are tried and tested, but you may wish to avoid the polarized capacitor by using a more modern regulator. Consider TLV75733PDBV "Stable With a 1-μF Ceramic Output Capacitor"
5) One of your AMS1117 seems to have a 100nF output cap and the other doesn't.
6) There's no sensing of CP or PP. I imagine this will be needed both for plug detection and ultimately for AC charging.
7) The STM32F103 seems to meet the requirements and it's popular but be aware that it has a basic ARM core with no FPU.
8) The QCA7005 is labelled QCA7000.

I haven't checked the rest of the design for the QCA7005 itself, that's a little beyond my knowledge for now.

[Pete9008]

Any chance you could add a pdf of the schematic to the repo? I'd like to have a look but don't have Kicad 7 installed.

[Bigpie]

Attached

[Pete9008]

Had a look over the schematic (not exhaustive but fairly thorough). Looks good

I agree with all Catphish's points, in particular the processor, the F103 is getting a bit long in the tooth now (I first used it 15years ago!). Is the idea that this board is only powered up if the charging flap is opened or is it permanently powered? If the former the power consumption probably isn't that critical and the F103 should be fine. If the later then it would be worth looking at something like the stm32L4 series which offer similar (or better) performance but MUCH lower power consumption. If going this route it could also be worth changing the QCA7005's linear regulator to one that has an enable line so that the modem can be powered down when it is not needed (pulls about 190mA on 3V3 in Rx mode and 250mA in transmit, does it have a software power down mode that could be used instead?).

QCA7000 page:
What's the plan for the PLC coupling circuit between J7 and J1, have you found a suitable transformer?
Seems to be missing the JTAG port pull down resistors that are included on the Sergey's reference schematic and recommended in the data sheet. It will probably work fine with them pulled directly to gnd but..?

Controller page:
Could do with one more processor decoupler so that a cap can be dropped right against every package power pin.
Would be worth adding lines to the QCA7005 Reset (so that is can be reset without a power cycle) and the Int pins (just in case it's needed in the future).
What's going on with the missing processor Gnd pins, are they mapped within the part (sorry just looking at the pdf not kicad)?

Output Drivers page:
I'd remove the HCT04. The input threshold of the NCV is 2.2v max so it will be quite happy being driven direct from the 3v3 micro. Also with the HCT there if the micro power supply fails for any reason it could conceivably (unlikely I know) energise the contactors!
Not sure what the expected contactor current will be but I reckon that a 50C part temperature rise will be produced by around 1.75A with a standard footprint or around 2A if you add a 1" square heatsink pad.

Power Supply page:
The TPS54302 is only rated at 30V absolute max input, for automotive I usually like a little more (but appreciate that EV suffer less from load dump and noise normally associated with ICE cars). As long as the TVS is suitably rated it should be fine. Also assume that the 470u input cap is an electrolytic? (not a good idea to use a tant in this position)

Inputs page:
Might put a Schottky protection diode on the temperate sense inputs. The 47k and 100nF should do the job but just to make sure?

Edit - the Pulse BMU6201NL transformer might do (it's qualified for the qca700x, see https://www.pulseelectronics.com/wideba ... sformers/ ) and is available from Mouser

Edit2 - can't see a pull up for the processor reset line?

[catphish]

I'd remove the HCT04. The input threshold of the NCV is 2.2v max so it will be quite happy being driven direct from the 3v3 micro. Also with the HCT there if the micro power supply fails for any reason it could conceivably (unlikely I know) energise the contactors!

I'm inclined to agree. Datasheet says they should be happy doing 1A at 3.3V gate drive.

Power Supply page:
The TPS54302 is only rated at 30V absolute max input, for automotive I usually like a little more (but appreciate that EV suffer less from load dump and noise normally associated with ICE cars).

I gave this some thought and decided that 30V is more than enough for an EV +14V line, however I also decided to add a SMAJ18A at the input to the TPS54302 to protect against spikes.

Edit2 - can't see a pull up for the processor reset line?

I'm fairly certain the correct configuration for an STM32 is to leave the NRST line floating with a 100uF capacitor.

[muehlpower]

Can you add an analog port like this? In the LIM it is used to measure the DC voltage at the connector. So my board or that of larsrengersen could be used as an alternative to your Dieter.

[asavage]

Can you add an analog port like this? In the LIM it is used to measure the DC voltage at the connector.

This would certainly simplify my install.

[Pete9008]

I'm inclined to agree. Datasheet says they should be happy doing 1A at 3.3V gate drive.

Apologies, I hadn't read the datasheet thoroughly enough, the on resistance at 3V is a bit high and 1A is about the limit. Still not keen on the HCT04 though so would look for another device better suited to a 3V3 gate supply if more current is needed.

I gave this some thought and decided that 30V is more than enough for an EV +14V line, however I also decided to add a SMAJ18A at the input to the TPS54302 to protect against spikes.

I'm still a little cautious here, too used to ICE, but agree it's probably fine on an EV. Anyone know what the specs are for EV's here, doubt the standard load dump waveform would be applicable as the DC/DC would react a lot faster than an alternator so would assume a smaller amplitude?

I'm fairly certain the correct configuration for an STM32 is to leave the NRST line floating with a 100uF capacitor.

Just checked and you're absolutely right, the F103 has an internal 50k pullup on the line! (to me it just looks wrong without the resistor but it is the recommended circuit!).

Assume you mean 100nf

[mikeselectricstuff]

Apologies, I hadn't read the datasheet thoroughly enough, the on resistance at 3V is a bit high and 1A is about the limit. Still not keen on the HCT04 though

NPN transistor and pullup to 12V?

[Pete9008]

NPN transistor and pullup to 12V?

Vgs max for the device is only 14V so danger of over volting it

Wondering whether the HCT gate is the way to go after all, just use something like a HCT08 (AND gate so no inversion) instead so that a low on the input defaults to the contactor being off?

[muehlpower]

How about taking an NCV8401, it has 15A at 3V

[catphish]

Meh - I think uhi's original design is absolutely fine. No need to overthink this. A buffer IC powered from 5v with a pulldown on both sides will drive the thing just fine

[Pete9008]

Not really fussed but to me a 3V3_uC rail fail = contactors on is not ideal. Uhi's design but with a non inverting buffer instead of the 04 sorts it so seems worth doing to me?

How about taking an NCV8401, it has 15A at 3V

Not sure what you mean, are you looking at Fig6 in the data sheet, if so that's Vds not Vgs??

Edit - ignore me there I'm taking rubbish, that graph is just the VI curve, doesn't allow for power dissipation which is the limiting factor here.