work in progress Common Rail MC33816/PT2001 dev board - 4 injectors GDI

[960]

Ah, ok..

Why not use the gerbers and bom from the OEM eval board?

It's all on NXP's site..

[AndreyB]

@Noxz found the issue - top side 10R resisters had 10K parts mounted by mistake. Once 10K was replaced with 10R and J5 was jumped with J6 (no idea why we have this) and DRIVEN pin was hand operated (we do not have FW for DRIVEN right now), we've got 60V on 0.3 board!

I am hoping that @Noxz would restore the damage board and add a couple of code around DRIVEN soon

[Noxz]

Not J5 to J6, but J5 to specify to use the xtal onboard, and J6 to use supplied VccIO (which is 3v, not 3.3v, from discovery).. these are carry overs from the eval board. They will be removed for next revision (or just remember to solder-jumper/0R them and not install headers)

[kb1gtt]

NXP files would make the product you can purchase from NXP. It's cheaper to just buy NXP's then it is to build your own. The goal is to have a validated KICAD project which has BOM information populated and validated footprints, which we can eventually make into our own PCB layouts.

[AndreyB]

We have seen these boards produce 60v while in Washington state.

Nich, Chris and Matt are now trying to get 60v produced in New Jesery. Some validation code was added today.

[AndreyB]

Finally we have same board generate 60v in NJ!

This video would not have happen without Chris, Jared, Nick and Matt!

[AndreyB]

[AndreyB]

MC33816 boards keep travelling back and forth between New Jersey and Washington

So far a number of lessons were learned in relation to all this MC33816 drama:

10K and 10R resistors are very not interchangeable.

MC33816 likes it's 5v to be pretty close to 5v, it even has relevant status bits to report it's unhappiness about 4.x volts instead of 5v. spec'd limit of 4.75-5.25

SBR10200CTL diode while it shares package with a MOSFET does not work like a MOSFET

There is progress! it looks like we will have soon two or three boards which look and act the same!

[AndreyB]

A couple HW issues reported, that's a lot of head banging

https://github.com/rusefi/rusefi/issues/1434

[Noxz]

This is a very useful graphic showing the main configuration points of an injection cycle

[AndreyB]

Q: how special are mosfters or diodes we use on the board right now? Is there a safe opportunity to replace more components with JLC parts?

We really need to simplify our assembly, we have proven that we mess things up

[AndreyB]

Version 0.3.2 of the board pushed! The board is now single-sided (still 4 layed PCB but all components are now on one side) and JLC is assembling some MOSFETs now. $130 for 5 or $190 for 10 so looking for some review before I spend the money?

[AndreyB]

Arrived. One board goes to WA.

[sordfish]

Nice progress!

Are there any boards available to the UK?

[AndreyB]

Are there any boards available to the UK?

All the files to fab are published but those boards are a bit useless until the software is done.

[kb1gtt]

The attempt of a boost voltage valid LED on R0.3.2 does not work. It appears the diodes do not drop the voltage enough and U1 gets destroyed. The below circuit was suggested in Slack by @mck1117 and was drafted in LTSpice. This circuit or similar should probably be Incorporated in future spins of this board.

DI_Boost_V_Indicator_25V_and above_indicator_LED_LTsim.png (273.2 KiB) Viewed 32090 times

DI_Boost_V_Indicator_25V_and above_indicator_LED_LTsim.jpg (984.38 KiB) Viewed 32090 times

[AndreyB]

The attempt of a boost voltage valid LED on R0.3.2 does not work.

Would anything have to be removed from 0.3.2 assembled to reduce risk of failure?

[kb1gtt]

Remove U1 would disable the VBoost indicator and would prevent issues on R0.3.2

[mck1117]

Remove U1 would disable the VBoost indicator and would prevent issues on R0.3.2

I just removed one of the zeners since they're easier to desolder. The working board I've shipped to Andrey already has a zener removed.

[AndreyB]

And we've promoted the board to PT2001 - this chip was confirmed to be 100% pin compatible, with minor differences on software side.

[mck1117]

missing zener

image.png (212.05 KiB) Viewed 32058 times

[mck1117]

Here are the details of how I had it connected:

image.png (17.65 KiB) Viewed 32057 times

And some photos of that:

image.png (7.52 MiB) Viewed 32057 times

image.png (5.99 MiB) Viewed 32057 times

[AndreyB]

wiki page https://github.com/rusefi/rusefi/wiki/MC33816-PT2001-dev-board

19) Can we connect connect TP6 to VCCIO since both are 3v?
20) silkscreen font should be larger for DRIVER RSTB etc https://raw.githubusercontent.com/wiki/rusefi/rusefi/Hardware/MC33816_test_board/MC33816_test_board_3_2_assembled.jpg
21) silkscreen for FLAG0 etc is covered by header plastic https://raw.githubusercontent.com/wiki/rusefi/rusefi/Hardware/MC33816_test_board/MC33816_test_board_3_2_assembled.jpg

[AndreyB]

Next step of hardware should probably be a standalone board that daisy chains off the high-z injevtor outputs of a normal ecu, and contains its own mcu and canbus. In Borch shell case with 121 pin connector please looking more or less like https://rusefi.com/forum/viewtopic.php?f=4&t=1564 just for 4 cylinders to keep things simpler.

[AndreyB]

It would be amazing to share some parts with MED 9.1 just to make it easier for people to make our board without access to a fancy distributor.

[mck1117]

It would be amazing to share some parts with MED 9.1 just to make it easier for people to make our board without access to a fancy distributor.

Not worth it. You need the MC33PT2001 anyway, and the caps/inductors/etc are only a few bucks total. I don't think "can harvest parts from assorted Bosch OE ECU" is a good thing to optimize for.

[azm]

Hi everyone,

Hope you doing well

I would like to take part in MC33816/PT2000 testing/using. We have a 1.2 TSI with all components (down to including even the low pressure pump) and complete installation laying around and was thinking, why the heck not.

Couple of questions:
1) Any website suggestions where to order the PCBs + Components?
2) There are PT2000 and MC33816 gerber available. Which one is more mature?
3) Any other tips/hints would be welcome.

Best regards!

[AndreyB]

jlcpbc for PCB+80% assembly
mouser/digikey for the rest
MC33816/PT2001 is in better shape

[AndreyB]

I've somehow burned microRusEFI while bench testing direct injector with a fused black 12v 5A wall power supply. General layout like on this photo just with all wires and oscilloscope attached.

Step 1: MRE+PTE2001+test injector: got three clicks as expected. So injector #2 channel was at least clicking things.

Step 2: hooked up my Rigol, clicked "bench test #2" again and got two weaker clicks instead of three. Next thing microRusEFI went dark.

1Ohm between GND and 3v on MRE, same 1Ohm between GND and 3v on MRE with 3v regulator removed. Sounds like a short inside stm32?

My Flir was not charged but I believe TLE8888 was really warm and power supply area of TLE8888 was pretty hot - but we know that's kind of expected without heatsink. No idea if related or not.

[AndreyB]

TIL that oscilloscope GND on the probe is connected to my house GND which is connected to my desktop USB cable GND. This combined with the fact that I've incorrectly attached oscilloscope probe to one of injector wires is currently the best explanation of MRE damage.