[info] Frankenstein - cheaper ECU shield / official / legacy

[russian]

Status: version 0.1 is fully functional!
2019 Status update: you really want Frankenso which is the newer and more capable design viewtopic.php?f=4&t=569

Frankenstein board wiki
pinout
Available @ Tindie







https://github.com/rusefi/rusefi/tree/m ... ankenstein
(older source forge links have same exact content anyway)
frankenstein.zip
Schematic.pdf
BOMs: digikey detailed / digikey aggregated


This board consists of:

12 channel analog input - LMV324ID
12 channel low side - VNS14NV04 (injectors & solenoids)
6 channel hi side (ignition modules) - TC4427
VR input - MAX9926
CAN bus- SN65HVD230
microSD & USB TTL - FT232RL

So this 10x10cm board would be sitting above or underneath the discovery board like this

I guess it's better to have the board underneath the discovery so that I can still have access to the reset button & the LEDs. On the other hand I need access to the components on the Frankenstein board so soldering the boards together is probably not an option.

ibom available at https://rusefi.com/docs/ibom/

[russian]

Q&A

Q: I see Frankenstein has a USB serial adapter on it. Will I be able to communicate with the ECU without that particular port?
A: You can use any $2 USB-UART from ebay to replace the FT232 functionality (worth noting that CP210x have lame reputation, I know from my own experience that CP2102 Windows 7 drivers are giving me BSOD too often), you can also get away with just the micro-usb on stm32f4discovery (worth noting https://sourceforge.net/p/rusefi/tickets/46/) See also: http://rusefi.com/forum/viewtopic.php?f=5&t=210




"m3 standoff" that's what these are called... gotcha

Attaching stm32f4discovery gerbers so that later I can open them using http://sourceforge.net/projects/gerbv/files/gerbv/ and find the best spots to drill...

[russian]

First post updated with the new picture - I think I have a alpha2 version!

Known issues for tomorrow:
1) merge 3V and 3.3V nets
2) more mounting holes

Other than that, I believe I have a board which combines 12 analog inputs with 12 low-side outputs with cam/crank with CAN. I think it's pretty cool.

[hasse.69]

Damn you have been working.........
Looks impressive.

Best regards , Hasse.

[russian]

It's probably too early to think about it, but I wonder if maybe we should keep this design forever - 10x10 CPU board on top with some super-common stuff like SD card / USB interface, and IO board bolted from the bottom?

Question is what kind of connection between two board would withstand automotive application?

[kb1gtt]

I don't see much reason why it can't all fit on 10cmX10cm. Long term I don't think it will be 2 boards, when it can be on board the size of your hand.

[russian]

I don't see much reason why it can't all fit on 10cmX10cm.

I probably agree that a board which would suit 80% of cars would fit on 10x10cm (4x4 inches), but I doubt the same 10x10 PCB would work for _all_ cars. So, either we would need a number of boards, or we would need a 'base' board and 'expansion' board?

[russian]

todo: need to write a script which would produce a full BOM

This module relies on external 5v power supply, you can use something like this one from eBay

[Number-One]

Price for this module?

[russian]

Price for this module?

Too early to say, I need to get the PCBs in my hands & test before I even think about it.

[kb1gtt]

I agree with russian, it's to early to really put a price tag on it. It will vary quite a bit by what is populated and what is not. I would gut feel a price for a typical setup to be between $50 and $150.

[russian]

Status update:
The biggest issue for me is probably the silkscreen references font size: it is pretty hard for me to read x=0.5mm/y=0.5mm font



On this picture R263 is the size I cannot read. We have two options:
1) increase y to 0.762mm while keeping x at 0.5mm, that's how C250 is on this picture.
2) increased both x and y to 0.762mm - but this would require re-arranging the references because they would not all fit in one raw.

Looking for opinions.

[kb1gtt]

Best option is to get better eyes

C240 and C250 are not likely to be readable in the real world. Silk near the edge of a pad doesn't work very well. You can probably get R263 a bit taller. Also by eye I'm guessing the thickness is .008 in, it would probably be a bit cleaner if it were .005 instead. It might be worth considering that you simply need to stuff the board when either have a magnification glass, or a PC such that you can open the KICAD. I've not good luck with larger text. You'll end up putting the text in really odd places, which gets to be a real bugger. i'm sure with larger text you'll miss read the silk and place a component on the wrong pads.

[russian]

Also by eye I'm guessing the thickness is .008 in, it would probably be a bit cleaner if it were .005 instead..

[kb1gtt]

Perhaps we can get away with
x = .030
y =
offset x = .0025
offset y = 0.00

that's about as large as you can go, even then I expect the tops and bottoms to get trimmed slightly. Basically the silk screen doesn't generally register perfectly with the PCB, so it's often slightly high or low, such that all tops will get clipped by the pads or mask clearances.

[russian]

More pictures on the wiki page!

[kb1gtt]

Sweet.

[puff]

congrats! have you got the toast oven? please, shoot the video!

[russian]

congrats! have you got the toast oven? please, shoot the video!

I've got a toaster but learning this soldering technique is a separate issue altogether. For now I need to assemble the board step to step while testing everything at each step so no toaster for now.

[sturovo]

Wow, that looks great!
I hadn't appreciated how compact the board is until seeing the photo.
Hopefully I will get my hands on one soon

[russian]

SD works, USB TTL works. Top post updated.

Time for a glass of port.

[russian]

First feedback:
We need W210 and W209, the way we have W211 and W212 - that's so that we can redirect op-amp channels to digital timer inputs. todo: figure out if we can get events right on the existing pins without routing them to timer input capture pins

P401 - the three pin connector 5v 3v GND - a minor inconvenience is that a 3 position terminal is required even while we do not really have use cases for the 3v pin. reordering the pins to 3v 5v GND would make it possible to use a 2 position terminal if you do not need the 3v, but I guess it might complicate the traces.

[russian]

I've verified injector driver & I've verified analog inputs - at this point I am comfortable to offer this for sale @ tindie

[sturovo]

ordered

[russian]

non-fatal issue with rev 0.1: hi-side driver: the part numbr is for SOIC package but the board has MSOP package. Part number should be TC4427AEUA
In the next revision we will probably go to a larger package

[russian]

ordered

Shipped

[russian]

I cannot believe it. I've assembled it and it JUST WORKS!

[kb1gtt]

Very nice very nice. Now I just need to convince you to break the 10cm by 10cm rule so I can put that connector all on one PCB and get rid of that spaghetti. Also it would be nice if we added an optional 2 pin jumper such that you don't have to run that red wire along the edge of the PCB. We have 5V down near the high side selection pin for the high low driver. As well I'm sure we can change the green and yellow wires into little 2 pin jumpers. At some point we should see how the injector drivers are doing thermally. It's winter now, so no problems, but on a hot summer day, with a long drive, I wonder how it would preform.

Good to see it's working cool.

[russian]

Frankenstein can host a character LCD screen

[russian]

Soldering chimp mode off.
#2 & #3