[info] Frankenso - ECU shield compatible with OEM case / official

[AndreyB]

We use early Miata (NA) pinout but these cases had 48 pin connectors. We use a 64 pin connector which 1999-2000 miata uses, but that could be the different case/different generation you are talking about. Frankenso needs an older 64 pin case, like the one in 1991-1994 Mazda Protege (323?) Automatic

Just added http://rusefi.com/wiki/index.php?title=Manual:Hardware_Frankenso_board#Case_compatibility

[AndreyB]

PS: is the case you are trying something like

? that's the 99 miata case, the WRONG case.

[Roo-Barb]

Yes that is exactly the case.
and interestingly the other that you show looks the same as the 1996 Rav4 case.

[AndreyB]

Yes that is exactly the case.

My messages were confusing, I've just edited the previous message. My previous message shows the wrong case, the right case is on the wiki page - this one:

[kb1gtt]

Take caution when you open up holes. There are 2 flooded planes internal to the PCB, one is GND the other is power. They flood pretty much everywhere up to about .010 inches from other traces or PCB edges. When you drill through the PCB you'll be drilling these planes and creating a potential place for an accidental connection as the flooded copper will go right to the edge of the PCB. Normally that .010 inch setback prevents this kind of accidental connection.

[Roo-Barb]

Ooop's

[kb1gtt]

Probably not a huge deal. I would check how many ohms you measure between 5V and GND. In that area you have 3 GND planes which don't really matter if they short together or if they short to the chassis / frame. However you also drilled through the 5V power plane. If that shorts to GND you'll have some problems. You can measure your ohms at the below noted location.

GND-5V.PNG

If you end up drilling through the PCB, I would suggest doing a close visual inspection to see if copper has been smashed and could be shorting the planes. As well if you are using a mounting screw, I would suggest adding some kind of sleeve to protect the threads from shorting to the exposed copper.

[AndreyB]

Schematics notes glitch:

[kb1gtt]

Oops. That will be updated on the next release. Thanks for the feed back.

[Roo-Barb]

Is there a straight forward parts list to cover the full population of both boards (i.e: Frankenstein & Frankenso).
Along the lines of R1 = 10k etc , and a summary like 20 x 4K7 0805 smd resistors , blah blah?

As Not many of us if any use kicad as it is unlike almost any other PCB Cad and won't transfer to DesignSpark or Eagle or Circuitmaker.

I am used to Eagle and DesignSpark as they are super easy to get started with and have everything already.

I have tried running Kicad on my mac and it runs like a dog and periodically crashes.

Yet all I want is a parts list and to see where to place them.
This project Is one of the most awesome on the net, but for it to be taken up by car enthusiasts and part time electronic tinkerers it needs the bog standard simple to follow stuff to be easy to locate.

I know that when you design something like this or have it grow as it has, it all looks easy as far as assembly, but when someone jumps into the project part way along with out the prior experience it seems daunting.

If you could create an export tool to handle more common cad formats or just create the listings and update them with changes that would be great.

I have downloaded the BOM's that are posted but they are near useless for part listing.

Sorry if I have been offensive in any way, but I would like this project take off and nuke the others.

Cheers.

[AndreyB]

Is there a straight forward parts list to cover the full population of both boards (i.e: Frankenstein & Frankenso).

Current documentation is not great, one day someone would create a HOWTO

For now we have by-ref 'longer' version of the BOM for both boards like https://svn.code.sf.net/p/rusefi/code/trunk/hardware/frankenso/frankenso_by-ref.csv

And there are some notes on the wiki like http://rusefi.com/wiki/index.php?title=Manual:Hardware_Frankenso_board and http://rusefi.com/wiki/index.php?title=Manual:Hardware_Frankenstein_board

[kb1gtt]

See these files for Frankenso.

https://sourceforge.net/p/rusefi/code/HEAD/tree/trunk/hardware/frankenso/frankenso_by-ref.csv
https://sourceforge.net/p/rusefi/code/HEAD/tree/trunk/hardware/frankenso/frankenso.csv

You say where to place them, do you mean physically? KICAD can export a list that indicates XYRS coordinates. I think that's correct, XY for position, R for rotation, and S for side. That file is typically handy for PNP machines. I have not created that list, but it could be generated from the layout tool. See file -> fabrication output -> *.pos

I see what you are saying for Frankenstein. It's been a couple years, so I'm not sure of Frankenstein BOM status. In KICAD you can export a BOM in various groupings. With Frankenso, I created a BOM with the grouping as REF and a different one grouped by value. I believe those various Frankenstein BOM's were because either we couldn't get all the parts from one vendor, or it was lower cost to do the multiple source thing. I've had that happen before were paying the extra for shipping from 2 suppliers costs less than ordering from one supplier. In general I put the MFG properties in the schematic, which means as you copy and past parts it copies the properties.

Oddly enough, I originally started with KICAD because I found it to be the simplest to start with. I found with pretty much any other tool, I had to first learn how to turn off all sorts of automatic stuff. As well there are things that I can't do with other packages, but I can with KICAD by editing the raw text in the source files. So I've liked KICAD because I like it's simplicity and I like it's flexibility. About the only thing I see as a problem with KICAD which might be solved on newer versions, is the efficiency tools and integration into electromagnetic simulation tools. Things like bulk component changes and mass properties updating would be handy. As well, I get eager to see integration into openEMS, or similar tools. I'm currently running a version that's several years old, as I haven't spent the time to learn / port to the new library system. I'm using KICAD build 2013-07-07 BZR 4022 stable.

I agree it can be daunting and such. I hope it's less daunting than starting from scratch, and I agree it's still daunting. It seems most people also get scared from SMT soldering. I general think the optimal plan as time and resources permit, is to make a wiki tutorial showing clearly how to use a purchased assembled Frankenso, which is applicable to all the PCB efforts, then work our way to a better detailed instruction on how to assemble a Frankenso which is applicable to a smaller audience, then eventually the same for Frankenstein boards. For now we have at least gotten them out there for people to get their hands on. As well we have a wiki which can allow others to try to help make the assembly process more user friendly.

I enjoy the feedback, and from my stand point, don't hesitate offering some potentially confrontational criticisms. However keep in mind that keeping a good attitude in such confrontational situations goes a long way. I look at it this way, I'll extract content which will help make it stronger, and generally ignore emotion.

Thanks for the feedback, you offered some criticisms that were done in a nice way. Feel free to keep the criticisms coming, they can only help make this thing stronger.

[Roo-Barb]

Awesome thanks

The CSV's are great.

I will do my best to document in detail the assembly and use the files to help with the wiki.

Thanks again.

[AndreyB]

todo?

109) Move D12-D23 schematics from the main page to the inj12ch page
110) Part number for D19 (PB9-INJ09/PB9) as Miata Idle Valve Solenoid

I am currently using a through-hole 1N4007 hanging on a wire for my idle valve solenoid and I would love to upgrade to something SMD, just need a suggested part number.

[Rhinoman]

Have measured the inductance and resistance?

[AndreyB]

Have measured the inductance and resistance?

http://www.wikihow.com/Measure-Inductance is above my pay grade but resistance is about 10 Ohm I believe.

[Rhinoman]

I've used a slightly different method where the rise time of the current is measured and the time constant is then used to calculate the inductance. When you know the inductance then you can calculate the stored energy and size the flyback diode to suit. You need to calculate the energy based on the highest voltage that the part will see which is usually during load dump. You can also use an LCR meter to measure the static inductance and allow 50% to allow for the change in inductance when the solenoid is open. If you use an LCR meter then its a good idea to use a 100 Ohm or similar resistor in series because some meters don't have enough drive when measuring low resistance inductors, I've seen this a lot with LCR bridges and big injectors.

[AndreyB]

On the left is the high-side module for alternator control (I am planning to replace it with a different one later that's why it looks like this), in the middle on the back we have two coil drivers mounted on the prototyping area.

These hanging red and black wires are GND and +12v rails, so

111) GND and +12v rails? at least some 1.1mm holes in the prototyping area.

The holes we already have on the prototyping area are just too small for 20 AWG wires. Some of these holes need to get bigger I think.

[AndreyB]

I was playing with my Thexton THE404 Spark Tester set to 30 000 v gap and I have a 100% reproducible serial port disconnect issue.

0.4 Frankenso board with http://rusefi.com/forum/viewtopic.php?f=4&t=286 mounted at the prototyping area. Car is parked, car battery is being charged by a wall charger. Laptop on my laps is running from it's own battery, no wall power. Engine is not running, I use spark bench test feature of rusEfi console.

Every spark test my serial connection is dropped with an I-O error. Happens with both stm32 virtual serial over USB and FTDI ttl USB device. Bench testing injectors does not cause serial I-O issues.

[puff]

I had the same issue when doing spark tests on the bench. I guess, the cause for this is that EMI from sparks is significantly higher than when you test injector. I believe, proper shielding (Faraday cage?) could improve the situation.

[Rhinoman]

The simplest solution is to use an external igniter, the current set up is going to very prone to issues and increasing the spark voltage will increase the spikes on the primary side until the output driver clamps at around 450V.
Have you tried separating out the ignition wires from the rest of the loom? Its a susceptibility problem but you need to determine whether its an issue within the ECU or outside it. Have you got ignition wires crossing over the serial port wires? Is it only the serial port that is affected?

[puff]

I was using the external igniter, just sending logical signals. The spark plugs were in a 3ft distance from discovery board.

[Rhinoman]

Then its probably just the emissions from the plug itself radiating into the wiring. How well grounded are the plugs? you may need to mount them in a metal box.

[puff]

that's what I am talking about. putting the plugs into the metal box (or a cage). my setup is in this shitty video

not only it breaks the connection, but it also affects rpm. didn't have time to build a metal box ;-(

[kb1gtt]

For puff, do I see a wire between the frame and the battery negative that is how they are arranged in an OEM vehicle. Perhaps that yellow wire is this wire. Also is the discovery board GND connected to the battery - that's a direct connection, not a shared wire. Discovery to battery - and no other wires shared with discovery gnd.

About russian's problem, that is also likely caused by a bad ground. I suspect your OEM harness is starting to have problems. I suggest for diagnostics purposes you solder a wire to the frankenso board GND pad, then run a ring terminal to the battery - this will help identify if you have a potential GND issue. The OEM had 3 GND's, and I'm sure that each one carried different currents, one was probably the ignition and injection GND currents the second was probably misc power currents, and the third was probably the analog and smaller MCU currents. In your case I believe you are trying to run all GND currents on one of the three wires. You likely need either more wires or one fatter wire. I suspect that you are fatiguing the one GND wire which is likely also causing problems with the OEM ECU.

[puff]

I was running it off a power supply. The yellow wire was indeed a direct ground wire coming from the power supply. There was also a ground wire coming to ignitor and a separate ground wire coming to discovery. Those were the coming from the same post of the PSU, so in some sense they were sort of shared. I still believe this is an EMI issue as in an OEM vehicle these sparks happen in a shielded space - inside cylinders. If it were a ground issue, I was lucky it hadn't affected the usb port of my laptop (but for that port disconnect problem).

[AndreyB]

Random question: what if we _remove_ the "native" chip option and go with external brain board only. Would it be possible to make Frankenso a two layer board?

I am thinking brain board assembled in China soldered to a simplified Frankenso with larger prototyping area as a more flexible solution. I think EGT and knock can go away/module option only.

[puff]

While it is a nice idea from the standpoint of flexibility, it is probably not that good when it comes to rigidity and reliability. On the other hand, we had some successful experience with this sandwich type, so why not?
I'd say, we may think of the brainboard with all the obligatory stuff, and the 'shields' which might be even custom made, i.e. several boards containing various combination of circuitry. So, when you need these two functions, you choose model a, when you need those three - you choose model b, when you need something else - you might even create your own. I'm wondering, why haven't I built my vr board as a next layer in this sandwich) - this would probably eliminate wires
But this doesn't answer the question if this makes it possible stay with 2-layer design (if I understood correctly, you are speaking of 4-layer pcbs?)

[kb1gtt]

I would say it might be possible, but from a technical standpoint, probably not a very good idea. Currently the low frequency analog circuits are on the back side, and the higher frequency digital circuits are on the front side. They are separated by the internal ground plane, which helps isolate noise form getting into the analog circuits. If we got rid of that ground plane you would be likely see elevated levels of noise on the analog signals. Also the copper helps spread the heat from the injector drivers. If we get rid of the copper, we would increase the chances that we need a heat sink. Other than that, it might be possible. I see several traces have short runs on the internal planes, so I'm not sure it can be done as you would have many more traces that need to be routed to get the power around. So I would say it would be difficult to do, but plausible.

[AndreyB]

flyback diode area is a bit problematic - one can only mount a couple of these simultaneously. Because of the connector these have to be on the back. Because of only ~3mm of space to the back cover these have to be flat.

I really wonder if we need to re-clutter the board
1) take CAN away to an add-on board at the prototyping area
2) take HIP away to an add-on board at the prototyping area
3) shift analog stuff so that we can re-arrange low side chips and have a through-hole for the fly-back diode next to each low-side chip?