[rusEfi] Mazda 1996 323f 1.5l #45

[alexander-n8hgeg5e]

Picture of the car is coming soon.

I got my ignition transistor from ebay. Some Bosch one, got no datasheet.
I mounted it under the hood, build special small case.
Original one works the same way,wiring is the same.
The original one was inside of a part that contains camshaft sensor, distributor, igniton coil and the igniton transistor.
I use this part only for camshaft sensor and distributor.
My coil is from some Honda.

This is my wiring/setup a bit more reusable and exitsting rusefi schematics excluded.

setup.pdf

(198.09 KiB) Downloaded 606 times

I assembled all parts of the frankenso 0.4, exept canbus (not need it) and had not enough leds.
Picture of the board:

reduced_IMG_20181029_205258.jpg (212.92 KiB) Viewed 34651 times

reduced_IMG_20181029_205334.jpg (236.5 KiB) Viewed 34651 times

The big resistors are for driving the igniton transistor properly (2 watt ones).
I wanted 8Ohms, like in the schematic, but they are 11.

I had a testing brakeout harness on my car, because of this i use the connectors on the car that i already attached to it.
So i can sniff the signals while the engine is running with old ecu.
This looks like this:

reduced_IMG_20181029_205445.jpg (173.88 KiB) Viewed 34651 times

Connector:

reduced_IMG_20181029_205504.jpg (147.31 KiB) Viewed 34651 times

My setup is mostly tested now and seems working.
All cables from the old ecu I have split out to connectors.

[kb1gtt]

You probably want to wash the board with some isopropyl alcohol, or similar cleaner. You appear to have lots of extra flux covering pretty much everything. Flux is quite aggressive when hot, and is still corrosive when cold. Over time it will likely cause corrosion issues.

Good job in getting it populated.

[alexander-n8hgeg5e]

damned this thing looks cool now!

reduced_IMG_20181030_034159.jpg (152.37 KiB) Viewed 34634 times

reduced_IMG_20181030_154236.jpg (95.61 KiB) Viewed 34634 times

I attached standoffs and all wires attached.
ftdi works now, were not working.
Maybe the bluetoothmodule had pulled the lines high somehow...

It is disconnectable now:

reduced_IMG_20181030_035033.jpg (143.95 KiB) Viewed 34634 times

The ign signal measures good. The 4Watt capable pullups gets hot now. 5v 22Ohm.
I had to move the trigger pin, could not select pd5(or pb5.. some pin with 5) as trigger.

Things tested and working:
######################
- bluetooth serial at 112500baud
- ftdi serial
- fan
- fuel pump
- brake pedal signal.
- throttle positon sensor works and adjusted
- battery voltage sensor works and adjusted
- connecting the dev console at the bench is working
at bench: micro usb connection and all gauges work
at car with netbook: only ftdi works , only 3 gauges do work, console seems somewhat broken
- I can see my trigger signal on oszi connected to the car. not tested if it reaches the input pin.
- Iat shows some value
- mass air flow gauge reacts to the engine, if it turns up, while the old ecu runs the engine.
- Air-fuel-ratio gauge shows something narrow-band-O2-ish. Seems Working.

Things not yet working:
#################
- micro usb in the car. one time i got a usb device for a few sec.
- with ftdi: dev console loads but shows only few gauges.
- could not get it to trigger or sniff the trigger signal somehow.. maybe i need the microusb for fully working dev console ?

My next plan is triggering the thing by doing bench testing.

[alexander-n8hgeg5e]

You probably want to wash the board with some isopropyl alcohol, or similar cleaner. You appear to have lots of extra flux covering pretty much everything. Flux is quite aggressive when hot, and is still corrosive when cold. Over time it will likely cause corrosion issues.

Thanks for the warning. I had problems in the past with some flux that was terrible...
Hard to clean, even with acetone, conductive and becomes coal if it is heated by the current...
Then it conducts even better...

I wanted to buy some flux but could not find good information about the fluxes.
So i mixed my own out of "Rosin", also called "colophony" , and some isopropyl alkohol.
This stuff is mildly activating. Its non conductive and it dries out to some hard stuff.
It protects the components. As far as i know its good to leave it there.
My board is not yet dry everywhere... its a little bit a sticky, slimy mess...
but i got used to it.
I use it to secure/clue the big elektrolytic caps to the board.
If a usb port is slimed, its important to wash it out first befor use. if not, the port gets damaged... and the cable.
With a small screwdriver you can pick up smd parts , you can clue them on the tip.

[alexander-n8hgeg5e]

My car was running at idle for aprox. 15 sec !
I switched to the vr-decoder chip instead of useing the op-amp channel.
My vr chip was soldered in the wrong way... it survived...
Now I try to tune this thing and take some photos.

[AndreyB]

I love rusEfi videos, first start or second start does not matter

[alexander-n8hgeg5e]

video will come , for sure

[alexander-n8hgeg5e]

I proudly announce , with the help of the forum, my engine config compiles , loads , and maybe triggers on real hardware.
Oh that will be fun. I yust have to plug it in.
Because I have "Y-style" wireing and much connectors i can run it in parallel while the stock ecu is trying to run the engine.

At the moment not all analog sensors are working if the middle connector of the stock ecu is not plugged in.
But i don't care. The rusefi ecu will do it if it needs to.

[alexander-n8hgeg5e]

I roasted my brain chip prob. via micro usb cable. ... again
This is the second one i roasted.
First one i roasted a years or so ago by pushing the pinheader on frankenso board in , with offset/not alligned.

Nothing was hacky style connected, and it was in it's plastic box.
It happened while swiching car power on and of and cranking and charging my netbook it was connected to.
I disconnected the display and hoped it would be ok to let the usb cable connected.
Micro usb voltage goes directly to pa9 or so on the discovery board.

[alexander-n8hgeg5e]

I fixed it by solderring on all the parts for the onboard stm32f407vgt6.
Then i added missing led's and removed some ugly troughhole caps from the board.
Not much ugly looking stuff remains.
No dangers that killed my chip are left, no micro usb, no pin header.
Now i will plug it in, maybe it burns.
Here some pics:

reduced_IMG_20181202_133122_upload.jpg (204.93 KiB) Viewed 34417 times

reduced_IMG_20181202_133231_upload.jpg (147.43 KiB) Viewed 34417 times

reduced_IMG_20181202_132958_upload.jpg (165.71 KiB) Viewed 34417 times

some through-hole stuff left:

reduced_IMG_20181202_133302_upload.jpg (147.44 KiB) Viewed 34417 times

[AndreyB]

You are not looking for the easy route

Do we need all the fly-back diodes, or do they just not hurt? I usually only have one for idle solenoid valve

[alexander-n8hgeg5e]

Actually I ' dont know . I installed them to be sure/not to forget one , so i can use all the channels.

[kb1gtt]

Most channels do not want the flyback diodes. The flyback diodes are typically only for circuits that specially need the slow reaction times.

Nice to see the native stm. Keep in mind you do not need the discovery like most people have done.

[alexander-n8hgeg5e]

Good to know, my ignition-transistor!-driver definitely do not need these flybacks.
Propably the power chips on the board can handle the injection coils.
My cheap oszi says that it spickes to aprox 30v up and down,..
but not sure if it can show that properly.

My board powers up, no fire. Programming haven't worked yet.
It does something on reset.
Next thing i will try is to enable ,possibly existing,
internal bootloader by boot0/boot1 pins and
flash it via serial. Maybe the "prog" labeled usb port is for that.

I tried to do it via swd.
I used a cheap usb logig analyser "usbee pro clone" to check what going on.
Software pulseview decodes the swd protokoll for me.
Seems the chip does not reply or so.

Then from that i had the idea to capture all the channels in my car with that logic analyser.
It has 8 channels , one of them is a analog .
I will wire to the mazda mcu's low voltage lines.
They are not hard to find. There are big mosfets on the board.
Trigger i can take from rusefi channel.
Then i can capture MAF , inj-coils , ign , and Cam-Sensor/Trigger.
Temperatures i can measure with multimeter and stopwatch and write it down.

So this should be enough to get good startup and idle.
Maybe i wire one channel up to a button, and ... during messurement
i press it together with, ... full throttle.

In this case i have all common use cases covered: Startup, Idle, Full-Throttle.

So rusefi video's progress is nearly as good as uploaded.

One thing i forgot...
I have to slightly modify the trigger code and make a pull request.
Somehow my measuring device got inaccurate.
But i can fix that.
On my netboot is the picture from what rusefi console recorded while the car running
on rusefi ignigion and on mazda-ecu injection.
It had to run on 4tooted no missing tooth...
i measured with,... tapeline, while pulling the car in 5gear and marking
it on the ground when the trigger changed voltage.
OT i got from sticking something in the sparkplug hole.
True timing is 10 degree later or so...

Security Warning: Need to disable the cars injection and igniton.
The car could fire up and run over you.


Add: I discoverd my logig analyser has no analog channel , so i need to get another one.

[kb1gtt]

Sounds like your making progress.

[alexander-n8hgeg5e]

Yes, programing problem, issue found.... It's the ... It happened again... look at biggest ic on my board...
2nd good thing, maybe my another stm32 survided , because during search for the issue i found out that the
current regulator on the discovery was blown. Luckily i desoldered the stm32 by flooding it with liquid metal.
The one on the frankenso i need to be much more gently. Don't want to damage the board...

I fixed the trigger and made a PR yesterday.
Here the image, looks nearly as befor.

trigger_42.png (6.57 KiB) Viewed 34341 times

captured png:

2018-11-23_14_06_44rpm_1267_maf_NaN.png (50.83 KiB) Viewed 34341 times

[alexander-n8hgeg5e]

I got swd reponse and it's flashing! It's alive! Blue Led flashing! Display works! It survied being soldered the wrong way on the board.
Definitely my chip is no sissy chip.

reduced_IMG_20181204_192820_upload.jpg (236.15 KiB) Viewed 34334 times

reduced_IMG_20181204_192738_upload.jpg (194.72 KiB) Viewed 34334 times

Prog port works.
I flashed via dfu-util.
(jumper between vdd and boot0. then reset, it comes up on usb in dfu mode.)

[ZHoob2004]

Bit of a side question, but do you have a hot air gun? If not, they're totally worth it. $30 on ebay plus $10 or so for a syringe of solder paste and you'll wonder how you got along without it. I know it's possible to get along on these boards without one (I did most of mine without), but it's significantly less headache to just buy the correct tools for the job.

[kb1gtt]

Sweet, that's one of the very few with a native chip. Good job.

[alexander-n8hgeg5e]

I have no elelectronic hot air gun. Some time ago i tried desoldering parts with a all-purpose hot air gun
and problem was that other parts got desolderd.
So this time i tried a new desoldering technic with a big soldering iron.

This could be true, maybe i would wonder why i not got it earlier. So thanks for the hint.
These hotair guns could be really good for electronics.
Smaller outlet and maybe better temperature and flow.

I think the most time is consumed by depacking the part and searching for the right place.
My last innovation was to do all with the kicad pcb new prog.
I found out that if i click at the right place i can see the part number and reference.
So i don't have to switch between bom and pcb layout.
The layout i need to check some references that are not clearly readable.

[Abricos]

I found out that if i click at the right place i can see the part number and reference.
So i don't have to switch between bom and pcb layout.
The layout i need to check some references that are not clearly readable.

Check
https://rusefi.com/forum/viewtopic.php?f=13&t=1424

[alexander-n8hgeg5e]

Cool kicad addon!
I will remember next time.

[alexander-n8hgeg5e]

I use rusefi to stimulate my old ecu. I wired it up on the bench.
So i can get timings out of it. First time, it i had some sequential stuff on the screen.
Now it don't want to do it again.. Only simultanious stuff. Without ign-key transponder it don't want do do it long ,too.
This measuring device is a logic analyser. I used a logiglevel converter to
convert some signals.
And a mosfet was needed to pull the trigger channel properly low.

reduced_IMG_20181209_035344.jpg (193.42 KiB) Viewed 34199 times

1-4 from top is injection.
green line is unused channel
then igniton.

Last ist the trigger coming from rusefi.

Screenshot_20181209_042411.png (6.59 KiB) Viewed 34199 times

Now, as i posted it , i see whats wrong... my trigger signal is inverted... because of the pulldown mosfet... (missing on picture)

[alexander-n8hgeg5e]

I added a 2nd mosfet stage and had to use a 500 Ohm pullup because my logic analyser sucks the signal away...
Now i can squeeze all timings out of it.
Trigger signal is measured at the inverted middle stage so it shows up inverted but does not mess my shiny last stage signal up.

Screenshot_20181209_063324.png (8.54 KiB) Viewed 34196 times

[alexander-n8hgeg5e]

Holy crap, i got the thing running! It drives, you can drive shopping with it. Have to eat now. Later I will take my crapy phone and try to fire it up again. I can accelerate whole first gear full throttle , the thing takes gas nearly all rpm.

[AndreyB]

Holy!

[alexander-n8hgeg5e]

I recorded it. I'ts a bit unshiny, during night. My phone is a piece of crap! My nice video , hope youtube can read/repair it.
Should be possible somehow, 500 mb should be something readable in.

edit: i hate my phone, the important one is broken somehow.
Have to wait until tomorrow.

[kb1gtt]

Sweet, and good job. Can't wait to see the video.

[alexander-n8hgeg5e]

I recorded a bunch of video, this time with proper cam. Now i upload them.
Takes some hours I think.

edit: one of them is uploading , mostly all things covered , 2 min talking about gauges and idle, then driving.
On situation included firing it up again while rolling , but not used the clutch to crank.
It is uploaded in 15 min or so.

All issues seams resolved now:

• cranking takes long and comes up hard issue: I forget to set mechanical idle adustment screw properly.
  So the engine was at limit with air. The iac was fully open and it nearly died if not warm.
  I had disajusted the screw because my old ecu was turning in idle 3000 rpm if cold...
• coolant sensor...)
  During video takeing , i propably figured out why the old mazda ecu had problems.
  I think it's the coolant temp sensor ... sometime it falls down , but not to unreasonable low values.
  Several times today this sensor made problems , especially during cranking.
  The last settings bring the engine up good, if the sensor is good for a moment.
• strange popping sound: its yust the idle valve i know, I got it popping again.
  It popps during rpm falling and catches it up. Idle was 10 times stronger at last.
  More air, more popping.

And thousand times i tought the tune is not right, because of acceleration interrupted...
hmm what will rusefi do if it thinks the engine is cold...

Conclusion:
I will do someting , "beyond the press like" to this sensor.

The coolest thing of all:
It feels like the thing goes of like rocket, (it's yust the feeling , i know, but good one)

[alexander-n8hgeg5e]

There it is : Rusefi video:


Now i watch some of the vids and look for some other good ones.
edit:
driving with cam attached to the sunshield

inside the car with the wiring

at day: outside of the car and engine bay

why, propably, my car made problems since this summer


Pictures

reduced_P1230669.JPG (478.45 KiB) Viewed 34088 times

reduced_P1230671.JPG (552.16 KiB) Viewed 34088 times

reduced_P1230674.JPG (627.66 KiB) Viewed 34088 times

reduced_P1230676.JPG (508.8 KiB) Viewed 34088 times