1994 MX5/Miata/Eunos 1.8 NA (Japanese Import)

[deviator]

Today's update, not good afraid. Although it could have been worse, a lot worse!




The situation so far...
As reported the only difference since the last time it was connected are the firmware updates. I can't imagine this being the issue. I've followed Audrey's instructions and metered out the board. I'm finding 5v and VDD on both boards are near enough a dead short. This is both boards independent of each other.

The Franko board was tested on it's on with the car and 5v was where it should be.

Time to order a new board! Arwell, I fully understand this is an experimental solution and there are risks, if I wanted an off the shelf solution, I would paid $400-1000 for one.

On this occasion, I've been lucky - although it took my laptop down, it doesn't appear to have caused any damage, I hope so as the laptop was more expensive than the car! The car appears fine back on the standard ECU. I think in the future I'll use an older laptop - just in case

[AndreyB]

Right, they do have some sort of protection on USB since they know people are sticking weird stuff like coffee warmers into the USB, but an older laptop would add a layer of safety

We have a bulged U1 regulator and at this point we have zero explanation for this

[deviator]

Before I order replacements, can I just check these are 100% the right boards. As there is a large black square in the upper right of this picture (oddly where my black mark is!***) which mine doesn't have.
http://uk.rs-online.com/web/p/processor-microcontroller-development-kits/7458434/

*** EDIT, no the socket is on the other side, my mistake.

[AndreyB]

http://uk.rs-online.com/web/p/processor-microcontroller-development-kits/7458434/

This is STM32F4Discovery and STM32F4Discovery is exactly what you need. The audio jack does not fit into the case vertically so I took some sterile pliers and removed it. Hopefully this did not contribute to the bulging.

[deviator]

Boards ordered.

The Frankenso board has a short between the 5v and the VDD line (taking both points from the header).

The Discovery board has a short between 5v, VDD, 3v and GND. (Which is hardly surprising given the voltage reg is black.)

Obviously it's not worth connect the new Discovery boards up until the issue with the Frankenso is resolved.

[puff]

replacing that u1 thing might resurrect the whole board?

[AndreyB]

The Frankenso board has a short between the 5v and the VDD line (taking both points from the header).

Obviously it's not worth connect the new Discovery boards up until the issue with the Frankenso is resolved.

Yes and no. Frankenso 0.1 has traces between 5v and VDD, that's a known defect - but I claim that this defect is not affecting us since your (and mine) headers are populated without the VDD pins in the header! Without the VDD in the header, Discovery is not connected to the faulty VDD lines on Frankenso 0.1, and that totally takes care of the problem for us.

The purpose of the VDD lines is to provide 3v to the chip in case it would be directly soldered on Frankenso. Since you are not using a chip directly on Frankenso, it's not a problem. Discovery does not need 3v power supply, U1 regulator on board is what would produce 3v from 5v. Once you have a replacement Discovery we would confirm that Frankeso 0.1 VDD is not touching Discovery VDD. We might also shorten Discovery VDD pins by a couple of millimeters just for paranoia. But, once again, I am confident that VDD is not the issue which has caused the fire.

Which means I have no idea what has caused the fire. I guess at some point we need to connect Frankenso into the Miata, turn ignition on and go over the header with a multimeter to confirm there is nothing above 5v on the headers. That's unless someone has a better theory.

Obviously putting the blame on a faulty regulator is the simplest thing to do but I simply do not believe we are so "lucky" to get a defective Discovery board. So far there was not a single defective Discovery.

[puff]

I am very glad to have an over-current-protected power supply - at least I saw my adc opamp went dead (because of reversed polarity, my fault) at somewhat 150-200mA. If I only knew what current consumption should I expect, I could have saved that chip. It was couple of seconds before I smelt it burning.

Powering the daughter-board alone without discovery and everything connected is yet another good approach.

Could it be that you missed the pins? (once I was close to ruining everything by not placing discovery correctly into the socket). There is a chance of shorting pins by placing discovery onto some metal surface...

[kb1gtt]

Hmmm, could the trimmed pin(s) on the discovery be close enough that they could touch?

Was laptop plugged in to the wall? I wonder about potential ground loops, or mains voltage spikes. I understand the key was in the on position, is that correct.

We have some diagnostics notes found here, perhaps they will help identify what went wrong and perhaps they might help identify a fix.

http://rusefi.com/wiki/index.php?title=Hardware:stm32f4discovery_repair

I understand the jumper wire for the hi/low was moved from 5V to 12V, can we get a picture of that? I wonder could the wrong end of the wire have been moved? Also for verification, this is a Frankenso R0.1, isd that correct? CAn we also get a picture of the jumper wires section? That was populated by russian, I've not see it, so a picture would help me.

[AndreyB]

Actually I took the pictures of this board before shipping it to the UK - it's now my reference picture for the 1994 setup

http://rusefi.com/images/Frankenso/frankenso_set_for_miata_1994.jpg
http://rusefi.com/images/Frankenso/frankenso_set_for_miata_1994_back.jpg

[kb1gtt]

I'm a bit confused by this one. A dead short from U1's output to GND will typically open circuit D3. U1 has thermal and over current protection, so I don't expect it to break. To make U1 break I expect you need to pull U1's output to above 3V, or you need to pull the input above 6V. The best guess I have is that D3 must have gone short circuit instead of open circuit, then 3V much have gone higher than the input V. Can you check D1 and D3 with a multi-meter on the diode setting? Do this with the discovery completely disconnected from everything.

[deviator]

D1 is measuring 248 ohms in one direction and no connection in the other.
D3 is measuring 10 ohms is both directions.

Still waiting on the Discovery delivery.

[kb1gtt]

Hmmmm, 10 ohms, that seems odd to me. I think D3 broke turning into a 10 ohms resistor. I'm not sure if that was before or after U1 broke. Best guess I have is an ESD failure, either that or some how higher voltages made it on the the 3V and or 5V lines. However I'm a skeptic that either was root cause of the issue. I guess I should suggest trying to avoid touching the pins to help avoid ESD. If you are going to touch something, touch a GND pin first, to help get you and the board equalized before you touch something else.

[deviator]

Thanks for the suggestion, I will ground myself before working with the board. Although I was no where near the board at the time of failure, it had been connected to the car for at least 3 minutes (probably more like 5-10) with the ignition on whilst I was talking to Audrey.

The last actions I performed was a firmware update and Audrey was checking the messages to make sure everything was within tolerance.

I wonder if flashing the firmware has caused an issue? As the laptop was running off mains electric from the house and obviously the car has it's own power and grounding system. I think on this basis. I will flash the Discovery board whilst it is away from the car and possibly even the Frankenso.

I'm hoping it's just one of those things and a complete one off.

On a positive note, the new boards have arrived and I have a board connected now flashing anyway. The firmware is updated and the Console title reads "20141112; firmware=20141113@5326 MX594d@COM4"

I've not yet connected the two boards together, what I am thinking as a precautionary measure is to trim the VDD pins from the header on the discovery. I'm thinking all the way back to the black backing. I don't believe these are required? As such I'd rather completely remove this as a problem area.

[AndreyB]

I've not yet connected the two boards together, what I am thinking as a precautionary measure is to trim the VDD pins from the header on the discovery. I'm thinking all the way back to the black backing. I don't believe these are required? As such I'd rather completely remove this as a problem area.

VDD pins are not used at all. If you trim them all the way back to the black backing that would give us a safe 4mm gap, should be fine for 5 volts

Right, no more mini-USB while the board is powered by +12v. I think it's safe if it's connected to the car while ignition is off since +12 comes to the board only while ignition is on.

Last test just to be sure - can you plug Frankenso without discovery with ignition on and go over all the pins on the header to make sure none of them are above +5v? None of them should be above +5 but who knows if they've shorted some traces at Her Majesty Revenue & Customs?

[deviator]

Next update.....
Another blurry photo, pins bent and broken using snipe nose (needle nose) pliers. I tested both cutting and snapping on the toasted board and found bending the pin back and forth tidy and closer.


The Frankenso has been connected to the loom with ignition on as request. All header pins tested from the underside of the board, nothing above 5v at all.

All looking good for the next stage.

[AndreyB]

there are three VDD pins overall, the 3rd one is in the middle on another header.

When/how do you want to processed?

Next step is a brief test of outputs: http://rusefi.com/wiki/index.php?title=Manual:Software:dev_console_commands#I.2FO_testing

fuelpumpbench
fuelbench 5 2000 3
sparkbench 5 200 3

And if this works as expected next step is trying to start it. More or less like

[deviator]

Ah, nearly forgot about the 3rd pin, glad I updated now before getting carried away.

To make the testing easier I am going to try and rig up another WAP at the back of my house so I can get internet access out there. This will give me more room to work and more importantly, should be quieter than last time.

With the output tests, will the console tell me if they are successful? Or will I need to go and physically check the spark, fuel, etc;

[AndreyB]

With the output tests, will the console tell me if they are successful? Or will I need to go and physically check the spark, fuel, etc;

Fuel pump you should hear, fuel injectors are surprisingly loud so you would hear them also. Spark you would need to hold a spare plug and look at it. The extended version of the command

sparkbench2 10000 1 5 200 10
would be helpful since it gives you 10 seconds to get the spark in your hands.

[deviator]

It's been a busy old day!

Fuel pump test - nothing, no sound.
Injector test - I think I can hear the clicking, so happy with that one.
Spark test - Not yet done
Fan test - Nothing, no fan.

The reason I didn't spark test was a noise heard whilst listening for the injectors.



The video catches the sound quite well. I'm pretty certain this isn't there on the standard ECU. Images captured from the screen, new messages....

[AndreyB]

no_pin_inj is a false warning, let's ignore this for now

your fan relay is probably not wired - let's take care of this later but let's not forget about it

the sound is idle solenoid valve, that's a work in progress issue. Since your board was assembled, some progress in this area was an additional diode. I am an idiot, I should have asked you to order the diode while you were ordering discovery boards. let's take care of this later but let's not forget about it #2
Let's turn idle solenoid off for now, here are the two commands you would need to execute consequently:
set_idle_pin none
writeconfig
and then reboot the chip with either the black button or power cycle
this should eliminate the sound

to be continued

[AndreyB]

1) Let's test all injectors individually, this version of the command has a delay:

fuelbench2 5000 1 4 600 3
fuelbench2 5000 2 4 600 3
fuelbench2 5000 3 4 600 3
fuelbench2 5000 4 4 600 3

2) with idle off, let's test spark - that's if the noise is gone once you assign idle pin to none

3) Fuel pump: any chance to open the trunk and ask anyone to listen closer to the tank? Fuel pump always gets on for 4 seconds on reboot.

Fuel pump is controlled by pin PE4. The corresponding transistor is the leftmost one in the larger group. All four pins on the left side of the transistor are routed to your harness wire 2T which is your fuel pump relay. If you want, you can carefully ground these pins with something and this should turn your pump on.

[deviator]

TO DO LIST:-
1. Fan relay
2. Idle solenoid valve, order the diode
3. Power steering

Right I'll give it another go tonight. BBIAB

[deviator]

Firstly, Idle solenoid fix worked fine. No more buzzing.

Test results...
1. Fuel Pump - Still nothing. Even with the wife in the boot, yes she was alive before anyone starts thinking I'm a hitman
2. Injector Test - All good, I can hear them clicking and also feel the relevant injector clicking. I would be happy at this stage to cross this off the list.
3. Spark Test - Nothing. I can't get a spark from it at all, I've tried 2 spare plugs and tried it on cylinder 1 & 2.

This command
sparkbench2 1000 2 4 200 10
Returned
"Output no ID" (paraphrase) on console.

Whereas when I tried "sparkbench 5 200 3"
It found an output Pin ID.

[AndreyB]

One day I will make it less obscure, but now your wasted spark is implemented as "coils for cylinders #1 and #3" (that's related to 1-3-x-x firing order), so

sparkbench2 1000 1 4 200 10
and
sparkbench2 1000 3 4 200 10

Can you make some not-so-blurry pictures of your harness plugs?
For coils we are expecting 1G BRN/YEL - that's fourth from the left and 1H BRN - the other fourth on the left.

For pump relay we are expecting 2T LT GRN Fuel Pump relay

Let's focus on the fuel pump I guess just because this should be simple as 1-2-3. I am looking @ http://rusefi.com/wiki/index.php?title=Vehicle:Mazda_Miata_1994 - does it have anything to do with your vehicle? It's just a relay, you should hear the relay make the click when you ground this wire. On some Mazdas that relay is actually right next to the ECU. If we cannot figure out the fuel pump we have no business... Ok, if I cannot figure out the fuel pump I have no business playing with all this!

[puff]

1). there might be something else blocking the pump in between - e.g.another relay controlled with immobilizer?
2) once i spent an hour checking for circuit continuity in my frankenstein, while the problem was in settings
what is the output of fuel pump info command in dev console?

[AndreyB]

http://www.mx5nutz.com/forum/index.php?showtopic=110149

[tomiata]

On the fuel pump relay wiring, it looks straight forward. Did you try to ground the Lt grn wire to the relay manually, you should hear the relay to under the dash click.

[deviator]

I've not got as far as bridging any wires yet! I will but I'm still double checking the basics before I commit another discovery suicide.

The wiring for my car with BPF3 ECU is similar to that shown http://rusefi.com/wiki/index.php?title=Vehicle:Mazda_Miata_1994 however there are 3 differences, highlighted in bold.

This is looking into the back plug, IE the harness wires going into the plugs

The pins different are...
2F MAP Signal Ground
1O Stop light switch
1E Check engine light

I think the latter two can be dismissed all together at this stage. It still doesn't explain why I'm not seeing a spark. The wiring appears correct for this.

[AndreyB]

It still doesn't explain why I'm not seeing a spark. The wiring appears correct for this.

Yesterday another 1994 was started once we've realized the coil wires were flipped in rusEfi configuration - see http://rusefi.com/forum/viewtopic.php?f=3&t=729&start=10#p12354

You would need to update to latest version, this particular version would reset to default by itself - just make sure it is still the same Miata1994d configuration.

Where are we with the fuel pump? It's not on the list of known wiring differences, we should be able to hear the relay and pump operate.