[info] microRusEFI

[960]

Do you have a microscope now, to check it? Use something sharp to check every pin sits, and there are nothing between pins.

I use hobby-knives for this

I have no idea about his soldering or checking, so no offend

[AndreyB]

Do you have a microscope now, to check it? Use something sharp to check every pin sits, and there are nothing between pins.

Check what exactly? I do not understand the instructions. I now have "Portable Digital Microscope With Stage 1-600X 4.3"LCD Screen PCB Soldering Tool" from eBay. Things were working until I touched something with something, how would this be related to soldering?

[960]

Do you have a microscope now, to check it? Use something sharp to check every pin sits, and there are nothing between pins.

Check what exactly? I do not understand the instructions. I now have "Portable Digital Microscope With Stage 1-600X 4.3"LCD Screen PCB Soldering Tool" from eBay. Things were working until I touched something with something, how would this be related to soldering?

Everything seems to be fine with components. You have voltage at KEY pin, the rest needed are directly connected to the TLE8888.

Put it under that microscope, and check all TLE8888 pins are soldered, none loose.

Also check there are nothing laying between pins.

It can still have worked with a loose pin.

[960]

You can also measure if Q1 pin 1 are shorted to gnd.

[mck1117]

This board was soldered with a stencil, paste, and reflow oven. They came out pretty perfect:



(This is the STM32, but the 8888 looked similar. I think this is a pic of the board that Russian has, if not, his is the other identical one I assembled at the same time)

I only had to had rework one spot, but it was over in the analog input area, so it's unrelated. Absolutely worth peeking around the 8888 and surrounding components to make sure there isn't anything wrong.

I think the board fab was probably okay to begin with, but russian bumped/burned/etc something with the meter probe...

But this means gate voltage for prereg missing?

Correct. No voltage at Q1 pin 1 means that the 6v prereg isn't alive, so the tle8888 likely isn't alive... And I agree, with the board off, checking resistance from Q1 pin 1 to ground is a useful check.

[AndreyB]

checking resistance from Q1 pin 1 to ground is a useful check.

Q1 pins 1 and 3 are not connected to GND, did not check resistance yet.

Here are some pin punctures, I believe this is the region of pads 12 and 13 on TLE8888

[960]

Yeah, looks good but I always check anyway.

I usually use a knife like this, scrape careful between pins, and twist to make sure pins are soldered.

I find this really strange.. I dont see how you can have killed a 8888. I have tried a lot, but never made it

knife.jpg

[kb1gtt]

Some test that I helped him with.
-- power supply set to 14V and 0.4A. However it was flowing 0A. This power supply has some front end capacitance, so short inrushes of more amps can Happen.
-- DMM black lead connected to PS -, as well MRE connected to PS -
-- DMM measured 0 ohms to via on several of the GND points of the PCB
-- DMM in V scale, we measured 14V at C1, C3, C16, and D1
-- DMM measured 7V at R5 both sides
-- 0V at R6 both sides
-- 0V at Q1 pins 1 and 3.
-- 13.9V at Q1 pins 2
-- C13 measured 0V both sides.

12V-switched is where most of the bulk caps are. We should have a PMOSFET to prevent back flow during cranking when the 12V will drop. Currently we loose the bulk caps during this cranking time.

Optical inspection might indicate some pads have cold solder joints. You can see a bit of a side to side line in picture 000 above. This might mean there is a not good connection on TLE8888 pads 12 -13. Suggest an attempt at reflow. Solder paste added. Still on phone with him now.

[960]

Do you have another 8888 to try?

Now, with vision you are able to solder it.

[kb1gtt]

This pad concerns me.

[AndreyB]

This microscope likes to hang up during video recording

[960]

Add a lot of solderpaste over the pins, and pull your iron over(laid flat)


You only need 12V at pin: 54, 87 and 90

And key at pin 49

Then it should be running

[mck1117]

12V-switched is where most of the bulk caps are. We should have a PMOSFET to prevent back flow during cranking when the 12V will drop. Currently we loose the bulk caps during this cranking time.

True - that will help with very brief dips, maybe 1ms at most. I think the actual issue with dropout during cranking is that the KEY pin is dropping below the enable voltage, so the 8888 turns off the regulator. We have a few options here, we can enable key off delay (to prevent brief dips from killing it), or change the resistors to reduce the ratio so it will stay on at lower voltages. I'll validate with my board tomorrow.

This has a github issue: https://github.com/rusefi/hw_microRusEfi/issues/93

Probably worth reflowing those pins, as they're the 6v regulator gate/input. If those have a bad joint it could cause issues.

[960]

I think that may be related to a halfway working power supply.

I have cranked until starter cant turn the engine, but never had a shutdown

[mck1117]

How low and for how long will depend on the engine, battery, vehicle wiring, etc, but it certainly wouldn't help if there was a dubious connection somewhere.

[AndreyB]

I have reflown around tle8888 but no improvement, still nothing and 0.00A power consumption. I guess chip replacement is the next step.

[960]

I have reflown around tle8888 but no improvement, still nothing and 0.00A power consumption. I guess chip replacement is the next step.

You have hot air and flux? Do you have solderbraid also?

[AndreyB]

Forum is not best chatting platform in my opinion, we have https://app.slack.com/client/T4NJHQ8QZ/CKNTZMK27/

[AndreyB]

One step closer to world domination.

[MHTSOS]

@russian Could you try some more measurements? Set the DMM in voltage drop mode, the one with the diode symbol. Connect the red lead to a ground and take measurements with the black lead. Measure both ends of C3 that connects to pin 95 and pin 90. Do the same measurement on pin 1 and pin 3 of Q1.
@kb1gtt C13 has 0V on it because BAT_STBY is connected to ground instead of 12V so it can't output V5V_STBY.

[kb1gtt]

I understand V5V_STBY the key we are faced with is how did it die, and should we take preventive measures to prevent it from dying this way again? Right now we do not know why the TLE8888 died.

[MHTSOS]

Russian said that it died while he was taking voltage measurements of 12V and 5V but there is no way he could have killed like that. Even if his body had built up static charge and it discharged through the board it should not have killed it since the chip was soldered on the board and all its pins are ESD protected. The only two faults that I can think of is either the charge pump does not work and there is no voltage to drive the mosfets for the outputs or the driving circuit of Q1 gate somehow failed. By measuring those 2 parts we will know what has failed but we won't know why.

[JRD McLAREN]

Wow .. it's lively there ...

So, I'm read last posts, and see video also ...
and not so happy, with board failure ...

In my opinion...
russian has wrong / low / undersized starting battery ..
(or wrong starter, or wrong starter wiring - high transient resistance)

If battery voltage drops below 10V when we cranking ..
then we have problem ...
"input voltage" pulses and "many in-chip voltage stabilisers" don't like them.

DMM is not right tool to measure "battery condition" ..
Old "analog" voltmeter is the right one.

[AndreyB]

Set the DMM in voltage drop mode, the one with the diode symbol. Connect the red lead to a ground and take measurements with the black lead. Measure both ends of C3 that connects to pin 95 and pin 90. Do the same measurement on pin 1 and pin 3 of Q1.
@kb1gtt C13 has 0V on it because BAT_STBY is connected to ground instead of 12V so it can't output V5V_STBY.

top side of C3 - DMM says 0.5
bottom side of C3 - DMM says 0.7

Q1 pad 1 says 0.666
Q1 pad 3 says 0.52

We are tracking resolution of this under https://github.com/rusefi/hw_microRusEfi/issues/97

[kb1gtt]

Maxxim notes 4V - 5V for 40mS is normal design specs for automotive cold cranking. AKA cold temps, small wires, etc. We should survive a short low blip. They also note 6V for 20 seconds.
https://www.maximintegrated.com/en/app-notes/index.mvp/id/4240

[kb1gtt]

Wow .. it's lively there ...

Yah, that slack think is crazy. I often can not keep up with the messages. It's common that it tells me there are 500+ message from after lunch and before I get home from work. Most days it's like 50 to 100 messages a day. It's crazy and it's usually good to see the conversations happening.

[MHTSOS]

Those measurements look fine unfortunately so this dead ic remains a mystery.
How can I take part in slack chat? It says I need an invitation.

[AndreyB]

How can I take part in slack chat? It says I need an invitation.

https://rusefi.com/forum/viewtopic.php?f=13&t=1198

[mck1117]

In my opinion...
russian has wrong / low / undersized starting battery ..
(or wrong starter, or wrong starter wiring - high transient resistance)

If battery voltage drops below 10V when we cranking ..
then we have problem ...
"input voltage" pulses and "many in-chip voltage stabilisers" don't like them.

Battery voltage can get pretty low during cranking. Remember that in the first instant that the starter contactor closes, it's a dead short. I've seen battery voltage as low as ~8.5 volts for ~50ms just as the starter engages. And that was on my car with a normal size battery, mounted in the front. My race car has a tiny lawnmower battery in the back, and a relatively long cable going to the starter. Looking at a random datalog, the voltage dropped to 7.38v for one sample. Frankenso survives this without an issue.

Anyway:

This one is easily solvable. I did some testing over the weekend, and my microRusEfi board would shut down at just under 8 volts. When a voltage is applied to the KEY input on the tle8888, it turns on its regulators to power up the ECU. This happens at a voltage of 3.5-4.5 volts, per the datasheet. Our board has a 2:1 voltage divider (r20 and r21), so the threshold for the tle8888 is 2 * (3.4 to 4.5) + 0.45v, which gives 7.45 to 9.45 volts. So I'm entirely not surprised that russian was seeing dropouts during cranking, as I've seen cranking voltages below that before.

I removed r21 (the low side of the divider), so now the KEY pin is essentially directly connected to the power supply. My board now shuts down at 4.8 volts, not 8 volts. Even though the 5v rail was all the way down at 4.35 volts, the MCU was still running normally, since the 3.3v rail was still stable.

We should either increase the value of or remove r21 to solve the low voltage dropout issue.

[AndreyB]

I am 100% sure that board#2 (v0.1) used to have both CAM and CRANK channels working before I got TLE8888 damaged (TLE8888 was since replaced and it works now). Now I am unable to get trigger events into the firmware while USB works and LED blinks.

Board#3 (v0.2) had bad solder joint on cam stm32 pad, just fixed. This board has sweet LEDs on crank & cam which respond to cranking as expected. Still no trigger events in the software.

I've compiled software with PAL/EXTI event capture instead of ICU peripheral - still no trigger events.

I am a bit confused. Did I somehow manage to damage only specific peripheral on TWO different boards? While not killing stm32 completely? Sounds unrealistic?

PS: trigger event capture works on Nucleo F767 board so F7 codebase has trigger working in some setups.

This was a software issue, this is now resolved under https://github.com/rusefi/rusefi/issues/898