pgmfi.org

The reason the fuel pump start delay is, TA8903 is a voltage regulator with watchdog timer. Some how the internal circuit/capacitor problem. It delay to give 5v output to MCU. To bypass these you just need to replace with modify fixed 5v to the MCU.

and need to solve feedback logic to MCU so it wont trough CEL.

Great to see this topic lives, thanks research continues.

So, if I understand correctly, dirty repair could be made by desoldering TA8903 entirely and make a wire bypass of +5V to M66207 MCU and figure out if MCU code requires any kind of feedback.

According to the only one schematic I found related to TA8903 and its surroundings (found on hackedecu website), it could be easy to do. Unfortunately I was not successful when looking for TA8903 datasheet and so I have the only pinout information from this image.



Some slight information could be extracted from schematics on this image (i would love to give a credit but I really don't know who made it - most probably someone from pgmfi but can't find.)

I also found a datasheet of TA8000S - it's the other part with other pinout but as some labels are quite similar and purpose of this IC is close to TA8903, I consider this as a very rough inspiration how things work in TA8903. (datasheet /pgmfi/forum/media/18/18075fe2275acf26f8b661af2699c8cfca65e18a30613e1ac216f4beab07a11f.pdf )

So, what could be worth trying:
- desolder TA8903 from PCB
- pin 1 was Vcc of TA8903 - not needed anymore
- pin 2 was BIAS of TA8903 - connected to +5V, not needed anymore
- pin 3 was ??? of TA8903 - connected to bypass cap 10n (C82) and to pin 13 of 74HC00 (IC20) - probably would be OK connect to ground
- pin 4 was ??? of TA8903 - was connected via 10k resistor to +Vcc - not needed anymore
- pins 5 and 6 of TA8903 - ground, not needed anymore
- pin 7 seems to be reset watchdog timer (resistor pullup, capacitor to ground = typical reset pulse circuit), not needed anymore
- pin 8 seems to be quite complicated, see later.
- pin 9 was ??? of TA8903 and was somehow related to D11 pin of ECU (backup power) - most likely this is only an input for TA8903 so it's not needed anymore
- connect pin 10 to ground so Q6 completely opens (however I don't really understand why Q6 works if D7 de facto bypasses +5V to "output")
// in second schematic the transistor is called Q4 with different surroundings - but the "core" logic is the same, GNDing pin 10 opens transistor and transistor puts +5V into MCU pin 64 (VDD) and IC20 74HC00 power pin (VDD)
- pin 11 is not mentioned on any schematics - let's omit it, if it is not okay let's try different logic levels to pin 9 of 74HC541 (and figure out which pin of 82C55 is powered by this IC - output of 74HC541 input pin 9 is on pin 11)
- pin 12 was some kind of output, my guess is "power good" signal to MCU pin 47 (port P4.1 = TM1CK?) - probably would be OK connect to ground so MCU P4.1 gets logic L

Pin 8 have some close relation to NMI trigger of MCU.
It's connected to the following destinations:
- pin 12 of IC20 (74HC00) which is one of two NAND inputs (the second one is from pin 3 of TA8903), I wonder where output of this NAND gate is connected (pin 11), have to check on some ECU
- pin 12 of M7U016 - i have no information about this IC
- pin 33 of MCU 66207 = NMI (inverted logic)
- two bypass capacitors (1nF)
If NMI is not to be triggered, I assume that pin 8 has to have logic H so it's interpreted as "no NMI".

Do anyone have some ECU with fuel pump delay so this "TA8903 elimination" could be tested? Unfortunately I don't have any ECU with this problem left as I used those ECUs for parts

I have the delay fuel pump issue and will be changing my caps hopefully this next week, but having done the research that brought me here I'm expecting it to likely end up being ic17. My ecu is a p0b-a01 but it still has the same ta8903sn. If been trying to find replacement but the only options are harvest used parts just as old, or china. There's some us listings but they want like almost 100$ for what likely is a harvested part itself. In any event il run whatever tests I can when I'm doing all my soldering when the caps arrive later this week.

Have you tried to find the part being used in any other piece of electronics? If seen one mention on an all Chinese forums referring to it as like ic665. But anyway thought is chime in so hopefully you can see it in time to continue your research into this.

I have found a part identical to the 8000s datasheet produced by a different manufacturer than toshiba
I'm not sure if KIA represents the car brand but the part number was
"KIA8000S" you can look datasheet up on Google or view the link I included. My capacitors have arrived so I'm probably going to begin my repair attempt in the next day or two. Will keep looking for more info.
https://html.alldatasheet.net/html-pdf/ ... 8000S.html

I have brand new non-china TA8903SN's if needed that I can sell for not $100 each lol.
I bought a large quantity of them some time back from a reputable US electronics parts dealer. They work great and I've brought back several dead OBD1 ecu's I've sat on for years.

LMK

Hi, i have two ECUs whith same issues, delay in the fuel pump and CEL, afther replace de caps, the problem continus, but afther read este post, i tried the following, heating the TA8903 with a heat gun, and the ECU turned on right away. Now i´m going to replace them.
Thanks and a hug for all.

Hey guys, just wondering if we could get something out of this i saw on facebook, its a Bypass on the ic17

The diagram photo, is missing the beggining (the part with the resistor on the left of the capacitor, anyway we could discover this?

IF I SAW IT I HAVE THE SAME DOUBT WHETHER THIS WORKS OR NOT?

ITS OK?

Doesn't physically look like a "drop-in" replacement...