How the 4A-GE ECU works

[firehawk]

This page has nice scope plots of the IGt, IGf, Ne and G signals:

http://www.westfieldsportscars.com.au/Ju...tters.html

(scroll down about one third down the page)

[Bad_Dog]

thanks a lot

[Red]

Where have you folks been hiding.<G>

If anyone knows how or whether the original 85 ECUs were different from the later ones in the non-turbo 4AGEs, like my 85 GTS, I'd love to hear about it. Apparently the 86/87 models got higher EPA ratings and actual higher mileage in the US models--and none of us can find any differences to account for that, unless it is something subtle internal in the engine or the ECU. I was hoping ECU "improvement" might be possible, but since these engines have no knock sensor there are limits. I know there are some aftermarket ECUs for steep prices but with no turbo and no knock sensor, they're wasted on my stock car.<G>

A thought on why there are two boards joined by a ribbon cable: Back then, some of the Japanese electronics companies, including the big ones, sent work home to cottage workers. They might send home 100 people, each with six baskets of parts, and expect to get back 600 alternators the next morning for that famous "just in time" zero-inventory production. In a situation like that, for small parts runs, I could see separate circuit boards going home in the basket to be hand-soldered to the ribbon cable during ECU assembly. Odds are it was this or a similar mundane production reason.

Much of the electrical system (ignition and charging both) on these cars was semi-primitive for the time they were built, about 10-15 years behind many GM domestic products. Like using a cap and rotor, instead of a reluctor assembly, and an alternator with wide (loose) regulation. And using the same exact type of wire for both the alternator output, and battery voltage sense leads. Wasted on one, too thin on the other.

[JMR_AW11]

Red Wrote:Where have you folks been hiding.<G>

A thought on why there are two boards joined by a ribbon cable: Back then, some of the Japanese electronics companies, including the big ones, sent work home to cottage workers. They might send home 100 people, each with six baskets of parts, and expect to get back 600 alternators the next morning for that famous "just in time" zero-inventory production. In a situation like that, for small parts runs, I could see separate circuit boards going home in the basket to be hand-soldered to the ribbon cable during ECU assembly. Odds are it was this or a similar mundane production reason.

Much of the electrical system (ignition and charging both) on these cars was semi-primitive for the time they were built, about 10-15 years behind many GM domestic products. Like using a cap and rotor, instead of a reluctor assembly, and an alternator with wide (loose) regulation. And using the same exact type of wire for both the alternator output, and battery voltage sense leads. Wasted on one, too thin on the other.

Not sure I agree that the Denso ECUs are primitive.
I believe they went to great lengths to make these ECUs as custom and reliable as possible.

A casual glance at a 1985 4A-GE ECU might make you think these ECUs are primitive as they use old fashioned through hole components (no surface mount).

But if you look closer you will see some big tall black plastic custom ICs with labels like "ND HC120"

These hybrid ICs are made by Denso themselves and if you take one apart with a scalpel you will find they contain a little ceramic circuit board with thick film technology with SMD as well.

This is very advanced technology for the early 1980s.

No waywould Denso allow their ECUs to be assembled by semi skilled workers. These ECUs will have been built under strict quality control to ensure reliability. They are extremely well engineered for long term service. Take a good look at the quality of the soldering on these ECUs. It is very good indeed and I would guess it was done by a machine.

Here we are over 20 years later and these ECUs are still performing well despite having been exposed to the shock and vibration from the car chassis for maybe 150,000 miles.

Here's an image of a Denso hybrid IC using film and SMD technology.

This kind of explains why these ECUs were so expensive if bought as a replacement from Toyota.

[JMR_AW11]

firehawk Wrote:This page has nice scope plots of the IGt, IGf, Ne and G signals:

http://www.westfieldsportscars.com.au/Ju...tters.html

(scroll down about one third down the page)

I did a simple drawing of the waveforms from a UK 4A-GE ECU from an MR2.

The UK AE86 will be similar.

[Harvey]

Hi JMR;

I have only just stumbled across this forum recently and have just read this thread. Is very interesting work you have been doing!!

So, are you still able to modify people's ecu and supply them with the kit to re-program them??

I have a few Mk1 (b) (1986-89) ecu's about somewhere, so dont mind having one butchered . My aw11 is currently off the road but i am trying to sort it out! And wouldnt mind trying my hand at this ecu, rather then buying something like a Greddy emange piggy back ecu.

[Red]

JMR-
I never said the Denso ECU's were primitive. In contrast, I said the ignition systems in these cars (and that means everything from the battery to the spark plug) were SEMI-primitive.
In the 1985 US AE88 we have a distributor cap and rotor, both mechanical parts that need routine replacement. That is, literally, SEMI (half) PRIMITIVE. If it also had mechanical points, I would call it totally primitive, because points+cap+rotor were the way cars were built 30 years earlier.
But by 1980, the US car makers had largely stopped using this system in favor of an all solid-state system (i.e. reluctor caps and magnetic sensors) so there were no mechanical tuneup parts to replace in the distributor cap. Toyota was still the more robust 2/3 of the obsoleted system, so I call it SEMI-PRIMITIVE since it was already OBSOLETE.
Thank god they at least got rid of the points, the worst part of it. I can set points to better than 1/2 degree of dwell, with just a feeler guage, but I'm happier not to need them at all.<G>

On using home builders? I'm fairly certain that Denso did in fact use them, I know that the Japanese auto industry used them for assembling startes and alterantors and this was reported in US business magazines which detailed the radical new "Just in Time" zero inventory system that the Japanese car makers were using, when it was "new" to US industry in the 70's or 80's. In the 70's Toyota was still pretty much a joke in the US, in the 80's these were still called "riceburners" because they just weren't up to par with the market standards, i.e. Honda had a massive class action suit over rust problems because they still hadn't learned to provide galvanic protection to metals.

The usual process would be for a company like Denso to build each of the ECU black modules themselves--just the black module with SMD's inside it--in a high tech facility. An then, at 7AM when a cottage worker came in to deliver the last night's work, they would be given ten or twenty more baskets, each basket with two black modules, to circuit boards, one ribbon cable--all the component parts to make up an ECU module. That "gross" assembly work would be done manually by 20-50-100 workers as it was needed to fill the next days orders. Literally, pieceworkers who were given NO WORK if there were no orders, and more work if there were orders. SMD assembly can also be done by the same process--you just need workers with different tools, better eyes, and a more valuable skill set. SMD "kits" of that kind are commercially sold in the US for various projects.

I do not know for a fact if Denso handled the ECU assemblies this way, I do know that style of work, in that industry, was happening at that time. And the way the ECU's are laid out, makes them a good candidate for that time of work.

When you consider the even the Apollo moon lander modules were all HAND ASSEMBLED, and the International Space Station today is still being HAND ASSEMBLED, it can still be the right way to build high technology products.<G>

Again, I didn't call the ECU's themselves primitive. Although, I would certainly like to know if they could be improved (for a stock NA engine like mine) and I'd really really like to know if there are specific differences in the ECUs or the maps from 1985 to the later models.

There are performance differences (significant gas mileage increase) in the last two years that none of us have found an explanantion for, in the US. The tire size change can't account for it, and I don't think the combustion chamber design or heads changed internally.

[JMR_AW11]

Red Wrote:JMR-
I never said the Denso ECU's were primitive. In contrast, I said the ignition systems in these cars (and that means everything from the battery to the spark plug) were SEMI-primitive.
In the 1985 US AE88 we have a distributor cap and rotor, both mechanical parts that need routine replacement. That is, literally, SEMI (half) PRIMITIVE. If it also had mechanical points, I would call it totally primitive, because points+cap+rotor were the way cars were built 30 years earlier.
But by 1980, the US car makers had largely stopped using this system in favor of an all solid-state system (i.e. reluctor caps and magnetic sensors) so there were no mechanical tuneup parts to replace in the distributor cap. Toyota was still the more robust 2/3 of the obsoleted system, so I call it SEMI-PRIMITIVE since it was already OBSOLETE.
Thank god they at least got rid of the points, the worst part of it. I can set points to better than 1/2 degree of dwell, with just a feeler guage, but I'm happier not to need them at all.<G>

On using home builders? I'm fairly certain that Denso did in fact use them, I know that the Japanese auto industry used them for assembling startes and alterantors and this was reported in US business magazines which detailed the radical new "Just in Time" zero inventory system that the Japanese car makers were using, when it was "new" to US industry in the 70's or 80's. In the 70's Toyota was still pretty much a joke in the US, in the 80's these were still called "riceburners" because they just weren't up to par with the market standards, i.e. Honda had a massive class action suit over rust problems because they still hadn't learned to provide galvanic protection to metals.

The usual process would be for a company like Denso to build each of the ECU black modules themselves--just the black module with SMD's inside it--in a high tech facility. An then, at 7AM when a cottage worker came in to deliver the last night's work, they would be given ten or twenty more baskets, each basket with two black modules, to circuit boards, one ribbon cable--all the component parts to make up an ECU module. That "gross" assembly work would be done manually by 20-50-100 workers as it was needed to fill the next days orders. Literally, pieceworkers who were given NO WORK if there were no orders, and more work if there were orders. SMD assembly can also be done by the same process--you just need workers with different tools, better eyes, and a more valuable skill set. SMD "kits" of that kind are commercially sold in the US for various projects.

I do not know for a fact if Denso handled the ECU assemblies this way, I do know that style of work, in that industry, was happening at that time. And the way the ECU's are laid out, makes them a good candidate for that time of work.

When you consider the even the Apollo moon lander modules were all HAND ASSEMBLED, and the International Space Station today is still being HAND ASSEMBLED, it can still be the right way to build high technology products.<G>

Again, I didn't call the ECU's themselves primitive. Although, I would certainly like to know if they could be improved (for a stock NA engine like mine) and I'd really really like to know if there are specific differences in the ECUs or the maps from 1985 to the later models.

There are performance differences (significant gas mileage increase) in the last two years that none of us have found an explanantion for, in the US. The tire size change can't account for it, and I don't think the combustion chamber design or heads changed internally.

Sorry, my fault. I misread your earlier post.

The denso ECUs are certainly 'part' hand assembled as the final ops for fitting the power semiconductors require the fitting of some fiddly screws and washers/insulators. Some of the power semiconductors look like they may have been soldered by hand at the same time as the solder often has residues around the joints.

The big awkward components like ICs may well have been placed by hand as they often have some legs bent on the solder side to hold them in place prior to (flow?) soldering.

I would expect the main PCB soldering to have been done on a machine as it is so neat and even.

Those little black ICs are also filled with black sealant and they take a fair amount of work with a scalpel to get them open. The one in the image above is the main 5V filter/regulator IC for the ECU. It also generates the /RESET signal for the MCU chip and a few other things.

I've seen quite a few Denso ECUs and one thing they appear to do is have a manual inspection (pre solder?) rather than optical. The reason I say this is that the ICs in the ECU will usually have a little white pen streak on them.

It's as if someone checked all the ICs went in the right way round and marked them by hand with a white mark. Maybe I'm wrong about this as I'm really only guessing.

what a fun job that must have been... If labour was cheap they may well have had 20 workers (with a pen each) in a line. Each one checks their own specific IC orientation and marks it with a pen and the ECUs move down the line one step. This would be better than one person checking all 20 chips and going dizzy!

[Red]

"Sorry, my fault. I misread your earlier post. " Not a problem. If I was posting in Finnish, we'd have a problem!

"But I would expect the main PCB soldering to have been done on a machine as it is so neat and even." Well, there are folks who literally do whole-board soldering of SMP components in a kitchen toaster-oven! It can be done and if done well, it is identical to the soldering done on a production line. Except, production lines mean inventory and Japan Inc. has always been against inventory.

"Those little black ICs are also filled with black sealant and they take a fair amount of work with a scalpel to get them open. " Yes, sometimes it seems like the whole electronics industry thinks that will keep the competition from seeing what they are doing. Of course, it won't keep out anyone but hobbyists. But sometimes, the gunk is there to ensure water can't get it and loose solder joints won't form, and to dissipate heat.

"I've seen quite a few Denso ECUs and one thing they appear to do is have a manual inspection rather than optical. " I'm not sure I understand what you mean. Usually, after a board is assembled, a different person with a large magnifier will visually inspect it for obvious defects like solder bridges or cold solder joints, or SMPs that are out of place. Visual inspection is incredibly effective when done properly

"It's as if someone checked all the ICs went in the right way round and marked them by hand with a white mark. Maybe I'm wrong about this as I'm really only guessing. " Sounds like you are right, and that is visual inspection.

"(what a fun job that must have been...) " Well, it the guy who cleans out the toilet waste tanks on jumbo jets probably thinks it would be a step up.<G>

When SONY first attacked the US television market in the 60s/70s, they performed a "300% quality control" program. Literally, every component was electrically tested before assembly, then after sub assembly, then the entire TV was tested as a unit before it shipped. This was incredibly expensive, but it earned them a reputation for top quality. You never heard someone say they had bought a SONY TV which was DOA (dead on arrival, didn't work brand new out of the box.) So SONY got the reputation as "expensive but excellent" and their patent on the Trinitron picture tube of course gave them a unique good picture.

They have long since stopped this 300% QC program, and in fact their quality has dropped in many products made in many countries (I found four cold solder joints in one small FM radio, obviously no little white check marks<G&gt but market pressures to compete on price forced them to change. A friend of mine bought a Korean TV, Samsung I think, in the aerly 90's and he had to take back two to get a third one that simply worked out of the box. But, it was half the price of a SONY, and for half price he was willing to gamble on that.

Since the US car market was already used to the military-grade electronics made by Motorola and Delco in Detroit cars, I assume that Japan felt some pressure to make sure their electronics would be up to par. The same clever way they figured out that Americans wouldn't be so upset by "there's no local dealer if I need repairs" if the cars just didn't NEED the repairs very often.

A lesson our own carmakers *still* seem not to have learned.

[JMR_AW11]

Harvey Wrote:Hi JMR;

I have only just stumbled across this forum recently and have just read this thread. Is very interesting work you have been doing!!

So, are you still able to modify people's ecu and supply them with the kit to re-program them??

I have a few Mk1 (b) (1986-89) ecu's about somewhere, so dont mind having one butchered . My aw11 is currently off the road but i am trying to sort it out! And wouldnt mind trying my hand at this ecu, rather then buying something like a Greddy emange piggy back ecu.

Hi Harvey
I'm not doing much with ECUs at present as I'm taking a break from electronics. I did some work last Summer on the JDM 4AGZE ECU and produced a diagnostic reader and a (big ugly prototype) reprog board for this ECU with all the code/maps etc in external EPROM.

The big circuit board replaces the original MCU chip on the ECU and it connects via two big ribbon cables to the main ECU PCB underneath.
Basically, you can see it still uses the factory MCU chip but it runs it in a special mode that addresses external memory.

It also has the diagnostic reader built into the board as well.

I plan to look again at this stuff in the new year.

Datalogger PCB only (someone is beta testing this for me)



Reprog Piggy Board


Datalogged Data





Some genuine JDM 4AGZE ECU Code showing SC Relay control code