Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[colt45]

got a 4agze im building on the stand.  the ecu needs to be addressed before too long.  any ideas or suggestions?  i've considered megasquirt...

[slappynuts]

Haltech platinum sprint 500.

[92CXyD]

Use a P28 and check with the guy with the mr2 build.  http://www.realhomemadeturbo.com/forum/index.php?topic=15111.0

That will be your best bet.  :yes: ;D

[2slow]

Yeah it seems like honda ecus are where its at for toyota guys

[Joseph Davis]

The 4A distributor situation is the reverse of the 3S.  If I remember it correctly, the 3S needs the Honda housing (stubby part with the o-ring) turned down to a smaller diameter.  With the 4A you need to make a sleeve to make it a bigger diameter.

Nothing against Haltech in this case, Doctor B, but if you want a 4 cyl distributor solution there is nothing superior to an OBD1 Honda ECU.

Also, some of the JDM 4A-GZE have a cam sensor in place of a distributor, and run wasted spark.  I'm going to build the harness (look at if it needs built, seemed plug and play from looking at the swap on a pallet) on a FWD AE101 4A-GZE swap with that setup into a 1G MR2 this afternoon.

[TTC]

Shit, i had a giant post to give you but lost it.  If you are limited ona budget but can fabricate go honda, if you cant go with megasquirt. If i was to do it again, id dump MS and save for AEM or something similar.

[Robb]

AEM EMS are going stupid cheap right now.  I picked one up a few months back for $500 in perfect condition.

[Joseph Davis]

Rossnoggin gave me a 30-1000.  I'm about to pick up a 30-1052 box for silly cheap because it has "a burnt resistor" and is therefore worthless.  The GEMS designed and produced motherboard is really very nice for a ten year old design, it's the AEM adapter board that has all the gremlins.  Honestly, there is some raw talent in the MS scene, some of those guys could redo the adapter board and incorporate correct handling of VR sensors, etc, and clean up various application-by-application flaws with the AEM I dropped out halfway through community college implementations.

[TTC]

AEM EMS are going stupid cheap right now.  I picked one up a few months back for $500 in perfect condition.

Robb, that was the Series 1? I'm still waiting to find one for less lol cause im a cheap ass. Maybe over the wintertime or some shit.

IIRC you can buy any AEM unit and reflash its bios to wtf ever u want.

[Joseph Davis]

The EEPROM on the GEMS board is a 29C256 and if/when it bricks itself (happens) you can reflash with whatever v1.xx .gin firmware file and go.

I've seriously considered reducing the GEMS board to FPGA and releasing it to the public.  FPGA based 68HC11 emulations run ~2.5 time faster than any 68HC11, which is lolololol faster than v2.0 AEM.  I'd probably be free to do so as long as i didn't profit - look closely, the AEMPro software has exactly zero copyrights.

As Jared Karagen's father lectured him about, reverse engineering anything is stupid and reverse engineering something without pay is fucking retarded.  He has an essay on the subject typed up on pgmfi.org, and reading it was a turning point in my life.  Jared's still all about it, but I very much see his old man's wisdom.   I've reversed hardware and will continue to do it for simple things (GEMS "AEM" is very straightforward and honestly enjoyable to reverse, I hope for many such pieces of hardware until I've taken my hardware knowledge past *needing* to do such things) but taking on a whole EMS that I can't even profit off of isn't going to happen.  I may or may not design one myself in the future, hardware is getting so powerful for the dollar that you can write sloppy code in VB and have a world class EMS if you know what you are about.

On that subject, makes me excited about what Fred's doing.

[TTC]

How can i exploit this fact? With no electrical hardware  design knowledge.

[Joseph Davis]

Acquire electrical hardware  design knowledge.  Trust me, you will be bald and unattractive for a very long time before you are allowed to die, don't worry if it takes 4-6 years to get up to speed as it occupies time enjoyably until you are allowed to escape this wretched mortal coil.

[kgx]

The easy/dirty way to run the P28 on the 4AGE would be to waterjet a 24x/4x/1x cam trigger wheel, read it with hall sensors wired directly to IC10. I've got a wheel drawn up with the intent of going that direction with the 3SGTE installs. 3x allegro micro hall geertooth sensors would do the job for under $30.

[Joseph Davis]

You could probably sell the H1C1 boards you pull for $5-10 each plus ship to guys trying to repair their shit.

[colt45]

Acquire electrical hardware  design knowledge.  Trust me, you will be bald and unattractive for a very long time before you are allowed to die, don't worry if it takes 4-6 years to get up to speed as it occupies time enjoyably until you are allowed to escape this wretched mortal coil.

the profound truth in that satement makes me sad...

however

1. gain knowledge
2.
3. profit

[kgx]

1. buy books
2. gain knowledge
3. download simulation software
4. experiment
5. profit.

here's the cam wheel i drew up:
www.apexology.com/files/schematics/camwheel2.DXF

get it cut in 1/8" thick 1018 steel.

big blue saw waterjetting service quoted them at $10.50ea for 10 of them. it would just bolt to the face of the cam gear (pretty sure it'll clear adjustment screws for adjustable gears). single beam with 3 sensors inline would be necessary to read it. allegro micro ATS675 were the sensors i was looking at.

[TTC]

So could you just take apart a honda distributor and use those sensors with that one wheel? I don't know much about honda ignition, but arent there like 3 wheels in it? Are you substituting these 3 wheels with this one single cnc'd peice with 3 sensors on it?

[kgx]

you wouldn't be able to use the honda sensors because the wheel is designed to be read from the face, not the edge. my plan is 3 hall effect ICs implanted into a beam that goes across the face of the wheel. set the beam to pivot at the center with a second fixation point and you could even be able to adjust timing without having to stop the engine, loosen the cam bolt and turn the wheel slightly.

i do need to start working on this more, since i have like 5 people wanting me to convert their MR2s at the moment.

[Joseph Davis]

That would make converting MR2's more straightforward, and with the MR2 Guru across town I imagine some shit would get cooked up.

[crxvtec91]

After reading this my brain hurts

Shit I need to do more reading.

[TTC]

So the beam would just be aluminum with some holes for the sensors to sit in.  Would the honda ecu read these 3 hall sensors normally without any weird pots to tweak?

I can't see your wheel but it must go like this.  Its a wheel with 3 rings of "teeth" cut into it for the sensors to read, right?  The sensors sitting on the beam would just sit over the rings they are searching for.  Each ring will have the correct number of holes and spacing to give the oem ecu the proper readings.

[kgx]

link just had an extra period in it. it's not an image, it's a DXF file for CAD software. you'd need to download the file and open it with something like autocad or solidworks.

here's what the wheel looks like:


the sensors look like this:


the sensors would be wired to the output side of IC10, bypassing the HIC1 signal conditioning board entirely.  they're about a 10mm circle with the top and bottom flattened. i believe the spacing on the "teeth" on the wheel is 16mm center to center, so the beam would just be 3 sensors in a row (orientation matters) with the centers 16mm apart.

[TTC]

Yeah, that is exacxtly what i figured.

On the Ic10, is that a portion of the honda board that receives the ignition signals from the halls sensors? Excuse my ignorance, im not familiar with honda ecus.  Why do you have to bypass the conditioning units?  Are there 3 inputs on the ic10, 1 for each hall sensor?

Also What size is the center bore on your Cad file.


2nd q:

allegro micro ATS675

Those look very flimsy and breakable, how do you plan to house them? The leads on them are just that, leads.

[kgx]

IC10 buffers/inverts the outputs of the HIC1 board. HIC1 converts the analog VR sensor (in the distributor) signals to digital. the HIC1 board cannot read digital signals because the input voltage must pass below ground to trigger the board. IC10 is on the output side of that board and merely inverts the signal. since the signal we'd be generating is already "inverted" so to speak, there's no need for IC10 to invert it, so we connect to the outputs of IC10. there are 3 signals CKP, CYP and TDC.

the power and ground rails of the sensors can be tied together provided the proper bypass capacitors (1.0 and 0.1uF should be sufficient) are used. i would recommend running a separate 5v regulator dedicated specifically to the hall sensors, rather than trying to pull from the ECU's sensor 5v rail, since these will be switching and generating transient voltages and jitter that you wouldn't want present on the analog sensor rail. probably just me being anal though.

the sensors would be affixed to the aforementioned beam with some form of adhesive. the leads would be wired together and then potted to the beam with silicone or urethane to stress relieve the component leads and waterproof the assembly.

[TTC]

Sorry, ninja edited on ya earlier.

So basically, you would have power, ground and 3 signal wires going to your aluminum beam.  The caps are just to limit noise iirc?, and you just put those inline within the power and ground wires? Or do you put them, between its been a while and I forget simple electronics.  Where would u get a 5v power supply, and how do you decrease the voltage? Again, these are probably basic questions lol.

[kgx]

caps go parallel to the circuit. essentially, they absorb voltage spikes and provide power on voltage dips to keep voltage level. install between power and ground.

power source would be an external LM7805 voltage regulator connected between 12v and ground. you can get them at radioshack, along with the capacitors. wouldn't be a bad idea to heat sink the Vreg.

[TTC]

OH ya, that makes sense. I have a 25uf cap on my coil from the ground to the power source.  So the cap goes inbetween. Would the LM7805 just be able to draw current from w/e available 12v source. It will decrease the voltage im assuming since its a regulator.

It has 3 pins on it, 12v input, 12v output and  ground?

Have you tested this wheel, is the signal clean enough that the honda ecu wont take a shit fit?

[kgx]

not tested. can't see any reason it wouldn't work. the timing lines up well enough to the scope captures i've seen. we're just cutting out the analog middleman.

[patsmx5]

. wouldn't be a bad idea to heat sink the Vreg.

+1 They catch on fire when they get too hot....

If you can easily/cheaply hook up a honda ecu, I'd say do it. Megasquirt works, but it's not the best ecu in the world. If you do MS, I would just put a trigger wheel on the crankshaft and a 20 dollar VR sensor to read it. Mounting the trigger wheel requires some effort, but then you'll have accurate spark. On my MS2 setup with a 36-1 trigger wheel. I can set the timing fixed at 10* BTDC, rev the motor to 7K, and a timing light shows the timing sitting dead on 10* with no scatter.

[TTC]

Timing on my MS wth the EDIS setup has always worked pretty good.