SIMON HOLTBY

I finally got round to developing a propane prom for my car today. Just to remind you my car runs on propane but cannot provide enough fuel at high engine load. To combat this my plan is to add petrol/gas aswell to richen the mixture.
I burned a chip with all the values in the MAF tables set to zero and went for a drive with Graigs scan software hooked up. I identified at what airflow the mixture was becoming lean and have added values in tables 5 and 6. I set the values to about a third of the normal values. Went for another drive and found that I have not started adding gas early enough. As load increases the propane is OK to a certain point then it leans off and then goes richer (ie at the point where gas starts to flow). Early indications look good with some further tunning I may be able to crack this.
I have found that even with all the MAF tables set to zero some gas is still added but certainly not enough for the car to start or run but this has worked to my advantage as I have reduced the propane flow and the small amount of gas added is regulating mixture better than with propane only.
So with all the MAF tables set to zero what determines the gas flow? I'm guessing MAF is ther main factor in determining the amount of gas injected but what other factors are concerned, rpm, etc?
I am concerned that if the car goes into limp home mode that the default fuelling in the LHM chips will cause a very rich mixture. I have connected a relay to the SES light to kill the propane. I'll try putting the car in limp home mode tommorrow and see if the car will still run. I had a problem in doing this as when the car is started the SES light comes on while cranking. To fix this I added another relay which energises when cranking to still allow propane.
My next headache is to get the car to run on gas only for if I run out of propane, certainly my new prom will not allow this. I want to be able to flick a switch and switch to an original prom. I would really appreciate any help on this. My current idea is to run a ribbon connector from the memcal socket to a seperate box where I have 2 memcals one for propane the other for gas. As the proms we use are 28 pin I'll need 28 relays or transistors or something to swop over. Although pins such as Vcc and ground can be left connected on both chips. Do you think the switching could be done with the engine running or would I need to switch the engine off first?
Any feedback appreciated.

Simon

Craig Moates

Sweeet...

I know folks have been successful in putting multiple tuning programs on a single chip and swapping between them just by shorting some leads on the EPROM. That'd put you a mile ahead of the 28-transistor deal. If you're using the 165, that's great, you could just put one calibration on the high and the other on the low memory of a 27C256 or 29C256 flash. If you look in the archives, you should find a clear-cut deal on this. Maybe someone can provide a link?

If you get that going, you could just hook your relay up to swap over to gasoline AND switch programs when the program flakes out or whatever criteria you set up.

By the way, I missed the earlier chapter, can you get more power from propane? Cheaper? Cleaner? What's the advantage? I bet you get nice mixing efficiency having it flash across the injectors, that's just cool.

And the switching should be fine with the car running. You can pull an EPROM out and pop another one in without a problem, I have at least (disclaimer on damages) with the car running. The software selection I think is what you're after.

-Craig

SIMON HOLTBY

Craig,

Thanks for the reply. Propane is half the price of gas in the UK and that is the reason I converted. It does burn cleaner I'm told but performance is less. Not quite sure why that is as propane is 110 octane. I think if you raised the compression of the engine and advanced the ignition to take advantage of the higher grade fuel it may have more power. The latter I could easily try.

I am using a 29c256 chip and it would be great if I could burn 2 calibrations on it, I'll do a search. This would be the easiest method.

The idea I had last night was to connect the 28 pins of 2 chips together with the exception of one pin which I think is called 'chip enable' hence which ever chip I enable will be used.

I was looking at your ECM chip holder, very impressive. A small switch could maybe be added to swop between 2 prom chips using the method above. Hence guys could switch between 'performance vs economy' or 'I know this chip works vs '(latest burn) hope engine doesnt quit at stop light chip' .
Dont know if there would be a demand for this and I am not certain if chips can can be selected/disabled as easily as just switching voltage on one pin.

Cheers

Simon

'88 5.7 IROC

SIMON HOLTBY

Done a search and cannot find how having burned 2 prom configurations to top and bottom of a 29c256 chip I can choose between them.
I may have to connect 2 chips in parallel pin 20 is chip enable and is enabled at 0v and dissabled when connected to Vcc. 2 SPDT relays should do for switching. Just looking at the Atmel data sheet and at standby it uses only 300 uA so should be OK.
I'd still prefer to just use the one chip. Please can someone tell me how to do this or post a link? Thanks

Simon

Craig Moates

Found it, over here:

http://www.diy-efi.org/gmecm/projects/bigeproms.html

It's doable for a 29C256 which would hold 2 165 programs and you could switch between them. Now if we could get a 512 or 1024k chip that is 28-pin...not probable. Looks like when you go to the higher sizes, you need those 32 pins for addressing.

I know there's a 28-pin 27C512, I guess the flash has a need for action on other pins... Life's not easy I guess... I'd love to see 8 128k programs rolling through a 28-pin 010 chip, but oh well. 32-pin it is.

This is something that could be done relatively easily I think. For what you want to do, just load 2 programs into a 29C256 flash, add a resistor and a switch, tie into 3 legs, and you're off to the races. Right? If you need the schematics for the 29C256 layout, check the Atmel datasheet via a Digikey search.

-Craig

Craig Moates

To simplify, I took a look at some of the schematics, and I think this will work:

- For a 165 ECM, use a 29C256 Flash Prom.
- Lift leg#27 of the Eprom so it does not connect to the ECM.
- Connect one side of a switch and 4.7K resistor to the lifted leg#27.
- Connect the other side of the 4.7K resistor to leg#28(5v).
- Connect the other side of the switch to leg#14(gnd).

This setup will essentially allow you to set the data bit (A14) at pin-27 of the Flash chip either high or low.

To recap, lift pin#27, leaving others intact and connected. Patch in the switch and resistor as described, making connections to pin#28 & 14 intact and still connected to ECM. Flip switch to select programs.

This would be very easy to do by just modifying a 'socket' by grinding off pin-27 and then soldering in the stuff. Think I'll try it today, see how it works...

-Craig

Craig Moates

OK, I put one together, works like a charm!

Just took a 28-pin socket, clipped a lead short, soldered in a resistor and a 2-pin header /w/ jumper block (ie. switch). Then loaded my stock 16k into the lower and modded 16k into the upper half of a 29C256. I can now switch programs 'on-the-fly' by pulling the jumper block on and off. Could have done the same with a remote switch also. I have different idle speeds set on the 2 programs, so it was clear when switching between the two which was active. The engine stumbles for a second when switching, but recovers almost immediately.

So, go forward with joy, this is a 5-minute mod...



-Craig

KAOSRacing

How hard would it be to mod your adapter to do this? I'm planning to stick with a 165 ECM and using the 256K flash chips as long as I can and it MIGHT be perfect to have a "street" calibration and a "strip" calibration loaded into one chip, drive to the track on street, switch the jumper, race, swap jumper again when racing is done, and drive home.

Maybe you could include instructions or put it on a web page somewhere? Just an idea.

SIMON HOLTBY

Craig,

following your link to the DIY EFI site I worked out something similar to yourself. I lifted pin 27 (difficult to do without breaking leg) but instead of using the resistor I simply have a switch that switches the the leg between its original position to access the info in the top half of the chip and ground to access the bottom half easy!
My guess is that as the original chips in our cars are 128k pin 27 (address A14) was not used but by default was set to Vcc 5V. Hence the reason we have to use the top half of the chip when using a 256k chip.
So it follows a 512 chip can store 4 prom calibrations with the last 2 addresses of the chip selecting which one to use Ie values

00 01 10 11 going from prom in lowest address to highest.

I have had a good weekend, sorted a lot of problems thanks for your help Craig.


Simon

SIMON HOLTBY

Craig your adapter could theoretically be altered so that you could switch between an original memcal and a 29c256 chip using pin 20 'chip enable' and then further switched between 2 configs on the 29 chip.

I find it a PITA having to remove the ECU each time I burn a new chip. What I am thinking of doing is running a ribbon cable plugged into the ECU to a ZIF socket somewhere more accessable. If you dont smoke the ash tray would be a good spot.
How cool would that be to lift the lid for the ash tray lid and have the ZIF bang infront of you! Just make sure it doesnt get covered in *** ash.

Simon

Craig Moates

Cool, even simpler.

Hmm, using pin-20 to enable individual chips. Could get interesting.

I know you can run into problems with signal loss if the length of the ribbon cable you're describing is too long. I've had that problem myself, although with short lengths you can sometimes get away with it.

I looked at the pinouts of the 29C256 and 29C010A. They're pretty much the same except one is 28-pin and the other 32-pin, with the extra pins used for the added address bits.

I'm thinking of an adapter which allows the use of a 29C010A (8x 16k progs, 4x 32k) in a 165 or 730 system. Just a socket with a few dip switches, and that would take care of it. And if that enable switch works to let you swap between two physical chips, that would be an added feature.

Very interesting...

-Craig

SIMON HOLTBY

Craig,

Can you switch between cals while the engine is running? My ECU screws up if I do this. With the same cal in both halves of the chip it is OK but if the cals are much different I have problems. It is OK to switch with the ignition off.

Simon

Craig Moates

I was able to switch between BINs while running, but the ECM stumbled for just a moment (maybe 1-2 seconds). Swapped between a 32 and 6E, so they were a bit different for sure. It seemed to recover pretty quick. I'll have to test some more...
-Craig

junkcltr

iTrader: (1)

This is an interesting topic........switching to a different bin while running. I know a little bit about the 32 and 6E code. The code is mostly the same, but there are differences. Enough to cause an illegal bus which would through an interrupt to start at the beginning on the code. That is what causes the stumble. Once it starts executing with the new code, everything is fine again. When it comes to just changing the data in the same code, one could possibly through off other data by the local data change. I am not sure the ECM could recover from this. The whole problem is synchrozation of when switch the bins. You need to do it at a point when the CPU doesn't care. Just my thoughts......they could be way off.
cheers,
J

SIMON HOLTBY

Ok here is the latest update....
I took the ECU case apart to isolate A14 pin on the memcal socket, from the processor. This was difficult as the circuit board has 3 layers and the track that needed breaking was in the middle layer. I used a Dremel to drill through the track. Next I soldered a wire to the socket pin so I could switch this wire between Vcc (5v) and ground to choose upper of lower prom cals.
I used the resistor method as this only required 2 wires to be taken out of the ECU box to a remote toggle switch (DPDT) mounted on the dash bellow the steering wheel. Hence the wires were for ECU ground and A14.
This switched between cals fine but every ten seconds when using the upper cal I would get an error code 51 (Prom failure) for a second or so. Luckily I remembered what you said Craig about bringing ECU wires out of the screened box. It was picking up voltage spikes.
To fix this today I bought an optoisolator chip (electronic relay) which I mounted inside the ECU and use the ignition live circuit as my +V supply (pink/black wire circuit ) to power the LED in the opto isolator. Hence no ecu circuits were unscreened. This works fine.
The car will now switch between bins with the engine running I dont know why. I originally used a relay and resistor method and it caused problems.
Just need to do some tunning now.
Could be difficult as the scanning software does not show MAF voltage only MAF airflow and as most of my values are set to zero MAF airflow will not appear to change other than in MAF tables 5 and 6 where I added values.

Simon

SIMON HOLTBY

Also, I use the other pole of the DPDT switch to switch the propane on and off. Obvioulsy propane is on with propane bin selected. This is now my main switch for choosing between propane and gas. I cannot remove the original switch as it has an inbuilt fuel gague and has a circuit that stops propane flow if engine stalls or engine does not start after x seconds. But it can always be left in the propane position.
Finally I may have refered to gas and propane the wrong way round in past mails as in the UK propane is called gas or LPG and gas is called petrol. You can see the confusion.

Simon

junkcltr

iTrader: (1)

I think I understand what you did, but what is a 'screened box'? Are you calling the case a screened box because it acts as an EMI sheild?
J

SIMON HOLTBY

Yes, the screened box is the aluminium case housing the ECU.
I have been getting an error code today, on hard acceleration code 21, high TPS voltage. I'm sure this is due to my MAF table values being so low.
Does anyone know the hex address of a constant that is along the lines of "max TPS voltage vs Airflow" so I can increase the value. Thanks.
Junkcltr, I agree with your earlier point on syncronisation,. Its kind of like reading a handbook for a Camaro and a Corvette, fine if you read one and then the other but if you read first half of one and last half of the other it is confusing. Hence optimum time to switch is after reading one book.

Cheers

Simon

Craig Moates

Simon,

Could you give a little more detail regarding the comparative switching success using an external toggle -vs- the optoisolator? Was there still a 'stumble' upon switch, or was it a smooth transition from one program to the other?

Any other info you could provide with respect to relative success would be helpful.

Many thanks,
Craig

SIMON HOLTBY

If you want to switch between bins using a 29 eeprom the pins of interest are A14, Vcc and ground. My findings was that certainly A14 and Vcc (5v) could not be brought out using unshielded wires from the protection of the screened box of the ecu. I would get code 51 fault code appearing every 10 seconds. I never tried using screened cable this could be a possible solution.
So I concluded the switching had to be done inside the ECU.
A14 could be switched between 5v and 0v to access upper and lower bins or the 'resistor' method could be used. The latter has the advantage of only needing a single throw single pole switch SPST but the former needs a single pole double throw (SPDT) change over switch.
So a small signal relay (SPDT) relay can work with both methods but an optoisolator is only (SPST) requiring the resistor method. I prefer optoisolators because they are cheap about $0.5 for a 6 pin DIL unit with one isolator. Relays can have problems during cranking due to battery voltage dropping which can cause the relay to 'drop out'. They are better than transistors because they provide complete isolation between input and output. Input is an LED and output is a light sensitive transistor.
I chose the purple and black (ignition live) wire in the ECU as my supply for the LED and ran this to an external switch and back to the LED via a current limiting resistor (I think 600 ohm or so) and grounded the LED at the input ground on the circuit board. A14 was connected to Vcc via the 4.7k resistor and to the collecter of the opto isolator the emitter connected to ground.
It is difficult for me to comment on the stumbleing when switching bins as I have the problem of switching fuel supplies. If I switch from gas to propane the engine stalls at idle as it takes a short time for the gas solenoid to open and fill the inlet manifold (mixture is too lean). Switching from propane to gas is OK as the gas injectors start much quicker I think there is a very slight stumble but nothing serious. This could also be because there is still propane in the manifold causing a rich mixture untill all the propane is used up.
If I am driving the car I notice nothing when I switch bins.
I'm very pleased with how it works I thought a least I would have to stop the car to swop bins. My guess is that the processor is confused for a fraction of a second when swopping bins. Maybe a check sum error but it very quickly recovers and importantly does not give an error code.
I'm interested why you are concerned about this I'd guess you guys would have an economical cruising bin and a track bin so even if you had to stop the engine it would be OK. Maybe you want to continually switch bins during driving depending on driving conditions in which case a slight stumble could cause a problem.
I have solved several problems since my last post. Firstly I think that when I said in an earlier post I could not swop bins it was due to one of the bins being bad. No problems now.
Also with most of the MAF tables set to zero I got fault code 21 (high throttle voltage). I believe there is an undefined (in Winbin) value something along the lines of "Maximum TPS voltage vs MAF flow" this was solved by putting small values in table 4 to at least show some airflow.
Also using a relay to kill the propane when the SES light came on was a bad idea as the engine stalled and I nearly lost control of the car (no brakes or steering). The LHM chips did not provide enough gas as I thought.
I was still using a bit too much gas during normal driving, I only wanted it added at high engine load when the propane could not cope. To fix this I doubled the value of the injector constant and compensated by doubling the values in the MAF tables. Gas flow is now neglidgeable.
Let me know if I can help further but as I am not switching between 2 gas bins it is difficult for me to comment.

Regards

Simon

Craig Moates

Thanks Simon, good detail!

Yes, the RFI thing is bad news, but doable. I was thinking along similar lines. Placing an optoisolator (already have a quad-channel unit PCB layout designed) in the ECM box is a great way to go. Wonder if the ECM EPROM 5v supply can handle the transistor end of the optoisolator business? Might be a bad idea. Might want a separate ground too, really keep the isolation that way by going with separate supply. What do you think, use the existing EPROM supply? What about ground? This'll be important.

What I was thinking of is multi-part. Initially, I can set up a switching station via DIP switches external to the ECM for user control, and later I could add a transistor array to allow automated switching between programs via parallel port signals. For instance, one could load up a chip with a group of programs (say 8) and vary the programs to affect performance anticipation. Log the data from the car which characterizes performance under each program's control. When the needed data is collected (say, after a couple WOT passes or cruising), the program could be switched. By keeping track of the programs which are active, and given a little info about the program content, the 'mother' program could dictate the next set of program contents and thus strive toward 'auto-optimization'. It's a bit of a stretch, but it could be done with the right numerical methodology and data.

Anyways, in the absence of a real-time emulator, this might be an option that is very close at hand.

-Craig

SIMON HOLTBY

Craig,

Yes the 5v supply will handle the transistor side of the isolator each one will draw only 1 mA using a 4.7k resistor. The Eprom ground and the incomming ground are common so I dont think there is a problem there.
I'm guessing that you will use a seperate 12v supply to the LED side of the isolator going via the DIP switch and current limiting resistor. This is a bit of a nuisance as there is not a 12v supply at the memcal socket.
It would be nice to use the 5v supply to power the LEDs as you could build a circuit board that plugs in the memcal socket, holding the memcal or eeprom and has a lead coming off it that goes to the DIP switches. It would save taking the ECU apart to find 12v or finding 12 v elsewhere. But there may be noise problems that could possibly be solved by using screened cable and is the 5V regulator capable of supplying the current for the 4 isolators about 40 mA? Checking the spec of the regulator would give a clue. If it is rated at more than 1 A I'd guess another 44 mA is OK but if it is say 200 mA it may not cope. Having said all that finding a 12v ignition hot wire is no big deal.
You have lost me a bit towards the end of your last mail. Possibly you intend to switch bins automatically depending on conditions/drivers intention. Dont quite understand that, how can you anticipate a drivers next intention?
Good luck anyway

Simon