Understood it's been done(kind of), but there is enough talent on this board to make a low buck one.I am aware of emulators though.How do you think I play nes,snes,n64,saturn,ps(soso) and sega on my pc?I just don't see why we don't have a laptop connection to the prom socket on the memcal.Then after driving,burn the modded bin to an (e)eprom.Not even use a physical prom chip until the session is over.I'm not trying to get out of manually tuning.But nobody can deny the benifits of making changes and seeing results in real time.This would'nt be for lazy slobs.More for a seasoned burner,I could see a newbee blowing his motor fast.Or for people that start tuning with a highly modded engine.Then they could get up & running without a zilion reburns.

 

the biggest problem is if your PC crashes whilst driving, it could put you in a bad spot real fast. I'll settle for pulling over and swapping chips. cheers, Bob

 

I will settle too b/c I can't write software and don't want a hacked up Kalmaker setup.From what I hear his stuff works great though.It's just out of my price range.If it was all software though with a ribbon cable adapted to the prom(zif) socket it would only cost time to develop.

 

If you're worried about that just keep a regular memcal in the car with you. If the the laptop crashes you have a way of getting home.

 

Wonder if you can program specific memory banks of Flash Prom while reading other ones simultaneously...Hmmm... Probably not, but...

Wonder if you could run two Flash Proms in parallel, and switch between them when you want to program one and use the other? Just alternate? Hmmm... Pretty close to emulation, but you'd still need the Pocket Programmer... Or not... Flash programming can't be that difficult...

Sounding good?

Maybe I can take some new reading material to the can tonight...

 

I had the idea of flashing the prom in the computer like you flash the bios in your home computer but was told it takes to much voltage. I just want to be able to flash the prom without having to take it out all the time. If you could flash it through the aldl connector would save a lot of time. just an idea I had.

 

Well, I did some reading, and I don't think the reprogramming of individual banks of an 'in-use' flash chip would work. Looks like the chip goes into a kind of maintenance mode during the flash. Maybe some sort of static RAM? Either that or chip swapping via transistorized selection.

You're right on some of those programming voltages, requires 12-21v DC. However, the Flash chips only require +5vDC. Could be done...

 

Craig, xtronics has a "ad on" emulator with static ram.This would be the direction to go,I would think.Xtronics did tell me the results of their emulator varies.They also told me to wait for new version.That was a very long time ago.

 

Bob,

No need to worry about the computer crashing. when you are using an emulator you send the entire file to the emulator & it runs off of it. if the computer crashes, big deal, the ECM is running off of the image in the emulator.

To all,

There are a few ways to skin this cat, probably the cleanest way to handle this in an add on fashion would be to use twin memory banks & to load the image into one & then switch to it. Then load the image into the other & switch back to it. This is what Eric (of WinBin fame) was trying to do with his emulator, but I dont think that he ever really got it going.

I am using one from Tech Tools & they way it works is that you halt the ECM with the ~Reset line on the ECM, & then load the image into the emulator, it then lets go of the ~Reset line & the ECM boots back up. It is not very gracefull when the car is running, but it really does not matter. You are not running down the road when you do this anyway, so..............

Here is how I do it. I take a drive & collect data. stop & view the data & make some notes in my PromLog. make the changes to the bin, & store it. load the new bin into the emulator, start the car & go for a drive. This is really not any differnet than anyone else, I just cut out the erase prom, burn new prom step.

The cool thing is that the emulator does not need the LapDog to run, I pull it out of the car & go into the house & review all my runs over a cold beer. the next day I can go out & drive the car around W/O the laptop, because the emulator still has the image in it.

check it out if you want. http://www.r71camaro.homestead.com/Emulator.html

my2c
BW

 

Xtronics told me the theory of how their emulator operates a few months ago. It will not work properly with an ECM with the engine running. It will work if the engine is not running AND you re-route the ECM processor reset signal. Do a search, I talked about this before.

Craig, I have been working on an emulator(engine running) that uses 2 non-volatile devices for over a year now.

I think a lot of guys are missing the point here. The ECM software is an algorithm......if you change something, it effects other things and may cause problems. The hardware part is easy. You have to understand the ECM algorithm to make it work correctly.

J

 

So, do I understand this correctly, You CAN use the emulator, if you have the engine shut off, load the image into the emulator, then start the car and drive around? Or will that not work? It seems that the emulator is based on some latches and Static ram, so that if you were to completly store the code in the emulator, before the processor is running, then it should work. The only time you need the reset is when you are trying to do a live 'dump' to the ram, and need to make sure nothing is trying to read the code while being written. Is there something else preventing that from working? How did xtronics explain the operation?? Just seems like even being able to do that will save wear and tear on the eproms, sockets, etc.

 

Works for me, but I am not using a X-Tronixs.

BW

 

You are right about holding the processor in reset while loading an image. That is how to do it with the engine not running, but key is on.

Xtronics seems to work by using by stealing the first part of the read cycle. The ECM read signal is disabled while the emulator logic performs the write to the memory. Remember that the PROM is a slow device. The static ram used is fast and that is what it makes it all possible.
J

 

But what happens if you swap memory banks on-the-fly during ECM operation? Even if it's done instantaneously (ca. 20ns), does it need to be done at some point in the ECM's read cycle?

I mean, for instance, if you were to have two Flash chips and try to switch between them on-the-fly, what are the important points to keep in mind?

I'm working on a switcher that's about ready, and it does employ a 4-bit S-R latch circuit (without optoisolator however), but I'm concerned about the timing of the switching event. Does it need to occur at a certain point in the ECM read cycle? The same would apply to this two-chip 'emulator' concept, just add a flip-flop array.

-Craig

 

Craig,
What you are looking to do is almost a different topic in itself. You have two pre-programmed images so no writing is to occur.

What you need to look for is the rising edge on the chip select. When you detect it, switch over to the other bank.

Also, it terms of needing an opto-isolator.....you shouldn't need it. Think about it, the MAF, MAP, O2, etc. are analog signals and look at how far they are from the ECM. The ECM doesn't have a problem with these. You shouldn't have a problem with yours.

You probably have a switch bounce problem or something else. If it is switch bounce then you could corrupt the first read and cause a fault.

J

 

One other thing that I can't stress enough is that the hardware is the easy part.

The hard part is knowing what values you can change that won't cause problems with the ECM algorithms. If you change a value, it propagates changes to other values that could be 'un-natural' to the ECM and cause a fault.

You have to spend A LOT of time reading the code
to understand what you can and can't do.

J

 

Hmmm, maybe a timed pulse triggered by the rising chip select that, when ANDed with the user switch select would trigger the S/R latch array. Need this for the S-R reset-set transition as well, I suppose.

You're right about the pre-programmed selection being a bit off topic, just trying to plug into the discussion. I think the same concepts could be readily applied to a standby-chip-type switching array where one gets programmed while the other is selected.

If you could only write info to a given bank of a chip while keeping another bank active for ECM I/O, that would be sweet. Don't know if it's a can-do, though. From what I was reading last night in the Atmel pdfs, it looks pretty darn close.

 

It looks like you understand about the gating of the ECM read with your S-R latch.

I know what you are thinking about in terms of trying to have the ECM write a seperate FLASH will reading from another. That is possible.

You can't have the ECM write only one FLASH cause it takes too long. You can keep the watchdog happy but other things get messed up. The overall state machine goes wacky.

My initial attempt was to write a single PROM and learned that it is too complicated.

My device uses 2 FLASH parts and a lot of logic to make it happen. It was designed last year before I became part of this board. I thought I was the only one that wanted such a device so I have spent A LOT of time and money making it happen.

You are on the right track though
Have fun,
J

 

maybe the solution is as simple as resetting the ECM every time you switch to the new image & have it come back up as if it were just started. this might avoid problems of catching it amid making some change.


BW

 

Hitting reset will always work. Generally, it is poor design practice. It usually means that the circuitry is designed poorly. Hitting reset is known as a 'fix all'. It also requires that the ECM be hacked to bring out reset. I wanted a clean design that didn't require modifying the ECM ckt board.

J

 

The reset line seems to work for kalmaker.I'm not saying copy him but to use less hardware.Could the reset process be sped up somehow?If so then you could get any rom emulator to work.?

 

Like I said, using reset to change system settings is known as a hack in the engineering world. No offense to the Kalmaker? thing. It means they didn't want to take the time to make a clean system. Try making a PROM change at 6000 rpm and see what happens.

Anything that uses the reset line will not be an emulator. An emulator functions with the system under normal operation. The thing that uses a reset always re-initializes the system.....not a normal system operation thing.

Yes, you can use any emulator if you mess with the reset. Personally, I wouldn't call it an emulator when used like this. I would call it a hack.

J

 

I understand where you are coming from, but you are not desiging it you are re-designing it. the difference is that if you had done it in the first place, you would have done it right. you are trying to fish in GM's pond now, it might not be as clean as you want it to be.

IMO, the important issue is, does the design satisfy the desired results. if the desired results are to have it clean, then start from scratch, if the desired results are to have it work & work reliably (good eng practice #1), then this might be an acceptable solution to the problem & the design goals are met.

BW

 

I am sorry but I don't agree with you. I believe GM did it right by not allowing the engine to be re-programmed while the engine is running. It saves on law suites from crashed vehicles due to improper programming stalling the engine while driving.

An emulator should be a drop-in device and not require a board to be hacked. That is why I wouldn't call the device being discussed here an emulator. Also, it is not necessary to hack the board if a 'real' emulator is designed corretly. This involves a detailed understanding of the operation of the ECM. A quick easy design would use the reset line. Yes, it will work somewhat OK. My goal was a device that would work all of the time just like the way GM and other places try to design their stuff.

It is not an issue of having to live with what GM did. It is an issue of gaining enough knowledge and understanding of the ECM in order to design an emulator for it.

I hope you don't take offense to me disagreeing. I can kind of understand where you are coming from.

J

 

Tomato - Tomahto, Potato - Potahto,

if I get the car tuned in a very short period of time, I personally dont care.

I did not say that GM did a bad job, the ECM does exactly what they designed it for, which is my point. we are trying to make it do something that it was not intended to do. What is the intended goal & where are the compromises?

look up the definition of emulator, & look @ the device that you are using. this device is emulating a UV prom, it is an emulator. What you are implying is that you want to emulate the ECM, which is not what the majority wants. if they wanted that they could use a PC104 or similar & emulate the entire system. but THAT, my friend, is another game entirely.

you would have to call my mother a WHOLE lot of nasty things for me to take offence, lifes too short.

Have a good one.
BW

 

It looks like they were from what I've seen, they (some, to include the P4s) are programmable on the fly. If you saw some of the calibration gear they used in developement work you'd see that it was.

I've seen two of the developement setups now, but without any software. They used an emulator, and the HUD had a set of dials they could change settings with. The only ties in were thur the Prom holder (memcal) and ALDL. How everything interfaced was beyond me. I have talked with a guy that was part of the drivibility team for a number of years and the stuff was all plug and play. The one clue I can offer is in the one version, there was an additional ecm case that had a microprocessor in it.

Also in a number of hacs they mention the HUD. The HUD isn't what you'd think of nowadays. It was a set of led read outs and rotary switchs. The emulator allowed for calling home to the mother ship for patchs when they incured problems.

I don't know if they could make code changes on the fly but it sure looked like they could dial in calibration changes from the HUD. And the one guy verified they could in the series of cals he did.

All the above is C3/P4 specific, you mileage may vary, not to be attempted with out adult supervision, off road use only, yada yada...

 

That does'nt sound much different than a rom emu(the way you described the switching).What if I loaded multiple roms on an emu and switched between em?It might need to use the reset line,I think.The emulators I have seen makes switches w/o upsetting the program.



Come on grumpy,cough some more up.

 

I guess this one will die again.I just can't program,at least not yet,but here is my idea>Have the ecm read the bin from the laptop.Then have a program like tunercat to make the changes realtime.When you're finished tuning then burn your prom.Pow there it is in it's simplest form.Junkcltr I can see you have alot of time in this.So I understand it would be a hard process.I just wish the people that can program get together and hammer this out.It's only a matter of time before some company makes it.The sad thing is the price will be more than it's worth.Considering most people have only one engine to tune.There I have said all I need to on this subject,for now.

 

When I designed my stuff I did it for myself cause I thought it was a pain to keep popping out chips to fix stuff up. I thought about selling it last year. The cost to make them would be the same as buying an EPROM burner, tunercat, an ALDL cable, and a few chips. So you spend the same amount of money, but you can program on the fly (essentially for free).

What I didn't want to do is have the hassle of supporting the device. That is a lot of work in itself. Nevermind, the upgrades people will want. Look at the wideband project. It is an excellent idea and great contibution to GM EFI. But people complain about the service. I didn't want that with my stuff. Also, look at how techedge ripped it off. I would rather burn my design and run over my ckt board with a steam roller before someone else makes a buck off it.

Anyway, mine works off two EEPROMs so you never have to pull chips to burn after you make changes. It burns them for you. Lots of software to do it. Quite a bit of hardware too.....about half the size of the ECM.

Also, all this talk about having a ckt that is remote to the ECM is BS. If you design it right it will work with out the random SES light.

One more thing I thought about last year when I was thinking about selling it. These GM computers are about 15 years old now. How much longer will they be around?

I also thought about selling it complete or as a kit for cheap money. The problem is that with kits, you need a lot of support cause people will buy the kit and mess it up. Then they need loads of help to fix it.

It is a great idea though. I hope this one doesn't die. I'll try to give as much advice as I can.

Have fun,
J

 

junkcltr,If you put up some pages with pics you could just let the project speak for it's self.I feel what your saying about supporting a project like this.I also agree with letting something burn to the ground before someone rips you off.But there is a way to do it without someone getting screwed.Posting the info free on the net will stop thiefs cold.Once you have the info on the net people will learn how to do it.Then they can help others.It sounds like you've been frustrated by it all though..Junkcltr,what about my idea though.Emulate the prom with all software.I guess this would take too much programming time?This is the firs step I would think.To be able to read the bin from your hard drive to run the ecm.If that could be done then it possibly could take a simple save from a bin editor to change on the fly.This seems to be the easiest way to do it.No hardware asides from a ribbon cable from the ecm to pc.

 

I didn't see any way to do it all in software using a parallel port or serial port on a PC. They are just too slow. You also have transmission line problems. Some sort of extra hardware is needed. Yeah, the old ECMs are slow, but not this slow.

Even with the extra hardware, a serial port or parallel port is required for the FLASH updates. Yes, this requires a lot of software. So, in reality there is a lot of programming even with the hardware.

One thing I don't recommend is trying to make updates through the ALDL link. It is too un-reliable. If you ever had your ECM lock up and stop spitting out data then you know what I mean.
Also, it requires a code change to the GM code image.

J

 

On the fly tuning requires the new updates come from somewhere. I save them on the PC / hard-drive and dump them to the tuning device.

Do a calculation on the speed at which you can realistically send the bin to the ECM when transmitting the entire image. You will probably be surprised. It takes a long time unless you do it in a parallel fashion. It is un-necessary. You are better off just sending updates as needed and offer a function to send the entire bin.

J

 

What about this,just getting to run the ecm from a bin on a laptop.I'm not saying to hook up to the aldl.But to plug in to the eprom socket on the memcal to A port on the pc.You mentioned that the pp/sp is too slow .What about a conversion through the usb port.Or an ethernet card or some type of new port device?So how hard would that be,just to get the ecm to read the bin?Later would be figuring out how to open and modify the bin with no ses.

 

If you use a serial link such as 10 Mbit / 100 Mbit, USB, RS-232 you will need some sort of hardware at the ECM to convert it to parallel data. You could build a ISA or PCI card that does the parallel communications to the ECM. But now you are back to building hardware and it is cheaper to build it at the ECM than at the PC level.

Just as an example.....look at how fast the FLASH programmers(e.g Xtronics) can program a device. It takes a long time in the nanosecond world.

For the heck of it I just looked at parallel port speeds and you can't use it because it is too slow in it's normal modes. You might be able to figure out how to make it faster. Personally, I can't see it working. You might want to dig deeper into it.

The reason I stayed away from using parallel port stuff is due to 'Windows' and having to write a device driver for it. It takes a some detailed knowledge to do it. It was more than I wanted to learn about controlling the parallel port in 'Windows'. If you use only Win 95/98, you don't need to write a device driver. But Win2K is here to stay so you might as well make it work under that or go straight DOS on a boot disk. But then you loose the graphing capability while tuning. That is a nice feature you probably want......realtime graphing.

I personally think the best way to do it is two use either 2 or more FLASH devices or a single dual port RAM. The RAM method requires less hardware and software, but you need to program the PROM afterwards. I went with the 2 FLASH parts because I don't have to do anything when I shut off the key.......plug and play whenever you want.

Or you could just go with an existing data card for the PC. They are spendy though. And the worst part is you don't learn much about how the ECM works. I think that was the best thing I got out of messing with it.

J

 

The xtronic emulator uses ram,but do you have to burn before shut down? This sounds like your device,aside from the flash prom(s?).I guess this is good enough for me for now.Really tc/winbin,flash rom,pp and emulator is light years ahead of gm's way of tuning the 165 ecm.I have heard of the rotary switch type tuning equipment,that grumpy speaks of.Really it's low level crap with little real world tuning ability.I am going to learn the code in my bin.But can't learn everything so J,I guess we'll bury this post huh..

 

Nah, it doesn't work at all like mine. It requires you to do a burn afterwards and you can't tune smoothly with the engine running.

It is quick and easy though. It is good if you don't want to spend a lot of time designing stuff.

Yeah, this one sounds done. Buried.

Have fun,
J