DFI vs FAST ?

[Two Lane]

USEFUL FEATURES FOR NON-NECROMANCERS

****INFO ON GEN 7: (March 2002)****

http://www.mrgasket.com/maildirect/md_archiveweb.asp?ID=310


****INFO ON GEN 7+: (April 2003)****

http://www.mrgasket.com/maildirect/md_archiveweb.asp?ID=444

_________________________________________________

HTH

 

[Reggie West]

The thread was about which was best.

No it was not.

The thread was about switching from a Gen 6 to a FAST unit and if the stumbling problems G McCall is having would be fixed by purchasing a FAST system.

 

[aDFIguy]

Ok, Here goes nothing...

First, for Bruce:

...is the code running continuosly or on a series of interupts.

Definitely interrupt driven. Timing is even more crititcal in aftermarket applications. How many OEM ECMs see applications on engines that spin to 12K or 15K RPMS? Especially with strange crank signals like a 60-2 tooth wheel. Think about the measurement and calculation interval for the crank signal on a 15KRPM * 60 Tooth application.

Are they actually doing PW corrections as the injector is firing.

That and much more.

Are any of the programs really turbo specific?.

How about a strategy that allows control of MAP via a programmable rev limiter in order to cause a turbo to spool up instantly on the starting line?

If you read thur what GM has been doing for years, are they even trying to match them?. GM has had a Power enrichment multiplier for time in PE since the GNs came out.

GM Racing has been using the DFI 7+ box to run their FWD drag cars since their inception. Here is how the last race went:
PRO FWD Class
Hot Rod FWD Class
Note that the Civic that beat the PRO FWD Cavalier also runs the DFI 7+ system.

TCC stuff has been ignored just until lately by the aftermarkets.

DFI has had TCC control in the Gen 7 box since it box came out.

I see from the last post the 8 to 32 bit advantage seems to be in datalogging...

32 Bit processors are very rarely used for engine controls. Tried and true 8 and 16 bit mcus are much more cost effective, and provide plenty of speed/resolution. Advantage of 32 bit procs is floating point math, but that requires much higher clock speeds, and quickly becomes financially unfeasible. On the other hand, Analog-to-Digital converter resolution is very important, and it takes a decent processor to give you at least 10-bit A/Ds. This has limited the use of 5 BAR MAP sensors due to the coarseness of MAP measurement that would be seen with 5 BARs at 8 bits.

OK, how many frames a sec will the new aftermarkets record at?, and how many parameters?. Will they accept extra channels.

Depends on whether or not the system has internal SRAM ($$$) for onboard data storage, as well as the number of free analog and digital inputs on the processor. Higher end systems record internally at 50-100 Hz, for short periods of time. These systems usually offer a few unused analog and digital inputs to be configured as extra datalogger channels. In the case of the DFI class systems, no SRAM internally, so no onboard data storage. Likewise, no extra pins for datalogging, so data is xmitted via serial link and RFI limits the effective bandwidth.

Why don't they release the Source Code, for people to look at?.

Ask Microsoft, Apple, Sun, Adobe...etc the same reason, and you'll get the same answer. Proprietary Intellectual Property. OEMs make no profit specifically on their engine control systems, so they lose no real money by releasing the details of it's operation to the public--long after the fact.


Some other random thoughts about things brought up in this thread:

On the original topic of this thread, Mr. McCall, your car is not tuned properly! Look at your warmup enrichment, Air Temp correction, and Fuel Map for starters.

OEMs typically take several thousand hours to generate a calibration for a single platform. Aftermarket calibrations, when done by someone who knows their stuff, can be finished in a day.

On the DFI Gen 6-FAST link, maybe someone from FAST can explain why DFI's parent company is CURRENTLY paid a royality by FAST's parent company for each and every FAST unit sold today. Not trying to slander/badmouth anyone, but that is just how it is. Draw your own conclusions.

 

[JDEstill]

bruce, I'm curious as to how you tune your system. When you reprogramed your 148 over to speed density, I presume you added a VE table? Is your car still open loop? If you want to add, say, 5% more fuel at 5400 rpm and 22 psi, how do you do it?

John

 

[bruce]

Originally posted by aDFIguy

Definitely interrupt driven. Timing is even more crititcal in aftermarket applications. How many OEM ECMs see applications on engines that spin to 12K or 15K RPMS?

That and much more.

How about a strategy that allows control of MAP via a programmable rev limiter in order to cause a turbo to spool up instantly on the starting line?

GM Racing has been using the DFI 7+ box to run their FWD drag cars since their inception. Here is how the last race went:

DFI has had TCC control in the Gen 7 box since it box came out.

32 Bit processors are very rarely used for engine controls.


Higher end systems record internally at 50-100 Hz, for short periods of time. These systems usually offer a few unused analog and digital inputs to be configured as extra datalogger channels.

Ask Microsoft, Apple, Sun, Adobe...etc the same reason, and you'll get the same answer. Proprietary Intellectual Property. OEMs make no profit specifically on their engine control systems, so they lose no real money by releasing the details of it's operation to the public--long after the fact.

Some other random thoughts about things brought up in this thread:

OEMs typically take several thousand hours to generate a calibration for a single platform. Aftermarket calibrations, when done by someone who knows their stuff, can be finished in a day.

Geeee, Answers, Thank you!.

If we're going to go off and about, the motorcycle oems do it all the time. But yes, in needing to be versitile they need to.

Much more?.
Such as?.

Rev limiter based on VSS, I guess does qualify, but I was thinking more like a time in PE enrichment. Actual over run antilag code.

Makes sense on the 32 bit.

GM uses the latest Bosch in the C5 program.
Rumot has it in part due to the knock detection strategies.

They sell software, as their *only* business.
Getting the code in the domain might just be refreshing and allow inventive folks to play. Seems tho folks like to play ego games about being better, rather then product refinement, just my opinion.

If the aftermarkets had to go to the lenghts to please the EPA, as the oems, they'd take alot longer, and I think other then in some exotic programs nothing takes thousands of hours to do.
Without going into details, they can fabricate an ecm from compnent pieces in a day. I've seen a Break Out Box used for pcm developement, that was just simply too cool.

 

[bruce]

Originally posted by JDEstill
bruce, I'm curious as to how you tune your system. When you reprogramed your 148 over to speed density, I presume you added a VE table? Is your car still open loop? If you want to add, say, 5% more fuel at 5400 rpm and 22 psi, how do you do it?

Yes,
I have a high resolution 20-100 K/Pa VE table (5 K/Pa steps).
Then 400 RPM entervals.
For boost I just use a multiplier, at 15 K/Pa setps, starting at 100 K/Pa and going to 300.

With the tuning I've done so far, I haven't seen the need to do what you mention. Adding precisely 5% in one step like that is an odd circumstance, IMO. Even with such a simple system, I can still hold the AFR very tight. Which in my opinion seems to be alot more accurate then the what some of the resolution gets to be in some of the aftermarkets. ie ~ 25 K/Pa steps, is alot courser then my 15 K/Pa steps.

Once you get to not having to run as many routines as the oems do the whole world changes, IMO. If everyone is running interupts, then that more of less levels the playing field. There is just so much computation that needs to be done for an injector pulse width or timing calc.

 

[JrTuner]

aDFIGuy said;
On the DFI Gen 6-FAST link, maybe someone from FAST can explain why DFI's parent company is CURRENTLY paid a royality by FAST's parent company for each and every FAST unit sold today. Not trying to slander/badmouth anyone, but that is just how it is. Draw your own conclusions.

Why don't you explain. You seem to know an awful lot about the subject considering your employment by Mr. Gasket/Accel. What you speak of has nothing to do with the quality of the systems or how they work. You are treading on US Patents here. It is what happens when one person developes some technology then his company is purchased by a larger company, along with said technology and any applicable patents. Then said large company feels they don't need to employ said person. Said person then comes out with another system that is light years ahead of said large companies system and takes said large company over 5 years to catch up. Actually who says they ever did.

 

[Craig Smith]

Well, now that I am back in town and have internet access again, I figured I'd throw in a piece myself.

First off, I got off track early on with respect to the original post. My response is that you should be able to tune your car to your liking with the hardware you currently have. Many of us have seen some fairly radical combos controlled by an old DFI system that work well. Newer hardware and software available today would possibly make it a bit easier for you to get things to your liking but I don't think you are beyond the capabilities of your current system.

I would have to agree with aDFIguy on most of his points. Some of them are product specific things that I'm not aware of, but I definitely agree with the technical info presented in his post.

With regard to his last bit of info, our owners do pay the DFI owners. The reason for this is because early on, DFI was able to patent nitrous oxide control within the ECU. Since our ECU also provides nitrous oxide control features, our owners reached an agreement with their owners and here we are. We don't pay a royalty because we lifted DFI's design. With two people being key players in the design and development of both systems, there's bound to be some similarities between the two. This is one of them. However, earlier implications in this thread that one is a copy of the other are untrue and not based on fact.

Another thing worth pointing out is that the FAST unit (or any other aftermarket system I am aware of) does not use a 32 bit microprocessor, as was previously posted. As spelled out here already, 8 and 16 bit processors offer plenty of horsepower for automotive applications.

As for some differences between the FAST and the old DFI system, about 10 minutes of research would answer that for almost anyone. Given that the systems are both programmable engine management systems, of course there are similarities! There are similarities between the Gen 7 and the FAST, and Motec, and EFI Technologies, and Electromotive, and anyone else! If there weren't, we wouldn't compete with each other. As far as differences, if they need to be spelled out, here's a few things to look into:

Wide band O2! (Much more than a bell or a whistle, for anyone interested)

Programmable on the fly.

An actual speed-density algorithm, not a lookup table based on RPM and manifold pressure.

Eight injector drivers.

Sequential capability and individual cylinder tuning capability without adding hardware.

Adjustable crank reference angle.

There are many design characteristics found in the FAST ECU that are not in the Gen 6 which help quite a bit with weather resistance, durability, and immunity from secondary ignition noise.

There are others, but before I get too far on this, I think it's more important to note (as others have) that it isn't important to compare them. One came out after the other; of course there will be things that are better about it! Now the Gen 7 stuff is out and it offers some improvements over anything either of us have done to date. It isn't an admission of guilt or inferiority, it's the nature of the business. Soon we'll have a new system out with many new features and improvements as well, and back and forth we go. It's to be expected, and it's what keeps everything rolling. It happens in every industry and we have that to thank for newer, better products all the time.

Bruce, why would we release source code for our product? The source code and the innovations contained within are a big part of what puts bread on our tables. We rely on our ability to market the entire package to be profitable. Showing everyone who wants to know how to do it simply isn't in our best interest.

Time for dinner. This is turning out to be a rather interesting thread to say the least.

 

[JrTuner]

Craig,
I seem to remember the original FP Performance box saying 32 bit on it. Am I correct or smoking something?

 

[JDEstill]

Originally posted by bruce
Yes,
I have a high resolution 20-100 K/Pa VE table (5 K/Pa steps).
Then 400 RPM entervals.
For boost I just use a multiplier, at 15 K/Pa setps, starting at 100 K/Pa and going to 300.

With the tuning I've done so far, I haven't seen the need to do what you mention. Adding precisely 5% in one step like that is an odd circumstance, IMO. Even with such a simple system, I can still hold the AFR very tight. Which in my opinion seems to be alot more accurate then the what some of the resolution gets to be in some of the aftermarkets. ie ~ 25 K/Pa steps, is alot courser then my 15 K/Pa steps.

Interesting...

So, you have a VE table up through atmospheric pressure. After that it sounds like you (in essence) assume a constant VE. If you just have a multiplier for MAP levels above atmospheric, then that pressure correction would be adding the same % of fuel at 3000 rpm as it would at 6000 rpm (for a given boost pressure of course).

A simplistic model, but you say it works good...

And this is all open loop, right? It's not using a wide band is it? Narrow band?

It sounds to me like your control of fueling is right around the level of a DFI gen VI then. A guy with a wide band and a DFI VI could get very good control of his AFR too, with some good testing.

On the other hand, a FAST system with wide band, that can see the AFR and actually adjust the fueling real time to hit it, that seems like a big step up to me. Filling in a VE table, and you are 5% off, in an open loop system (like the DFI VI) you end up 5% off on fuel. FAST, it sees it and adjusts it.

You seem to be asking everyone to tell you why the aftermarket ecm is better than your setup. Well, why is your setup better than the aftermarkets?

I think one of the big draws of the aftermarket ecms is ease of tunability. If you have a VE table to fill in, and % fuel per MAP steps to enter, it sounds like your setup takes a lot more effort to tune than a FAST system. If you were to say, swap cams and add some ported heads, how long would it take you to retune your car? A guy with a FAST can do it pretty quick. When TurboTR has swapped cams he's had his car pretty much nailed in an evening. Can you do the same?

Lets face it, two average car guys, one goes and buys a FAST, the other sits down to learn how to program assembly, learn how to solder up these mods to his ecm, learns all the mysteries of the inner workings of the oem program, makes the same helpful friends you apparently have made so they can get him squared away when he gets stuck, all things required to duplicate your system... I think the FAST guy is going to have his car running a year or two before the other guy. You're harping on how good your setup is. I'm sure it is awesome, I wish I had it myself. But there is no way in hell I could duplicate it with my current knowledge unless I had a few years (at least) of free time on my hands!

Reminds me of a story JayC told back when the FAST first came out. As I remember he had a program running on a 5.0 and decided to use it to fire up a GN (or was it vice versa?). All he had to do was change the # of cylinders and the injector size and fire it up. Drove it off with the laptop going to start tuning it, and had it running pretty good later that day. I'd challenge you to do something similar.

John

 

[bruce]

Originally posted by Craig Smith

Wide band O2! (Much more than a bell or a whistle, for anyone interested)

Programmable on the fly.

An actual speed-density algorithm, not a lookup table based on RPM and manifold pressure.

Eight injector drivers.

Sequential capability and individual cylinder tuning capability without adding hardware.

Adjustable crank reference angle.

There are many design characteristics found in the FAST ECU that are not in the Gen 6 which help quite a bit with weather resistance, durability, and immunity from secondary ignition noise.

Bruce, why would we release source code for our product? The source code and the innovations contained within are a big part of what puts bread on our tables. We rely on our ability to market the entire package to be profitable. Showing everyone who wants to know how to do it simply isn't in our best interest.

I dunno, while I stand 100% the WB as a tuning aid, running a engine WOT with no real redundancy for protection, just spooks me. From my understanding of things, if they fail they fail with a lean reading, and that would force the engine rich. Extremely rich, washing a cylinder with fuel at WOT, even possibly causing the sensor to read even more falsely.

While on the fly is fun, it removes, the time and motivation to take notes, and without notes you can fly right on by the best setup. I've watched too many guys get lost on the tune, with on the fly.

Speed density without a table. How does the ecm target an AFR then?. Oh wait, you said rpm and density. While there might be no VE table, it would be just a word game to have commanded AFRs, instead.

SEFI vs batch, so far from what I've done, with somewhat big injectors, batch seems to work best. Firing the injector 2x as often makes for better atomization. Individual cylinder trim, seems like for on a helmhotz or wave type N/A engine might be a good thing, but this board is about the buick and with it just being a 6, and little to no wave tuning seems like overkill.

Better packaging is always better.

Why not?.
There are just so many ways you can run the code, and there are some that might be alot more prone to buy it if they had the OPTION of being able to tweak the code as they see fit.
To some folks selling an ecm without the code, is like buying a car with the hood welded shut. There are a number of people that have gone to great lengths to correct that....

 

[Craig Smith]

Originally posted by JrTuner
Craig,
I seem to remember the original FP Performance box saying 32 bit on it. Am I correct or smoking something?

That's the software application, not the ECU.

 

[Craig Smith]

There IS a target a/f table, Bruce. The purpose of the VE table is to align the actual a/f ratio with the target a/f ratio.

If your method works well for you, I'd hate to talk you out of it. But I really don't see the average performance enthusiast or even a serious racer being able to do it. It's not like you can buy a book on writing source code in the checkout lane at the supermarket.

People don't want to make science experiments out of their cars. They want to drive them and have them run well. An aftermarket ECU makes that easy. And that is why an aftermarket setup is better than an OEM setup. Even with programming knowledge, it'll take most people an hour to make a tiny little tweak that an aftermarket system lets you make in seconds and then move on.

 

[Taffy]

I am sure that Bruce has a bitchen setup that rivals the best in aftermarket systems. The problem is that the average idiot like me can not even come close to getting to the level that he is at. I sure wish I could (hat off to Bruce), but it just ain't gonna happen for me any time soon. So off to the aftermarket systems I go. I have personally never used either. I just want to see what FAST has up their sleeves. Either way, I would have another "Bruce" set it up anyways. Later, ERIC.

 

[bruce]

Originally posted by Craig Smith
There IS a target a/f table, Bruce. The purpose of the VE table is to align the actual a/f ratio with the target a/f ratio.

If your method works well for you, I'd hate to talk you out of it. But I really don't see the average performance enthusiast or even a serious racer being able to do it. It's not like you can buy a book on writing source code in the checkout lane at the supermarket.

People don't want to make science experiments out of their cars. They want to drive them and have them run well. An aftermarket ECU makes that easy. And that is why an aftermarket setup is better than an OEM setup. Even with programming knowledge, it'll take most people an hour to make a tiny little tweak that an aftermarket system lets you make in seconds and then move on.

If it's doing a look up, it's doing a look up, if it's a 3d table then it's doing 3 math interpalations. How you want to label the axis, is immaterial.

I think you're selling a few folks real short. As time goes on you'll see more and more folks looking for answers that are more then PR statements. Look at the effort the import crowd puts into their ecms. And I dare say that is sprading.

A Hotrod is a science experiment!.
Understanding the science of it all makes the car faster more effeicent, etc,etc.

Please stop with the ancient technology stuff. The Romulator as well as other emulators have been around for a while now, and while you might have to read a page or two to find a spot, in a paper book, it's hardly that much more effort then scrolling screens with a laptop. And shortly there should be a much friendlier interface for using the Romulator.

 

[bruce]

Originally posted by Taffy
I am sure that Bruce has a bitchen setup that rivals the best in aftermarket systems. The problem is that the average idiot like me can not even come close to getting to the level that he is at. I sure wish I could (hat off to Bruce), but it just ain't gonna happen for me any time soon. So off to the aftermarket systems I go. I have personally never used either. I just want to see what FAST has up their sleeves. Either way, I would have another "Bruce" set it up anyways. Later, ERIC.

It all takes time.
No ecm is any better then the user.
Stock or aftermarket it's about getting 100% of whatever your using.

And it is about experimenting and taking notes.
Notes are the only way that you can note the trends of what really works. Unless your using a particular system 24/7 and have just that fresh in your mind 24/7, you will miss details, and details are what seperates 1st from being an also ran. Well, unless you have a perfect memory. I used to have a photographic memory, but ran out of film years ago.....

 

[808 87 WE2]

Bruce,

Much kudos for your hard work and accomplishments...not to put you down...but like someone posted earlier many of us here dont have the time or knowledge to do what you have done, and the aftermarket such as accel, fast, motec makes tunning easier and faster...which means more time for racing(street or strip)!!!...rather than staying in front of a computer, and write computer codes!!!

to you and all the others,

thanks for all the very deep technical info...all though some of us wont understand any bit of it!!!

have fun
Jeff

 

[Evil RT]

Yea ,


Seems kinda of a mute point, the gen 6 is "very"old news and I believe that for every fast system sold they still pay a royality to the dfi guys per the settlement of the lawsuit.

Comparing current product lines and based on features and benifits , the DFI Gen 7+ unit is the best out there period.



Scott

 

[John Wilde]

I don't agree that the Gen 7 is the best unit.
The MoTeC has several more andvanced features!
The Gen 7 and Fast have a much prettier GUI though!
I have a Fast in my car and it works pretty well.

 

[Evil RT]

I agree The Motec is an excellent choice .

but this thread was the labeled fast vs dfi..?



Scott