Crystal Mod A9L (long story) - Ford Mustang Forums : Corral.net Mustang Forum
 1Likes
  • 1 Post By FOXRODDER
 
LinkBack Thread Tools Display Modes
post #1 of 20 Old 08-22-2014, 11:56 PM Thread Starter
Registered User
 
TomR's Avatar
 
Trader Feedback: (2)
Join Date: Aug 2009
Location: Macon, GA
Posts: 4,960
Garage
Crystal Mod A9L (long story)

About a year ago I started building a test fixture to run the Ford EEC on my test bench. After severely fouling some plugs with a new non-Ford system, and having problems getting it working, I decided to finish my test fixture enough to run my EEC.

Basically what it does is modify the output of a signal generator to look like a Ford distributor PIP signal, plus it has a bunch of voltage sources from pots to simulate the MAF and other things. Some of the controls are electrically tied together so I can simulate a throttle opening with increasing MAF in different proportions.

I had already bought a broken A9L from Craigslist, and decided to repair it tonight. So I fixed some blown traces, a bad resistor, and a bad transistor. That A9L now works perfectly in my car.

So I decided to see what happens. With the stock A9L and Tweecer tune I use, everything went nuts at about 6800 before reaching the fuel cutoff that was set at 7200. The timing went to -2000 degrees or some other crazy number, and the injectors closed. Reducing RPM to about 6600 or so would bring the system back on line, and it would flake out at 6800 or so. This was simulating my engine's conditions at that RPM from a log. If I wasn't a dummy with software I could probably figure a way to play a log back into the test EEC, but I can do it manually OK.

So...knowing how things work electrically I decided to use a 125% over sample crystal. This would, in theory, make all count functions or time based functions scale by the reciprocal of 1.25, or 80%. 1.25*15 MHz was a 18.750 crystal, and I had a 18.68. It actually measures, through some fate, 18.72 when running in the A9L. This gives a scaling of 1/1.248 times or 80.128%.

When I set the generator so my Tweecer read 6800 on the dashboard, my "ignition" was at 8500 RPM. Perfect. 8500 * .8 was 6800 RPM

Injector PW had no significant error increase.

Timing showed no significant error.

Now the A9L, with the 18.75 MHz crystal, nuts up at about 600 Hz from the generator. This is 600*15 = 9000 RPM. 9000 RPM shows as 7200 on the Tweecer logging program, because the A9L thinks it is 7200. The injectors drop 50% of injector periods at the half fuel setting, and then go into full cutoff at the full fuel setting.

This would not make the engine lean, it would simply drop half the cylinders before quickly dropping all of them. It would show lean on a wideband, but that would be because clean air without fuel would pump through and "fool" the wideband.

So, it looks like a clock running around 18.75 MHz pushes the A9L up to where it is a 9K RPM capable system. Somewhere above 9K RPM everything goes nuts. Things that are off relate to RPM and count functions. In my car with the idle set at 1100 RPM, the actual idle is around 1400. Because I stretched its rubber watch, the computer thinks an engine at 1400 is .8*1400 = 1120 RPM. So it sets it for 1100, but the engine is really at 1400.

This is why people are successful clocking the A9L a bit higher. I have crystals (I do hardware design) so I'll eventually look for the minimum overclock that allows me to go 7600 or so (where my cam runs out) that is an even multiplier for easy scaling. I'm going to try a 16.750 which get me close to 90% or 1.111 times. 7000 in the A9L's brain would be about 7700 in the real world, although the clock now looks OK (but I'll never use 9K engine RPM).



To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
89 LX. 363, single turbo, Super Vic EFI, TFS high port heads by TEA, solid roller, glide. Holley HP EFI. (exact combo varies)
TomR is offline  
Sponsored Links
Advertisement
 
post #2 of 20 Old 08-23-2014, 12:43 AM
Registered User
 
RBELL's Avatar
 
Trader Feedback: (21)
Join Date: Dec 2004
Location: Marion, TX
Posts: 2,252
I would leave the 18.72 MHZ crystal on the board. I did and the car picked up about 11 whp everywhere below the point where the computer shut off. The faster operating speed was good for some hp. Then it was not hard to tune with clock speed being off. Just know what the actual rpm and the indicated rpm, then it is not hard to tune. I did it for a couple of years. I was shocked and very happy the first time I took the car out for a drive and it went straight to 8000 rpm, very quickly.
I would suggest you cut a hole or two in the case of the computer and install a fan for a lap top, possibly more to let the air flow through across the board and heat sinks. It will run hotter. I am going to do another one in the near future.


If I buy a part or parts from you do not send me those parts through USPS, I would actualy like to get the parts I purchase.
RBELL is offline  
post #3 of 20 Old 08-23-2014, 02:26 AM Thread Starter
Registered User
 
TomR's Avatar
 
Trader Feedback: (2)
Join Date: Aug 2009
Location: Macon, GA
Posts: 4,960
Garage
Quote:
Originally Posted by RBELL View Post
I would leave the 18.72 MHZ crystal on the board. I did and the car picked up about 11 whp everywhere below the point where the computer shut off. The faster operating speed was good for some hp. Then it was not hard to tune with clock speed being off. Just know what the actual rpm and the indicated rpm, then it is not hard to tune. I did it for a couple of years. I was shocked and very happy the first time I took the car out for a drive and it went straight to 8000 rpm, very quickly.
I would suggest you cut a hole or two in the case of the computer and install a fan for a lap top, possibly more to let the air flow through across the board and heat sinks. It will run hotter. I am going to do another one in the near future.

I measured processor temperature and it is well within limits of typical processors. I am going to make a mounting bracket to get the EEC up out of the kick panel to make swapping units easier, and that will be a much cooler location for the case.

EDIT: added info

Temperature measurements show no measurable increase in processor temperature when clocked at 18.72 MHz. I can't measure the difference between 15 MHz or 18.72 MHz clocking in a half hour run time with a temperature probe on the processor's case. The processor essentially runs a few degrees over case internal temperature, and is far below any thermal limits.

Injector transistors and other parts make up the vast majority of case heat.

Other than the RPM going off, and other time based functions changing, nothing seems to change.

I will do a comparison between my unmodified A3M1 and the A9L and post up actual comparison numbers, but things looked OK inside after running at 7000 RPM with fake injectors (10 ohm 10 watt resistors) for a half hour. I run a lot of semiconductors much hotter than it ran.

Other than the RPM errors, which are predictable when the clock is properly chosen, I'm stoked about this. This is all based on test bench and idle tests.

I'm going to install an absolute timing monitor when I have time to get one built. You don't know of anyone who already makes something like that, do you?

I should have done this weeks or months ago. Now I can run EEC's on my bench instead of stressing my engine and car while testing.


To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
89 LX. 363, single turbo, Super Vic EFI, TFS high port heads by TEA, solid roller, glide. Holley HP EFI. (exact combo varies)

Last edited by TomR; 08-23-2014 at 05:56 PM. Reason: added temp info
TomR is offline  
Sponsored Links
Advertisement
 
post #4 of 20 Old 08-24-2014, 10:39 AM
Registered User
 
FOXRODDER's Avatar
 
Trader Feedback: (30)
Join Date: Aug 2004
Location: Orlando
Posts: 1,540
Pretty cool stuff. I am still trying to figure out why nobody is making a modern A9L unit with new components. Is it really that crazy of an idea?
BDASPNY likes this.

87 FoxRod, 00 Towbra, 06 Tree-Fiddy Much Duty
FOXRODDER is offline  
post #5 of 20 Old 08-24-2014, 11:04 AM
Registered User
 
RBELL's Avatar
 
Trader Feedback: (21)
Join Date: Dec 2004
Location: Marion, TX
Posts: 2,252
The two computers I modified definitely ran hotter. I also talked to some engineers, after I did so and found the computers ran hotter. Both are electronic engineers, that design computers or components, they both said the computers should run hotter when upping the clock speed. I have had some people who were also into electronics and computers, that told me it was not a good idea to change the crystal, because it would make the computer run so much hotter and be very likely to fail. The fact that the car was mainly a race car, for drag racing, not any form of endurance racing, I did not worry about it much.
The reason I did two computers, the first died quickly, I suspected it had problems to begin with. The second ran hot but never died. I originally had the computer under the passenger's side seat, after figuring out how hot it was running, I made a bracket to mount it behind the dash. With no evap or header case, there is a lot of room around the computer, so I never had any problems with the second one.
Now I have a EEC V computer in the car that came from the factory with a 27 MHZ crystal, I am going to turn off all of the OBD II diagnostic functions, all O2 sensors, canister purge, EGR, fuel pump driver module, closed loop, torque control, PATS, trans controls, speed sensors. The only things that will be working are the TP, crank sensor, cam sensor, ACT, ECT, MAF, injectors, COPS and IAC, open loop only.

I very likely will modify one or two more A9Ls or similar computers with a 18 MHZ crystal. There are two cars, that I need to finish, that have engines that will easily rev past 7000 rpm, both are turbo cars. If I do so, I am going to try to install a fan in each computer. Both of those cars, still have the computers in the stock location and still have evap/heater cases in the cars. I can make some more room around the computer by removing the sound deadener and the computer bracket. I think a fan would be a good idea.

If I buy a part or parts from you do not send me those parts through USPS, I would actualy like to get the parts I purchase.
RBELL is offline  
post #6 of 20 Old 08-24-2014, 03:01 PM
Registered User
 
Blown88GT's Avatar
 
Trader Feedback: (2)
Join Date: Nov 1999
Location: Palm Beach Gardens, FL
Posts: 3,554
Quote:
Originally Posted by TomR View Post
...

Basically what it does is modify the output of a signal generator to look like a Ford distributor PIP signal, plus it has a bunch of voltage sources from pots to simulate the MAF and other things. Some of the controls are electrically tied together so I can simulate a throttle opening with increasing MAF in different proportions...
I have crystals (I do hardware design)...
....
I did this for Pratt & Whitney, they called them Verifiers, different ones for different engines, F100, F117, F135.

Megasquirt has them, they call them Stims.


Quote:
Originally Posted by RBELL View Post
...I also talked to some engineers, after I did so and found the computers ran hotter. Both are electronic engineers, that design computers or components, they both said the computers should run hotter when upping the clock speed. I have had some people who were also into electronics and computers, that told me it was not a good idea to change the crystal, because it would make the computer run so much hotter and be very likely to fail.....
Tom is an engineer & so am I. I designed many different computers.
Technically, upping the clock speed on any cpu will make it run hotter, how much hotter is the $64,000 question. A 20% increase is generally not a problem. Remember, these cpu's were designed to run in a very hot environment, well above 90degF ambient.

If you want a good mounting location that requires no modification to the factory harness, see where I mounted my Megasquirt. Under the glove box, there's a steel channel that goes all the way across the cockpit.

Attached Images
File Type: jpg IMG_1741.jpg (78.7 KB, 364 views)
File Type: jpg IMG_1601.jpg (118.9 KB, 363 views)

1988 GT, 57k mi, Orig Owner
ProCharger P600B, FRPP Headers, Flowmaster F2, 3G Alt, Mk8 Fan/DCC, 3.55's, Prog Springs, ST Brace, G/Trac Bar, Subfrms, UCA, LCA, FCA, Tokico 5-ways, C/C Plates, CSA 16x8, Bridgestone RE760 Sport 225/50R16, Megasquirt DIYPNPF60, LC2 WBO2, C&L76mm MAF w/LU47's
Blown88GT is offline  
post #7 of 20 Old 08-24-2014, 06:43 PM
Registered User
 
RBELL's Avatar
 
Trader Feedback: (21)
Join Date: Dec 2004
Location: Marion, TX
Posts: 2,252
Quote:
Originally Posted by Blown88GT View Post
I did this for Pratt & Whitney, they called them Verifiers, different ones for different engines, F100, F117, F135.

Megasquirt has them, they call them Stims.



Tom is an engineer & so am I. I designed many different computers.
Technically, upping the clock speed on any cpu will make it run hotter, how much hotter is the $64,000 question. A 20% increase is generally not a problem. Remember, these cpu's were designed to run in a very hot environment, well above 90degF ambient.

If you want a good mounting location that requires no modification to the factory harness, see where I mounted my Megasquirt. Under the glove box, there's a steel channel that goes all the way across the cockpit.

I made the old harness that was in my car when I had EEC IV and the new harness now that I have gone to EEC V. I originally was trying to mount everything in the car as far to the rear as possible, to improve the front to rear weight ratio as much as possible. After the first A9L died, I pulled the computer from under the front seat and mounted it behind the dash. There is nothing behind the dash on my car, just a big open space. So I was able to mount the computer there, there is a lot of open space around the computer, nothing that gets hot, but then no insulation on the fire wall or any air flow behind the dash. That is where I previously mounted the old EEC IV computer, the new EEC V computer is mounted in the glove box. So I don't have the extra weight of a mounting bracket, that and I can open the glove box and get to the ECM quickly.
The newer EEC V computers with the 27 MHZ crystals, I assume have more heat sink than the older EEC IV computers with the 15 MHZ crystals, that or the components used, don't produce as much heat. Because I have not noticed the EEC V computers running hot.

If I buy a part or parts from you do not send me those parts through USPS, I would actualy like to get the parts I purchase.
RBELL is offline  
post #8 of 20 Old 08-24-2014, 06:59 PM Thread Starter
Registered User
 
TomR's Avatar
 
Trader Feedback: (2)
Join Date: Aug 2009
Location: Macon, GA
Posts: 4,960
Garage
I directly measured the processor with a thermal probe pressed against the case and grease over the probe. Difference with clock increase to 18 MHz is absolutely insignificant. I'm positive of that. This is not a GHz processor driving many dozens of lines with capacitance with a square wave. Is is a snail, so the power wasted to drive or sink external lines a little faster is not a concern.

Also, most of the case heat (and that heat gets conducted to the CPU and everything in the box) is from transistors that switch high current loads. The processor is mostly heated by external devices, and it just does not get that hot. I was a little surprised how little beta the Fomoco transistors have and how light the base current is. They would be cooler if saturated when switched, but I'm also using 10 ohm load resistors to simulate injectors.

I expect it would be a whole lot worse when buried under the kick panel when stuck in Atlanta traffic. But clock speed is not an issue at all for heat.

88. Your car setup is virtually identical to mine. If that was an IBM ThinkPad instead of farmer in the Dell, it could be my interior.


To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
89 LX. 363, single turbo, Super Vic EFI, TFS high port heads by TEA, solid roller, glide. Holley HP EFI. (exact combo varies)
TomR is offline  
post #9 of 20 Old 08-24-2014, 07:39 PM
Registered User
 
Blown88GT's Avatar
 
Trader Feedback: (2)
Join Date: Nov 1999
Location: Palm Beach Gardens, FL
Posts: 3,554
Quote:
Originally Posted by TomR View Post
...88. Your car setup is virtually identical to mine. If that was an IBM ThinkPad instead of farmer in the Dell, it could be my interior.
I'll bet you don't have the GT seats of the era, i.e. inflatable lumbar support, side & front bolster adjustments.

I found the Dell at the curbside trash, missing the HDD, battery & power brick. It now has an SSD, larger capacity battery, etc. It took a swan dive off my friend's car when it was being connected to his OBDII Mercury, & the wind gust picked it up off the hood. It was on at the time & it survived with a couple gouges repaired with JB Weld. My next door neighbor has the exact same one & I had a Macrium WinVista32 image of it. Do you like my laptop stand & mouse pad?

1988 GT, 57k mi, Orig Owner
ProCharger P600B, FRPP Headers, Flowmaster F2, 3G Alt, Mk8 Fan/DCC, 3.55's, Prog Springs, ST Brace, G/Trac Bar, Subfrms, UCA, LCA, FCA, Tokico 5-ways, C/C Plates, CSA 16x8, Bridgestone RE760 Sport 225/50R16, Megasquirt DIYPNPF60, LC2 WBO2, C&L76mm MAF w/LU47's
Blown88GT is offline  
post #10 of 20 Old 08-24-2014, 09:26 PM
Dean Scheidt
 
SLO 5-OH's Avatar
 
Trader Feedback: (53)
Join Date: Sep 2006
Location: Sandston, VA
Posts: 3,374
I have my 18mhz crystal modded A9P in the kick panel. Because there was potential for additional heat (and because it only cost $7.50), I installed a computer fan on my case. The fan is a Delta model AFB0512MB.

I drilled several 7/16" holes and there is the large opening at the top where the chip is installed. The fan pulls quite a bit of air out of the case. I accidentally mounted the fan in a great spot. With everything installed back to normal, the fan blows the air over the plastic kick panel cover and into the passenger foot well area so there are no obstructions to air flow.








1989 Mustang GT (Sold)
1/8-mile: 6.57 @ 102.67
1/4-mile: 10.40 @ 128.13 (1.37 60')
425 RWHP:
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
, TW 205 heads, tmoss TFS Box-R intake,
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
AOD,
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
8" converter, 4.10's, Team Z Suspension, M/T 26" ET's, full interior, 3,360 lbs race weight. Tuned by Willie at
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
SLO 5-OH is offline  
post #11 of 20 Old 08-25-2014, 08:32 AM Thread Starter
Registered User
 
TomR's Avatar
 
Trader Feedback: (2)
Join Date: Aug 2009
Location: Macon, GA
Posts: 4,960
Garage
Any worry about the EEC running hotter is completely unfounded. Clocking up 20% makes no discernible change in heat, and heat is already **far** below safe limits. The bulk of heat change with speed in a digital semiconductor comes from transition time when a device is switching states. The switching time is extended in PCs because they drive long bus lines with multiple banks of memory and I/O devices with lots of stray and parasitic capacitance, and drive them at very high speeds. Something has to absorb the energy stored in those lines, and it is the device driving the lines.The CPU has to do a lot of work to move lines up and down. It is a struggle to switch anything with a 2 GHz clock, because the smallest amount of capacitance just from copper traces looks like a big heavy load.

The Ford EEC is totally different. It is a lightweight. It clocks at turtle speeds and clocks very lightly loaded I/O ports with short bus lengths. The fan won't hurt anything, but it won't help anything either. That is just a cold hard fact of life.


To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
89 LX. 363, single turbo, Super Vic EFI, TFS high port heads by TEA, solid roller, glide. Holley HP EFI. (exact combo varies)
TomR is offline  
post #12 of 20 Old 08-25-2014, 10:09 AM
Dean Scheidt
 
SLO 5-OH's Avatar
 
Trader Feedback: (53)
Join Date: Sep 2006
Location: Sandston, VA
Posts: 3,374
Thank you very much for sharing your information. I just entered the crystal mod world a few weeks ago. I wanted my cpu to have the ability to rev cleanly to at least 7,200 so I know I'm good to go with my 18 MHz crystal.

From reading various posts on corral, I thought it was a "race car" only mod since most people were describing idle and part throttle issues. My car, tuned by dirtydirtyracing, is driving just like the stock computer so I couldn't be happier that I tried it.

Thanks again,
Dean

1989 Mustang GT (Sold)
1/8-mile: 6.57 @ 102.67
1/4-mile: 10.40 @ 128.13 (1.37 60')
425 RWHP:
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
, TW 205 heads, tmoss TFS Box-R intake,
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
AOD,
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
8" converter, 4.10's, Team Z Suspension, M/T 26" ET's, full interior, 3,360 lbs race weight. Tuned by Willie at
To view links or images in signatures your post count must be 10 or greater. You currently have 0 posts.
SLO 5-OH is offline  
post #13 of 20 Old 08-25-2014, 07:08 PM
Indo-Canuck-Yankee
 
Shaker666's Avatar
 
Trader Feedback: (4)
Join Date: Sep 2002
Location: Tucson, AZ
Posts: 2,736
Garage
I've got to say however, I do like the aesthetics of the cooling fan on the EEC enclosure!

What do you guys think of my EDIS8 mounting bracket with integral (and heatsinked) EDIS8 module? I know it's overkill to cool the EDIS module, but whatevs... that AMD Athlon heatsink has been sitting around for far too long!