Carpokes

Hi Everyone,

I just completed a new DME design specifically for the early 944 and I am in need of a Beta tester to verify form, fit and function. If you are interested, please carefully read the description below to understand exactly what this new DME is about. To qualify as a Beta tester, you must have a an early 944 that runs and drives normally. You must also have a sufficient skillset to remove and replace the DME as well as the ability to troubleshoot any issues that may come up. The ideal candidate will already have updated their early 944 to the ML3 generation of Bosch DMEs used in 1985.5 and newer 944's quipped with 8-valve engines. In exchange for detailed feedback and pictures of the installation, you can keep this DME as compensation. However, I may need to replace the Beta test DME with the final production version. Beta testers also have special standing that includes a lifetime warranty and priority support. If you would like to be considered, please send me an email at joe@ftech9.com with some background information about yourself and your early 944.


Project Overview (F9T-DME-ML3-MK1)
It has always bothered me that I could not accept an early style 944 Bosch DME (ML2) as a core for my late model F9T ML3 compatible DMEs, and this is what I came up with. While it looks very different from my current Sport DME products, it is actually nearly identical. On the electronics side, the main difference is I designed a new flywheel sensor circuit to replace the S100 flywheel sensor chip that is not present in the early ML2 DME. For the enclosure, I am using an off-the-shelf extruded aluminum body and custom 3D printed adapter for the 35pin system connector. It is specifically designed to be a direct-fit replacement that will mount in the early 944 factory location without modification. That said, it is still a late model ML3 DME, so you must update the AFM to Bosch 0 280 202 064 or upgrade to a Lindsey Racing MAF kit. The only parts needed out of the ML2 DME are the injector driver microchip and the spring clip that retains the 35pin System connector. Everything else in this DME is brand new!


Shown above is an advanced prototype that is very near production quality. Some more work needs to be done on the 3D printed adapter, but it is perfectly functional at this point.




The above picture shows what the circuit board looks like. All electronic parts are brand new except for the injection driver microchip that must be reused from the early 944 core DME. A core ML2 DME is not required to participate in Beta testing.



Above, the side-by-side picture shows the outline of the new F9T Early 944 Sport DME (ML3 Compatible) compared to the original Bosch ML2 early 944 DME. The mounting holes near the connector are designed to be in the exact same position as the original ML2 DME. The F9T DME is also 1.3lbs (590g) lighter than the original Bosch ML2 DME.



The 3D printed adapter shown above perfectly mates the factory 35pin system connector. The spring clip retainer is the only other reused part from the original ML2 DME core.



Shown in the above picture, the F9T Early 944 ML3 DME is designed to match the outline of the early Bosch 944 ML2 DME, however it is slightly taller.



Shown above is the back of the F9T “Early” 944 Sport DME. All F9T DMEs are faithfully based on the factory Bosch ML3 design and therefore have an FQS (Fuel Quality Switch) just like the Bosch equivalent.

Very impressive! Love the 3D printed connector! Did you replicate the early, pre-S-100, DME speed/ref sensor circuit, or did you create your own design (with one of the more modern chips perhaps)?

If you don't get a good beta tester by next week, I'll include this in the weekly #newsletter. That will get thousands of eyes on it all at once.

Tom wrote: Fri Feb 09, 2024 1:32 pm Very impressive! Love the 3D printed connector! Did you replicate the early, pre-S-100, DME speed/ref sensor circuit, or did you create your own design (with one of the more modern chips perhaps)?

If you don't get a good beta tester by next week, I'll include this in the weekly #newsletter. That will get thousands of eyes on it all at once.

Hi Tom!

Thanks, I appreciate the positive feedback! It took a lot of effort to get to this point, not the least of which was getting ASA filament to print complex parts that are both strong and look good. I went through quite a few spools of filament and 3D printer upgrades before I got acceptable results.

The new flywheel sensor circuit is a completely redesigned. I did think about reverse engineering the ML2 pre-S100 circuit, but everything in that DME is late1970's era technology, it just didn't seem like a good path forwarded. Instead, I choose to use a purpose-build automotive VR sensor chip but implement all the S100 chip's logic using a separate microcontroller. I feel kind of strange using a modern 32bit 68Mhz STM32 microcontroller that is likely hundreds of times faster than the 8bit 6Mhz 8051 microcontroller that runs the whole engine computer, but that's modern technology for you. Also, the new VR sensor chip is amazingly sensitive and should be a solid upgrade over the factory S100 flywheel sensor chip.

That would be great to get some exposure in the weekly #newsletter! I am excited to get this new DME into the hands of a capable early 944 guy. I have tested it extensively in my 1985.5 944 NA, but can only make an educated guess that it will bolt properly into an early 944. Once I have confirmation that is meets all design requirements, I will button up some loose ends and go into full production.


-Joe

I got a Bambu X1C about a year ago after struggling for years with ABS. I'll never go back -- I can (and do) print ABS and ASA over and over without any lifting or issue. And it's fast. And the prints look great every time. But I digress...

Are you using the MAX9926? I played around with that chip a few years ago and was convinced it could do the job. I had previously tried the LM1815 but didn't get far with it -- but I'm a hack so maybe in the right hands it would do the trick.

Hope to see you around Carpokes more often. We don't have the reach of RL or FB yet, but we are growing fast and have a pretty seasoned group of 944 owners. What we lack in sheer numbers, we make up in quality and enthusiasm. When something makes its way to the weekly email, it gets a lot of high-quality eyeballs on it.

Tom wrote: Fri Feb 09, 2024 5:37 pm Hope to see you around Carpokes more often. We don't have the reach of RL or FB yet, but we are growing fast and have a pretty seasoned group of 944 owners.

I concur Joe has developed some amazing products for 944s and 911s. I'm sure there are owners here that could take advantage of the product lineup. He has a 944 NA MAF kit in beta testing now, and the 951 will follow (hopefully soon). Checkout the website https://www.ftech9.com/

FTECH9 wrote: Fri Feb 09, 2024 3:52 pm


The new flywheel sensor circuit is a completely redesigned. I did think about reverse engineering the ML2 pre-S100 circuit, but everything in that DME is late1970's era technology, it just didn't seem like a good path forwarded. Instead, I choose to use a purpose-build automotive VR sensor chip but implement all the S100 chip's logic using a separate microcontroller. I feel kind of strange using a modern 32bit 68Mhz STM32 microcontroller that is likely hundreds of times faster than the 8bit 6Mhz 8051 microcontroller that runs the whole engine computer, but that's modern technology for you. Also, the new VR sensor chip is amazingly sensitive and should be a solid upgrade over the factory S100 flywheel sensor chip.

I was just about to ask this! Interesting that emulating the S100 required additional circuitry beyond an LM1815 plus some glue. I'd love to hear more about that, should you be willing to share.


Have you considered replacing the Bosch 0127 chip that handles the peak-and-hold for the injectors? Hard to get away from burning power in a big ol transistor in linear mode for the "hold" bit, but I've seen some designs use pwm to decent effect.

@FTECH9

Thanks for bringing this awesome product to market. Can you build a new DME for the 968? I may be able to lend you mine if needed

Tom wrote: Fri Feb 09, 2024 5:37 pm I got a Bambu X1C about a year ago after struggling for years with ABS. I'll never go back -- I can (and do) print ABS and ASA over and over without any lifting or issue. And it's fast. And the prints look great every time. But I digress...

Are you using the MAX9926? I played around with that chip a few years ago and was convinced it could do the job. I had previously tried the LM1815 but didn't get far with it -- but I'm a hack so maybe in the right hands it would do the trick.

Hope to see you around Carpokes more often. We don't have the reach of RL or FB yet, but we are growing fast and have a pretty seasoned group of 944 owners. What we lack in sheer numbers, we make up in quality and enthusiasm. When something makes its way to the weekly email, it gets a lot of high-quality eyeballs on it.

I came to the same conclusion regarding the X1C, that is the 3D printer I am running currently. It does print most ASA geometry without issue. However, I nearly returned it because of print issues with the system connector adapter. With ASA filament, it doesn't like to transition from tall supports to lower walls. It took a lot of support emails and trial and error to tweak settings so it would print well. My next printer will likely be commercial grade with an actively heated chamber.

I am using the MAX9926 chip to translate the VR sensor analog signals to digital. It is a very flexible chip with lots of different modes and programmable references. I remember considering the LM1815, but I don't recall why I decided against it. I had an S100 flywheel sensor circuit replacement based on the MAX9926 and a PIC microcontroller more or less working many years ago on an old Rogue Tuning DME prototype. So, when I decided to peruse this project, the MAX chip it seemed like the obvious choice to get up and running quickly.

I have been looking for a reason to post on Carpokes and this project seemed perfect. Last year I was continuously overloaded with orders, which is a great problem to have. However, there was very little time for development much anything else. I had to put my CEO hat on and have a very candid conversation with my engineering self about priorities. What came out of that was a realignment to a production focused company. This was the only way to free up time for development and to be more involved in the Porsche community. I really like what you guys are building here and I want to support Carpokes in any way I can.

Given that the module supports both the newer AFM and the Lindsey AFM, from which I conclude that your software includes selectable mapping, I'm wondering why it does not support the earlier AFM as well. After all, this is your target audience for this module.

four0four wrote: Fri Feb 09, 2024 6:23 pm

FTECH9 wrote: Fri Feb 09, 2024 3:52 pm


The new flywheel sensor circuit is a completely redesigned. I did think about reverse engineering the ML2 pre-S100 circuit, but everything in that DME is late1970's era technology, it just didn't seem like a good path forwarded. Instead, I choose to use a purpose-build automotive VR sensor chip but implement all the S100 chip's logic using a separate microcontroller. I feel kind of strange using a modern 32bit 68Mhz STM32 microcontroller that is likely hundreds of times faster than the 8bit 6Mhz 8051 microcontroller that runs the whole engine computer, but that's modern technology for you. Also, the new VR sensor chip is amazingly sensitive and should be a solid upgrade over the factory S100 flywheel sensor chip.

I was just about to ask this! Interesting that emulating the S100 required additional circuitry beyond an LM1815 plus some glue. I'd love to hear more about that, should you be willing to share.


Have you considered replacing the Bosch 0127 chip that handles the peak-and-hold for the injectors? Hard to get away from burning power in a big ol transistor in linear mode for the "hold" bit, but I've seen some designs use pwm to decent effect.

The S100 chip has a number of features on the digital side such as arming the mark signal and a minimum RPM. It is possible to implement these features with digital logic, however it would likely take quite a few chips and discrete components. To save cost, complexity and board space I decided to put a microcontroller in charge of it. Software also has the advantage that if you discover an oversite or mistake, the fix is a simple code change and recompile. Whereas if this happens with a purely electronic design, it usually means a whole new circuit board,

I am working on a replacement for the Bosch 0127 (i.e. S400) fuel injection driver chip. Like my S100 flywheel sensor replacement, it will be based on a microprocessor. The factory injector driver does use a "Peak and Modulated-Hold" schema, presumably to reduce the heat generated in the "hold" phase. I could switch it over to a traditional "Peak and Hold" schema, but as you pointed out, additional heat would be generated. However, this can be mitigated with a larger heat sink. I manufacture my own heat sinks in-house, so this is really not a big deal.

-Joe