t vago's MPGuino workspace thread, continued

[FlashStopFall]

Mpguino is looking good!! I’m having a strange issue where the rpm bounces up and down when not at idle. At idle the rpm is correct, but when I’m at 2000rpm on the highway the mpguino bounces from 1100 to 1500. I’ll monitor the injector signal with an oscilloscope and see if anything looks weird there.

[t vago]

That is strange. If you don't find anything and your issue persists, I'll post an older version of my MPGuino code that I know works, and we'll go from there.

[FlashStopFall]

Originally Posted by t vago

That is strange. If you don't find anything and your issue persists, I'll post an older version of my MPGuino code that I know works, and we'll go from there.

I had put a random small resistor inline with the injector signal instead of a 50k, because 50k made the signal too weak. I replaced it with a 10k and it works perfectly now.

What did you say your vehicle was again t vago? All I had to change in the settings were my speed pulse count, I didn't touch usec/gallon but so far the mpguino seems spot on!

[t vago]

Originally Posted by FlashStopFall

I had put a random small resistor inline with the injector signal instead of a 50k, because 50k made the signal too weak. I replaced it with a 10k and it works perfectly now.

That's great to hear!

Originally Posted by FlashStopFall

What did you say your vehicle was again t vago? All I had to change in the settings were my speed pulse count, I didn't touch usec/gallon but so far the mpguino seems spot on!

I had a 2005 Dodge Magnum SXT with the 3.5L V6. It's very possible it shared the same fuel injector part number with your Jeep, at least from the factory. Not surprised about the VSS count.

Keep us posted about your MPGuino! As always, it is a work in progress.

Speaking of which - I have begun to split apart the file into smaller chunks, stored in the mpguino_tav directory of my repository. It is available here -> https://github.com/tvago1972/MPGuino

[FlashStopFall]

Originally Posted by t vago

That's great to hear!



I had a 2005 Dodge Magnum SXT with the 3.5L V6. It's very possible it shared the same fuel injector part number with your Jeep, at least from the factory. Not surprised about the VSS count.

Keep us posted about your MPGuino! As always, it is a work in progress.

Speaking of which - I have begun to split apart the file into smaller chunks, stored in the mpguino_tav directory of my repository. It is available here -> https://github.com/tvago1972/MPGuino

That’s sweet!! Thanks for putting the code on GitHub, that’s way easier.

Do you happen to know off the top of your head what your v6 used g/h at idle? I’ve got the 4.0 inline 6, relatively similar, but I use 0.8+ g/h all the time at idle. I think my first ecomods to the jeep besides the mpguino will be lowering that number.

[t vago]

Originally Posted by FlashStopFall

That’s sweet!! Thanks for putting the code on GitHub, that’s way easier.

Yep, that's why I moved the code there. Imagine taking a 4 year long break from working on a program with 10K or so lines of code, and then trying to dive back in! Ugh...

Originally Posted by FlashStopFall

Do you happen to know off the top of your head what your v6 used g/h at idle? I’ve got the 4.0 inline 6, relatively similar, but I use 0.8+ g/h all the time at idle. I think my first ecomods to the jeep besides the mpguino will be lowering that number.

My Magnum originally had a 3.5L V6, and it had an idle consumption of around .6 to .7 gph, IIRC. I later on swapped in a stick shift to replace the automatic tranny that decided to grenade itself, and then later on swapped in a 4.0L V6. When I sold the Magnum, it was getting idle figures of 0.55 to 0.6 gph. I think that was mainly due to my ditching the automatic.

[FlashStopFall]

Originally Posted by t vago

Yep, that's why I moved the code there. Imagine taking a 4 year long break from working on a program with 10K or so lines of code, and then trying to dive back in! Ugh...



My Magnum originally had a 3.5L V6, and it had an idle consumption of around .6 to .7 gph, IIRC. I later on swapped in a stick shift to replace the automatic tranny that decided to grenade itself, and then later on swapped in a 4.0L V6. When I sold the Magnum, it was getting idle figures of 0.55 to 0.6 gph. I think that was mainly due to my ditching the automatic.

Gotcha. I’m going to be starting a new project this summer, a ZJ, and I’m planning to manual swap that one. On my current Jeep the alternator squeaks a bit, I believe it’s causing some drag, I think it will be interesting to see the g/h with the serpentine belt removed.


On another note, here’s my mpguino setup! I would like to design my own pcb to put the whole thing together compactly, but I’m just learning how to do that.

[t vago]

Originally Posted by FlashStopFall

Gotcha. I’m going to be starting a new project this summer, a ZJ, and I’m planning to manual swap that one. On my current Jeep the alternator squeaks a bit, I believe it’s causing some drag, I think it will be interesting to see the g/h with the serpentine belt removed.

Say... that reminds me...

I had successfully converted both my Dodge Dakota and my Dodge Durango to use electrohydraulic power steering pumps out of a Toyota, with resultant rises in fuel economy. I was able to get something on the order of 19 MPG combined with the Dakota and 17 MPG with the Durango.

I had planned to do that conversion with the Magnum, but fate intervened.

You might want to look into doing an EHPS conversion on your Jeep. It will definitely bring down your idle fuel consumption rate.

Originally Posted by FlashStopFall

On another note, here’s my mpguino setup! I would like to design my own pcb to put the whole thing together compactly, but I’m just learning how to do that.

Sweet! That's old school right there!

[t vago]

It's been a little while since I last updated this thread, so...

• Speeded up SWEET64 math. Old mult64 would take around 1 ms to complete, and old div64 would take up to 10 ms to complete. New routines both take about 200 usec to complete. This is about an 80% improvement for mult64, and a 98% improvement for div64.
• 32-bit isqrt, used in Chrysler fuel correction, is also fixed to be faster and more accurate. Old isqrt took about 1 ms to complete, and had an... interesting interpretation of various input numbers at times (square roots would be up to 5% off from what they were supposed to be). New 32-bit isqrt now takes about 70 usec to complete for a 92% speed improvement, and actually returns accurate square roots for all 32-bit cases.
• Made general decimal number output auto-ranging. Previous output was limited to output display values up to 9999999.999, and decimal rounding was inconsistently applied to the output. New output function allows for output values 1000x larger (of up to 9999999999), can be constrained to any window size between 2 and 10 characters, and properly handles decimal rounding.
  
  Example debug monitor "display decimal output" command for a window size of 10 with 3 decimal places:
  

  ]a<3.u
  00: 00 0A 03
      0000000000000005 -      0.005 -
  01: 0000000000000037 -      0.055 -
  02: 000000000000022B -      0.555 -
  03: 00000000000015B3 -      5.555 -
  04: 000000000000D903 -     55.555 -
  05: 0000000000087A23 -    555.555 -
  06: 000000000054C563 -   5555.555 -
  07: 00000000034FB5E3 -  55555.555 -
  08: 00000000211D1AE3 - 555555.555 -
  09: 000000014B230CE3 - 5555555.56 -
  0A: 0000000CEF5E80E3 - 55555555.6 -
  0B: 0000008159B108E3 -  555555556 -
  0C: 0000050D80EA58E3 - 5555555556 -
  0D: 00003287092778E3 - ---------- -
  
  ]

• Added a graphical status bar which can display between 0-100%, or indicate overflow. Currently being used for INST FE vs CURR FE, and INST FE vs TANK FE. For these two pages
  • If INST FE is equal to CURR/TANK FE, then the status bar will display 2/3s full
  • If INST FE is less than CURR/TANK FE, the status bar will be less than 2/3s full
  • If INST FE is greater than CURR/TANK FE, the status bar will be more than 2/3s full
  • If INST FE is at an overflow condition (EOC for MPG or KPL, IDLE for 100km/L or 100mi/G), then the status bar will show that, too.
• Refined debug command-line terminal to be able to directly edit stored parameters, major in-program variables, and trip measurements. Also measures execution time of SWEET64 programs and mult64/div64 instructions, and measures all interrupt execution times. Also can act as a simple 64-bit calculator from the command line, to directly test output of all 64-bit SWEET64 math routines.
• Continuing to slowly push original source code file into numerous smaller sub-files for ease of readability, while fixing the inevitable bugs that show up from splitting it up.
• Almost all features have been completely refactored. Only features left to refactor are drag race, and coastdown testing (which was in work anyway when I paused development several years ago)

Currently working to get my version to support the MPGuino Colour Touch. I powered up the unit I purchased from abbalooga to verify operation, and will modify it to incorporate an ICSP header for programming purposes. Also studying the datasheets associated with the TFT and touchscreen controllers being used.

The code is currently being stored at https://github.com/tvago1972/MPGuino

[t vago]

I have added a icsp programming header to my Colour Touch unit, and have successfully programmed it with both my code (which of course doesn't yet support the Colour Touch), and the original v25 abbalooga code. I also verified BT operation with the MPGuino Blue Android app.

Now the fun begins!