|
To me it sounds exactly like the Gauge that sits at full for 1/3 of a tank, and then drops like a rock. The ECU uses the other table to know exactly how many gallons are left in the tank. I kind of wondered how they did that, it would be neat to see what the factory tune had in those tables and if they were the same also it might be that the gauge isn't linear so the idea is to try to make it so the gallons line up with the line on the dash.
|
#
?
Sep 8, 2023 19:58
|
|
|
|
| # ? Apr 28, 2024 06:17 |
|
|
I'm pissing in the wind, in the dark here, but my impression is the fifth table is what you're suggesting for linearization of the gauge. That leaves the mystery purpose of 2 of the remaining 4 tables, which I intend to learn about shortly.Commodore_64 posted:It doesn't have something like a temperature sensor on the fuel rail used to interpolate between the two tables for temp? Is the second table some sort of second order coefficient for reasons? Captain Cool posted:What do the values look like on the two tables? Does it look like theoretical + correction or fuel tank part number A vs B? If the values are similar, what do they look like if you plot them?
|
|
#
?
Sep 9, 2023 00:00
|
|
|
Alright, here's the factory calibration. Main fuel system tab in hptuners:  My interpretation: Capacity is the actual capacity of the overall fuel system, including both tanks (if used.) Capacity gauge is the gauge displayed capacity of the overall fuel system, including both tanks (if used.) Primary Capacity is the actual capacity of the primary tank. Primary Capacity Gauge is the gauge displayed capacity of the primary tank. Not sure why there is no Secondary Capacity/Secondary Capacity Gauge, maybe it'll show up on a system with a second tank populated. Guess I'll find out. Sender Max A/D is the value above which the ECU is supposed to trip P0463 (I think. Untested.) Sender Min A/D is the value below which the ECU is supposed to trip P0462 (I think. Untested.) Gauge Output is a button that opens the gauge output PWM correction factor table. Primary Volume opens this table. I'll only show it as a graph because it's wider than Siberia on foot in the winter in table form and I'm way too lazy to patch screenshots together right now. Basically the sender ADC value indexes into this 128-entry table and out pops a number of gallons. This table contains values ranging from 0.00 to 31.44 with a substantial dead band at the top and bottom (entries 128 and above are all 31.44, entries 36 and below are all 0.00.)  Primary Volume Gauge opens this table. Sender ADC value indexes into it as well and out pops a (different) number of gallons. This table contains values ranging from 0.00 to 31.00 with all entries 36 and below containing 0.00 and all 128 and above containing 31.00.  Secondary Volume and Secondary Volume Gauge are all 0s in the tables since the donor vehicle had only the primary tank, I think secondary tanks are only an option on truck/SUV platform (gmt800) rather than vans (gmt600.) Gauge Output opens this table:  And the same data graphed:  My interpretation is: - notice that the gauge table contains less volume than the other one. Maybe it's only indicating usable fuel, and there's 0.44 gallons left in the tank once the pickup starts sucking air? It's also significantly rougher curve than the non-gauge table. I'm not sure what the implication of that is. - all 4 of the linearization tables have columns numbered from 0 to 256, even-only. I'm interpreting that as they know the LSB of the ADC is noisy as hell (it would represent approximately 20mV, that's to be expected without extensive RFI filtering) and I'm not sure why they bothered with cell 256 since only cells 0 through 254 should be reachable. The ADC value(s) definitely index into these tables and give you a gallon reading output, but I'm still not sure why exactly there are two per tank. I'm debating setting one table to totally nonsense data and the other to real world values and just seeing which outputs and parts of the ECUs behavior start acting crazy and which ones work, as that would very clearly tell me which is what. - once the data is linearized, it is reported via OBD2 (I'm monitoring mode $22 PID $12C5 using Torque Pro as my fuel gauge right now, though eventually I'll be building a custom dash adapter that will use SAE J2178 status ID $83 secondary ID $12 / PRN 6005 instead) over J1850 VPW. It is also used as an index into the Gauge Output table, which converts a percentage into a PWM percentage to drive the ground side switched PWM output on C2 pin 38, which on SOME GM vehicles - mostly early enough that they were still using the older instrument panel designed to work directly with a resistive sender or via PWM from the TBI/spider injected ECUs - is wired to the fuel gauge. I'm mostly curious about the differences between the regular and "gauge" linearization tables going from raw ADC value to fuel quantity. I don't understand why there need to be two tables for each sender.
|
|
#
?
Sep 9, 2023 02:46
|
|
|
After looking through what I would describe as a buttfuckton of tunes from the hptuners tune repository, it looks like all 02- operating systems used a single table per tank while all 03+ operating systems use two tables per tank. Can anyone think of anything fuel and emissions related that changed from 02 to 03 that would be affected by this? Curiously, some of these tanks have smaller capacity listed for the gauge table, and some have larger capacity listed for it. I am not sure why. Some show enable and disable fields for the gauge pwm mode as well as a center frequency field for it, some don't, all seem to have the duty cycle table though. The 01 suburban tune I found with dual tanks is very interesting - the second tank calibration table isn't even monotonically increasing. There's a jump in it, it rises to 13.74g at ADC value 136, then drops to 11.89g for 4 cells, then to 14.53g. I suspect this was just never noticed because all of those cells are above the point 250 ohms will get you to, which is ADC value 128, roughly. That lines up well with the fact that it has the primary tank capacity maxing out at 26.42, total system capacity at 38.49 (12.07 gallons above primary tank), and secondary tank table entry 128 is... 12.15 gallons. E: alright, just found a 03 suburban 8.1 tune (OS 12579405, fuel segment 12577990, cvn 86c7) that has dual tanks and claims to be a factory tune, totally unmodified. The gauge and main tables for each tank are identical and the capacity and capacity gauge fields are identical as well. The lower dead band on each table goes up to cell 36; the upper dead band starts at exactly 128. The max values of the primary+secondary tanks tables add up to the exact value of the overall system capacity cell. I think the intern handling fuel system calibration at GM finally learned to get everything correct in 2003 and I'm just going to set all my tables like this and forget about it because it makes sense to me this way. kastein fucked around with this message at 06:16 on Sep 9, 2023 |
|
#
?
Sep 9, 2023 06:00
|
|
|
kastein posted:After looking through what I would describe as a buttfuckton of tunes from the hptuners tune repository, it looks like all 02- operating systems used a single table per tank while all 03+ operating systems use two tables per tank. Can anyone think of anything fuel and emissions related that changed from 02 to 03 that would be affected by this? "We need to finalize the firmware soon. We're going into production with fuel tank A, but we might switch to fuel tank B in a few months. So put both tables in the firmware and then if we switch we can send a message on the production line telling the firmware to use tank B's table." And then the issue never came up again and the firmware team started putting the same values in both tables, just in case that message gets sent accidentally. It's easier to change data than code. The "gauge" label probably came from hptuners' reverse engineering when they noticed one table was a slightly smaller capacity and assumed it was a pessimistic value for the dashboard. But the difference is 0.44 gallons at most which doesn't seem like a reasonable reserve.
|
|
#
?
Sep 9, 2023 17:26
|
|
|
I don't think it's a production line thing - by that stage of the assembly you really shouldn't be futzing with configuration, and the config is stored in the same EEPROM as the firmware on these beasts, much to my annoyance. I could be wrong but with how just-in-time inventory and production floor systems work since the early 90s it's been very unlikely something like that would happen, at least without signs of it in the GM parts fiche as to which tanks were available for what models and when. However, I certainly do not know for sure. Now that I have seen at least one allegedly factory ECU with all the main+gauge tables equal I'm just gonna roll with that unless it causes issues though. Today I was busy with other stuff but I managed to find some time and installed my aftermarket Dorman universal fuel sending unit (#55818). It's 240-33 ohms instead of 250-40 but since I'm doing the calibration, that's fine. I guess next time I fill it up I'll go 2 ADC counts at a time recording the amount of fuel required and put it all in my tune. Temporarily wired it to the ECU input and I'm seeing variation in Torque Pro on my custom PID. I also cleaned all the nastiness off the fuel line fittings on my junkyard e250 external fuel pump and hooked it up to the lines on the fuel tanks. None of this is permanently mounted yet... at some point I need to just empty the bed out and sit down and make a bunch of custom brackets to mount the evap canister, fuel filter, aux fuel pump, evap purge solenoid, and all the stuff I'm forgetting and then run all the lines nicely because it's just all piled up in the front of the bed and it bugs the poo poo out of me.
|
|
#
?
Sep 10, 2023 05:53
|
|
|
Well. I have received news from the owner of the piss Forester. He is stripping it for parts after it sat in his yard for a few years and found... ... a load bearing rear seat back! I'm glad I never dug in deeper after hearing a weird pop from the back seat area on a pothole in early 2021 and noticing that the strut tower trim had parted ways with the cabin wall trim, I might have lowered the rear seat to investigate and never gotten it to go back into place. I got every mile out of that car, I was daily driving it up to the moment it got parked before loading it to the gills with spare parts and driving it 240 miles to central Maine to sell it to him. LloydDobler posted:If you're not in need of making the best time, and your cooling system is up to snuff for a hard hill climb, it'll only add about an hour to 1.5 hours to go over I-70 out of Denver, and the scenery is worth it. Beats the crap out of Wyoming and you just touch the Utah badlands before turning up 191 to SLC. Just don't do it on a weekend or near afternoon rush hour. There might still be snow when you go. CO is known for freak snow dumps in March-April so watch the weather. This will also affect WY so don't skip it just due to weather. I was looking for something else and found this post and yeah, you were right. It was also one of the scariest drives of my life because I was pretty much fully loaded and holy poo poo those downgrades are really something. And I wish I'd done it during daylight instead of leaving Denver at dusk headed up i70 on a whim because why the hell not try and meet CSB on the other side of the Rockies tomorrow? Then his turbo exploded, again. Oh, and while my cooling system did fine (in fact it over cooled most of the way, and still does, I think my thermostat is stuck open) it was definitely close to its limit after what seemed like an eternity at 3500-4000+ rpm trying to maintain 25 up the pass. I'm glad I never have to do that fully loaded again, but can't wait to go back for vacations and actually enjoy it instead of just driving basically straight through minus a stop in Denver, SLC, and John Day. kastein fucked around with this message at 06:18 on Sep 12, 2023 |
|
#
?
Sep 11, 2023 23:41
|
|
|
I don't want to live in a rusty area like that and yet I'd like the experience of rolling entropy.
|
|
#
?
Sep 12, 2023 02:08
|
|
|
Holy poo poo that strut tower.
|
|
#
?
Sep 12, 2023 04:44
|
|
|
StormDrain posted:I don't want to live in a rusty area like that and yet I'd like the experience of rolling entropy. I mean, come on out here and we can do one of those "adventure vacations" where rich people get eaten by lions. Except it's you and whatever we can find for under a grand.
|
|
#
?
Sep 12, 2023 06:28
|
|
|
This sounds like fun. Game plan, all contestants must appear in New England with 3000 dollars and whatever you can fit in your luggage. You buy a vehicle and must repair, fuel, and drive it to the Pacific Coast. Whoever arrives with the most money remaining wins. May the gods of oxidation be blind to you. https://www.youtube.com/watch?v=PWIbtYWbL6E Today's project was a neighbor's Jeep Patriot manual transmission, needed a new bearing or new transmission, I threw a new bearing in it since a transmission is way out of the budget he wanted to spend on it and we couldn't find a junkyard one locally, though I have my eye out for a spare. He drove on it for 6 months with that much play and grinding so we have no idea how long the gear teeth will last, though it sounds nice and quiet now and we're much more hopeful about its remaining life than we were going into the repair. He had been quoted somewhere around 3500-5k for a rebuilt transmission installed by a local shop. Don't buy a Patriot, compass, or Dodge Caliber IMO. I'm not very impressed with the transmission or for that matter much of anything about them. There are many other competing vehicles that do everything they do, better, with more reliability and parts availability. These NVT355 transmissions have been known to do this with as little as 39700 miles on them - no, I didn't miss any zeros there. From what I can tell the problem is the design, either the bearing chosen was too small or it is not lubed well enough. More info in the video description, I should note I'll do it a lot cheaper for goons, or simply walk you through how to do it and you can do it yourself for the cost of about $200 and 3 to 4 hours of semi leisurely wrenching.
|
|
#
?
Sep 13, 2023 08:58
|
|
|
kastein posted:This sounds like fun. Game plan, all contestants must appear in New England with 3000 dollars and whatever you can fit in your luggage. You buy a vehicle and must repair, fuel, and drive it to the Pacific Coast. Whoever arrives with the most money remaining wins. May the gods of oxidation be blind to you. Done  Side note, I think summer would be the ideal season for this contest.
|
|
#
?
Sep 13, 2023 10:37
|
|
|
kastein posted:Don't buy a Patriot, compass, or Dodge Caliber IMO. I'm not very impressed with the transmission or for that matter much of anything about them. There are many other competing vehicles that do everything they do, better, with more reliability and parts availability. These NVT355 transmissions have been known to do this with as little as 39700 miles on them - no, I didn't miss any zeros there. From what I can tell the problem is the design, either the bearing chosen was too small or it is not lubed well enough. I had a Compass as a rental and... christ, you couldn't pay me to own one of those piles. Had <30k on it, already telling knock knock jokes, CVT was already making GBS threads its pants (it liked to bang the rev limiter when giving it the beans), absolutely loving gutless and could barely keep up with highway traffic in Dallas (I suspect that's partly because the CVT was on its way to having all neutrals though - trying to stay with the ~80 mph traffic resulted in it sitting around 4500 RPM). Somehow the radio pissed me off the most I think; a solid 3 second lag when streaming anything to it over bluetooth. So would it be LeMons style where safety related stuff (tires, brakes, avoiding Mopar unless it's a 4.0 Jeep, etc) don't count toward the total? randomidiot fucked around with this message at 11:33 on Sep 13, 2023 |
|
#
?
Sep 13, 2023 11:30
|
|
|
Phone posting, so no citations included here… I recall seeing that Chrysler argued (with these CVTs) that they aren’t in breach of warranty because they’ve either fixed them or (this cracks me up) they aren’t actually defective if/when an individual dealership claims to be unable to reproduce this universal (and universally known) issue. The funnier (because it feels like the automotive section from the Onion) one is with the 9-speed transmission lawsuit that they’re trying to get tossed via a method as novel as it is short-sighted: FCA is claiming (by way of lawyers to whom they’re giving real money) that the transmissions in question are, almost to a one, free of manufacturing defects. They say that the warranty isn’t being violated because the components have been manufactured precisely according to their design and that the flaw is, in fact the design, and that a terrible design, while terrible, isn’t a manufacturing defect per se. “You should know by now that our stuff is poo poo even before it makes the jump from paper to reality,” is, as it turns out, not as appealing a turd sandwich to a federal judge as the aforementioned lawyers had hoped it would be.
|
|
#
?
Sep 13, 2023 13:00
|
|
|
Yeah, the CVT is just absolutely garbage. Under no circumstances would I recommend ever buying one of those for any reason. This one was manual, I had my hopes that they had not managed to gently caress up designing a manual but it seems like NVG did that on their behalf. The loving up part, that is. I'd honestly rather they had just copied Toyotas hybrid synergy drive and used that on all of them instead of either the CVT or the unreliable manual. HSD has no magical rubber band, no conical sheaves, no clutches (automatic or manual), no converter, no main/countershaft gears, no... nothing really to fail. Not even a starter motor on most models. Just one planetary gearset, maybe a chain drive, and a couple big brushless motors. It's really my favorite ICE transmission design now, having looked at them in detail. Elviscat posted:Done I dunno if I'd want to go through Chicago with that little crash safety, but then again I'm not sure any of my cars I've driven through Chicago had any more impact safety than that so I guess you win that one.
|
|
#
?
Sep 13, 2023 16:43
|
|
|
I think Chicago's a fantastic place to visit but if I'm driving cross country and not actually stopping there, I'm definitely picking a route that says away from there. Absolutely the scariest driving I've seen in my life.
|
|
#
?
Sep 13, 2023 16:50
|
|
|
It didn't really scare me at all but I learned to drive in eastern Mass while living in Worcester and Boston and Worcester are #99 and #100 on the list of most dangerous cities to drive in iirc. What was not great was uh, the roads. What is left of them anyways. I think the Chicago skyway shook all the remaining body rust out of the holes on the Honcho, and it and the surface streets being a cratered mess are probably what finished off the right front tire on the Justy. I missed Chicago in the Comanche but I have no doubt it would have broken something loose.
|
|
#
?
Sep 13, 2023 17:13
|
|
|
I'd probably break North and take the Muskegon/Milwaukee ferry. Or do it like the motorcycle trip I did take across the country, and avoid cities as much as possible. I made it across all of Texas without stopping in a town with more than 10k people. I've also ridden bikes in NYC and Boston, so I feel pretty confident. I love that CVT legal defense, "no no no, we just designed them all hosed up, they're built perfect!"
|
|
#
?
Sep 13, 2023 19:59
|
|
|
IOwnCalculus posted:I think Chicago's a fantastic place to visit but if I'm driving cross country and not actually stopping there, I'm definitely picking a route that says away from there. Absolutely the scariest driving I've seen in my life. I drove from NYC - Seattle twice and went through Chicago both times (intentionally, because I unironically love Chicago). The first time the potholes were so bad they caused the trailer to pop off of my U-Haul, but the second time it was really lovely right at sunset and well worth detouring a little bit. Elviscat posted:I'd probably break North and take the Muskegon/Milwaukee ferry. This is also a good plan. I think the ferry is $200ish if you have a car, but it's a gorgeous ride especially if you've never been at sea before.
|
|
#
?
Sep 13, 2023 20:08
|
|
|
kastein posted:It didn't really scare me at all but I learned to drive in eastern Mass while living in Worcester and Boston and Worcester are #99 and #100 on the list of most dangerous cities to drive in iirc. I've never been to Massachusetts, but so far Chicago is the only place where I've witnessed someone driving the wrong way on a street and then rallycross style through a park's walking paths to make a left turn a minute sooner, and then been surrounded by a biker gang of dirt bikes and quads that did wheelies and donuts around our Uber.
|
|
#
?
Sep 13, 2023 20:08
|
|
|
IOwnCalculus posted:been surrounded by a biker gang of dirt bikes and quads that did wheelies and donuts around our Uber. That's a semi-frequent occurance in every city I've lived in.
|
|
#
?
Sep 13, 2023 20:18
|
|
|
IOwnCalculus posted:I think Chicago's a fantastic place to visit but if I'm driving cross country and not actually stopping there, I'm definitely picking a route that says away from there. Absolutely the scariest driving I've seen in my life. I had a coworker whose first experience ever driving in snow was driving from O'Hare during rush hour in a blizzard
|
|
#
?
Sep 15, 2023 01:25
|
|
|
IOwnCalculus posted:I think Chicago's a fantastic place to visit but if I'm driving cross country and not actually stopping there, I'm definitely picking a route that says away from there. Absolutely the scariest driving I've seen in my life. I94 through Joliet to I80's not too bad, but you could cut further south and take highway 24 if you want to stay out of pretty much every urban area between Ft. Wayne and Peoria. I can do Chicago traffic, I'd rather nail my foot to the floor than ever drive through Dallas during rush hour again.
|
|
#
?
Sep 20, 2023 10:03
|
|
|
When I rode my bike through Texas I avoided Interstates and cities entirely. Lots of lovely straight, flat highways with high speed limits and no traffic out in the boonies. A+++ would recommend again.
|
|
#
?
Sep 20, 2023 10:51
|
|
|
I regret to inform you all that I am on my bullshit again  Waiting on a response from the connector and enclosure vendor (they provide no mechanical drawing for the board outline and I want this to fit on the first shot) before I order the boards, but this should be a drop in way for people with any 96-06ish (not entirely certain what switch type 95 uses, need to find one and look as the documentation is inconsistent and self contradictory) Jeep and probably most other Chrysler products, possibly even other models, to use the factory resistive multiplexed cruise control buttons to make factory GM LS cruise control function. The only competing product from Novak is not exactly clear on what years it supports (they changed the resistances every 1-3 years for the hell of it, and it doesn't list which years work), more expensive than the price I'm targeting, and doesn't appear to be fully waterproof, like this one will be. Oh, it'll be fully configurable via wifi or bluetooth from any smartphone, and I have space left in the enclosure and connector to add support for driving most years of Chrysler instrument panels without the piggyback/dual ECU setup most people use. That's a long term goal though, got some research left to do on that. Don't mind the URL on the boards, it goes nowhere right now, though I do own the domain.
|
|
#
?
Sep 25, 2023 07:04
|
|
|
"KEEP OUT ZONE"
|
|
#
?
Sep 25, 2023 18:37
|
|
|
EWAF?
|
|
#
?
Sep 25, 2023 21:10
|
|
|
The keep out zone is so I remember to not put parts on the board under other parts - it's in a documentation layer that doesn't get printed on the board at all.Elviscat posted:EWAF? I'm terrible at naming things but eventually expect to build things for all 4 Greek elements. (4x4, boats, aircraft, and some kind of fire related thing.)
|
|
#
?
Sep 26, 2023 00:57
|
|
|
Which BLE/micro module are you targeting?
|
|
#
?
Sep 26, 2023 02:39
|
|
|
ESP32 at the moment. We'll see how I like it, if I decide it's hot trash I'll be going RSL15 most likely.
|
|
#
?
Sep 26, 2023 06:31
|
|
|
kastein posted:I'm terrible at naming things but eventually expect to build things for all 4 Greek elements. (4x4, boats, aircraft, and some kind of fire related thing.)
|
|
#
?
Sep 26, 2023 14:55
|
|
|
kastein posted:The keep out zone is so I remember to not put parts on the board under other parts - it's in a documentation layer that doesn't get printed on the board at all. Oh, I understood it. Just funny is all (but very much gets the point across unambiguously, so kudos. And funny is fine!) Krakkles posted:I would like to request the first sold model of the kastein flamethrower, please. Name your price. Gotta be better than Elon's lame-rear end effort.
|
|
#
?
Sep 26, 2023 18:03
|
|
|
It's more likely the first fire project will be some sort of a jet engine data monitoring and control system TBH. Meanwhile on the Honcho front - my kingdom for a bin or hpt file from a 12593058 GM ECU OS from a vehicle with factory dual fuel tanks. It turns out the loving fuel system segment is not the same between single and dual tank OS fuel segments and I cannot find a dual tank one to save my life. If they ever existed they were rare. No shortage of 12579405 with dual tanks, but apparently I can't segment swap from a mismatched OS ID. If I have to pcmhammer, binary diff, and manually edit this poo poo I am going to be rather cross.
|
|
#
?
Sep 29, 2023 09:59
|
|
|
kastein posted:It's more likely the first fire project will be some sort of a jet engine data monitoring and control system TBH. Oh, which Jeep is the jet engine going in? edit: or is that for the wagon?
|
|
#
?
Oct 2, 2023 21:55
|
|
|
Probably for Jack's j79 he got for the lols. It looks like hptuners tech support has turned down the idea of them changing 8 bytes of ASCII in a binary, updating the checksum, and emailing the resulting hpt file to me, so seems I'll be learning to use pcmhammer and tunerpro next and forcibly shoehorning a 12577990 fuel system segment into a 12593058 OS for science shortly. And as a direct result will be able to tune all my future projects without paying them any ECU licensing fees anymore. Suit yourself I guess. I wanted to be lazy and keep paying $100 a tune but if you insist...
|
|
#
?
Oct 2, 2023 23:56
|
|
|
kastein posted:Probably for Jack's j79 he got for the lols. *bricks ECU and walks to work for science* I need to get a spare ECU and do some playing with pcmhammer and tunerpro as well, it appears from my minimal playing around that I have codes for things like EGR and want to do some other tweaks. Probably fun since the 8.1 has a slightly less common OS... I guess I can flash whatever the common P01 OS to my ECU and copy the tuning tables, this is what a sane person might do.
|
|
#
?
Oct 3, 2023 13:51
|
|
|
I'm guessing you'll be on 12202088 from the info I've got but who knows. It looks like 12212156 and 12587603 are considered the most flexible desirable OSes to work on, but try to verify that yourself, I'm just getting started on this. There are still roughly weekly requests for xdfs for 12579405 which matches up with the fact that I found a LOT of images with that OS when I started putting my spreadsheet together, so if you happen to find out you're on that, you'd be making life harder for yourself by figuring it out and building an xdf for everyone to use, but you'd be extremely popular on gearhead-efi for doing it, probably. Personally I'd probably recommend using hptuners to read the tune out of that, then use pcmhammer and tunerpro to write one of the more common OSes with all the data from hptuners extraction copied into it manually, though.
|
|
#
?
Oct 3, 2023 22:00
|
|
|
Well I just realized I haven't updated this in forever. I've been doing some mobile mechanic work for the last few months and while business isn't great, it is partly because I suck at getting my name out there. I've got a number of repeat customers now though. No real pics on any of that because it's mostly boring basic maintenance and repairs on uninteresting cars. Steering column and 4L60e park pawl linkage fix on an 08 GMC 1500, nvt355 input shaft bearing on a Patriot that I think I actually did post, oil pan on a TDI someone went over a railroad track at warp 9 with, a few repairs on a 1970 GMC k10 with a tbi swap (ongoing), some modern Dodge power wagon maintenance, Subaru diag, blah blah blah. Other than that I finally got the binary file for the fuel segment donor I mentioned above and have begun picking it and a random 12593058 OS binary I got my hands on apart and comparing them. It's not ideal since the 12593058 binary I have doesn't have the same fuel segment as my existing tune on the Honcho, but I'm still learning mc68332 assembly and don't have an adapter pcmhammer will work with yet so it'll have to do for now. Also I've gotten the cruise control adapter project a lot further than most of my electronics module projects go, it's actually to a marketworthy point already. I have discovered a few feature improvements that will significantly increase potential market share and sale price so I don't think I'll be making more than 3 of my existing design before respinning the board but this was an excellent way to learn esp32 and get this project done. First I finally got the housing in hand for the prototype. It came here on the slow boat from China so that took 2 weeks right there while I fine tuned my PCB artwork. Here's the board artwork with the theoretical outline of the housing...  As soon as the housing arrived I did a few measurements to sanity check my work and then sent the artwork to the board house. About a week and a half later the boards got here along with all of my parts orders so I put it together. Test fit:   Looking good, time to solder!  (Pay no attention to that oops wire. You saw nothing.)  And rigged a ftdi232 and some transistors for io0/en programming to it for first testing and software.  Amazingly, no hardware issues other than me putting an sc70-5 footprint on the board in one place and a sot23-5 on the BOM and parts order. Thus the oops wire, the part didn't exactly fit and I had to take some liberties with it to recover, but it worked first shot.
|
|
#
?
Oct 29, 2023 11:27
|
|
|
Next up software. I had never developed for esp32 yet. I had a large degree of disdain for Arduino Studio and if anything, this experience has increased that. It is not a serious development environment, many basic features are missing or broken, it teaches extremely bad programming and debugging practices as a result that will take a concerted effort to unlearn once people are given access to real, serious development systems. I strongly recommend not using it. If people want to argue about this we can please do it somewhere else though  Anyways, despite that all, I needed an idiots introduction to this new platform that could get my hardware at least vetted and ready for a real development environment ASAP so I installed it. And then started programming. And got fairly far on the project before I ran into the wall. First, I knew since I don't trust my programming on this to be perfect on the first shot and I may ship it with bugs that I would definitely want to be able to have customers update their firmware over the Internet without me being present. So I started by using an OTA programming example project I found. But it was acting as a Wi-Fi client instead of a WAP because most people want that, and using mDNS because that's the best way to get home networking clients off the ground quickly. The mDNS didn't work at all once I modified the example to be a WAP instead of a client of course, but it did get me as far as having a WAP, basic web server, and hello world working...   After that I hosed around with it a lot more and got a "real" DNS server working. I may change the fake domain name to one that dummyproofed web browsers will recognize as a domain instead of trying to Google for the user such as device.ewafmotorsports.com or something, but I haven't decided.  Then got the SPI ADC working and reading values, SPIFFS working and storing my configuration and calibration files, an XML parsing library to parse the config and calibration files, etc etc etc. And after login you get:  And then on the status page:  And on the configuration page:  On Bunsen's recommendation I used POST for getting the data from the form here instead of GET. it took me a hot minute to figure out how but it was the right choice, thanks dude. Getting the OTA system to actually function required far more web dev fuckery than I've done in about 15 years. It turns out the example I used made the assumption that since it was a Wi-Fi client, so were you and you could still reach the Internet. That's not a valid assumption when the device is the WAP you're connected to. So it was trying to use a jquery library on a Google server somewhere to help with the firmware update web page, and failing silently. Which had me ripping my hair out for a few hours. But I figured it out and half rear end fixed it by ganking the library off of Google's server, sticking it in my SPIFFS file system, and changing my HTML to refer to the local copy instead. Let this be a lesson to you or something. I also have put my mad sexy css stylesheet from my awful mid 90s era ham radio nerd-esque website on this whole mess to make the interface a little more cohesive since I took these screenshots. At this point I think it's ready for final wordsmithing, adding a bit more to the user manual (conveniently hosted on the device itself, though I'll be including a hard copy as well), creating proper calibration files for it and shipping it, and it's on to adding the list of basic fixes for issues I found during first assembly and adding all the new features I've come up with and had requested. Oh and ship 2 units before I get started on rev 2, I don't want to ship the third till after I've got parts in for rev 2 so I can keep using it as a development mule. If you have vehicles you'd like supported or other feature suggestions I am all ears. This one will remain targeted at vehicles with digital bus driven gauge clusters and engines, but I also have another related design planned for running factory voltage, frequency, and thermal gauges off of a module that pulls data from the GM ECU too.
|
|
#
?
Oct 29, 2023 11:52
|
|
|
|
| # ? Apr 28, 2024 06:17 |
|
|
kastein posted:Next up software. I had never developed for esp32 yet. I had a large degree of disdain for Arduino Studio and if anything, this experience has increased that. I had pretty decent luck with Visual Studio Code. I wrote my libraries and unit tests in C++ in VSCode, and did the bare minimum in my .ino file, only using Arduino Studio to flash my ESP32.
|
|
#
?
Oct 31, 2023 17:14
|
|