Gripless

Finished a pcb design for the audio section and ordered yesterday. It should arrive in a week. No time to make a nice case etc but should have something usable in a day or so after it arrives. I spent a few days working on a spreadsheet to work out what the management/control over canbus looks like for these projects. eg any packet with the first byte as 0xFF will be a command, next byte is is which command eg test or reply with current status. Initial setup to give it a canbus address will be via the web page. But after that the ecu can send a canbus packet to change the ID, so if you are in the ecu software or want to change it you don’t have to go back to speed setup. PCBWay has 24 hour turnaround after they check the design. So far other than the first order it’s been under an hour to be checked and approved. I tried JLC PCB, which is part of LCSC which is a huge parts/components supplier. If they had a better way of combining shipping or parts and pcbs id likely use them more. Other than a few cheap bulk parts every chip and part has come from LCSC.

Maybe add a brake switch as well. If you’re going to contort yourself into the foot well may as well swap all the old electromechanical parts.

Updated code to latest canbus v5.5 functions. These are a little more complex up front but way easier and consistent once working. Not sure yet how well it’ll work for dual canbus networking but that’s not needed for these projects. I was thinking of building some bridges to allow canbus translation between different generations or ecu to body control modules etc. Second biggest things was I found a small solder blob that had been stopping canbus working on one of the boards. Which was a test one, so likely wasted a bunch of time trying some can stuff on a board that wouldn’t have worked. tried a new library for led display that doesn’t require a commercial licence if you sell anything based on it. This is a real issue when they want thousands of USD for a single licence. Accidentally stumbled onto a LIN chip for the older K or L line pre canbus that cost 50c US. Makes it cost effective to make stuff for the older cars especially since it’s still in production. Spent a day getting some parts and building a NAS and making a single storage point for all the work. I was spreading work across a fast desktop, an old slow 2014 laptop (it has all the can bus test rig attached) and a modern laptop.

Could have been one of those quick fixes on a trip that never got properly fixed.

Try a multimeter on the plug to the pedal. Maybe the switch failed and someone just bypassed it to save $

Between heating the metal weights and applying ice to the case I now have 3 more cases to use for prototypes. the resin glue still has higher melting point than the ABS plastic case, so I tears a thin layer off from the inside. Spent time rewriting all the storage and variables to use #define names not random numbers. Everything now makes sense and sits in arrays. Found a bug that stopped the wifi from being set to always on or off. The button on startup to activate wifi had always worked, but the default is now wifi on so initial config can be done. That required wifi on/off to work so it can startup off and the shiftlight started. Was halfway through swapping the led display to use a timer and interrupts rather than the dodgy timing hack I made up (It’s simple and fast, but I have no idea how it works or how I came up with it). But the hack can stay as too much effort to fix with the new canbus upadates. Espressif that make the esp32 these are all based on released v5.5 which redoes the canbus system almost completely. There are no examples outside the Espressif ones, so it’s all new code. While searching for some fixes to the newer code AI keeps offering code snippets. The code is all old. If you say that and ask for updates using new library names it just swaps those names and keeps the old data. It’s a mess and hallucinates the changes between the old and new system and makes a list that isn’t real. AI coding is pretty poor and keeping up with changes, it likes old examples that everyone repost. All examples it gives for canbus are version 1, that’s very old as few examples even use version 2. That lastest release replaces v2 with a whole new design.

Well the dinosaur board with custom tiny LEDs got assembled then other things for in the way. All parts arrived for another project after a typically painful aliexpress not accepting returns for mislabelled products. the first of a series of 10 moisture sensor on leads with a central temperature, pressure and humidity combo for plant monitoring. Which connects to apple HomeKit so can be used in automation and remote monitoring. So far nothing seems to remove the glue for the metal blocks. Acetone, glue remover etc all do ore damage to the case than the glue. The plan is still to have the shift lights by the end of November along with another ECU canbus expansion. But that might not be on a custom pcb by then just early prototype multi board. But the main code will be done which will make it fully functional.

AVCS only moves the outer sprocket rotation slightly compared to the camshaft. Even if it’s dead the cam should still spin as freely as the others. And if turning using the cam removal tool or turning from the camshaft bolt the AVCS part is irrelevant. The AVCS has a position sensor and solenoid so the ecu knows its position if it was lagging or failed. Even then I’m not sure it’d hit a valve if fully out, plus the ecu would have been in limp mode. Any resistance is bad bearings or bend valve etc. The chance of belt snap for a hard to turn by hand cam seems low. it is possible but unlikely there was oiling issue. That could cause a belt to snap but then the cams are usually rock solid stuck and hand turning them you’d notice then crack like a stuck bolt before they turned more freely. Much more likely it’s from post belt snap and valve to piston. As @boon said any valve contact should show on the piston. if there is any then no chance it was avcs. Don’t all Gates belts have their logo on them so if it didn’t then it’s was likely a cheap copy.

https://www.romraider.com/Documentation/GettingStarted#toc8 older cars can use the older 1.3 cables which maybe easier to find.

What model car as the early cable work fine in many cases. Are you going to just read the ecu or do you want to write new rooms to it?

Yeah I was with 2 rally drivers in a Land Cruiser so zero f***s were given on the way home. Others had 4-5 hour longer journey,

Only if your looking at leadfoot, the lower highway has been one way during road works before and backs up for hours with just normal local traffic.

You could ring the council out that way and check if they have any big events or road work projects planned.

Sorry if you have thought of this already. Before you approach anyone make sure you have a good written presentation. With cost breakdown and what sponsors get for the money. I know some companies get dozens of applications a week for sponsors and having seen some, most are utter s***. Like I’m great, give me money and they haven’t got better in decades. There are good ones in a few pages but the best are a dozen or more with easy summary first then details broken down. Its a what we can do for your sponsorship dollar sales brochure really. For race teams they have driver page with history of racing events even if it’s a kart racer getting into real cars etc. list of social media accounts or platforms they are active on. Good excuse the rework a history topic or create a dedicated sub forum for sponsors to see without login. If you want track day etc don’t forget to have a map with areas of signage and named location. The banner sizes they can display in each etc. Tent areas if they wish to bring and setup a stand etc. Anything like a PA system and MC can then use the sponsor name when talking about eg the LINK panic area bbq is now serving… Can you have cars on track require a big sticker on a window etc so the at every photo helps promote the sponsor long after the event. That also includes things like sending the top sponsors little awards or framed certificates thanking them. For them to display in their showroom or meeting areas etc. Oh and a timetable for setup and breakdown if you have need for them. More generic sponsorship like tires etc maybe and option. Then there’s just generic like spark who did early lead foot sponsorship with extra cell towers etc since cell coverage was non existent. If you are at any events before go ask the people at sponsor booths etc what they lie and dislike about the event.

Link hit up Adamw on their forums he’s the most active one there. https://jmms.co.nz Looks after RCM parts on nz @Niran organised the group buy there a few years ago.

Found night lights make good shift lights as they have good front light diffusers. Will solder up a new board tomorrow and fit a 24 led strip inside. Using some new smaller 3mm LEDs. This is mostly a joke for a friend, but they come in green, blue and yellow. So may have to make a set and respray them to WRC blue, gravel express green and legacy yellow. When idle or low rpm the LEDs show the same colour as the body.

Restored all the code that was commented out for ready for real canbus testing. Still need to stop the watchdog timer running when in wifi mode. The watchdog reboots everything if it does not see can messages in 10 seconds. This was as during cranking the voltage dropped and the canbus receiver became unstable.This isn’t likely now but the first prototype has no capacitors in the power supply. Added a webpage button to check canbus driver status along with stats on queued packets and errors. Fixed a mistake displaying the flash colour in webpage. soldered some light sensors to wire and add heat shrink to keep it tidy ready for when needed. The case I have spare has a couple of metal blocks in it that need to be removed. The early ones the glue was easy to remove, the current ones seem to have melted plastic not glue. It doesn’t dissolve and won’t melt before the case does if the blocks are heated. Guess to save cost they melt other products that fail QA and use them as glue. This one just happens to have been from higher melting point plastic product. I ordered the other model with the battery to see if it has the metal blocks as well.

Well it’s 1am and fixed all the current issues. Added the colour option for when it flashes which should have worked already but I had a read 16 bits in the 32 bit read code from a copy paste. Code to make canbus speed adjustable added, though the option was in web portal already. Started to add some formula options, default is STI but have Porsche and GR Yaris. Different model use different byte orders and modifiers to get rpm. Eg sti is just byte6 and byte5 joined to make 16bit number Porsche uses two bytes joined divided by 4. Which in binary jist removes the last two digits making it 14 bits. Yaris may use two bytes joined multiples by 0.8 I still have to figure out how to do the formulas while running, I may have a seperate function for each, then when it boots that function gets set as the canbus interrupt handler. It saves having a switch if bunch of if statements in the code that would run every can message. makes the code larger in storage, but faster in the most critical part. Canbus code all needs a testing as the current chip has power off and listen only mode pins that can be controlled. Old version didn’t have this. But the old chip can be used on the new board if needed. Once it all updated and running it’ll need a complete rewrite as the Esp32 has new release of its core code to 5.5 which has all new canbus functions.

Found and fixed an issue with the RGB LEDs control dropping out. This was due to the pin used this time. The full incase build was pin 26, current test rig was pin 2 and now pin 4. There may have been an issue with power as the LEDs that show the wifi password are white and full brightness so draw 3x more current than when acting as a shift light. They also get toasty warm when full white compared to ambient when showing typical colours. Finally added a test led function for the custom colour set. Once set in the web interface you can click a button to show those colours love on the display. Since you can only set half the LEDs color as it mirrors each side, when testing it uses the whole display with every other LED blank so it’s easier to interpret color and work out which number each one is.

Learned that wire isn’t always copper anymore. for a project I got some 3.5mm audio cables to cut in half to make the wiring have nice quick change sensors.. Since I needed 30+ cables I ordered so that looked like thick cable but they were cotton braid over tiny wire. Turned out the wire was copper coated aluminium and is absolutely s*** to solder. Wire is not even thick enough to have insulation on each wire just enamel coating. For the car projects all the wire I use has braided shielded which seems to only come in oxygen free copper OFC. The braid helps prevent electrical noise but more importantly prevents damage when pulled through the back of the dash etc.

Managed to flash a microcontroller module in a generic test rig I brought with spring loaded sides. It worked fine except that you need to time the grounding of the boot pin and reset at the same time. The board has a button to do that automatically but the first flash timing doesn’t match. After the first flash though the automatic button works consistently. Flashed another blank module that was already soldered to my pcb and its far from ideal too. The 4 pin for power, ground, send and receive connector was great initially when pin header was soldered in for prototyping. For assembly into a case it can’t have the header fitted. I ordered a spring loaded 4 pin pad to deal with this and clipped it on with a clothes peg. But the fast timing on first flash it needs the ground wire and power to be connected after the serial starts. I split power and ground out and it works ok now. So once the pcb is final I’ll likely need to make a test rig pcb that lines up with all the serial and power points. That will have pogo pin so you just place the real pcb on top with a little pressure it’ll connect. Otherwise it wastes 2-5 minutes trying to get it connected and flashed. It will also reduce cost allowing the led that indicate different voltages to be left off each pcb and only fitted to the test rig. After flashing I left the pcb running in wifi setup mode for hours. The heat from the voltage regulator was the highest as expected. It was still only warm and efficiently spreading heat into the pcb via the large copper pad and holes that were added for exactly that purpose. I retested again the next day and max regulator temp was 42.4c and 24c room temp. Microcontroller was 41c with wifi idle. Wifi takes 300mA max where no wifi drops as low as 40mA so this should be the maximum power use. Worked more on another pcb design for a different canbus project. Between design and selection or parts it takes a lot of time. Seems making sure there will always be stock when you don’t use basic parts is important. Adding little extras to the pcb to allow flexibility will hopefully be worthwhile. Been trying to find my spare steer wheel controls as I can’t remember what the button inside where. They feel like silicon pads and not switches. So that could be a nice quick project to have a board made up with nice micro switches.

So many little things still need work. While not needed for functionality I’m making a testing portal over wifi for testing the connections to canbus and led display etc. This will allow both initial testing and end users to check the device is fully functional and test any repairs or changes they make. There are many variations in how RPM is encoded in canbus messages and they will need to be added as options in settings and the above test portal. Mounting is another area that has to be tested as so far it’s been black foam tape. There are 2 places for a mount either using alloy riv-nuts or square nuts fixed with resin. The riv-nuts need with only 2 of the 4 petals so they all need cutting before installing which is slow and they take up a lot of space. The other option is square nuts that are glued in place and I have used them in the past with god results. The other advantage is that it leaves enough space for a layer of protection so if a longer bolt is used it will have some cushioning before it hits the electronics. Most of the testing side came from a way to reduce cost by using a more traditional tube case with a smaller number of LEDs.

Forgot to note that ordered pcb in different colours to see what the difference is. Each colour flows differently and has a minimum width between component pads it can flow, with green being best. I tried white and blue this time as the text colour is white on green and blue, but black on the white board. The black text is not as thick or clear as white. The white text is close on both but still best on green. Worked out option to add a buzzer to play annoying tone when shift light is flashing. Overall little progress, spent a few days doing custom apple HomeKit projects for environment monitoring. This was an early project that was put on hold as the person who wanted it had to make a bunch of choices. It has some value for automotive where the car when it arrives home connects to apple homekit and sends data. Depending on the car it could be milage and fuel use or just last battery voltage and temps. Realised I need a better test rig as laptop and stack of usb canbus dongles isn’t the best. So made some initial plans for a stand-alone canbus network similar to the below but with SD card for adding real logs. Plus web portal to setup the canubus IDs for the buttons for lights, gears and pedals to emulate along with a rpm and speed estimate. It will have a obd2 plug with 12v power so it is easy to test devices. https://www.aliexpress.com/item/1005009344611851.html Have some work for custom canbus gauge cluster upgrades for older Subarus that have no obd2 or canbus but run aftermarket ecus. All stealth or OEM+ look inside the stock cluster.

Just ice the inter cooler for the power run on the dyno. It’s just a piece of paper, it doesn’t show gearing or real performance anyway.

If they want a real solution it’s not based on anything they have said. It’s all in the contracts, repair costs are covered by the contractor for the first year. These already exist for some infrastructure along with very strict times when work can be done eg only low traffic times, often at night. Lower the cost of roads rather than bill users. Ban luxury good from road transport, low use high status items are not allowed on road transport and must be sent via train. Supercars just add 10% duty on import for roads. Hybrids and electric weigh more than petrol cars so I’m not sure how complex their maths is to make anything balance out. The other part is if people are to pay the same or less how does that increase the income. Even if it is to cover the development costs which being software and/or hardware will be many millions of dollars. plus isn’t there going to be more of an imbalance with local roads in cities not getting money. Isn’t Auckland the only city that gets income from road tax now? let’s also throw in Auckland has more congestion now than before 1)ride share apps have studies showing they increase city traffic and congestion by 13%. Add to that the 3 lanes roads that are now 1 lane for bus or cycle lanes that are pretty much unused.