Street Build 95 Eclipse GSX 700bhp on 95-Oct

[m_0ney_pit]

Did you notice it handle better with the ctsv brakes? the track width in the rear is narrower than in the front at least on a 2g found that out when I did the string alignment method. Using the spacers pushes the fronts even further out 5mm on each side. The wheel hub can not properly sit on the bearing hub and ends up being supported by the wheel studs which is not the best. Watched a youtube video of a 1g guy milling 6mm off the caliper mounting tabs and he was pretty happy with the results it seems using a hubcentric spacer. I have a set of the older strictly spacers (orange) and had to mill the inside diameter of the spacer to let it sit properly on the bearing hub but decided to not run that setup because of the wheel studs/hub issue.

 

[Moonlight GSX]

Did you notice it handle better with the ctsv brakes? the track width in the rear is narrower than in the front at least on a 2g found that out when I did the string alignment method. Using the spacers pushes the fronts even further out 5mm on each side. The wheel hub can not properly sit on the bearing hub and ends up being supported by the wheel studs which is not the best. Watched a youtube video of a 1g guy milling 6mm off the caliper mounting tabs and he was pretty happy with the results it seems using a hubcentric spacer. I have a set of the older strictly spacers (orange) and had to mill the inside diameter of the spacer to let it sit properly on the bearing hub but decided to not run that setup because of the wheel studs/hub issue.

I have not yet tested these out yet so I am unable to determine that but I will definitely let you know if I notice anything - thank you for the heads up.

 

[Moonlight GSX]

A work in progress, just need to make some minor adjustments, ensure it won't damage anything, and then publish. I am still adding components as well, definitely not the finished schematic.

 

[curt-s]

What about using a smaller discrete microcontroller, like a nano or, even smaller, an esp32-s3 with a multichannel PWM controller like a pca9685?
Optocouplers can feed logic level signals from a 7805 and provide isolation

I did this same thing on my 1G, 7 segment tails with 272 HPWT LEDs on either each side.. I started with a 2560 during dev but went to a nano for size and coupled with a 9685 using i2c. The pwm frequency is stable and configurable. The biggest problem is inrush current control when you start turning on 9A worth of LEDs .

 

[Moonlight GSX]

What about using a smaller discrete microcontroller, like a nano or, even smaller, an esp32-s3 with a multichannel PWM controller like a pca9685?
Optocouplers can feed logic level signals from a 7805 and provide isolation

I did this same thing on my 1G, 7 segment tails with 272 HPWT LEDs on either each side.. I started with a 2560 during dev but went to a nano for size and coupled with a 9685 using i2c. The pwm frequency is stable and configurable. The biggest problem is inrush current control when you start turning on 9A worth of LEDs .

This is exactly what I need! Damn I wish I had know this before designing the above... either way this is a game changer. Although, I do have one question that you might be able to answer since you have already dabbled in this. Would you think it possible to operate all the switches named above on 5V instead of 12V for simplicity? My thought is that why convert the logic signal from 12 to 5 when I could potentially run 5 to the switches and back. Most of them are cockpit switches anyway and would have a 5V power supply created just for that purpose. In theory it would make the car half and half for 12V and 5V operation. Again, curious as to your thoughts on that.

Thanks for the PCA9685 idea, that I will be using. It looks like a Nano would be perfect as it has more digital I/O pins than the Uno (I have one from a physics class), and it is a smaller footprint.

 

[curt-s]

It's an either-or situation with regards to 5V. You still need to convert 12-14 to 5 somewhere somehow. On my specific implementation I did a 7805 on the same board as the rest of the components and just fed that to an optocoupler triggered by the OEM harness inputs of brake, running, and both turns. Those then fed four digital pins on the nano.
You could do a voltage divider for signalling but I chose the isolation. The 7805 also ran the rest of the 5v components on the board, I including the nano itself. Might as well just limit yourself to one regulator instead of two and one four channel optocoupler handles the 12V signals (with appropriate resistors on both sides)

Code handled the rest for debouncing and timing since turn signals are essentially square waves.
Doing it again today I'd use an esp32, as they're the same cost (if not cheaper) and they have more oomph. Of course, my segment animations were complex and some math-heavy. It also had Bluetooth integration with an android app I wrote as well, so the integrated BLE of the esp32 would be beneficial. It would have saved from having to use a slaved HC05

 

[Moonlight GSX]

I see. What makes this situation unique is that all of the stock wiring has been removed. There will be nothing OEM about the wiring, the ECU will be an EMU Black, the main fuse block is a modular one (seen earlier in this journal), and all the amenities are to be controlled by the Arduino (except the windows). However, I intend to use higher current handling MOSFETs for the power relays in the fuse box for the push-button ignition and other high current components that I want to isolate from the Arduino. Those chips won't directly drive them just the relays that will.

I figured since this will be the case I could run the 5V to all of those switches and remove the need for a voltage divider and optocouplers - I have researched both and was still undecided on which.

The board that the now Nano will control will have its own dedicated 5V power supply to be powerful enough to possibly run those switches (no load so it should be fine), the Nano, and all the necessary LEDs - mostly LED bulbs of the original type (i.e. 1157, etc).

 

[miliman13]

pca9685

+1 on this module, I used it to operate the joints of my robot manipulators.

OP
I went back and read that your basically rewiring the entire car , that’s cool and courageous someone’s a EE.

I noticed in one post you mentioned building the turn signals like modern cars, I’ve planned on something similar.

Consider the shift register 74hc595 this will allow you to compact your circuits more and the difference is normally turn lights use some form a cascading effect and don’t normally use a PWM signal.

Unless of course you’re looking for variable intensity.

Mosfets require heat sinks and relays work well in similar applications, I always recommend using optocouplers when ever possible, it’s always easy to burn circuits.

But you appear to be more than capable, and I’m looking forward to more.

 

[curt-s]

I see. What makes this situation unique is that all of the stock wiring has been removed. There will be nothing OEM about the wiring, the ECU will be an EMU Black, the main fuse block is a modular one (seen earlier in this journal), and all the amenities are to be controlled by the Arduino (except the windows). However, I intend to use higher current handling MOSFETs for the power relays in the fuse box for the push-button ignition and other high current components that I want to isolate from the Arduino. Those chips won't directly drive them just the relays that will.

I figured since this will be the case I could run the 5V to all of those switches and remove the need for a voltage divider and optocouplers - I have researched both and was still undecided on which.

The board that the now Nano will control will have its own dedicated 5V power supply to be powerful enough to possibly run those switches (no load so it should be fine), the Nano, and all the necessary LEDs - mostly LED bulbs of the original type (i.e. 1157, etc).

By ammenities, I'm not sure what you mean. Going by the pictures, you mean the lights.
MOSFETs to control all of them is definitely possible and, depending on current, may not even require heatsinks. You'll have to consult the datasheets for their thermal and rdson characteristics for whichever ones you plan on using. If youre going full 5V, you're looking at logic level as you know. TO220s or SOTs, doesn't matter but TO can support better heatsinking options. GOFORD semiconductor makes an interesting logic level mosfet I just got a tube of. Super low rdson at 4.5v, gate threshold voltage is around 0.7v.. I'm currently using them for switching on my ecmlink RPi and big assed touch LCD. We'll see how they go. Coming from 5V relays, I support the mosfet move where possible.

 

[Moonlight GSX]

Sorry, meant to say light and other amenities (i.e. push button start, remote entry, RFID immobilizer). So far, without looking at the datasheets, I think I might be okay without heatsinks as the highest draw I .ight have would be the 1157s at approximately .6A. The headlights might have something different but those will be externally triggered by a relay (mechanical), the high beams will probably be triggered via the MOSFET. Looking into the bulbs I am stuck with 12V, which is fine. I may keep the above setup and just redesign it to accept the Nano and PCA9685 - more PWM for led control.

miliman13
Funny enough I wanted to become an EE but then changed my degree (halfway done) to Secondary Education in Mathematics and Special Ed. My knowledge here is 17 years, and counting, of F15 and B1 Avionics.

But yeah, most of it is to have variable intensity and automated control via a microcontroller.

 

[curt-s]

Just don't forget to design for an automotive environment: Expect load dumps, high voltage transients (invest in a stock of appropriately sized TVS diodes), current sags, etc.
You'll also probably want to protect those mosfet gates, especially if going logic level

 

[Moonlight GSX]

Most definitely, I appreciate all the inputs. I am hoping I can get this all figured out before I start building the engine and everything else mechanical as the harness will essentially be built around it entirely save for the ECU.

More to come.

 

[Moonlight GSX]

It is still a work in progress, as seen with open nets, but it is starting to come together and make sense. I may make it to PCB fabrication on this one, need to finish this up and then do my checks and double-checks.

Yes, there is CAN in this. An interesting turn of events called for a unique module that will support the ECU of choice (solve the current requirements).

Thanks for the PC9685 recommendation; it has proved its worth.

 

[Moonlight GSX]

It beat my ass, but I am finally finished with my product. I still have to build the power supply (not too hard) for this, as it will keep the Nano powered for the push-button start feature. Once I build the power supply, I might send this one for fabrication... posting here to see if I missed anything or need to make changes (passes DRC verification).

 

[92AWDHX40]

Very nice man above my level.

 

[Moonlight GSX]

Thank you, still kicked my ass - if only I were an EE.

So for some reason my power supply is the harder one. Or I am just over thinking it, one of those two. Either way, started the power PCB and had to re-evaluate the design due to some flaws in the pcb creation.

 

[Moonlight GSX]

The power PCB schematic is complete and has some neat features. It has reverse polarity protection, OV and UV protection. It also determines if the input power is too low for the buck converters, which will turn them off before they operate erratically due to adverse power conditions. PCB drawing is currently in progress.

 

[Moonlight GSX]

They are done. Now I need to have them made, build the code, and test them to ensure they won't catch anything on fire (safety first).

I had to make a change to the control PCB. I changed the GND and HC_GND pours (green and blue respectively).

 

[Moonlight GSX]

Thanks ChatGPT... you screwed my whole design by helping me find flaws in my design.

So, I am having ChatGPT help with writing the code—I can read and modify it, but I am not super familiar with it (I am trying to self-teach)—and it just told me that I am missing a component that, unfortunately, takes up 4 additional pins on the Nano. What does this mean? I have to redesign my PCB to accommodate another component. Luckily, I have the space, and nothing on the right side has to be replaced.

Correction: I fixed it and added the required chip... I think it's time for me to go back to my dorm as I have been up for a while now.

 

[Moonlight GSX]

I guess writing here is keeping me motivated because these redesigns and code alterations are starting to get old quickly. BUT, I think it will turn out nicely once I get everything all squared away. I think this will be my final redesign, and I should be good to go.

 

[Moonlight GSX]

Update:

I've done some intensive research on a new MCU that is more powerful than the Nano and has a built-in CAN bus. With that being said, my initial plan was to purchase ECU Master's EMU Black and PMU16 as a package and use the Nano to create a push-button start application that could control the lighting for the car. With my newly found research, it looks as though I will be designing a PMU that will communicate with EMU Black via CAN bus and control the lighting and push-button start features. It will also communicate over CAN to a keyboard panel that will control various things such as windows, locks, fog lights, and hazards - I haven't defined the complete list yet. As always, I will post my progress here - not trying to be competitive with any of the PMU manufacturers, but my setup is starting to become too unique not to create my system for it... Stay tuned.

Power input is the biggest issue, so we shall see if I can pull this off. Otherwise, I will settle for a pre-built PMU.

For those of you who knew about the Nano, it's a Teensy 4.1 - removed components and added others due to onboard CAN controller, but lower operating voltage (using logic level shift registers).

Edit:
I found a way to get the RADLOK connectors onto the PCB... looks like this is a go-flight situation now - time to start re-planning and designing. Maybe if I am lucky, I may be able to freelance for the DSM community so they have fresh harnesses for the whole car (if wanted) or at least a non-mechanical relay system that doesn't break the bank (they are good, just expensive). They won't be generic, though, like the others, and would be built to order, but we shall see how this turns out first.

 

[Moonlight GSX]

Depending on the placement of components it seems that I may be able to make this an all-in-one unit. I found a place in Indepence, MO that can do one off CNC designs so that I have a solid case for mounting and something to help with heat dissipation.

As of right now I moved the original 5V power supply over to the new PCB and started working on the input power monitor for the high current loads. Similar to the 5V, it has reverse polarity, OV, and UV protection. On top of that it also has fault and reset pins that can send a message to the MCU let it know that there is a power issue, and then in turn allow the MCU to then output a trigger to the reset removing the fault. My plan with this is to have the code configured to have time delays between automatic resets and then eventually a manual reset to have the operator inspect the fault before resetting the automatic reset counter and the lockout fault.

More to come, seems this might be similar to the current PMUs on the market. Not bad I guess.

 

[miliman13]

I may be confused but I thought the Nano was atmels Arduino base board, didn’t know the teansy had a nano version.

The thing about any GPT “Generative Pretrained Transformer “
Is that they are unable to be honest about what they don’t know, so in their best efforts to be useful they will lie to your face.

I use them plenty to fast track all my graduate research and work, but you must be already familiar with the topic , so you can catch it in a lie!!!

Otherwise it will steer you wrong .

Also could you do a tiny short brief summary of what you’re attempting to do ?

I know you have discussed it , but I get so lost in the text. - I’m curious to know more.

 

[Moonlight GSX]

I may be confused but I thought the Nano was atmels Arduino base board, didn’t know the teansy had a nano version.

The thing about any GPT “Generative Pretrained Transformer “
Is that they are unable to be honest about what they don’t know, so in their best efforts to be useful they will lie to your face.

I use them plenty to fast track all my graduate research and work, but you must be already familiar with the topic , so you can catch it in a lie!!!

Otherwise it will steer you wrong .

Also could you do a tiny short brief summary of what you’re attempting to do ?

I know you have discussed it , but I get so lost in the text. - I’m curious to know more.

So Teensy is a bit different as it doesn't use the ATMEL chipset; it uses an ARM Cortex-M7 - a slightly different beast. The main reason for switching to this was that it has CAN (multiple, actually) built straight into the chip. Additionally, it also has multiple UART ports for my GPS and RFIDs (immobiliser, door locks). The size of the Teensy 4.0 is about the same size as a Nano but slightly wider, I think (or narrower).

I know that it has its quirks, but since I am "now" somewhat more familiar with understanding C++, reading the code and making edits won't be as bad/daunting. In saying that, it occasionally does bring up some good points that I may have overlooked and makes relatively okay suggestions, so I find it useful for helping me build a decent base or starting point. As for the circuitry, I have to correct it a few times, but I get the general idea of what I am building and merely use it for a part reference, so I have somewhere to start.

ECU Masters, like a few others (i.e. Haltech), has an ECU/PMU (Power Management Unit) combo that matches up with what I want to do, which is create a small and form-fitting "relay box" that is more digital than mechanical. The ECU I am planning to use due to the CAN bus and DBW (among other tunable things), but the PMU had me intrigued because as I started to build my previous setup, I felt that maybe I could build the PMU myself - lo and behold, I figured out how. I had to incorporate some NMOS drivers to drive the N-Channel MOSFETS on the high side, but that was one of the few changes I made.

TLDR
I wanted to save money on the ECU Masters PMU16 and build my own. However, I wanted to still communicate to the EMU Black ECU over the CAN bus, so I had to switch to Teensy to save on PCB space. Not to mention, I was calculating potential comm issues with the Nano and EMU Black/PMU.

 

[Moonlight GSX]

The other side of that coin is I may consider helping the DSM community have its own custom digital relay box that automatically resets itself and detects faults with indications of where the fault is (possibly). All will be in a small form factor similar to the PMU Black and a few others.