Working e-Dash with a Raspberry Pi and ECUGN

I like it too but why that versus an Android tablet?

I love how you are always pushing the envelope and trying new things.

Sent from my SM-G965U using Tapatalk
 
I like it too but why that versus an Android tablet?

I love how you are always pushing the envelope and trying new things.

Sent from my SM-G965U using Tapatalk

Check the blog for a longer explanation, but Android pissed me off.
Granted, it was Amazon's hacked up Android on a Fire HD. But the experience was frustrating and shitty. Perhaps dropping $500 on a high-end Samsung tablet would be better, but that's halfway to an AIM or Race-Tech Dash2 Pro, which is best.

But I had a pile of Pi 3's just laying around (sixteen, actually), so I figured I'd give it a shot. As a low cost way to prototype, it's magnificent. I'm looking at other single board computers with more gonads as possibilities for an upgrade if this survives this weekend. Though, with the amount of time I've put into this mess, a CANBUS connected bespoke race dash would have been a better value. Maybe. The Dash2 I tried initially worked, but just couldn't get enough channels out of the ECUGN over CANBUS to do what I wanted.

But if I'd done that, I couldn't have made a cool video and claimed I made an eDash out of shit I had laying around. The burn-in you see at the end of the video when the thing shuts off showing Adele is left over from the screen's previous life as a media station for the stereo on my back porch. I think Adele was the graphic on the display for something like 16 months straight. And now I get to see her every time it powers down. Oops.

The thing is, I can pull the SD card out of this unit and shove it in a new Pi 4 and get double the performance with zero additional work. So the load time goes from 38 seconds to 19 seconds, and there's enough horsepower left to do some other crap, like run a Nav app or some kind of in-car audio thing. And this setup is a cellular modem away from live streaming telemetry off the car to the mothership like an F1 car. That has limited actual value, but huge nerd points. Other SBC's offer even better performance, though at a premium. The difference between a $35 SBC like the Pi and a $150 SBC like an Edge-V is HUGE.
 
Check the blog for a longer explanation, but Android pissed me off.
Granted, it was Amazon's hacked up Android on a Fire HD. But the experience was frustrating and shitty. Perhaps dropping $500 on a high-end Samsung tablet would be better, but that's halfway to an AIM or Race-Tech Dash2 Pro, which is best.

But I had a pile of Pi 3's just laying around (sixteen, actually), so I figured I'd give it a shot. As a low cost way to prototype, it's magnificent. I'm looking at other single board computers with more gonads as possibilities for an upgrade if this survives this weekend. Though, with the amount of time I've put into this mess, a CANBUS connected bespoke race dash would have been a better value. Maybe. The Dash2 I tried initially worked, but just couldn't get enough channels out of the ECUGN over CANBUS to do what I wanted.

But if I'd done that, I couldn't have made a cool video and claimed I made an eDash out of shit I had laying around. The burn-in you see at the end of the video when the thing shuts off showing Adele is left over from the screen's previous life as a media station for the stereo on my back porch. I think Adele was the graphic on the display for something like 16 months straight. And now I get to see her every time it powers down. Oops.

The thing is, I can pull the SD card out of this unit and shove it in a new Pi 4 and get double the performance with zero additional work. So the load time goes from 38 seconds to 19 seconds, and there's enough horsepower left to do some other crap, like run a Nav app or some kind of in-car audio thing. And this setup is a cellular modem away from live streaming telemetry off the car to the mothership like an F1 car. That has limited actual value, but huge nerd points. Other SBC's offer even better performance, though at a premium. The difference between a $35 SBC like the Pi and a $150 SBC like an Edge-V is HUGE.
Yes, I recall your Fire HD issues. I was leaning toward a Samsung Galaxy Tab S5e. Do you think I'll have any issues with that? I keep an eye on them as the price slowly drops.

Sent from my SM-G965U using Tapatalk
 
Yes, I recall your Fire HD issues. I was leaning toward a Samsung Galaxy Tab S5e. Do you think I'll have any issues with that? I keep an eye on them as the price slowly drops.

Sent from my SM-G965U using Tapatalk

I haven't personally dealt with any Android tablets beyond what I have, so that's the FireHD and I bought a cheap Hyundai with base Android on it, which also didn't work out due to FTDI incompatibility problems. And it was slow as balls.

I know my fellow Autocrossers swear by the Samsungs to run Solostorm. So the performance is there. The wildcard is USB port access for applications or how it behaves on a wifi network with no internet access. From what I've seen, that's a very inconsistent set of behaviors. Different vendors apply different policies to those settings in their Android customization. Amazon has to push ads to the FireHD, that's why they're cheap. The Ads subsidize the cost of the tablet. But to push ads, it has to be on the internet, and so the device aggressively seeks out internet access.

A tablet without such encumbrances may fare better.
 
I haven't personally dealt with any Android tablets beyond what I have, so that's the FireHD and I bought a cheap Hyundai with base Android on it, which also didn't work out due to FTDI incompatibility problems. And it was slow as balls.

I know my fellow Autocrossers swear by the Samsungs to run Solostorm. So the performance is there. The wildcard is USB port access for applications or how it behaves on a wifi network with no internet access. From what I've seen, that's a very inconsistent set of behaviors. Different vendors apply different policies to those settings in their Android customization. Amazon has to push ads to the FireHD, that's why they're cheap. The Ads subsidize the cost of the tablet. But to push ads, it has to be on the internet, and so the device aggressively seeks out internet access.

A tablet without such encumbrances may fare better.
Did you ever go any further with this? I have a ecu-GN coming; I have found a ton of Megasquirt videos and threads but they are mostly 2+ years old.

I have a Pi 4 with a 7" touch screen that would be great though my concern with using a Pi is the boot up time. I know it can be lowered by stripping out anything not needed bit I don't know off hand what all is required by tunerstudio yet. It would be nice to get it down to under 10 seconds boot time like some of the embedded linux systems.

Other thoughts that come to mind; as I am not really looking to edit the tune through the e-dash is something like RealDash; although again like many of these projects it seems to be semi-dead. Shadow dash may work too.

If the ecu-GN supports all the CAN bus stuff the MS3 does another thought is a arduino based solution using a CAN hat; I have found some examples of communication with the MS3 but I don't know if the arduino has enough power to run a larger screen (5") even though I would just be doing text for the most part. I have a few arduinos and a can bus hat I might start playing with when the ECU gets here.
 
Did you ever go any further with this? I have a ecu-GN coming; I have found a ton of Megasquirt videos and threads but they are mostly 2+ years old.
What I settled on was a Pi4 in a case with an Adafruit GPS sensor and external antenna. I got the boot and TS startup down to about 48 seconds. The speed limitation really is the SD card it boots from. Get the fastest one you can find. I also built a timer based UPS circuit that provides power to the Pi for 30 minutes after key-off. When the car goes key-off, it sets one of the GPIO pins low and a timer starts. If the key doesn't come back on in 29 minutes, it shuts the Pi down gracefully.

But if the car's keyed back on within 29 minutes? No waiting for it to boot. It's already up. So far it works like a charm. It's all stuffed into the glovebox.

BUT! No display. What I did was set the Pi up as a wifi hotspot. Instead of cluttering up the interior or hacking the dash with a screen, I use a suction mount and a phone connected to the PI via VNC when I need the gauges to be visible.
 
What I settled on was a Pi4 in a case with an Adafruit GPS sensor and external antenna. I got the boot and TS startup down to about 48 seconds. The speed limitation really is the SD card it boots from. Get the fastest one you can find. I also built a timer based UPS circuit that provides power to the Pi for 30 minutes after key-off. When the car goes key-off, it sets one of the GPIO pins low and a timer starts. If the key doesn't come back on in 29 minutes, it shuts the Pi down gracefully.

But if the car's keyed back on within 29 minutes? No waiting for it to boot. It's already up. So far it works like a charm. It's all stuffed into the glovebox.

BUT! No display. What I did was set the Pi up as a wifi hotspot. Instead of cluttering up the interior or hacking the dash with a screen, I use a suction mount and a phone connected to the PI via VNC when I need the gauges to be visible.
That is a good way to do it. Simple and there when you need it.

The gauge cluster in this car was originally digital, then changed over to a cyberdyne digital dash then to a AVC GNX style dash. The VDO gauges are suspect at best. I replaced the speedo with a electric one last year because the original one stopped working, the tach is erratic at times, other times it's fine. The oil pressure gauge (electric) reads about 5# low compared to a manual gauge, the boost (mechanical) doesn't match the AEM 3.5 bar but it is closer than the oil pressure. Gas gauge is just as bad, even with a new (correct for analog gauge) sender; could be a ground issue on that. The coolant temp reads high by 5-10deg vs the ECM but that could be different placement of the sensor.

The tldr; version is my current gauges suck.

Ideal for me would either be a full digital cluster like the Holley to replace it but to work with the MS based ECU's it going to be a DIY thing. DD-EFI makes a dash setup and while nice, it's not anything I can't DIY since it is Pi based. My other idea is something like the speedhut dual gauge with the speedo/fuel level and a tach with a 4"-5" screen between them display all the other sensor information. I'll probably order a small TFT screen to experiment with since I have everything else to do CAN based info.
 
Doing a bit more digging, the arduino route will actually be pretty easy. There is a github repo of someone who is using an arduino to read the CAN data from a MS3 (srenner/AutoCAN_Meguinauge) For the UI I should be able to use a Nextion lcd (https://www.amazon.com/gp/product/B079172C9G); the display does the heavy lifting for the gui so the arduino just needs to send values to it..

It will be a cheap experiment:
Arduino Uno $10
Arduino CAN Hat $7
2.4" Netxion TFT $26

If it works, a 5" screen is only $86
 
Doing a bit more digging, the arduino route will actually be pretty easy. There is a github repo of someone who is using an arduino to read the CAN data from a MS3 (srenner/AutoCAN_Meguinauge) For the UI I should be able to use a Nextion lcd (Amazon.com the display does the heavy lifting for the gui so the arduino just needs to send values to it..

It will be a cheap experiment:
Arduino Uno $10
Arduino CAN Hat $7
2.4" Netxion TFT $26

If it works, a 5" screen is only $86
If you want to have some fun we have done a 12 inch display in the stock dash with a 3D printed holder with an MS3 Ultimate. Cost was about $350 depending on which tablet you buy.
 
If you want to have some fun we have done a 12 inch display in the stock dash with a 3D printed holder with an MS3 Ultimate. Cost was about $350 depending on which tablet you buy.
What did you use for the pc/tablet/?? hooked up to the display? I have seen the 12.3" displays for around $120-160. Playing around today I was able to get a pi to go from plugging it in to tuner studio loaded in about 45-50 seconds. The arduino route will be a few seconds from power to ready but it will just be able to display what I can get out of the CAN bus.
 
Top