Opensource LS Flashing(It's Closer Than You Think)

**Xnke** · 02-22-2018

So you're keeping this under wraps...just like every other LS1 flash tool that has been done so far, except Dimented24x7's, it's all "free" until it's complete and works...then it's for sale.

HPTuners, EFILive, Antus's flash tool...all of them started right down the exact same "Yeah let's get this done for free" path you're on right now. And all of them said "it's too complicated, I'll release it for free when it's done."

I'm hopeful, and still willing to work on this, but won't be surprised when this comes out to be another few hundred dollar subscription, instead of a free flash tool.

**PeteS** · 02-22-2018

Originally Posted by Xnke

So you're keeping this under wraps...just like every other LS1 flash tool that has been done so far, except Dimented24x7's, it's all "free" until it's complete and works...then it's for sale.

HPTuners, EFILive, Antus's flash tool...all of them started right down the exact same "Yeah let's get this done for free" path you're on right now. And all of them said "it's too complicated, I'll release it for free when it's done."

I'm hopeful, and still willing to work on this, but won't be surprised when this comes out to be another few hundred dollar subscription, instead of a free flash tool.

Here's some pepper to go with your salt

https://github.com/LegacyNsfw/PcmHacks

Publishing every thing myself and others have done that lead down this path would be well over a year of collective information. Forgive us for not having that much data formatted and laid out for others to understand right away. There are a number of people involved in this project and every one has the same goal. What makes this different is that no one person holds all the information. Loosing one of the core would seriously hurt our work BUT every thing they have done has been shared with others and WILL be made public before this is all said and done. If this goes belly up before we finish out work you can be sure at least ONE of the people involved will throw EVERYTHING online regardless of how clearly it's formatted just so that all our work will have been for nothing. The idea was that others could finish what we started if we are unable to.

So..... no matter what happens SOMEONE WILL BE ABLE TO FINISHING THIS.

**NSFW** · 02-24-2018

If you search for my alias in the RomRaider.com forums you'll find a handful of open-source projects for Subaru, from tuning utilities to a speed-density implementation that has since been ported (by someone else) to a bunch of other Subarus besides my own, and a rev-matching patch that is still a work-in-progress but is also up on GitHub for anyone who wants it. Everything I contribute to this effort will be open source as well.

I got involved in this effort a couple weeks ago because I bought a Corvette a few weeks ago. If you want to see the code I've written so far, it's all at the link PeteS just posted above. It's mostly just some simple dev utilities and a prototype that we're using to discuss the design of the real app, but hey... gotta start somewhere.

There are basically three main holdups right now.

1) The flash kernel we have now was written by a guy who can't be reached. He posts as Dimented24x7 on some other forums. His job takes him offline for long periods. We have permission to USE it, but nobody thought to ask for permission to post the source publicly before he left, and it wouldn't be right to share the source without that permission. But it works. So it will appear only in the form of a long string of bytes until he gives us his blessing to post the .asm files. (Or tell us to knock it off, in which case we'd just have to write a new one. But I don't think that's likely. And he did give us a nice reference to follow...)

2) We want an interface that doesn't cost an arm and a leg. The proof-of-concept uses an AVT interface, but it's hard to come by and rather expensive. PeteS mentioned the Arduino project, that's got great potential. Hopefully it will be up on GitHub soon as well, but that's up to the guy who is developing it. I also just ordered a Scantools SX in hopes that we can make it work. If anyone else wants to pitch in with a better / cheaper / faster interface, that would be fantastic!

3) The existing app needs to be rewritten for maintainability, flexibility (it's too tightly coupled to the AVT interface), and reliability. That's where I'm focusing my time.

This will be open source when we're done. Parts might not be open source at first, but that just means that we're not done.

**Xnke** · 02-24-2018

Can you use raw binary data instead of the AVT data interface? To be clearer, are you attempting to emulate the AVT852-E and are looking for less costly hardware, or are you willing to work with a direct serial translator instead?

If so, I'll have a 1x/4x capable hardware unit posted here shortly. It'll be atmel based so you can run your current VPW implementation if you'd like, I'm putting the old design through the wringer right now to make sure it'll do what you need it to do. Also, you could simply use the Harris or Fairchild bus interface chips to do the hardware, they're discontinued but available through second-source vendors. I have been pulling them from GM body control modules, they're surface mount so you need to be good with a hotplate.

Also the AVT interface is available directly from AVT, you can buy singles at 250ish dollars each. That's where mine came from.

**buddrow** · 02-24-2018

Xnke,

just how soon is "shortly:? :-)

Buddrow

**NSFW** · 02-25-2018

Originally Posted by Xnke

Can you use raw binary data instead of the AVT data interface? To be clearer, are you attempting to emulate the AVT852-E and are looking for less costly hardware, or are you willing to work with a direct serial translator instead?

If so, I'll have a 1x/4x capable hardware unit posted here shortly. It'll be atmel based so you can run your current VPW implementation if you'd like, I'm putting the old design through the wringer right now to make sure it'll do what you need it to do. Also, you could simply use the Harris or Fairchild bus interface chips to do the hardware, they're discontinued but available through second-source vendors. I have been pulling them from GM body control modules, they're surface mount so you need to be good with a hotplate.

Also the AVT interface is available directly from AVT, you can buy singles at 250ish dollars each. That's where mine came from.

Raw binary is fine, we're designing the next app(s) so that the raw binary payloads are totally separate from the details of the different interfaces. That should make it relatively easy to support different interfaces. Antus wrote the proof-of-concept and has an AVT, so that will probably be the first interface to work, but I'm not planning to buy one just yet. As a software developer I'd prefer the "direct serial translator" approach. But as someone who wants there to be an active community of people hacking 0411 PCMs, I really just want there to be an interface that is inexpensive and readily available. $250 could work, but that price point seems to leave a lot of room for improvement. DIY SMT isn't going to work for very many people though, so I'm not real excited about that approach.

What do you think about getting productizing your design through service like seeedstudio.com? Or PCBs through oshpark.com, a case through shapeways.com, OBD2 connector with pigtails from Amazon, etc, etc, so anybody with a soldering iron could build one?

**Xnke** · 02-25-2018

The problem with things like oshpark and seeeedstudio is cost. You can't buy just 1 board through those guys, and oshpark is 3 boards for close to 100 bucks all told.

You can't buy a lot of the parts and bits you need in through-hole anymore, you'd be buying breakout boards and main pcb's and pin headers...SMT is very easy to do at home, but people see it as scary so no one tries. It's actually much cheaper and faster to do than through-hole!

gotta go, second and possibly third tornado today just sounded the warning-if I have power in a few hours I'll start posting schematics.