DIYEFI.org Forum

Why not blaze a trail and help develop such a decoder? :-)

tpsretard wrote:with that resolution i could get my afr's at idle to stop playing ping pong...
Humm, i am thinking once there is a wheel decoder for the 6/7 subaru i might have to spend some money in hardware..

are you talking about the LMPI 6 tooth thingys that are used with the JECs ECU ? I could help testing whatever you come up with .. I have a late 90's forrester that uses it

I also have two late 90 scoobaroo's, and am willing to help with decoding efforts. I seem to recall the last code I saw posted did not include a functioning RPM decoder, and I didn't know what the plan was for how to pipe the signals around ect. That is now obvisouly functioning, so I'm eager to read over it again once it's out there.

Not one of you suggested I get off my lazy ass and implement the fix myself?
Something's wrong here.



Aaron, my car is so close to getting back on the road. Hell, it runs now, which is insane. A couple quarts of transmission fluid and some charge pipe re-install and I'll be back on the road. When that happens I'll be back into the python. I miss coding.

BenFenner wrote:Not one of you suggested I get off my lazy ass and implement the fix myself?
Something's wrong here.



Aaron, my car is so close to getting back on the road. Hell, it runs now, which is insane. A couple quarts of transmission fluid and some charge pipe re-install and I'll be back on the road. When that happens I'll be back into the python. I miss coding.

Good stuff mate Can't wait.

Lots of tidy ups and refactoring to do, as well as some new features. The sprint over the last week has been a bit messy lol

The load at the top is a wxPython thing, not sure if I can move the col headers

Aaron, you're a legend! Not only did you fix the headers you also added more features! :-)

One of which I just hadn't complained about yet - the fixed table size. Or does it just not reload fully? I didn't test that, but I did notice that it was better :-)

The other was adjust cell = connect dialog :-)

I tuned a little last night and drove 130 miles on it :-) Wicked work mate, keep up the good work :-)

Fred.

Fred wrote:Aaron, you're a legend! Not only did you fix the headers you also added more features!

One of which I just hadn't complained about yet - the fixed table size. Or does it just not reload fully? I didn't test that, but I did notice that it was better

The other was adjust cell = connect dialog

I tuned a little last night and drove 130 miles on it Wicked work mate, keep up the good work

Fred.

Hehe, yeah if I see something that makes sense I try do it while I'm there

The table will grow in size, but it won't get smaller yet.

Keep finding me bugs!

OK, so, issues, we would like some of the below :

when you hit burn, get confirmation that it occurred and display this to the user, eg, light something up red and when you get the ack turn it green. maybe orange for unburned changes as a status thing.

when you hit load, light a red led and then turn it green if and when you receive a packet to update the tables.

the above things are annoying as piss because you cant possibly know if what you did worked. why it doesnt just always work, i dunno, but this is the point, maybe my rs232 is dodgy... we need to know what happened. along the same lines the cell updates dont make it in some of the time. i dnot know what the solution is, but we MUST keep the table and ram synced so the user knows what he has. this we should discuss. it may just mean that we need to put in some sort of delay system for sending

the cell you are adjusting isn't obvious enough either. the current cell the engine is in is plenty obvious. rob raised the point that the movement makes it more obvious by default. He thinks the one you tune should be white with black text, maybe bold, and the moving one should just be surrounded like the current cell you are adjusting is at the moment, but more obviously.

another comment was that the general contrast of the rest of the cells is too poor. he suggested a pastel colour scheme rather than saturated, not sure iif that is hard or not.

as well as +/- 0.5 we'd like to be able to scale them by say 15%

along with that, groups of cells being adjusted, with that though, you're definitely gunna need to put delays in between the packets or you'll lose half/most of them.

in old video games you controlled them with ikjl and i / k were up/down - rob would prefer this and i'm with him, the left right thing is unnatural and hard to remember/find. perhaps j/l as a course 5 at a time adjustment too?

not sure if there are any others, i'll get rob to do some posting later - he did the tuning today while i explored tampa randomly and revved the car hard :-)

Fred.

Awesome! Thanks for the list.

I will start creating issues for each of them soon

First off- good job Aaron. It was a pleasure to jump right in to tuning.

As Fred mentioned those are a few worthy improvements above. The colors are great,they just need to be about 70% lighter to provide better contrast to the values (type). Then the active (current operating) cell should be a contrasting color and more intense- like a green or blue would contrast the warm colors of the other cells.

An off line mode that allows you edit and load the visible map in one packet is needed. We were itching to play with the prelim tune over some beers but were denied by the perpetual off line warning box. I'll post back after some more time with the app. - Rob