Page 7 of 7
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Sat Jun 21, 2008 8:31 pm
by Fred
OK, cool. I saw that, but wasn't sure, and the gtk sharp site links the novel one. Novel are the sponsors of Mono, so I thought that seemed the safest bet. Good to know it works.
As it turns out, I have some work to do to downgrade my GTK stuff to 2.8.3 permanently for building stuff. All my c# apps require the later version which if built against precludes win use. Anyway, at least I finally sorted out why it didn't work as it should have...
Fred.
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Sun Jun 22, 2008 8:13 am
by AbeFM
Fred wrote:MotoFab wrote:p.s.: Did you get the part about programming the slopes into the processor, and perform the divides in the PC instead of the processor?
Of course, however it was decided via logic about 4 pages ago that the space was a non issue and the lookup table thumped all other options in every way except space. If we run out of space later [(512 - (2k + 2k)) = 508k left to play with = not likely] we can do something like you suggested. (good suggestion, thanks).
I wouldn't go quite as far as to say you showed it was better in all ways. Without interpolation, it's going to be less accurate, and with it, it will be slower. You might not think it's worthwhile... Can't argue that a whole lot.
There's a way out of the make-the-pc integral, or not have precaculated slopes, let the EMU calculate those tables, either on startup (second choice) or during the burning process (first choice) which will retain those benefits of speed...
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Sun Jun 22, 2008 9:11 am
by Fred
8InchesFlacid wrote:I wouldn't go quite as far as to say you showed it was better in all ways.
I didn't so much show it as rather found it. It always was that way.
Without interpolation, it's going to be less accurate, and with it, it will be slower.
This is half true. The last half. You have 1024 points of data. You transfer them to their final values in some fashion or other. You still come up with the same 1024 points either way, and at the same output resolution. One gives you that figure exactly in a pretuned way. The other is interpolating a CURVE between points. IE, less accurate. One is instant almost, the other even with the slopes requires multiple lookups, sorting, and calculation. There is no competition. I'm open to you proving me wrong, but I can't see how. You have to use technical stuff and logical reason to do it. Saying something like "there is a god" will not win me over :-)
There's a way out of the make-the-pc integral, or not have precaculated slopes, let the EMU calculate those tables, either on startup (second choice) or during the burning process (first choice) which will retain those benefits of speed...
Now, there is a problem, we ARE shortish of ram, so even though that is fast enough, we have no where to store it without writing flash on every boot up which is a terrible idea as I articulated to austinbob recently.
EDIT, if we were writing to flash, we would be using the same flash as doing it at the PC which is totally pointless
Fred.
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Fri Jul 04, 2008 8:04 pm
by shameem
Some pages back we had this discussion about how to do the measurement/conversion etc.. Here is how MS handles it.....
it has some good links to sensor data....
http://www.msefi.com/viewtopic.php?f=38&t=5522
The link may not work if you click on it - you will have to open up a new browser window/tab and copy-paste the link....
P.S. - Fred: Do you think we can get around the "referrer" issue if you change the link handler to automatically open in a new window/tab when clicked?
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Fri Jul 04, 2008 9:15 pm
by Fred
shameem wrote:Some pages back we had this discussion about how to do the measurement/conversion etc.. Here is how MS handles it.....
it has some good links to sensor data....
http://www.msefi.com/viewtopic.php?f=38&t=5522
The link may not work if you click on it - you will have to open up a new browser window/tab and copy-paste the link....
It's not too much of a pain as it won't be too long and we'll have all of our own data from scratch :-) I hadn't actually seen that thread or explanation when I started thinking about this or right up until now actually. I just googled thermistors and things until I understood how to achieve it. The I posted the spreadsheet, you implemented it in C, I made a gui, fixed a few things, and ta da, we have our own ;-)
What's described in that thread is exactly the same thing, just for ASM. Peter Siebold's code, MSAVR, and MS2 all use C arrays to achieve the same thing, just like we do. The difference is, ours are in SI units, and theirs are all imperial ;-)
P.S. - Fred: Do you think we can get around the "referrer" issue if you change the link handler to automatically open in a new window/tab when clicked?
I'm not sure. I doubt it, but there may be a way to bounce it off some other website like a search engine with a "i'm feeling lucky" button. Google respects robots.txt so that won't work (I noticed the spam artists use that technique or used to). I just tried an illegal redirect through google and I think that could work. Let me paste your link and see how it goes.
http://www.google.com/url?url=http%3A%2 ... 8%26t=5522
Code: Select all
http://www.google.com/url?url=http%3A%2F%2Fwww.msefi.com%2Fviewtopic.php?f=38%26t=5522
%2F is /
%26 is &
%3A is :
It might require an extra click and some effort to post, but I think it should avoid the 403 ;-)
Fred.
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Thu Nov 20, 2008 12:05 pm
by jharvey
I just downloaded FreeTherm 0.5, unzip, click click .exe, and I get a error message. It notes "FreeTherm.exe has encountered a problem and needs to close. We are sorry for the inconvenience." When I run it from a command prompt nothing is echoed to the shell. Perhaps I need something installed? I've got GIMP and the GTK thing, perhaps I need something else?
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Thu Nov 20, 2008 12:18 pm
by Fred
You need GTK# and the .net runtimes etc too. Not just GTK.
Gimp has a habit of doing dodgy things to GTK on winblows. That could be an issue, but I doubt it.
Fred.
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Thu Nov 20, 2008 12:32 pm
by jharvey
From add / remove programs I see
GTK+ 2.8.18-1 runtime enviroment
MS .net Framework 2.0 sp1
MS .net Framework 3.0
So GTK# is different than GTK+. I see it can be obtained here
http://www.mono-project.com/GtkSharp
But I also see wikipedia search for GTK# redirects to GTK+, so I should confirm the mono project is the right one.
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Thu Nov 20, 2008 12:36 pm
by jharvey
GTK# is 70+ megs. Bit larger than I was expecting. I won't have it downloaded until this evening when I return from work.
Re: Coolant and Inlet Air Temps Transfer Required Accuracy??
Posted: Thu Nov 20, 2008 2:14 pm
by Fred
Jared, see the last two posts on the previous page :