After a lot of thinking, wondering, and confusion I think I've figured out the problem with the aldl data stream in the 61 mask. It seems as if there is an lsb pa3! What this means is that the normal msb pa3 counter is pretty much worthless.
After looking at my mysterious SA retard (from knock) and not getting any actual counts, I've found that the lsb is moving, but this isn't in the aldl data stream. So if the ecm detects knock, the lsb will count, you'll get like 12 degrees of SA taken out but the msb won't move.
To better explain this funky code I'll upload a screen shot of the software I wrote for reviewing lockers data (all the ram at 15fps plus 6 extra a/d inputs).
RBob, I know we had looked at this briefly and I'd like to point out that I've revamped my software to read in the raw code, do the checksum, etc. It's great! Still working on the dynamic stuff, seems that my Radio Shack USB cable doesn't like the mscomm.ocx being in development stages. Windows XP will also blue screen if I'm not datalogging with your HDS.exe but have data coming into the port, go figure.
Anyways, I thought this would be helpful to those of you tuning TBI with the 61 mask. Although I think there are only 3 or 4 of us, lol.

I noticed the same thing in my '8063 as well... Only the MSB of the couter is sent out through the ALDL. It only reports KNOCK and not, like, knock.

Yeah, the msb is all you'll see in the stock aldl data stream if you're restricted to winaldl or scan tools of the such. Lockers lets you look at it ALL. As grumpy would say (and I hope he doesn't mind me quoting him), "too easy."
Modify the winaldl to spit out useful data like final SA, knock retarded SA, and the lsb instead of msb for PA3. I believe doing that will help any TBI tuner. For the longest time I couldn't understand where the SA was going because from the msb PA3 there was no increase but SA was being removed... I thought there wasn't any knock but guess again. It's fun looking back on old data and being able to say that even with knocking and having the ecm pull 12 deg of timing out after the 2-3 shift and then spending half my trip down the 1/4 in 3rd wondering why it isn't pulling very well STILL to run a 13.6 @ 103 with a slipping transmission and full weight street trim puts a on my face. Now it's time to return to the track with all of this knock removed and see what it'll do in street trim and then drag trim. Got me a pair of DOT slicks :lala: . With the 2800 stall and 3.73's along with the 4L60's 3.06 first gear I shouldn't loose any mph, but we'll see how she does. Looking for that 13.5 @ 105 in street trim. Which isn't bad for a TBI powered stock truck block with a mild bump stick.

What would be even better is if WinALDL was modified in some way that removed the PROMID from the 8746 datastream and replaced it with the knock counts that you are referring to.
I looked at Lockers...I think...expen$ive!!!

If you go into the prom you can just type in the addr. of the stored counter in place of the prom ID in the aldl xmit list and itll be outputted to the aldl.

Id thought id try it since I had to go somewhere and as I thought origionally, its not of much use alone since the LSB just keeps rolling over faster then the ALDL can sample it when detonation occures so its only a good indicator of where detonation is and not what magnitude it is. The best way is to output the entire counter to the aldl using the prom ID.

Is WinALDL something you can change? Hmmm...I guess if I am asking that, I probably don't have any business trying to change it.

Winaldl can be changed if you have the source code but it isn't required. You can simply change what the ecm spits out through the aldl in the code on the eprom. Just change the code to display what's in __ ram location instead of battery volts, a good replacement would be the desired AFR. This way winaldl displays the decimal in the correct location even for desired AFR. Ex) desired AFR = n * 10 , battery volts = n * 10 .
None of this is required if you datalog the raw data with winaldl and then later convert the columns into useful real world values. The prom id is pretty worthless, so is battery voltage. Replace battery voltage with desired AFR, replace the prom id with the 2 knock count bytes, replace the existing knock count (msb) with oxygen x-counts, and then find something else you don't need like INT and replace that with the Final SA. Find another place for knock SA and you should have a much more useful aldl data steam for tuning. The current stream is tailored to diagnostics and not tuning.
I wrote a program that changes the code for you and you just need to select what you want in the datastream. You can even double up values like o2 voltage to be slightly more accurate. I've got a meeting tonight so I won't be able to post the program until late late tonight or tomorrow evening. Remind me if I don't reply to this thread with a link to craig moates upload server.

That is REALLY interesting. Please, post it. I would definitely be interested in something like that.

New idea, I'm just going to post the RAM up and through together an ecu file. I forgot that TunerPro has the little "help/comments" tag. This is a better idea than a program dedicated to just changing those few bytes.
Can somebody send me a datalog file from winaldl that has EVERYTHING datalogged.

Send me an email address and I'll send you my datalog from tonight...nice and fresh. I haven't hit the VE tables yet, but I'll have them adjusted in about an hour. I got some GREAT data off the thing tonight at idle and off-idle. The BLMs at idle are a little funky, but that's soon to change.

jon, I would be very interested in using this program to see what my knock retard is. you talked about changing code before to do this but I didn't understand well enough to try. Maybe this program will help. Look foreward to you making it available and I hope I can figure out how to use it .
Thanks........bob