Easier Access to VVE Coefficients?

[KillboyPowerhead]

Hello HP Tuners,

Just wondering if it would be at all possible/practical to create an easier way to access the VVE coefficients for a given zone - specifically, being able to see/copy all of the coefficients at once for an individual zone I think would greatly speed up and reduce the tediousness of tuning the different cam angle tables in the VVE. Right now one must go into each coefficient table (15 different tables for DOHC engines), copy the row/table, paste into Excel (or whatever), calculate the new coefficients, then copy/paste each of the cam tables back into HP Tuners. Not a concern if doing several zones at a time, but when only doing a few it would be easier/faster to see all coefficients for just the zone in question, copy, calculate, paste, bada bing bada boom. So maybe along with the coefficient tables, there'd also be 30 other tables ("Zone 0", "Zone 1", etc.) that have just the coefficients for that zone.

Just an idea...

[Alvin]

Someone has just made his own VVE tool and has been posting updates. Cringer I think is the user name.

[KillboyPowerhead]

I saw Cringer's tool but haven't used it... I imagine you'd still have to copy each coefficient table to Cringer's tool and then copy/paste back to each table in HP Tuners? I noticed that you can't just change the different cam tables directly because that changes the non-cam coefficients, and when "calculating coefficients" it messes up the base table (both cams at 0) big time.

I managed to figure out the formula for the VVE table values and from there made a calculator in Excel to figure out what the coefficients have to be to get the VVE values I want; really just looking for an easier/faster way to go back and forth between HP Tuners and the calculator (Excel, Cringer's, etc.); would be easier to deal with one table that contains all coefficients for the zone in question: copy table, paste in Excel, get the new numbers, copy/paste back into HP Tuners - just once instead of 15 times. Again, if doing many zones this wouldn't matter so much since you'd need a lot of coefficients anyway, but if doing just a couple/few zones one table would be a lot quicker.

[Alvin]

I do a lot of marco's in my excel sheet.. one button press to paste/copy etc.


off topic.. wonder how MS feels about me still using office 2000 in 2023..

[GHuggins]

I saw Cringer's tool but haven't used it... I imagine you'd still have to copy each coefficient table to Cringer's tool and then copy/paste back to each table in HP Tuners? I noticed that you can't just change the different cam tables directly because that changes the non-cam coefficients, and when "calculating coefficients" it messes up the base table (both cams at 0) big time.

I managed to figure out the formula for the VVE table values and from there made a calculator in Excel to figure out what the coefficients have to be to get the VVE values I want; really just looking for an easier/faster way to go back and forth between HP Tuners and the calculator (Excel, Cringer's, etc.); would be easier to deal with one table that contains all coefficients for the zone in question: copy table, paste in Excel, get the new numbers, copy/paste back into HP Tuners - just once instead of 15 times. Again, if doing many zones this wouldn't matter so much since you'd need a lot of coefficients anyway, but if doing just a couple/few zones one table would be a lot quicker.

It shouldn't be unless something happened to it. Was working fine the last time I used it. If you have one of the "bugged" versions it had some issues. Like everything else it requires you to log correctly and input the data correctly to use it.

[smokeshow]

I saw Cringer's tool but haven't used it... I imagine you'd still have to copy each coefficient table to Cringer's tool and then copy/paste back to each table in HP Tuners? I noticed that you can't just change the different cam tables directly because that changes the non-cam coefficients, and when "calculating coefficients" it messes up the base table (both cams at 0) big time.

I managed to figure out the formula for the VVE table values and from there made a calculator in Excel to figure out what the coefficients have to be to get the VVE values I want; really just looking for an easier/faster way to go back and forth between HP Tuners and the calculator (Excel, Cringer's, etc.); would be easier to deal with one table that contains all coefficients for the zone in question: copy table, paste in Excel, get the new numbers, copy/paste back into HP Tuners - just once instead of 15 times. Again, if doing many zones this wouldn't matter so much since you'd need a lot of coefficients anyway, but if doing just a couple/few zones one table would be a lot quicker.

Unfortunately it isn't as simple as that. It is response surface methodology; you can't approach it piecewise or you'll just break something else nearby in the surface. The only 'seamless' way to pull it off is by weighting all of the error data you collect against it's distribution and number of samples. For example, if you only hit one virtual 'cell' and nothing else nearby, but you hit the cell 1000 times and it's 25% error...it still doesn't justify a blanket adjustment to the entire area. Likewise with all the cells hit in a given zone, but only one time each...still not a good case for making the adjustment towards your error. There needs to be a balance between surface coverage and sufficient samples to motivate the surface correction. GM does it in house using a regression approach, so the entire 'virtual' VE can be corrected to the greatest extent possible while preventing the correction from causing increased error elsewhere. I do it in matlab (if I am feeling frisky). Wouldn't trust excel with that sort of task though. Basically...HP tuners doesn't have the firepower to get this right, least not yet.

[KillboyPowerhead]

It shouldn't be unless something happened to it. Was working fine the last time I used it. If you have one of the "bugged" versions it had some issues. Like everything else it requires you to log correctly and input the data correctly to use it.

Not sure if you are referring to what I said about the base table getting messed up, but if so I was referring to using HP Tuners, not Cringer's tool as I haven't used it yet.

Unfortunately it isn't as simple as that. It is response surface methodology; you can't approach it piecewise or you'll just break something else nearby in the surface. The only 'seamless' way to pull it off is by weighting all of the error data you collect against it's distribution and number of samples. For example, if you only hit one virtual 'cell' and nothing else nearby, but you hit the cell 1000 times and it's 25% error...it still doesn't justify a blanket adjustment to the entire area. Likewise with all the cells hit in a given zone, but only one time each...still not a good case for making the adjustment towards your error. There needs to be a balance between surface coverage and sufficient samples to motivate the surface correction. GM does it in house using a regression approach, so the entire 'virtual' VE can be corrected to the greatest extent possible while preventing the correction from causing increased error elsewhere. I do it in matlab (if I am feeling frisky). Wouldn't trust excel with that sort of task though. Basically...HP tuners doesn't have the firepower to get this right, least not yet.

Not 100% sure what you mean but here's an example of what I've been doing:

Based on a somewhat recent discussion in your Simultaneous thread I've been adjusting the VVE table based on simply comparing the "VE Airflow" to the MAF airflow and adjusting the VVE table by the % difference between the two (and compensating for EQ ratio error; also made sure my MAF was pretty close first). Since you can only modify the base VVE table (both cams at 0) directly, I set both cams to always 0 to dial in the base table first, and then after playing with the cams and finding where they work best at cruising I then again compared the "VE Airflow" to the MAF with the cams at their cruising positions and adjusted the coefficients for that specific cam table.

So I'm only adjusting one table at the moment and not necessarily every table in between (and since changing the coefficients affects that zone in every table I don't know how practical/possible it is to tune every table, but figured the table that I'll be using most often would be a start). Also only adjusting based on steady state driving data, so I haven't been adjusting too many zones, just the ones in which I typically cruise, and then I'll adjust the adjacent zones manually if there's a significant difference so it transitions better.

Does what I'm doing make sense or am I way off?

[smokeshow]

You'd at minimum need to save all of your data for separate cam positions and then make all of the corrections at once. But you can't just apply the error to the response surface - having an actual '3d' surface means cam position is assumed to be a constant. You can't generate coefficients and then move onto the next cam angle because it will generate errors at nearby cam angles. Only way to avoid that with hpt is copy pasta into excel for each row of coefficients without making any vve changes in between, then dump the tables in and start smoothing. Pain in the ass... Matlab is preferred

[GHuggins]

This is the way I do it and honestly sounds like you're trying to do something similar although I use Cringer's, but basically you have to make sure things are logged correctly. I set the cam to 0. Dial in the base VVE table with the cam at 0 and pay attention to the fuel errors I'm getting at 0 for what cells you're going to be using the variable cams in - best to save and have this log open at the same time you're dialing them in for the next step.

Next step is I set the cam in the tune to the highest value it's going to see and then repeat the logging process. Using Cringer's tool you can plug this new log in and make corrections to shape the new coefficients into working the best you possibly can. Like Smoke said you also need to keep your transitions as smooth as possible. Then all you have to do is plug those new coefficients back into the editor. If your not going to be running variable cam timing in certain zones - then 0 those zones out.

At least that is how I do it. Turns out really well. You will sometimes get some goofy looking zones, but even a lot of GM's tunes have weird looking zones in some of these cells. Wish I had matlab, but I can't justify the $ to get it when I already have tools that gets me close.

I know it doesn't exactly look right, but in the below this one had a 1 to 2 percent error for wot and light pedal high rpm throttle. I left the lower zones higher for decel fueling errors to be right. Now honestly this is showing me something else - most likely cam timing is wrong for the lower loads - looks like it was flowing better at 0, but I left it alone because I've noticed some GM tunes as of late dropping out in these same zones, so for whatever that's worth...

This is with the cam at a constant 0 position - VE @ 0.jpg
This is at 5 degrees even though it shows 10 for the crank references conversion for the gen 4's - Cam @ 5 degrees.jpg

You probably can't tell with this, but it raises up at peak rpms and load and then drops in the midrange before popping back up in the lower zones. I used this one as an example as it's a fairly aggressive variable cam with an interesting ouput. Most of them just raise or lower 5 or so percent difference wise from 0 whereas this one was changing up to 30 percent and required more twisting of the zones than most others I do.

[KillboyPowerhead]

Regarding making changes to multiple VVE tables at once, I think the only way I can do this is by changing the coefficients directly rather than the tables, like I've been doing. When I make a change to a table directly I can't go to another table without calculating coefficients first which of course changes every other table. And as mentioned changing the non-base tables directly and then calculating coefficients changes the base table as well.

So I need to change the non-base tables via the coefficients directly, but in doing so each zone in question gets changed in every table, so I can't really change multiple individual tables (or I guess I could, but that would call for some sort of compromise between them all rather than focusing on one) - I can adjust the base table with cams set to 0, calculate coefficients, and then I basically have to pick one non-base table to adjust (via the coefficients), and just let the software do what it does to the rest of the tables (I think what GHuggins described as how he does it?). Working with an E39A ECU if that makes a difference (Gen V, 3.6 V6 DOHC with VVT on both intake and exhaust, so potentially 10's or even 100's of tables between the base and "final" table...).

If I were to create a table in Excel (or Cringer, Matlab, etc.) and paste it into that particular table, I'd have to calculate coefficients first before moving to another table which would affect every other table (including the base), and then when doing the same for the next table would change every other table again, including that previous one I just changed.

Or am I still missing something? I won't be offended if you explain it like I'm 6 years old.

[GHuggins]

I'm still not sure I'm following what you're describing. If I understand what you're saying correctly, you're saying changing the coefficients for "ONLY" the cams is changing the base tables. It shouldn't be doing this until you change the cam positions in the editor at which time it's supposed to do that as it's showing the amount of fuel it's adding or taking out based on those changes.

Cringer's tool allows you to change the individual zones and shows you the expected fuel corrections after the fact. It'll also show you real time changes to the 3d map, so you can twist it or whatever to keep things as smooth as you like. It's a small pain, but a big time savings vs the way I used to do it.

[KillboyPowerhead]

When I said that the base table was changing I meant it does so when you "calculate coefficients" on any of the non-base tables, so to change any of the non-base tables without affecting the base table I've been changing the cam-related coefficients (and not the non-cam coefficients). I figured out how to manipulate the formula to figure out what the coefficients must be to get the numbers I want in the VVE tables; I guess my concern/question is whether or not I'm going about adjusting the VVE tables properly.

So I got the base table dialed in with the cams both set to 0 all the time, and then after playing around with the cams and seeing where they work best at cruising (just been focused on cruising for now), I began adjusting the coefficients for the one particular table (4 deg intake, 10 deg exhaust in this case) to make "VE Airflow" the same as the MAF airflow (just like I did with the base table).

If I understand Smokeshow correctly he said that I should be changing all of the VVE tables in question at once (so I guess from cams at 0/0 to 4/10...?), but that's where I got lost because from what I can see I can't necessarily adjust each cam table exactly how I want it, because when changing the non-cam coefficients for a given zone every table except the base table changes at that zone. So I thought I would just change the 4/10 table (where I'll be spending most of the time driving) to make VE Airflow and MAF match, and then I guess just hope that the rest of the tables in between are good enough because I can't change them without affecting the 4/10 table at the same time.

Hopefully that makes sense.

[GHuggins]

I think I follow what you're saying now. If you don't mind post your tune - or even a stock one of yours and a log with the cams in position. I will say that you really need to have the cams at their max position for the corresponding areas, so if 4 and 10 are the max positions for the cruise area then post the log with that and really best to post your current tune - even if you only put the VE portion into it for viewing.

I haven't done any dual cam tune ins, but smoke and what you believe are right - you need to have both operating at that point I believe to get it right. I'm curious to try it out in the tool to check it.

[Cringer]

Hello HP Tuners,

Just wondering if it would be at all possible/practical to create an easier way to access the VVE coefficients for a given zone - specifically, being able to see/copy all of the coefficients at once for an individual zone I think would greatly speed up and reduce the tediousness of tuning the different cam angle tables in the VVE. Right now one must go into each coefficient table (15 different tables for DOHC engines), copy the row/table, paste into Excel (or whatever), calculate the new coefficients, then copy/paste each of the cam tables back into HP Tuners. Not a concern if doing several zones at a time, but when only doing a few it would be easier/faster to see all coefficients for just the zone in question, copy, calculate, paste, bada bing bada boom. So maybe along with the coefficient tables, there'd also be 30 other tables ("Zone 0", "Zone 1", etc.) that have just the coefficients for that zone.

Just an idea...

KillboyPowerhead, the new version of the VVE Assistant handles DOHC stuff. You can manipulate the coefficients (base VVE, intake, and/or exhaust values) as well as the cam angles in real time and watch the effects of the shape. Doing this in HPT can be done, but you have to close the VVE editor, edit coefficients, and reopen the VVE editor...then you really lose sight of what happened to the shape. Plus my tool allows you to match the histo log so you can see if you are making the final shape better or worse. Without the histo its a shot in the dark IMHO. As you know, once you click calculate coefficients...it changed...but was it a good or bad change and by how much?

Yes, the downside for right now you have to copy each coefficient table out of HPT manually. It is a pain, but its really not that bad (then again as the programmer for this thing I guess you could say I have had to do this a billion times and have some practice lol). However, I am going to work on automating this in future version. In the meantime, you can save your work in my tool and re-open it later, so that should reduce the amount of times you need to copy it out of HPT.

In case you missed it, this is the latest demo. This video was done on a slightly older version (like before the exhaust cam support was added), but once you understand this, it should be obvious how it works. If you have any questions let me know. Otherwise, GHuggins would be the foremost expert on my tool.
https://youtu.be/_Eiyy460C_k

[liviner]

Unfortunately it isn't as simple as that. It is response surface methodology; you can't approach it piecewise or you'll just break something else nearby in the surface. The only 'seamless' way to pull it off is by weighting all of the error data you collect against it's distribution and number of samples. For example, if you only hit one virtual 'cell' and nothing else nearby, but you hit the cell 1000 times and it's 25% error...it still doesn't justify a blanket adjustment to the entire area. Likewise with all the cells hit in a given zone, but only one time each...still not a good case for making the adjustment towards your error. There needs to be a balance between surface coverage and sufficient samples to motivate the surface correction. GM does it in house using a regression approach, so the entire 'virtual' VE can be corrected to the greatest extent possible while preventing the correction from causing increased error elsewhere. I do it in matlab (if I am feeling frisky). Wouldn't trust excel with that sort of task though. Basically...HP tuners doesn't have the firepower to get this right, least not yet.

Been working the last couple of weeks on a poly regression script in python that will give best fit coefficients per zone. Still need to incorporate cell weighting, which can be accomplished by providing number of hits in a cell as the weighting is just relative to the other values provided. Should be an excellent tool for dialing in the base VE coefficients. Also working on using a linear regression model to back calculate cam coefficients. Not currently user friendly at all; still have to copy/paste a bunch of table data out of HPT.