Interpolating Voltage Bias values
12-13-2018, 09:54 PM
#1
Supreme Member
Thread Starter
Interpolating Voltage Bias values
I recently needed to swap brands of injectors on my LT-5. I was able to get the Voltage Bias values BUT of course, they don’t come based on the voktage units found in your calibration. So I found a way of using Excel for doing that. Here’s the video that shows how to use it.
Perhaps getting into the weeds a bit for some, but when u get Injector Voltage Bias from injector providers, u may not get the values for the voltages specified in the calibration. So u need to do interpolation. I found this utube vid to be very helpful.
This is not specific to Voltage Bias and is helpful for any form of interpolation u may need.
Perhaps getting into the weeds a bit for some, but when u get Injector Voltage Bias from injector providers, u may not get the values for the voltages specified in the calibration. So u need to do interpolation. I found this utube vid to be very helpful.
This is not specific to Voltage Bias and is helpful for any form of interpolation u may need.
12-14-2018, 12:48 AM
#2
Re: Interpolating Voltage Bias values
Good trick. Don't forget if your bias values are @ 39.15 psi (Ford pressure), you need to first multiply the values by the target pressure voltage bias offset, then do the interpolation. There are also quite a few free apps for doing interpolation and curve fitting of a non-linear nature such as for low pulse width compensation, etc.
GD
GD
Last edited by GeneralDisorder; 12-14-2018 at 08:25 AM.
12-14-2018, 01:41 AM
#3
COTM Editor
iTrader: (22)
Join Date: Mar 2001
Posts: 9,899
Likes: 0
Received 1,855 Likes
on
1,270 Posts
Car: '89 Firebird
Engine: 7.0L
Transmission: T56
Re: Interpolating Voltage Bias values
I don't think Microsoft Excel does interpolation. I've always had to write my own equation.
I think the forecast function is for extrapolation beyond the range of data and it can give you junk answers if you try to use it for interpolation. The only reason it worked in the video is because the dummy data had a predictable pattern.
It's actually a real pain in the *** to create an all new table in Excel if the whole table is interpolated values from another table. You have to create two different lookup tables of data sorted different ways, then do a match and index function to lookup values, then run the interpolation calculation. I know how to do it but like I said it's a pain in the ***.
I think the forecast function is for extrapolation beyond the range of data and it can give you junk answers if you try to use it for interpolation. The only reason it worked in the video is because the dummy data had a predictable pattern.
It's actually a real pain in the *** to create an all new table in Excel if the whole table is interpolated values from another table. You have to create two different lookup tables of data sorted different ways, then do a match and index function to lookup values, then run the interpolation calculation. I know how to do it but like I said it's a pain in the ***.
Last edited by QwkTrip; 12-14-2018 at 02:07 AM.
12-14-2018, 06:44 AM
#4
Re: Interpolating Voltage Bias values
can also use one of the many online linear interpolation calculators. it also works for extrapolating.
https://www.omnicalculator.com/math/linear-interpolation
https://www.omnicalculator.com/math/linear-interpolation
12-14-2018, 07:38 AM
#5
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
Maybe I’ll try one of these other suggestions and see how close Excel got.Glad to see this brought up some discussion. Perhaps the topic of data interpolation ought to have its own sticky, if it doesn’t already.
GD,
I did confirm that the values I received were in fact @ 3bar.
GD,
I did confirm that the values I received were in fact @ 3bar.
12-14-2018, 10:23 AM
#6
COTM Editor
iTrader: (22)
Join Date: Mar 2001
Posts: 9,899
Likes: 0
Received 1,855 Likes
on
1,270 Posts
Car: '89 Firebird
Engine: 7.0L
Transmission: T56
Re: Interpolating Voltage Bias values
Show me the data you're dealing with.
12-14-2018, 10:31 AM
#7
Re: Interpolating Voltage Bias values
Here's a really simple online one if you just want to generate the data yourself or check that excel Forecast function....
https://www.johndcook.com/interpolator.html
GD
https://www.johndcook.com/interpolator.html
GD
Trending Topics
12-14-2018, 07:35 PM
#8
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
Here's a comparison I did. The xls is also compared to the johndcook interpolation. It compares the stock injectors w Accels and the Delphis.
12-15-2018, 07:52 AM
#9
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
So which of the two (.xls or interpolator) would u use in the cal? They get pretty close in the relevant range of 12-15 BATT v.
12-15-2018, 08:32 PM
#10
COTM Editor
iTrader: (22)
Join Date: Mar 2001
Posts: 9,899
Likes: 0
Received 1,855 Likes
on
1,270 Posts
Car: '89 Firebird
Engine: 7.0L
Transmission: T56
Re: Interpolating Voltage Bias values
What do you mean which one? There is only one right answer! If you have two answers then something is wrong.
Show us the original data table for both injectors (not the charts & not your calculated values) so somebody else can do it to check work.
Show us the original data table for both injectors (not the charts & not your calculated values) so somebody else can do it to check work.
12-16-2018, 01:47 PM
#11
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
Here is a table of Injector Bias values if u would like to demonstrate your methodology of interpolation. And a table of the BATT v values for the interpolated values.
12-16-2018, 02:36 PM
#12
Re: Interpolating Voltage Bias values
Ok the problem here is that you are using linear interpolation on a non-linear function. Additionally you need data outside your data set (I guess - practical use of voltage values below 5 seems unlikely) which is extrapolation not interpolation. You need a curve fitting tool to generate a polynomial that can be used to more accurately extrapolate (and interpolate) this data set.
GD
GD
12-16-2018, 03:49 PM
#13
COTM Editor
iTrader: (22)
Join Date: Mar 2001
Posts: 9,899
Likes: 0
Received 1,855 Likes
on
1,270 Posts
Car: '89 Firebird
Engine: 7.0L
Transmission: T56
Re: Interpolating Voltage Bias values
Spreadsheet is attached. You can take it from here. Beautiful interpolation between 8V - 18V where it matters. But you'll need to make up some numbers for the yellow colored cells that are beyond the boundaries of your original data.
At high voltage I am guessing the curve asymptotically approaches zero so I just made up some values that are 50% of the previous value. Good news is the areas where interpolation is not possible is outside the area of normal function so probably of no consequence what you put there.
I found that Excel can't deal with that curve to give an extrapolation. Forecast and Trend functions give total junk. I tried to clip off some of the data to make it easier for Excel to approximate but it still didn't work. That's why I suggest just eyeballing it and moving on.
At high voltage I am guessing the curve asymptotically approaches zero so I just made up some values that are 50% of the previous value. Good news is the areas where interpolation is not possible is outside the area of normal function so probably of no consequence what you put there.
I found that Excel can't deal with that curve to give an extrapolation. Forecast and Trend functions give total junk. I tried to clip off some of the data to make it easier for Excel to approximate but it still didn't work. That's why I suggest just eyeballing it and moving on.
12-16-2018, 06:20 PM
#14
Re: Interpolating Voltage Bias values
I think you guys may be over-thinking this a little.
just get as close as you can to the manufacturer's values and (if the manufacturer is accurate in their data) you should be good.
just get as close as you can to the manufacturer's values and (if the manufacturer is accurate in their data) you should be good.
12-16-2018, 08:26 PM
#15
Re: Interpolating Voltage Bias values
Here's a free curve fitting app that can do many types of regressions. You just plug in the data and select the type of curve....
https://play.google.com/store/apps/d...urvefitterfree
And yeah - values below 5v..... that will never happen in the real world so you may as well fill those with the same value. Just continue it down from the 6.4v value in case the software averages multiple values on either side or something crazy like that. It probably just picks the two closest values and averages them. I would have to check the HAC to see but I don't have a HAC for a ZR1.... I don't even know what ECM that runs.
GD
https://play.google.com/store/apps/d...urvefitterfree
And yeah - values below 5v..... that will never happen in the real world so you may as well fill those with the same value. Just continue it down from the 6.4v value in case the software averages multiple values on either side or something crazy like that. It probably just picks the two closest values and averages them. I would have to check the HAC to see but I don't have a HAC for a ZR1.... I don't even know what ECM that runs.
GD
Last edited by GeneralDisorder; 12-16-2018 at 08:33 PM.
12-16-2018, 08:42 PM
#16
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
Ok, so attached is a comparison of interpolation methodologies
Its clear that the John D Cook and QwikTrp methods agree at the very least within the most relevant range of operating volts. Using the Excel interpolation would also appears to be valid. The discrepancy is nearly irrelevant, it doesn't even register with TPRT. When u view the portion of the curve in the area of relevant voltage values, u basically have a linear function.
Last edited by Dominic Sorresso; 12-16-2018 at 09:27 PM.
12-16-2018, 10:22 PM
#17
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
Here's a free curve fitting app that can do many types of regressions. You just plug in the data and select the type of curve....
https://play.google.com/store/apps/d...urvefitterfree
And yeah - values below 5v..... that will never happen in the real world so you may as well fill those with the same value. Just continue it down from the 6.4v value in case the software averages multiple values on either side or something crazy like that. It probably just picks the two closest values and averages them. I would have to check the HAC to see but I don't have a HAC for a ZR1.... I don't even know what ECM that runs.
GD
https://play.google.com/store/apps/d...urvefitterfree
And yeah - values below 5v..... that will never happen in the real world so you may as well fill those with the same value. Just continue it down from the 6.4v value in case the software averages multiple values on either side or something crazy like that. It probably just picks the two closest values and averages them. I would have to check the HAC to see but I don't have a HAC for a ZR1.... I don't even know what ECM that runs.
GD
The LT5 ECMS are:
01228331 1990 to 1991.
16168001 Early 1992.
16163993 Later 1992 to about 1993 1/2.
16196394 About 1993 1/2 to 1995.
The Masks are $8E, $8EA, $AF, $AFA, $D0, $D0A, $F0, $F0A.
12-17-2018, 10:54 AM
#18
Re: Interpolating Voltage Bias values
I can find the BIN files and some definitions, but my few minutes of searching yielded no commented hac's. Disassembling and commenting the bin's is a serious commitment way beyond what I can invest in an ECM I don't use. If you can find a commented hac it would likely tell us exactly how the ECM uses this table.
GD
GD
Last edited by GeneralDisorder; 12-17-2018 at 03:58 PM.
12-17-2018, 01:05 PM
#19
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
I can find the BIN files and some definitions, but my few minutes of searching yielded not commented hac's. Disassembling and commenting the bin's is a serious commitment way beyond what I can invest in an ECM I don't use. If you can find a commented hac it would likely tell us exactly how the ECM uses this table.
GD
GD
U’ve been a big help already. W a commented hac, it would be nice to make a few mods to the program.
12-17-2018, 03:28 PM
#20
Moderator
iTrader: (1)
Join Date: Mar 2002
Location: Chasing Electrons
Posts: 18,401
Likes: 0
Received 215 Likes
on
201 Posts
Car: check
Engine: check
Transmission: check
Re: Interpolating Voltage Bias values
The injector offset compensation table value is added to the fueling PW value to make up for opening/closing time. The resolution of the table is 30.52 usec.
The ECMs of this era do a sixteen point linear interpolation between two rows. So a voltage of say 13.4 volts will use the rows at 12.8 V and 14.4 V to come up with the value to use.
RBob.
The ECMs of this era do a sixteen point linear interpolation between two rows. So a voltage of say 13.4 volts will use the rows at 12.8 V and 14.4 V to come up with the value to use.
RBob.
12-17-2018, 03:56 PM
#21
Re: Interpolating Voltage Bias values
Rbob - that is great information. Thank you. In your opinion - would there ever be a time when the values below 6.4v would be important? Given how the table is used I cannot see a scenario where the engine would still be driveable if the system voltage drops below 6.4v. Am I missing anything? If they aren't important can you guess at why the engineers included them in the table? Easier math for determining which row to pull data from maybe?
GD
GD
Last edited by GeneralDisorder; 12-17-2018 at 04:02 PM.
12-17-2018, 04:40 PM
#22
Moderator
iTrader: (1)
Join Date: Mar 2002
Location: Chasing Electrons
Posts: 18,401
Likes: 0
Received 215 Likes
on
201 Posts
Car: check
Engine: check
Transmission: check
Re: Interpolating Voltage Bias values
I use the same value from the 6.4 V entry down to the 1.6 V entry. Then the value at 12.8 V for the 0 V entry (something broke and the ECM isn't getting a voltage input).
GM uses a voltage divider then a 4.7 uf tantalum capacitor for filtering the IGN and PMP voltages to the ADC. I've seen instances where the cap goes bad and starts to drag down the voltage to the ADC unit.This is one reason not to go real high in the compensation values below 6.4 V. Don't want to drown the engine in the case of ECM failure.
RBob.
GM uses a voltage divider then a 4.7 uf tantalum capacitor for filtering the IGN and PMP voltages to the ADC. I've seen instances where the cap goes bad and starts to drag down the voltage to the ADC unit.This is one reason not to go real high in the compensation values below 6.4 V. Don't want to drown the engine in the case of ECM failure.
RBob.
12-17-2018, 06:36 PM
#23
Supreme Member
Thread Starter
Re: Interpolating Voltage Bias values
I use the same value from the 6.4 V entry down to the 1.6 V entry. Then the value at 12.8 V for the 0 V entry (something broke and the ECM isn't getting a voltage input).
GM uses a voltage divider then a 4.7 uf tantalum capacitor for filtering the IGN and PMP voltages to the ADC. I've seen instances where the cap goes bad and starts to drag down the voltage to the ADC unit.This is one reason not to go real high in the compensation values below 6.4 V. Don't want to drown the engine in the case of ECM failure.
RBob.
GM uses a voltage divider then a 4.7 uf tantalum capacitor for filtering the IGN and PMP voltages to the ADC. I've seen instances where the cap goes bad and starts to drag down the voltage to the ADC unit.This is one reason not to go real high in the compensation values below 6.4 V. Don't want to drown the engine in the case of ECM failure.
RBob.
Thread
Thread Starter
Forum
Replies
Last Post