New wiki and manual has launched! Check it out at http://wiki.speeduino.com
Any strange behaviour, crashing issues etc, post them here! Problems compiling the firmware should go in the General support section rather than here
By Black Knight
#41501
Just spent several days trying to find the best voltage correction.

This really is a big part of tuning.

I also set up an adjustable voltage circuit for my injectors to see if it made any useful change.

The answer is a BIG OH YES.

I got voltage correction to work with the car's 14 volts and tune very nice.
So that is all working as normal and is good. It works.


Now enters the voltage controller.

Real voltage is by far much much better than voltage correction.

I ran voltage from 9v to 16v into the injectors circuit.

I was able to change ARFs from 9 to 1 all the way to 23 to 1 with just real voltage change.
Yes this is with my NB O2 with my O2 controller and my test WB both.


I watched .1 volt, yes 1/10 volt make a change to the tune and I could see the O2 gauge and EGT change with that small voltage change.

I did see injector dead time go way down starting at 15v and up. 14.8 works good too.

I did see my idle tune smooth out and the AFR gauge had almost no swing at idle and AFR of 14.7.
Very stable. Because piddle and dead time went down so there is more time for good spray.

I did find the auto tune did lower the VE table value numbers a lot (tune leaner). I ended up needing to up the req fuel/CCs by 50% and the VE table by 50% and this helps make tuning better with higher numbers as I wrote about on another thread. It has to do with something like this:

Tuning 1% of a VE value of 5 is ten times more difficult that 1% tuning a VE value of 50.

Real voltage will do the same thing as changing the inj dead time but much more.

It really will lower dead time and piddle time not just try to cover it up in ms.

I want to set up a time laps vid of an inj test bed to see what is really happening.
That is a dream for another day.

All I can tell you is with real volts the tune is much better than with inj voltage correction.

Real volts makes the injector really work better.

Much less AFR swing at all points of tune.

More on this to come.

I am building a controller to make the ECU volts anything we want it to be, now I can see how good it works.


Black Knight
#41517
*******************************************************************************


Useful link.

Thanks

I have worked with boost bucks. I am using one now but they are bulky. Also, contrary to some views on bucks, I did not see any noise signal show up on any part of my logs from the buck.
The frequencies are above the ECUs range and the built in hysteresis of my circuit is greater than the noise level it creates. The buck actually cleaned up some of the alternator noise and stopped the regulator hysteresis swing of about .4v

I have created a voltage multiplier circuit that I then stepped down with linear voltage regulators.
Works great but too many added circuits and hard to get the amps I want.

I also used a separate power supply to test this. Good fast way to see if you like the idea.

I am now moving in another direction to set an adjustable voltage.

I will share my circuit only after I test it and prove it to work as well as I want.

I will find a good simple circuit to make this work because I really like what is has done for my tune.

I will be starting on the voltage control circuit right after I finish my NB O2/EGT final controller. Got the parts today and started testing the components so I get the best combo to get the best calibration.
Found them. Solder and build tomorrow and possibly in the car for tune testing.

All info is welcome. I always learn something.


Thanks

Black Knight
#41536
OK. I just did some more testing and playing with real voltage.

I went on four drive tunes and tried different voltages, dead times and req fuel/ CCs settings.

The table is now very flat now. No peaks or valleys on the 3D VE table. Just like a gentle rolling one hill at the peak load of my tune. This was done because the dead time and higher voltage to injectors has the effect of tipping the table lower on the peak load and lifting the table on the low load.
Then I could up the req fuel/CCs and the VE table by the same percent. Bigger numbers on the VE table. 32 near idle and 245 at peak load.

Now I have higher VE numbers and higher voltage to the injectors. Both of these changes did make tuning more accurate and tune faster.

So I can say with a proven test that voltage to injectors is far better that inj voltage correction.
The gains from real voltage tapered off about 15v. Still got some but not as much as from 14v to 15v.

Req fuel/CCs, VE table higher value numbers, dead time and real voltage are worth understanding and using as a major part of base tune.

One good bonus with the flatter table is, I can see the lambda delay on the VE 3D table. It shows up as peaks and valleys. I now know how to read this table and I can make delay change as I drive and see if it is correct by what auto tune is doing in that area.

Next is to create my new circuit for voltage control.

I have the parts for it now, I just need to finish my NB O2/ EGT controller circuit first.


Hope this helps some one out there.

Black Knight
User avatar
By PSIG
#41556
BK, I am not raining on your parade, but I am trying to understand what you are saying or stating. So, let me make a few comments, and then you can rephrase your responses so I "get it".
Black Knight wrote:
Wed Mar 18, 2020 1:19 am
… This was done because the dead-time and higher voltage to injectors has the effect of tipping the table lower on the peak load and lifting the table on the low load.
I'm not sure what "tipping" and "lifting" are. Higher voltage typically shortens latency (dead-time). Dead-time is nearly irrelevant, so long as the corrections are accurate in order to provide the exact same fuel delivery at any operating voltage. So, it is critically important that dead-time corrections be accurate, but made effectively irrelevant if they are. It should not matter if your dead-time is 0.5ms or 2ms, if your corrections allow it to deliver the exact same quantity of fuel on each pulse.

The primary effects of dead-time are limitations on operating range (operating quick enough for full pulses at max rpm) and a small extension of the non-linear flow range. So most dead-time effects (if properly and accurately corrected) are limited to the very minimum and (occasionally) maximum operational points. Note that other factors may appear to be dead-time related, but are typically external, such as voltage affecting fuel pressure or effects of stronger or weaker spark (coil voltage/dwell corrections), etc.
Black Knight wrote:
Wed Mar 18, 2020 1:19 am
Then I could up the req fuel/CCs and the VE table by the same percent. Bigger numbers on the VE table. 32 near idle and 245 at peak load.
This is a common manipulation in order to increase resolution and therefore relative accuracy. While increased resolution at idle is especially beneficial, be sure to leave enough room at the top for fuel adders, such as warmup, AE, etc. This is where the balance of resolution versus flow (injector "size" and pressure) becomes important.
Black Knight wrote:
Wed Mar 18, 2020 1:19 am
Now I have higher VE numbers and higher voltage to the injectors. Both of these changes did make tuning more accurate and tune faster.
Here I misunderstand, as the higher VE numbers increase resolution (as above), but the higher voltage (within reasonable range of correction) should not be a primary influence. Not splitting hairs here, as more stable fuel, voltage, temperatures, etc, are always beneficial to tune stability.
Black Knight wrote:
Wed Mar 18, 2020 1:19 am
So I can say with a proven test that voltage to injectors is far better that inj voltage correction.
Not sure what you mean by that, based on what has already been said. :? I wouldn't say anything is "better" than something else, but that it all should be accurate and stable as-possible. They are different factors. The stability and corrections to deviations work together in order to improve fueling accuracy.
Black Knight wrote:
Wed Mar 18, 2020 1:19 am
Req fuel/CCs, VE table higher value numbers, dead time and real voltage are worth understanding and using as a major part of base tune.
Absolutely! :)
Black Knight wrote:
Wed Mar 18, 2020 1:19 am
One good bonus with the flatter table is, I can see the lambda delay on the VE 3D table. It shows up as peaks and valleys.
When you're done, the delays should fairly mimic the VE curve shape, as a relatively linear delay based on exhaust pulse and flow (volume). The exhaust design greatly influences the flow speed, pressures, reversion, etc, at various loads and speeds. While variable, it should be smoothly transitioning between various load and speed points, e.g., from WOT redline to mid-range closed-throttle decel. Changes in torque should be relative and noticeable in the delay.

David
#41566
PSIG wrote:
Wed Mar 18, 2020 8:57 pm
BK, I am not raining on your parade, but I am trying to understand what you are saying or stating. So, let me make a few comments, and then you can rephrase your responses so I "get it".

David
********************************************************************************************

You will never offend me if I am learning from you. Rain away.



David, thank you as always.

Your timing of involvement is perfect.

Your comment about helping you to “get it” is exactly what I’m hoping you can help me to do - “get it.”

Just so you and everyone else are aware, I am only into tuning for 4 months now so in many ways I am still a beginner. The only thing I am doing that might be unique is I want to understand how each of the aspects of base tune work and interact. So I slowed myself down and tested the extremes of each one of the aspects of base tune. I have lots of tests done and lots of data starting to show results as in getting good tunes.

I need your help to make sense of all these tests to try to understand why they work. I’ve done the tests and experiments and know that they do work, but I don’t know that I have the background or the right wording and technical descriptions to have it make sense to the ECU community.

Help me validate or debunk this data and then help me find a way to communicate it so that everyone else can understand what I am doing.

You asked me to explain what table tipping is. I would like to find a better wording for that and I’m sure you’ll have a good answer.

Tipping, there are two aspects to it.

There are two things that get changed with this tipping idea.

One is, you will see that the table is flatter and smoother so it visually has less big peaks and much less deep valleys as is shown in one of the pics I have attached called “bad voltage and dead time” and you can compare this to the pic called “good tune” which is much smoother. Also these peaks and valleys are telling you that it is tuning erratically, i.e., the O2 sensor gauge swings wildly and never settles down to a targeted AFR.

The second one is, working with voltage and dead time you can change the actual VE value numbers as described next. This enables you to elevate the idle area for easier tuning.

Originally I had a value of 12 in the idle cell and my peak load area I had 180 as my largest cell value. Then I changed the voltage from 13.9v to 14.8v and I noticed that my idle value after tuning was now 34 which is a gain of 22. In the high load area it became 189, which is a gain of 9. So the idle circuit had more improvement from voltage than the peak load area. A gain of 22 at idle made low end tuning a lot easier and the 9 I gained at top end was not so large that I couldn’t then elevate the entire table. This then enabled me to add a percentage to the entire VE table and add the same percentage to required fuel/Ccs to end up with idle numbers of 42 and peak load numbers safely below 255. So I went from 12 at idle to 42. Tuning better in the whole bottom end. You will need to do another tune as these changes are not real accurate to the original tune.

Dead time has a similar effect. If you think about it, you can fool the dead time a little bit, one to two tenths off of best dead time is about it. But with voltage there is almost no limit. Changing your dead time corrects your fuel load around the dead time issue, but changing voltage looks like it is actually reducing dead time.

So there are two ways the table changes. One visually smooths out the table as shown in my “good tune” VE picture. And the other way the table changes is with voltage and dead time as described above so you can create higher numbers at low end which helps the low end tune better.

I could use some help here, David. I got the data, but I’m not absolutely sure what it is doing here.




And regarding leaving room for AE and Warmup, again I need your help.

Correct me if I’m wrong, that neither AE nor Warmup change any value on the VE table. Aren’t they supplementary fuel outside the normal VE tuning? If that’s true, then they are not part of the concern of the VE table 255 limit. But if you are concerned about running out of injector flow, that is always a concern. Please advise me, I am still learning.

There is also a third picture here that is called “lambda delay” and it shows a strange peak in the upper left area that represents a delay number that is too large. Over on the right side you will see a depression in the VE table, that indicates that the lambda delay number is too small. I have a fast, crude setup for finding lambda delay that I do early on in my base tune, but as you can see, as your base tune and VE table get better you will also be able to see more clearly the lambda delay faults. Just another bonus.

The “good tune” VE table is a result of auto tune on TS with no manual tune changes. The other two tables were recreated on purpose for demo purposes here.

I’m tired of typing and I’ll bet you’re tired of reading so we both need a break for now.

Any and all help is appreciated.

Black Knight
Attachments
wrong lambda delay.png
wrong lambda delay.png (49.07 KiB) Viewed 13516 times
wrong dead time or voltage.png
wrong dead time or voltage.png (52.82 KiB) Viewed 13516 times
good tune VE.png
good tune VE.png (46.49 KiB) Viewed 13516 times
#41573
I whould like to ask a few questions before I express my self on the matter.
First of all
Do you know the injectors caracteristics?(deadtimes at different voltages & pressure, staticflow)
All the data that you have is on a bench tester or actuall road condition?
If the data is actuall how is the actuall feel of the car on high load conditions?
#41576
george-angelosvoudouris wrote:
Thu Mar 19, 2020 10:42 pm
I whould like to ask a few questions before I express my self on the matter.


All questions are good.

First of all
Do you know the injectors caracteristics?(deadtimes at different voltages & pressure, staticflow)


*******************************************************************************************

All testing was done on a running motor with road auto tune and checked on MLV.

Dead time was found by best tuning behavior. By trying different dead times and seeing how well it tuned. It ended up needing .85 ms dead time. So I chose .9ms as there is no half ms.
I will be testing fuel pressure with dead times and real voltage to see what happens.
I think pressure will up dead times and higher volts will lower dead time.
I also think this will create a sharper inj opening and better spray pattern.

********************************************************************

All the data that you have is on a bench tester or actuall road condition?

********************************************************************

Road tuning and viewing the logs.

*********************************************************************


If the data is actuall how is the actuall feel of the car on high load conditions?

*********************************************************************

All ranges of the VE table were used for testing. I was full tuning on a test if it was working well.
Many test were real bad and got trashed.
Idle, light cruse economy, moderate load, high load and flat out going for it.
All good tune set ups were tuned to a level that I would daily drive the car with that tune.
The car just ran better every where. No problems at all.
The only time I had issues was when I went past the limits of calibration of dead time or had voltage that was too low. I even ran the car at 16v for a while but did not want to damage the ECU so I backed off to 15v and like it.

Near future I will be building an injector and ignition controller that can handle 30v so I can safely test much higher voltage. I did an ignition set up years ago with 18v. It worked very well.

I hope this helps and I hope to get more questions. It is questions from other people that got me started on this idea. I just put three people's ideas together and separate they work well but together they work great.

Thanks to them my tunes are better.

I am just a beginner that wants to get it right.


Thanks for the questions


Black Knight
#41581
The main reason I asked this things is because on a normal Speed density setup the VE usually follows the torque curve (on non heavily modified engines) and on Alpha-N setups usually you have a closer grouping when the TB is slightly open due to non linear air flow and the grouping is more widely spread close to WOT because airflow is more linear.

As for the injector dead time compensation the main reason for a setting like this is in order to compensate for the different time needed for the injector spindle to open on different voltages in order to stay within the VE table injection time. Some ECUs even have this as a setting that calculate the non-linear flow of the injectors. Pressure and temp of the fuel also plays a role.

I noticed also that you were saying that your Lamda readings were fluctuating this is normal you can expect to have the exact same chemical reaction on every cycle of your engine. You can get as close as possible but you have to leave a margin of error. Even a small hole before your Lamda sensor can change your readings and what about when your injector starts clogging up one your cylinders is going lean but you don't know which one so you bump more fuel to all of them.

Just saying scenarios not pointing fingers. I have seen a few of this happen over the last 3 years I've been doing stuff like that.

Best regards sorry for the long post.
#41593
george-angelosvoudouris wrote:
Fri Mar 20, 2020 9:21 am
The main reason I asked this things is because on a normal Speed density setup the VE usually follows the torque curve (on non heavily modified engines) and on Alpha-N setups usually you have a closer grouping when the TB is slightly open due to non linear air flow and the grouping is more widely spread close to WOT because airflow is more linear.

As for the injector dead time compensation the main reason for a setting like this is in order to compensate for the different time needed for the injector spindle to open on different voltages in order to stay within the VE table injection time. Some ECUs even have this as a setting that calculate the non-linear flow of the injectors. Pressure and temp of the fuel also plays a role.

I noticed also that you were saying that your Lamda readings were fluctuating this is normal you can expect to have the exact same chemical reaction on every cycle of your engine. You can get as close as possible but you have to leave a margin of error. Even a small hole before your Lamda sensor can change your readings and what about when your injector starts clogging up one your cylinders is going lean but you don't know which one so you bump more fuel to all of them.

Just saying scenarios not pointing fingers. I have seen a few of this happen over the last 3 years I've been doing stuff like that.

Best regards sorry for the long post.

George,

Thanks for all of this.

This is why I put my self on the edge of my tests and sharing here.

I get brilliant information like this.

Now I have more to think about and I can see it will help my tuning skills.

The first part of you posting is good info I need to learn from.
Thanks

The part about O2 fluctuating.

Yes there will always be fluctuations for the reasons you stated.

I am seeing that if your voltage is too low or your dead time is wrong (both too high or too low) the fluctuations are too large and a good tune is not possible. As you get closer to the correct dead time for the voltage at your injector input, your AFR swings become much less.
This is how I find dead time for a given voltage.
I have gotten the swing to less that .5 AFR swings. That is good enough. With the wrong dead times or low voltage I can see as much as 10 afr to 20 afr swings. That is proof of wrong settings.
This also shows up on the 3D VE table as shown in the image above known as bad tune.


Make your post as long as you want because the info you are sharing is all good.


Thanks

Black Knight
Ignition timing LED light

Jumper 1 is not set :( I will set it on 5v and rep[…]

Another Honda Build :P

Good day to every one As far as progress none so[…]

I think the main challenge here will be controllin[…]

R1 Thunderace trigger

Apologies for raising this old thread, but those a[…]

Still can't find what you're looking for?