To be honest I don't really care about power, I just need to work with torque for my personal work. But I've just noticed the omission makes people interested to the subject.
The only disappointment is that nobody yet noticed that no torque peak on my graphes reaches the ones specified in LFS garage
Are you sure the shift light can calculate turbo spool up time and engine moment of inertia for the instant you'll shift to the next gear? There are many other parameters involved, not just the 'generic' dynochart shape, which actually is altered all the time while racing, especially with tubocharged cars.
Last edited by Bogey Jammer, .
That's I want to study, the truth between 2nd and last gear. I have not found documentation about this, and I don't believe that gear spacing must be set up just to avoid shifting in corners. Many other parameters may be taken in account like acceleration, available pressure from turbos and torque range. I also want to know when the last gear should not be used.
I'm just curious about this, not only want to be faster. I also want to make it predictable since I have very short free time to play LFS, so I'm making programs to build automatically my setups as quickly as possible, relying most on theory with the help of very few but accurate enough driving feels, instead of full feeling corrected setups like most of people do.
Believe in god or not, that's why I need these charts. Why not building my own rocket? Everything I learn like this can be useful in general life. Another reason is that I'm fed up to wait for other people to make that I want, so I try to create it myself. And... I don't thrust the red shift light very much
Why not, but I'm don't really want to be the torque curve factory for the LFS community, especially for each car class created. Let me think about it for a while.
Sorry about the chosen colors, I knew that some people would have difficulties to read it. But this picture is for illustration only, the values are in the tables.
Many people who asked this data will probably study gear spacing.
Many people who asked this data will probably study gear spacing.
LFS cars dynocharts here
Finally here are all LFS cars torque curves charts.
I've polished my data logging technique and think that the data is reliable and accurate enough to publish them:
Download here
Sorry I've not added labels on curves, it's too painful with Excel®©™, and there is too few contrasted colors to choose for so much curves.
Feel free to discuss, comment, post your own work...
May this thread be dedicated to this subject.
I've polished my data logging technique and think that the data is reliable and accurate enough to publish them:
Download here
Sorry I've not added labels on curves, it's too painful with Excel®©™, and there is too few contrasted colors to choose for so much curves.
Feel free to discuss, comment, post your own work...
May this thread be dedicated to this subject.
Last edited by Bogey Jammer, .
Reason : added power curves
An optimal known torque curve is ideal to setup gear spacing quickly. I agree you have to deal with turbo lag and moment of inertia while racing but predicting these parameters requires engineer skills and lot of data of the car and engine design. I think I already look like a geek about all of this, most people don't need it to enjoy the game. Work with a filtered torque curve is a lot easier than with turbo lag and inertia, I wonder how you will succeed with this mess.
My values are attached in the rar file above. I had ≈338 N.m at rpm peak ≈4000rpm. Garage shows 345N.m @3896rpm, not bad but more measurement accuracy would be great.
andrand I said that aero drag is not involved in the shapes in your charts, and the curves are not scaled to show engine torque output.
Moreover as mentioned many time before, you are obviously affected by turbo lag and moment of inertia, that's why you've got delayed peaks especially at lower gears. The force peak at 5800rpm for the XRT means that there's indeed a big problem. Please read and understand what was already said about this method, give this aero drag shit up.
andrand I said that aero drag is not involved in the shapes in your charts, and the curves are not scaled to show engine torque output.
Moreover as mentioned many time before, you are obviously affected by turbo lag and moment of inertia, that's why you've got delayed peaks especially at lower gears. The force peak at 5800rpm for the XRT means that there's indeed a big problem. Please read and understand what was already said about this method, give this aero drag shit up.
The XRT dynochart
Here is the XRT dynochart shape:
Longitudinal acceleration filter helped a lot, but maintaining the car at 0±0.01G to log valid points is hard and measurement takes several minutes.
On the pic you can notice that the turbo boost can't be maintained at full below a critical engine angular speed, and produces cloudy curve below it.
Longitudinal acceleration filter helped a lot, but maintaining the car at 0±0.01G to log valid points is hard and measurement takes several minutes.
On the pic you can notice that the turbo boost can't be maintained at full below a critical engine angular speed, and produces cloudy curve below it.
Last edited by Bogey Jammer, .
A twin-turbo? which one?
About the torque curve glitch I think it is affected by engine's moment of inertia. Torque output is lower when accelerating and higher when decelerating. See how LFS physics are detailed! I wonder how far the devs modeled the engines.
The evidence of this phenomenon can help to setup gear spacing, but I now doubt that real life dynometers make accurate measurements, more than engine they must take the moment of inertia of transmission and wheels as well.
Concerning my program I'll add a longitudinal acceleration filter, then my curves should be good.
About the torque curve glitch I think it is affected by engine's moment of inertia. Torque output is lower when accelerating and higher when decelerating. See how LFS physics are detailed! I wonder how far the devs modeled the engines.
The evidence of this phenomenon can help to setup gear spacing, but I now doubt that real life dynometers make accurate measurements, more than engine they must take the moment of inertia of transmission and wheels as well.
Concerning my program I'll add a longitudinal acceleration filter, then my curves should be good.
The first test: not very good but not so bad
First test with the UF1000:
:wtf2: the curve is not perfect, what a disappointment :melting:
My magic torque value is influenced by longitudinal acceleration. When I apply or release brakes, or initialize a turn, the value is altered. Note that the throttle is always full opened when a new point (green dot) is created. I modulate engine revs with brake action.
It stills surprise me, LFS may have quite complex formula to compute the torque at engine output. I'm still investigating it but tbh I don't have an explanation to this yet.
:wtf2: the curve is not perfect, what a disappointment :melting:
My magic torque value is influenced by longitudinal acceleration. When I apply or release brakes, or initialize a turn, the value is altered. Note that the throttle is always full opened when a new point (green dot) is created. I modulate engine revs with brake action.
It stills surprise me, LFS may have quite complex formula to compute the torque at engine output. I'm still investigating it but tbh I don't have an explanation to this yet.
@AndRand: Check this topic, full of good words
The turbo lag delays the maximum theoretical available torque when you accelerate. The boost is not instant because the turbocharger has roll inertia and needs time to spool up (to around 300 000rpm depending of the device size). Twin-turbo layouts are designed to reduce this phenomenon by using two small turbochargers instead of a heavy lonely one.
You must not try to remove the aero drag force from your chart because it does not affect the tire longitudinal force value. This is actually the force developed by the tire surface on the ground against everything else, including aero drag. BUT you have to filter the tire resistance, which depends on pressure, car static weight and weight transfert, wear on the road surface, deformation constraints, AND AERO LIFT FORCE which pushes the car (or pull for street cars) on its tires depending on the speed, thus changing tire deformation.
The turbo lag delays the maximum theoretical available torque when you accelerate. The boost is not instant because the turbocharger has roll inertia and needs time to spool up (to around 300 000rpm depending of the device size). Twin-turbo layouts are designed to reduce this phenomenon by using two small turbochargers instead of a heavy lonely one.
You must not try to remove the aero drag force from your chart because it does not affect the tire longitudinal force value. This is actually the force developed by the tire surface on the ground against everything else, including aero drag. BUT you have to filter the tire resistance, which depends on pressure, car static weight and weight transfert, wear on the road surface, deformation constraints, AND AERO LIFT FORCE which pushes the car (or pull for street cars) on its tires depending on the speed, thus changing tire deformation.
Last edited by Bogey Jammer, .
Andrand, check Bob Smith's achievements on this kind of method, you seem to not take in account some parameters mentioned by Bob, turbo lag for example.
About my theoretical method, I grab the actual pure engine torque value, I don't need to post-process it. My current task is to write a program to plot it, and it would be more difficult than expected.
About my theoretical method, I grab the actual pure engine torque value, I don't need to post-process it. My current task is to write a program to plot it, and it would be more difficult than expected.
If you look at his past threads, you can notice he has a psycho problem, and he's almost always indifferent to other's comments.
Just forget him
Just forget him
It is possible, I'm creating it now.
I've just learned in 5 minutes how to read an application's process memory, and found the actual engine output torque value, expressed in a float value in N.m. The value depends of clutch state and temp, and engine damage. When applying full throttle it is really the max torque@rpm specified at garage. The value is negative while engine braking.
Tracing the charts with memory reading would be the most accurate method (except getting the engine algorithm from the devs...).
Tracing the charts with memory reading would be the most accurate method (except getting the engine algorithm from the devs...).
- Problem: this value's address is always different each time you start the race, so a search is needed each time.
- Problem: the common memory hacking softwares can't log the desired values continuously, so a specific tool must be developed. For my case, I want to do it but I have skills with the Python language only, and I don't think it is memory reading capable.
- I always thought that looking for such game values was a matter for geeks only. I realized it is very simple and I'm now frustrated that I could do that long time ago :doh:
Last edited by Bogey Jammer, .
Wonderful ! 
This seems to be very accurate
This seems to be very accurate
Even when the race starts?
My old method for dynochart generation with raf file, maybe it would inspire some peoples:
The rear tires don't touch the ground.
Engage a gear, apply handbrake and just press the long pedal.
Extract engine rpm vs time from the raf file, derivate it and voilà.
But the results for the XRT is very disappointing:
The max torque is not at 38xx rpm. I think the wheels acceleration is a lot too fast to erase the turbo lag on this chart...
Attached the layout I made for the operation.
The rear tires don't touch the ground.
Engage a gear, apply handbrake and just press the long pedal.
Extract engine rpm vs time from the raf file, derivate it and voilà.
But the results for the XRT is very disappointing:
The max torque is not at 38xx rpm. I think the wheels acceleration is a lot too fast to erase the turbo lag on this chart...
Attached the layout I made for the operation.
Last edited by Bogey Jammer, .
The curves I need must be processed by my program, VHPA's ones are estimations and I don't want to recreate the curve generation algorithm. I've also noticed a deviation between LFS and VHPA results, so I would like something more accurate.
BTW any replay analyzer can't generate torque curves...
PS: Breadfan, are you talking about this?
BTW any replay analyzer can't generate torque curves...
PS: Breadfan, are you talking about this?
Last edited by Bogey Jammer, .
Torque curve determination project?
Hi all
Like probably many other racers, I would like to get the LFS cars dyno charts. They would be used for precise gear setup creation. I personally started a custom program which could calculate gear spacings with my own criterion, but without these damn torque curves :gnasher: .. frustration.
I havn't succeeded to estimate them or extrapolate them from .raf files because of the turbo lag, so I imagined a (crazy?) method:
Like probably many other racers, I would like to get the LFS cars dyno charts. They would be used for precise gear setup creation. I personally started a custom program which could calculate gear spacings with my own criterion, but without these damn torque curves :gnasher: .. frustration.
I havn't succeeded to estimate them or extrapolate them from .raf files because of the turbo lag, so I imagined a (crazy?) method:
- We already know two points of the torque curve: max torque and max power.
- Is there another accurate parameter? the red shift light. It is switched on every time the current engine torque is lower than the one that would be generated on the next gear at the same gearbox output shaft rotation velocity. And since it's calculated by LFS it is always right, no estimation (I think).
- So, when the red light switches on, we can deduce from the gear ratios the equivalent lower rotation velocity with same engine torque.
- But until then on the dyno chart we only have rpm coordinates, not torque ones. So we need as much "mirrored" rpm values as possible to get the curve shape, probably hundred or thousand equally spaced on the torque axis, that makes the idea crazy...
Hi,
All the screenshots are scaled to 100%. Click on the thumbnails to bring them full size.
All the screenshots are scaled to 100%. Click on the thumbnails to bring them full size.
Well, corrected manual upload done. I should have more positive replies now. Sans Plomb doesn't have very much success atm, probably because people don't get it to work because of the manual mistake.
oops
:doh: Ok, ok...
I just noticed all this problems came from me..
In the LFS shortcut, put the LFS receiving port (800) instead of 801
The manual is wrong, I'll reupload a corrected one.
Despite of this, Sans Plomb is working fine
I just noticed all this problems came from me..
In the LFS shortcut, put the LFS receiving port (800) instead of 801
The manual is wrong, I'll reupload a corrected one.
Despite of this, Sans Plomb is working fine
Last edited by Bogey Jammer, .
Reason : big mistake
TBH I don't use any other InSim app, but try Virtual LFS Dashboard, it looks simple to use.
Try to run both SP and LFS on the same machine with default port settings and 127.0.0.1 as IP address everywhere.
This should work and if not, try another insim application to check if you can connect with LFS.
This should work and if not, try another insim application to check if you can connect with LFS.
I can't tell what's wrong, you can initialize InSim just with defaults settings on the same machine...
How do you activate the InSim port listening in LFS exactly?
How do you activate the InSim port listening in LFS exactly?
FGED GREDG RDFGDR GSFDG