991 Tune Issues / Concerns ... ?
You are over complicating the problem I think. The issue is to map the response of the car with the ECU over a range of operating conditions and sensor inputs and it only needs to be the group of sensor inputs that control what you are looking to change at the output. Before the piggyback is ever programmed it acts as a pass through and data collector during the test runs. The closed loop response of the system is measured under different performance regimes and with this data an I/O map of inputs to outputs can be created for a given performance profile. Now it is possible to begin to make modifications to the appropriate inputs by applying a filter to them. This is simply another control box in front of the ECU. The final result is a new closed loop response from the car. The closed loop response now includes the piggyback and the ECU. What is going on in the ECU doesn't matter since you already have mapped / estimated the transfer function of the car with the ECU. Now you have created a new transfer function with the addition of the piggyback in the loop.
It's exactly how you design a control for a system with unknown parameters. You model the transfer function and apply a control according to how well the system response fits the model. When the control fails certain criteria, a new model is generated and a new control is designed. This is called an adaptive control system and they are fairly common in autos and airplanes I believe. Of course it may be the case that instead of a new model being generated it may be selected from a group of models and the appropriate control subsequently applied until the control criteria are satisfied.
I don't know for sure if this is what the piggyback people are doing but I suspect that it is close to this in practice. It might be more hit and miss and there might not be as much real control system analysis involved but I suspect that the net effect is the same.
It's exactly how you design a control for a system with unknown parameters. You model the transfer function and apply a control according to how well the system response fits the model. When the control fails certain criteria, a new model is generated and a new control is designed. This is called an adaptive control system and they are fairly common in autos and airplanes I believe. Of course it may be the case that instead of a new model being generated it may be selected from a group of models and the appropriate control subsequently applied until the control criteria are satisfied.
I don't know for sure if this is what the piggyback people are doing but I suspect that it is close to this in practice. It might be more hit and miss and there might not be as much real control system analysis involved but I suspect that the net effect is the same.
There is currently only one tuner that can flash your ECU unit. And that is GIAC. Others offer piggy back units which trick the ECU by providing altered sensor data. To maintain all adequate safeties, it's probably best not to trick your ECU into doing something.
To flash your ECU, you will need to pull the ECU and do a bench flash. Currently, I am not aware of any tuner that can provide a flash via the OBD port...yet.
To flash your ECU, you will need to pull the ECU and do a bench flash. Currently, I am not aware of any tuner that can provide a flash via the OBD port...yet.
Agree with you!
That was a good reading, but in the end what I'm really missing is:
Does in the end the ECU revert to stock factory levels with a piggyback the like of a techart Powerkit for example?
I ask because in my place this is the only option to tune my TT, and I don't really want to trash 5k for something with an expiration date.
thanks for your input
Does in the end the ECU revert to stock factory levels with a piggyback the like of a techart Powerkit for example?
I ask because in my place this is the only option to tune my TT, and I don't really want to trash 5k for something with an expiration date.
thanks for your input
Thread
Thread Starter
Forum
Replies
Last Post
markski@markskituning
996 Turbo / GT2
2
Aug 24, 2015 10:07 AM




