Discuss all things tuning in this section. News, products, problems and results.
Been rushing to get my car (Project 7dere) set up for WTAC which is now only 4 days away and had some questions on the Haltech implementation of boost control.
At the shakedown at Sandown last week (in which the car suffered a minor engine bay fire, but that's another thing), I noticed that when coming onto the throttle at high rpm after already being at part throttle (eg, at corner exit) the car would accelerate very lightly then BAM suddenly would get a surge of power after a second or so. It felt too long to be a transient throttle lean/rich spike.
In investigating this, I noticed in my logs that the boost solenoid output would sit at 100% duty cycle (Spool assist), then drop down to 46% duty (base duty) and wait a second or so before the proportional component started increasing the duty cycle.
Obviously, this means I really need to better tune the base duty cycles, but as far as I can tell, the delay in closed loop operation must be due to the "Controller Start Delay" table for boost control. This has been left at the base map values which are about 1s at 4000rpm and taper off to 0.5s at 7000rpm.
My question is, what is the best way to utilize this table? I understand it's trying to wait for things to stabilize a bit before activating the close loop boost control to prevent integrator wind up, but isn't that sort of already being done via the Control Point Offset (that prevents the integrator from building up too early)?
What are some good values that people use for this table in a fast revving car like a rotary (understand that dyno tuning is the best way to tune this sort of thing, but won't have any dyno time between now and WTAC Practice Day)?
Excerpt from logs attached in a PNG - bottom chart has the boost control output. Also attached the full logs.
My personal preference with these kinds of delay timers is to turn them off completely and only use them if you can't get good enough control with the normal feed forward (duty cycle table) and feed back (PID gains). Only make it as a complicated as is necessary. So for example, tune the open loop table and bake all the 100% duty "spool assist" into that, use proportional gain only if you can get away with it, turn off delay timers, etc. Bake as much into the feed forward/open loop control as possible and ask closed loop to do as little as you can get away with. So for example, if you have feed forward gear based compensation, air temperature based feed forward compensation, rpm or throttle based feed forward stuff, etc - get as far as you can with that stuff before dialing in all the closed loop bells and whistles.
Clearly Haltech doesn't have this exact philosophy hence the various delays and activation criteria for closed loop not only being in the code but being tacitly recommended as you have seen. Now just because I don't do it the same was as Haltech doesn't mean that I'm right and they are wrong; it's more like two different preferences.