Ken Henderson wrote on Sun, 09 October 2016 15:40
> Keith,
>
> Very nice work!
>
> I'm curious about your chart of Ride Height Data. I presume that's the raw
> data from the ride height sensors. Otherwise I'd expect to see only binary
> signals to the valves.
>
> How much delay (hysteresis) do you have built into the system (like GMC's
> 4s.-18s.)? Is that operator adjustable as a configuration setting? I ask
> primarily because it seems to me that the hysteresis needed depends upon
> the characteristics of the air bags and the available pressure and volume
> from the compressor. If the system is set up for bags requiring high
> pressure and a slow compressor, the delay would presumably be shorter than
> for a low bag pressure/fast compressor vehicle. A setting suitable for the
> former might make the latter too sensitive to dynamic loads.
>
> Do you intend to preserve GMC's long delay or to shorten it for "real time"
> operation?
>
> Still Curious Old Cat. :)
>
> Ken H.
Ken, yes that raw analog data from the sensors, The controller only uses the sensors in travel mode, in manual or camp they are ignored ( except for
the low limit feature I'm thinking about adding ) It's only there for curiosity, I did spend a lot of time looking at charts and state change events
if you are really a sucker for punishment
there are no user settable parameters, everything is fixed.
Except low limit, travel height and the horizon level values, they are calibrated.
If I did feel the need to vary the value the system would measure the system response time and do an auto calibrate.
Theres no way the average person would be able to do that without mucho training
But the algorithm is quite complex, it's running a state machine, and multiple filters, both FIR and IIR.
When it's close it actually pulses the valve to get right on.
When you first put it into travel it uses a fast filter, then once it reaches ride height it goes to a very very slow filter. This allows it to come
up to height fast then not adjust on cloverleafs or grade changes. the slow rate is in the minutes range.
It also has different reaction times if it's filling vs if it's already at ride height.
It's control like this that make using a micro better than a mechanical switch!
The hysteresis is much better than the stock system, I measure it in ADC counts and don't have a good correlation to inches, I need to do that.
It's +- 10 ADC counts on a 10bit ADC, but the height is measured as sin(adc) 0' to 90'
someone do the trig for me ha ha
The pressure in the bags shouldn't make any difference, but I do want to test on a sully bag and a 4 bag to see if anything needs to be tweeked.
I really don't think the system in use will have any effect on the algorithm, the only delay Ive seen is the delay caused by the air lines.
If you run big valves and big airlines the residual pressure in the line could cause overshoot/undershoot as the valves open and close but the air
line still has to equalize.
This system is targeted directly at the GMC it's not a general purpose solution.