OK - this one is going to get a bit technical, and I am first and foremost wondering if anyone else has seen this behavior. Maybe it is simply "yeah that is what happens, work around it", but I want to know.
So -- here is the setup. I received, last week, an Accutrax II Speedometer. I requested it for my birthday (and it was delivered). So, I was all set to start adjusting speed tables in the different locos to try to set up all my roster so that like could easily be consisted with like. Of course, I have lots of locos, with many different flavors of decoders, and knew this would be a slow project.
As I started off, the first thing I noticed, of course, was just how fast Kato engines run at full throttle (N-scale). No problem, that is what the speed table or the Vmid - Vmax settings are for. Started with the first loco (Kato E9A with Digitrax DN163K0A installed. Easily found a Vmax setting for a reasonable maximum speed. And, with setting a Vmid to one half of Vmax, got a fairly nice straight line response (didn't play too hard at the bottom, but then, this is Digitrax, so slow speed control is problematic, anyway). So, I figured, lets go to the 28-step speed table, to help tweak the performance. And this is where we entered the Twilight Zone (see graph).
The chart shows the scale MPH against throttle settings. I used 3 different throttles: JMRI throttle set for 28 steps; JMRI throttle set for 128 steps; and D402 set at 128 steps. I set Step 28 the same as Vmax (84); Step 14 the same as Vmid (41) and Step 1 the same as Vstart (4).
While I am not sure why the JMRI 28-step throttle runs a bit high in the mid range of basic motor control, we see that all three using Vmid/Vmax basically react the same - as well as JMRI-28 step using Speed Table. However, the two throttles using 128 speed steps is a dramatically different curve, that flattens out the top 20% of the throttle range!
[ Guests cannot view attachments ]
I used the LocoNet monitor in JMRI to watch the commands, and speed steps are being sent correctly on the track. So, OK, I thought, maybe it is the decoder. While I don't have a pretty chart yet (wanted to post this, first, for comments and thoughts), I tried locos with Lenz and TCS decoders, and they have the same behavior! In 128 speed step mode against a speed table, the top 20% of the throttle is flat!
I had never really noticed this, since I had never done much worry over speeds (just matched overall performance of the locos I wanted to always consist). And, I never ran at the top of the throttle before. It was after spending a lot of time looking at this, that I decided to start the lab experiment of actually charting the speeds at lots of different throttle steps and settings
My first thought was "I wonder if everyone has an algorithm error, using speed step 100 as the denominator for calculating the voltage", but even when using a 28-step throttle, the command (at least on the LocoNet Monitor) is showing the corresponding 128 step value
Is this behavior something that is known? It is easily worked around, but is very disconcerting. By rights, all the lines of the chart should be superimposed (with only small deviations).
