Frimware glitch / problem with TCS decoders.
- I noticed that if you set CV 5 or CV 6 to anything other than zero, then the speed of the loco does NOT increase with increasing speed steps, but has a "glitch" approximately ever 4 speed steps.
If you set cv 5 nonzero, and then step up from SS 0 by increments of 1 SS, (use 128 ss, it's easier to notice) you will hear/see the loco increase speed with every step, except at the "glitch points", where increasing the speed by one step actually results in a lower speed than the previous step.
The glitches I noticed were as SS 5, 9, 14, 18, 23, 27, 32, 36, 41, 45, 50 (did not bother going higher)
I found this out because I was working on a different problem, in a Marklin Mikado, increasing the speed near max, results in a runaway at full speed, uncontrollable. It appears the motor noise masks the ability of the decoder to receive commands.
By setting CV5 to about 70, this tendency was almost completely eliminated, and when the "runaway" occurred, usually a second speed command brought it back.
I surmise the first problem above is caused by bad coding, a math error in the firmware. Thus I reasoned a custom speed table would solve the "speed step glitch". It did.