Hi Zack, and welcome to the forum...

When a loco runs at full speed, it can certainly induce "electrical noise" into the tracks through sparks from the slider or wheels that can be interpreted as signals by decoders in other locos... I do not believe this is a problem in the 6021, but rather "noise" from other loks... You say you have 12 loks on the layout. How many are run simultaneously ? More than 3 loks running at the same time will likely sink more current than the 6021 can handle, and voltage on the track will drop - which can also lead to disturbances... At full speed, the running lok/loks consume maximum current, around 1 A with lights on (more with older loks), and the 6021 is able to give about 2.5 A...

What transformer are you using?

Just to verify, only C-sine loks go haywire?
Maybe the decoder tolerances are too narrow, or maybe decoders are kind of faulty, it has happened to other members of the forum...

Has the problem occured since you have added more locos to the layout? In that case it is probably a "not enough power" problem...

Is it the same lok/loks that go haywire every time?

The problem you report is interesting and a bit worrying. As the Webmaster points out, overload could cause disturbances. But in your case with just one loco running it could hardly be overload. It couldn't be the 6021 sending wrong codes either, since both the other locos started at the same time.

The Märklin/Motorola code format has been used for a long time and is considered to be very reliable. Each code is sent twice, and only if the decoder sees two identical packets with the correct pause between them should it react on it. Thus, disturbances causing a wrong but valid code on the track is very unlikely (provided that Märklin implemented it correctly in the decoders). And still, if it happens, it will only be one wrong code. Therefore, your problem could indicate that the C-Sine decoders have some kind of weakness in the design, making them sensitive to electric noise. That's why I think it is worrying.

If there are certain "bad" addresses it would be interesting to know which they are so that we can avoid them. The problem reported by Zack appears to be rather independent of address since it happens with several different locos. Still it is interesting to know which addresses that are used by the involved locos (both the running one and the starting ones). This could perhaps give a clue. Testing with other addresses is a good idea , but if the error only occurs 3 times in 5 weeks it will take a long time before Zack will be able to tell whether this behaviour is address dependent or not.

Thank you for responding.

Here are details of each incident:
_____________________________
Addressed: Lok "01" HAG (850 mA) w/ LokPilot at full speed
Sudden Movers: Lok "07" Marklin 39361 and Lok "68" 39560
_____________________________
Addressed: Lok "70" Marklin 39700 at full speed/interior lights on
Sudden Movers: Lok "05" Marklin 39606
_____________________________
Addressed: Lok "70" Marklin 39700 at full speed/interior lights on
Sudden Movers: Lok "08" Marklin 39357
_____________________________

I couldn't draw any conclusion from this information. We need more incidents. Are there any reported problems about the use of LokPilot in Marklin Digital?

On the other hand, Marklin digital club just offered that a similar problem was observed the 6023 unit that was developed for educational purposes. The cure was to bring each lok on the track to a zero speed every time the unit was turned on. The suggestion is to use the same cure for 6021. Alternatively one can restore all decoders’ settings from 6021’s memory using the sequence “STOP-93-97-GO” during the first 5 seconds after turning the unit on. If all loks were set to zero speed before 6021 is turned off, then using the sequence will set all loks to a zero speed before operation starts.

I will follow the suggestion and see what happens. I am not an expert in the old or new Motorola protocol, but, if this is the solution, an explanation could be that by setting each lok speed to zero, 6021 is forced to address all loks (present on tracks) in the refresh signal that it continually broadcasts to the addressed loks. If 6021 is faulty and susceptible to noise, even if it suddenly includes a wrong new address in the refresh signal, no lok on tracks will have this address programmed. If the C-sine decoder is faulty and picks up two consecutive identically wrong speed packets, it will not have time to act because it will receive the correct packet in the refresh signal from 6021. In the new Motorola protocol, a lok would get double “speed packet” every ~0.7 sec in a refresh signal for 16 loks. However, if more loks are present, it will take longer to get the correct packets. Does this explanation make sense?

It was also suggested that using a computer interface would “help a lot”. But that I could not speculate how it may help. It is my understanding that 6021 is the unit sending the refresh signal even if the computer interfaced is used (?)

I am a little disappointed and concerned about Marklin Digital. My concern is not just because a lok or two go berserk sometimes, but because accidents/damage could result. I have already lost a buffer on a stationary Roco engine after a Marklin collided with it at full speed. In another incident, the Marklin engine pushed 2 freight cars off the rails ... and the board.