Hi All,

We were working on a DIY locdecoder, which works rather nice, but only with MM2 commands.
They were used to do a low cost upgrade for old locs.. see our example rejuvenating an oldie
Recently we adapted the software to be used with the basic MFX commands, like speed and functions.
In order to add the feedback for programming and other extended communications, we would like to know if someone already did some research on the RDS topic, as it must be possible to do it software with minor external components.

Also should someone be interested in the progress we made, questions are welcome.

thanks,
Walter

@perz we exchanged some info a few month ago, after seeing your site I was already convinced that using a transformer was better than a resistor.
I took an interference suppressor coil from an old PC power supply and removed a coil counting the loops, then added a 1.5 mm² as primary.
Result is a 1 to 25 transformer which gave good signal.


For now I used a simple passive RC filter, but I ordered some RDS chips to further examine the signals after filtering 52.632 KHz

the final goal is to implement some encoder software with filter and modulator in the locdecoder

more info later

Sorry, http://www.persmodelrailroad.net is down for the moment due to change of ISP. But all the content is also available at http://www.persmodelrailroad.com. I have edited my previous post.

Hello Walter,
a great project to have such a basic decoder to reuse old AC motor locos without converting the motor to a DC motor! I would be interested in testing one...

Moritz

Hello everyone,

I am reviving this old thread, hoping that someone has made progress on the topic of RDS.

I successfully developed an MFX command station directly using the track protocol and a program running on the ESP32 (WiFi).

Here is a demonstration video:

The program is available here: https://github.com/BOBIL...ristophe/directMFX_ESP32

However, I am not very skilled in handling the feedback (RDS) from decoders, which is why I found this thread particularly interesting. Unfortunately, it seems incomplete as no electronic schematics were provided.

Could anyone help me with this?

Thank you in advance,

Christophe

Hello,

Thank you for your response and for pointing me to Stefan Krauss's documentation. I am indeed familiar with his work, which has been instrumental in developing my MFX command station. However, while the document mentions RDS and describes its theoretical operation, it does not provide detailed information about the electronic implementation.

Specifically, I am looking for a clear electronic schematic or practical guidance on how to connect an RDS feedback circuit to an ESP32. For instance:

How to handle the PT2579-S chip for RDS decoding?
How to process the returned signals to ensure compatibility with an ESP32 input pin?
If you or anyone in the community has more information or examples of such an implementation, it would be incredibly helpful.

Thank you again for your time and support!

Best regards,

Christophe

Hello and thank you for your reply.

First of all, it should be noted that Märklin still uses the PT2579 component in its command stations, including in more recent models like the MS2 (I opened an MS2 to confirm this).



When searching for a dedicated component, my goal was to avoid the need for software-based filtering. I am using an ESP32, which is not particularly optimized for complex DSP tasks. Indeed, for operations such as intensive real-time filtering or frequency analysis, its performance can be somewhat limited.

Additionally, the processor of the ESP32 in my command station is already heavily utilized by the track signal encoding, which relies on a 50 µs timer. This leaves little room to efficiently handle a fully software-based solution for processing feedback signals.

That being said, I appreciate your suggestion. Exploring a software approach remains an interesting option, and it could be feasible if I were to add a dedicated microcontroller, such as an ATmega or an ATTiny, to the command station to offload this task.

For now, however, I am prioritizing the use of a dedicated component, such as the PT2579, TDA7330, or TSA5511, which could greatly simplify the process.

Thank you in advance to anyone who can provide additional information or share their experiences.

Christophe

Here’s where I got to with RDS testing before stopping to think about a move to DCC. It’s an old school analogue approach to the problem.



MFX feedback frequency is 52.63kHz (4Mhz/57).

PLL components were selected for a capture BW of 12% for inputs >=200mV p-p at the PLL input pin, this is the nominal operating condition for the PLL. Measured BW in practice at roughly 7kHz. 10k trimpot is adjusted to centre the response – see below.

BW is narrower at lower levels with about 70mV p-p being the lower limit for locking for both sine and square waves, BW is a little over 2kHz at this level. The high pass preamplifiers give a lower lock limit of <5mV p-p at the reduced BW and the full 7kHz BW for inputs >10mV p-p.

High pass filtering is used to couple each stage with each set for an Fc of around 36kHz. Signal is not attenuated at 80kHz, about 2dB down at 52kHz, about 20dB at 10kHz. A much fancier filter could certainly be used but I find them unpredictable with transient waveforms. Pick up is by toroidal current transformer.

---

To align the detector circuit, if you don’t have a signal generator, you can use a UNO’s hardware timer 0 to generate a precise RDS carrier frequency. Use a couple of resistors to drop the signal to a useful level. Also, using a bridge, MOSFET and a resistor you can superimpose an RDS carrier signal on the Marklin track current for testing.

Chip datasheets and .ino here

https://mega.nz/folder/U...A#xyaaqkKX0ypSpaHtTKZENw