I've got a KNX-like bus system in my house. (Specifically it's ABB/Busch Jaegers free@home, which I think is based on KNX, for this post I will keep referring to it as bus or KNX bus) This bus system carries power and communication for bus devices over the same two wires. It's nominally 24V DC, and in my real-world setting 28V.

I would like to tap into this power source to locally generate 5V. This would be used to power an ESP32 module with display, set into a wall box. Therefore, the step down module must be quite small.

I bought a couple of step down modules based on MP1584EN (https://www.openimpulse.com/blog/products-page/product-category/mp1584en-mini-dc-dc-step-module/). When I hook these up to the KNX bus, they produce a nice 5V, and the ESP32 module runs ok.

However, apparently the MP1584EN generates so much noise on the input line that the communication on the bus is disturbed and other bus participants cannot function.

My question: what can I do to extract power from the bus to power a 5V module (I probably need around 70 mA), while not disturbing bus communication.

I've hooked a simple scope I had access to to see what the communication looks like. It looks like to send information, you pull down the bus voltage by abound 8V for 50 ms at a time. These are images of the way the bus operated normally. Scope hanging on the wall communication

I didn't really succeed in capturing a good image of the noise generated by the step down module (this may be worth trying again), but I notice that when things operate normally (i.e. no step down module connected), on average only every few seconds a message comes by, whereas when the step down module is connected, there are lots of 'messages' each second (retransmissions?).

I have tried putting a diode and a 50ohm resistor in series with the step down module, and also a 47 uF cap across the input pins of the step down module, but that didn't help. (Why 47 uF? Because that was the only 50V cap I had...)

Any ideas?

  • \$\begingroup\$ Maybe a choke (instead of 50R) to isolate? \$\endgroup\$ Aug 15, 2021 at 14:37
  • \$\begingroup\$ Is it normal for bus-connected devices to draw power from the data lines of the bus? Does it not have a separate power line? \$\endgroup\$
    – user253751
    Aug 15, 2021 at 14:49
  • \$\begingroup\$ @DanielChisholm: Thanks for this tip, it lead me down a path to learning about chokes, and also towards finding this repository where they use a 150mH choke to accomplish the same. Unfortunately I haven't been able to find a webshop that sells such a coil to consumers yet. \$\endgroup\$ Aug 18, 2021 at 10:35
  • \$\begingroup\$ @user253751 This is a bus system with two wires, which in the meantime I've found commonly to be considered a 'one-wire'-type bus, so power and data share a wire. I understand where they're coming from but still I see two wires... ;-) \$\endgroup\$ Aug 18, 2021 at 10:36
  • \$\begingroup\$ I've had some success with putting a 10 mH coil in the positive line (together with the diode, resistor and capacitor I already mentioned in the post). It's still so noisy that I'm having occasional problems on the bus (commands not being received) but at least it's not totally unusable. I then had to devote my time elsewhere so that's still where I'm at. I probably need a higher rated coil (the link I posted mentioned 150 mH) but I haven't been able to find one yet. \$\endgroup\$ May 13 at 13:54

1 Answer 1


You could use an inductor in the feed line to your DC-to-DC converter or you could use a gyrator (something that mimics an inductor) and also drops a few volts. I mention dropping a few volts because your DC bus may be 28 volts but, with up to +/- 8 volt ripple on it from the data, you should probably assume that at any one moment in time you may see a peak voltage of 36 volts and a valley voltage of 20 volts. The 36 volts exceeds the input voltage rating of the converter you bought so, a gyrator will serve two purposes.

So, you should aim your gyrator to provide a DC output voltage of about 18 volts (and no more). You can use something like this: -


simulate this circuit – Schematic created using CircuitLab

The equivalent AC impedance presented to your bus line will be about 1600 Ω or: -

R1 || (R4 x β of transistor).

That impedance should normally be OK for not disturbing the data too much. You may need to raise C1 if the data rate gets quite low. I mention this because you didn't state what it is.

  • \$\begingroup\$ I'll give this a go; it will take some time to source the components. \$\endgroup\$ Aug 18, 2021 at 10:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.