I faced a strange behavior of a DALI controlled LED luminaires circuit. I have 48 LED luimianires with a MeanWell ELG150-C700-DA drivers connaected according to the scheme below:enter image description here I commissioned the system and it works OK but sometimes I can't switch some luminaires OFF via the DALI commands. When I send DAPC 0 or OFF command this luminaire go OFF, but in a second the go back to 100% of their output. Besides that this problem is floating, because in a different days different luminaires start to work in a such way. Moreover sometimes ALL luminaires go ON spontaneously.

I doubt that the problem is in LED drivers because when I tested some other drivers from the same bunch, everything was OK. I think that something is wrong with the DALI bus installation quality, but I don't know from where to start the thoubleshooting.

I am a novice in a DALI theme, could you advice me what parameters I should check to find the root of the problem?

Here is a DALI signal at the beginning of the DALI bus near the 905 routerenter image description here

By the way why the signal amplitude for 2 DALI packets is different?

  • \$\begingroup\$ Do they work on a short line? Do they work if only a few are connected? \$\endgroup\$
    – Andy aka
    Commented Sep 14, 2018 at 13:51
  • \$\begingroup\$ I checked several single luminaires connected directly to the controller with a 2 meters cable and they worked OK. \$\endgroup\$
    – Ilya
    Commented Sep 14, 2018 at 14:07
  • \$\begingroup\$ What is the spec of the control panel? If they are taking more than 52mA each, you would exceed the 250mA limit of the 905 router. \$\endgroup\$
    – Martin
    Commented Sep 14, 2018 at 14:37
  • \$\begingroup\$ What type of cable have you run the system on and is it run in such a way as to provide any shielding, such as through metal enclosures, metal conduit or armored cable? How long are the average runs and how is the bus connected (I don't know the standard, so this may be part of it), such as a line, a tree pattern or a loop? In one of your answers about volt drop being a potential issue you reply with the cross sectional area of your cable, but this has no meaning without knowledge of the cable length. Depending on frequency, a $20 oscilloscope may be in order. \$\endgroup\$
    – K H
    Commented Sep 15, 2018 at 0:34
  • \$\begingroup\$ The NYY-J cable is used. It has no any shielding. The maximum cable run length is 150 m, total cable length is about 250 m. Electrical circuit has bus topology. It shouldn't have any loops, but I am not sure about it. \$\endgroup\$
    – Ilya
    Commented Sep 17, 2018 at 7:50

2 Answers 2


DALI requires less than 2V maximum difference between the highest and lowest points on the line. To achieve that you must use sufficient cable gauge and a power supply with sufficient current limit. The luminaires will go on "spontaneously" if they see a system failure condition, which is less than minimum DALI voltage for more than 500ms. For up to 150m you need to use 0.75mm squared CSA.

You may be overloading the PSU in the 905 router. Although it has a current limit at 250mA, which is the maximum allowed by DALI, you have 48 luminaires plus 3 control panels. The luminaires (assuming one DALI driver per luminaire) are restricted to 2mA each by the standard, but the control panels do not have the same restriction; they are allowed to be up to 250mA each (although not a good idea to go that high). So measure with a current meter in series at the router end what you are drawing during non transmit times, and look at the DALI signal with a scope to see if the high levels are dipping below 9.5V.

  • \$\begingroup\$ Martin, thank you for quick reply. I haven't check voltage drop in a bus yet, but I used 1.5 mm squared cable. \$\endgroup\$
    – Ilya
    Commented Sep 14, 2018 at 14:05

How are you at debugging any communication serial port? for BER (bit error rate) integrity testing?

Data "0" = -4.5 to +4.5V ( 0V nom) new +/- 6.5V Data "1" = 10.5V to 22.5V (16V nom) new 9.5 to 22.5V

This is to allow "lots" of ringing when sending 0 to 16V. But high power RF bursts on long cables can also induce errors.

This enables reliable communications in the presence of significant electrical noise. But noise spikes can create false triggers and possible command data.

Detection of the problem may be hard if you have no tools or EMI skills.


If transient power or radiated transients are suspected, try an AM radio between channels in a quiet band. Then wait for the event.


Put AM radio near AC line with a lamp to indicate power level.

The ingress of noise ( input)depends on location near-interference and length of the cable. So isolation of an offending zone depends on isolation methods.

Debug tools

If you have access to DALI test tools, , then it is easier to test the 9600 Baud Bit error rate (BER) in a continuous mode in half duplex feedback.

not a testimonial, just an example

My hunch

RF interference from transmitter bursts.

Potential fix, choosing RF cap of right size that does not reduce slew rate of data. Maybe 100 pF to 1 nF across node connection.

Yet shielded pair is about 100pF/m so a 150m cable is 15nF with a low impedance driver (50 ohm?) My guestimate is that > 100nF may cause slew rate limiting or 6x 150m, but noise can affect this range.

Each Rx can load of 2mA which at 22V is equivalent to 11 kOhm . The signal is biphase-M (Manchester)

  • \$\begingroup\$ Not quite. IEC62386-101 Ed 2 has Low level voltage at receiver -6.5V to +6.5V, High level voltage at receiver 9.5V to 22.5V. \$\endgroup\$
    – Martin
    Commented Sep 14, 2018 at 14:00
  • \$\begingroup\$ I wonder if hysteresis thresholds were changed \$\endgroup\$ Commented Sep 14, 2018 at 14:02
  • \$\begingroup\$ 4.5V was and is the transmitter maximum low level, back to the EN60929 version. Not sure where you are getting the 10.5V from, minimum high level was 9.5V in EN60929, IEC62386-101 Ed 1 had 9.5V at rx and 11.5V at tx. Changed to 9.5V at rx and 10.0V at tx in Ed 2 unless you have an integrated PSU in which case 12.0V. \$\endgroup\$
    – Martin
    Commented Sep 14, 2018 at 14:20
  • \$\begingroup\$ Tony, thank you for such comprehensive reply! Unfortunately I nave nor experience at debugging communication serial port nor DALI debug tools, but I'll try to use capacitors. \$\endgroup\$
    – Ilya
    Commented Sep 14, 2018 at 14:23
  • \$\begingroup\$ What cable type is used? THis signal has uncontrolled termination impedance (10k is high) so ringing is pervasive. A Scope to check for signal integrity is best tool. \$\endgroup\$ Commented Sep 14, 2018 at 21:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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