2
\$\begingroup\$

Original schematics

I want to make the following DMX Splitter:

enter image description here

Credits by J. Mack, see Full article: DMX Splitter

Changes

Changes made by me (with help of Maple, Transistor and many others, thanks for that):

  • Four instead of three output channels
  • Power led
  • Signal led
  • 5 separate DC/AC converters
  • Separate 220 ohm resistors per output
  • Temperature fuse and varistor
  • TVS diodes

My circuit

The circuit I came up with is:

enter image description here Because not all components are present in DIYLC I used some conventions:

  • Light blue circles are header pins
  • Black dots are terminal boxes
  • There are 6 PCBs
  • The bottom middle PCB is overlapping slightly. On the left bottom and right bottom PCB there are two black squares, these are broken off to fit the enclosure.
  • The dark areas on the other two left PCBs are to make sure the fuses can be replaced if necessary (to have a gap to the fuse boxes)

Questions

I have a bunch of questions (hopefully I can ask them in one question since they are all related to the circuit).

  1. In the left bottom PCB there is a QuickBlow fuse, but in the original link it is a SlowBlow. I don't have these; will this be a problem?
  2. Would the HiLink HLK-PM01's have enough distance between each other?
  3. I'm putting everything in one enclosure. I know officially a double enclosure should be used, however, there is really no space for it. And I intend to put the AC220/240V signals UNDER the bottom PCB which will be screwed tightly, so there is no possible way to touch 220/240V. Is this 'acceptable'? I'm intending to use the device on stages with my music bands. The device will be under my keyboards stand, reasonably safe.
  4. The TVS types are 1.5KE18CA DO-201AD 1.5KE18 Bidirectional TVS diodes. Are these ok? I checked them, but I couldn't check the real usage.
  5. I tried to lay the wires as good as possible, although several overlap (some I can route a bit better not crossing an IC, but it cluttered up the picture more to use only straight lines.
  6. Is there anything I didn't think of (regarding safety/functionality?)

[Transaction Screenshot]

\$\endgroup\$
3
\$\begingroup\$

In the left bottom PCB there is a QuickBlow fuse, but in the original link it is a SlowBlow. I don't have these; will this be a problem?

That depends on the inrush current of the AC/DC adapters and the nature of any line faults your device might experience. Generally, as long as the inrush current isn't sufficient to pop the fuse every time you turn the device on, using a faster fuse than necessary will just mean you blow fuses more often/unnecessarily.

Would the HiLink HLK-PM01's have enough distance between each other?

This is a little dicey. You have mains voltage on perf board, which is going to be a challenge, and generally I just plain wouldn't recommend it. At the very least, you should remove as much of the unused copper around the AC connections as you can. Better would be to either use separate enclosed AC-DC converters for each port rather than PC mount, or use a single AC-DC converter and then use isolated DC-DC modules to power each port. Best of all would be to use an external DC wall wart and then a set of isolated DC-DC modules and avoid messing around with mains voltage entirely! That will be the smallest and probably cheapest solution, too.

I'm putting everything in one enclosure. I know officially a double enclosure should be used, however, there is really no space for it. And I intend to put the AC220/240V signals UNDER the bottom PCB which will be screwed tightly, so there is no possible way to touch 220/240V. Is this 'acceptable'? I'm intending to use the device on stages with my music bands. The device will be under my keyboards stand, reasonably safe.

If the case is non-metallic, then this is generally fine as long as everything is mechanically secure. If the case is metallic then it should be grounded via the ground pin on the power cord.

The TVS types are 1.5KE18CA DO-201AD 1.5KE18 Bidirectional TVS diodes. Are these ok? I checked them, but I couldn't check the real usage.

Those are 18V bidirectional TVS diodes, with a clamping voltage of 32V, so by the time they activate to clamp a fault it may be too late for your transceiver. RS485 requires that devices tolerate data line voltages of +12/-7V, so you should set your protection much closer than that. Also, TVS diodes will be much more beneficial with something to limit the current into/out of the device when they activate, generally a 50-100mA polyfuse on the data lines is used. Otherwise, a severe fault will just blow up the diodes before it blows up the transceiver.

\$\endgroup\$
8
  • \$\begingroup\$ Thanks for these elaborated answers. Guess when I do testing I will find out if the fuses are ok or not. About AC/DC converters: I thought it would be better to have isolated converters as early as possible (from the AC power). Guess this was not needed (still wondering what part you had in mind but I will do some homework first). The AC/DC converters will only be put on the perf board to be steady, I'm intending to use wires (24AWG?) directly to the AC pins of each converter so there will not go 220V through perf board copper. I'm using a 2-pin power plug (no GND/earth). \$\endgroup\$ – Michel Keijzers Sep 10 '18 at 21:04
  • \$\begingroup\$ About the TVS diodes, I need to check better, I thought if the voltage is higher, it's good enough, but guess not :-( \$\endgroup\$ – Michel Keijzers Sep 10 '18 at 21:04
  • \$\begingroup\$ Re: power supplies, it doesn't really matter where you isolate them. Most equipment with isolated interfaces will use a small isolated DC-DC converter powered from the local low-voltage supply because this is the smallest and cheapest solution. \$\endgroup\$ – ajb Sep 10 '18 at 21:12
  • 2
    \$\begingroup\$ Re: TVS diodes, the point of using them is to clamp the voltage under a fault condition to a low enough level that your device will not be damaged. At the same time, you don't want the TVS diode to start conducting during normal operation, so you have to strike a balance. Here's a fairly good overview: completepowerelectronics.com/tvs-diode-selection-tutorial \$\endgroup\$ – ajb Sep 10 '18 at 21:13
  • \$\begingroup\$ About the power supplies: I don't think a DC/DC (buck) converter is really isolated like such Hi-Link or am I missing something? (btw, I prefer using only DC voltages compared to AC if possible). \$\endgroup\$ – Michel Keijzers Sep 10 '18 at 22:16
1
\$\begingroup\$

Plastic enclosures generally are not good for devices with high-speed network signals (they do not provide shielding) and for devices with sufficient heat radiation (they block heat dissipation). As it happens, your splitter is both.

At the very least make many thin (2-4 mm) vertical slots in the sides of the enclosure and cover them from the inside with metal plate with about 5 mm between inner wall and the plate. This should provide some convectional cooling while preventing debris from getting in.

Ideally I'd recommend holes on top and bottom, but you might not want those in case of water getting onto splitter.

Also, try the following - place the splitter boards vertically with enclosure open or without enclosure at all (careful with mains power!) and run it with full load (controller and LEDs connected) for about 30 minutes. Then disconnect power and measure temperature of the DC-DC converters and transceiver chips.

Whichever board is hotter should be placed on top in the final device, for better thermal solution.

Of course, ideal solution would be to use metal enclosure with cooling fins, like this or this

\$\endgroup\$
12
  • \$\begingroup\$ I'm not happy with your remarks (don't get me wrong, they all are good remarks). But the ideas about the metal parts make that it will not fit in my plastic case. The vertical slots seem not a problem (wasn't intended to do but might be a good idea). Metal plating is more difficult, especially the 5 mm gap. Also placing boards vertically is hard to fit, and no possibility to fix them tight (unless for a test). I was under assumption that these ICs use very little power so heat would not be a problem (not even with 5 AC/DC converters inside). \$\endgroup\$ – Michel Keijzers Sep 12 '18 at 11:48
  • 1
    \$\begingroup\$ Placing boards vertical was only for 30 min test. You can simply do this on a table. the idea is to allow good airflow so that one board won't heat up the other. the result of a test will tell you what heats up more, DC-DC or transceivers. Whichever heats up more should be placed as top layer in the enclosure \$\endgroup\$ – Maple Sep 12 '18 at 15:18
  • 1
    \$\begingroup\$ The idea of tiny 5V 20-30mm fan like this was the first on my mind. I did not suggest it because these fans are often quite noisy. But you might get lucky and stumble upon quiet one. You can power it from any of the 5 converters, should not be a problem. I sometimes loop fan wires 3-5 turns trough toroidal ferrite choke, just in case. \$\endgroup\$ – Maple Sep 12 '18 at 15:34
  • 1
    \$\begingroup\$ You don't need 2 fans, one 30mm is enough. The problem with plastic is that the heat has no way to go, so it accumulates over time. Even a small airflow is enough to fix this. If you go with fan, don't forget to drill some holes on the opposite side and make sure it works on exhaust, not blowing in \$\endgroup\$ – Maple Sep 12 '18 at 15:42
  • 1
    \$\begingroup\$ "can I just connect the fan to one of the 5 AC/DC converters" I believe I answered this 4 comments above \$\endgroup\$ – Maple Sep 12 '18 at 16:03

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.