1
\$\begingroup\$

This question is related to the DMX splitter I'm building and shown in Prototyping a DMX splitter.

However, for testing a different question is appropriate.

I never really tested a hardware device, so be gentle please.

The intended use for the DMX splitter is to use on a music stage with my band(s). The device will be reasonably safe under my keyboard stand.

For completeness: for safety it has 5 optically and power-isolated input and four output channels, TVS's for each DMX output, a varistor and normal and temperature fuse for the main 220/240 AC, but no double enclosure, just one plastic enclosure.

I wanted to perform the following tests which might be typical for the intended usage:

  1. Normal behavior: power DMX splitter on (and check functionality).
  2. Hot-plugging input device: while DMX splitter is powered on, switch off and on DMX input device.
  3. Hot-plugging output devices: while DMX splitter is powered on, switch off and on DMX output device (per output).
  4. Removing the DMX input cable/plug and reinsert it, repeatedly.
  5. Removing the DMX output cable/plug and reinsert it, repeatedly.
  6. Power on a heavy power device on the same 220/240 V group (like a drill), repeatedly.
  7. Power on a 380V 3 phase device in the neighbourhood.
  8. Using 20 meters of cable per output (will be more than we ever need).
  9. Connecting 8 devices per output.
  10. Laying heavy power cables next/on the DMX cables (like the 380V phase cable).

Are these reasonable tests to perform or should I add more?

\$\endgroup\$
  • 2
    \$\begingroup\$ When its a one off unit, I would just start using it \$\endgroup\$ – PlasmaHH Sep 10 '18 at 20:27
  • \$\begingroup\$ @PlasmaHH The 'problem' is, we are kind of depending on it regarding the light. And it is not nice if I can fix something before a problem happens. If there are too many problems, the band can buy a commercial DMX splitter, but it's fun to learn and try to make one myself, especially if it works in most/all conditions. \$\endgroup\$ – Michel Keijzers Sep 10 '18 at 20:29
  • \$\begingroup\$ @Maple thanks for your comments. Right about the channel 4 optocoupler resistor (luckily there is some space for it). About the transceivers? I can rotate them, moving to the other side is no option, since I would use another few inches for wiring the outputs to the actual actual DMX panel mount plugs. And I also made a mistake with the 0.1uF (I used the right ones on my breadboard, but my picture shows the wrong ones). Thanks for these improvements. \$\endgroup\$ – Michel Keijzers Sep 12 '18 at 10:07
0
\$\begingroup\$

Before you start testing, here are couple observations on your posted "schematics".

1) You are missing current limiting resistor on channel 4 optocoupler.

2) The positioning of transceivers relative to DMX connectors is awful. The wires go across entire boards. Either rotate the chips or move output pins to the other side of the board so that the distance between transceiver pins and board connectors is minimal. I believe you've asked about exactly this here.

About the transceivers? I can rotate them, moving to the other side is no option, since I would use another few inches for wiring the outputs to the actual actual DMX panel mount plugs.

Other than 3 DMX pins all others are internally wired, i.e. not connected to panel jacks. If you move these 3 to the other side next to MAX487 outputs you can then rotate breadboards too, to put pins close to jacks.

Update: OK, I see it won't be quite that easy since your PCBs are stacked and wired through. Well, then I suggest rotating the transceivers. Especially because current placement puts all those DMX signal wires next to AC power distribution wires.

3) The decoupling capacitors in the picture listed as 1pF. That is way too low. Schematics suggests quite typical value 100nF (0.1uF), which is 100000pF or 100 thousand times more than you have.

4) On channel 1 there is no ground connection between DMX GND and DC-DC GND1

5) On ALL channels pin 1 (output!) of MAX487 is connected to VCC. It should be left unconnected. Only !RE connected to Vcc prevents them from instant burnout now.

6) there is only one 120 Ohm terminating resistor (on the input). You need termination on all 4 outputs as well.

Other than these, you have comprehensive set of tests planned.

There is one test that is not really fair to the splitter, but might expose vulnerabilities. It is powering off the splitter while controller remains active. Technically, input transceiver should survive this OK.

Also, let me re-emphasize the importance of using correct cables (either DMX or CAT5/6). The XLR cables have wrong impedance. While short runs can be used for testing, 20m of that would be quite taxing on output transceivers.

\$\endgroup\$
  • 1
    \$\begingroup\$ Added 4), 5), 6) \$\endgroup\$ – Maple Sep 13 '18 at 0:28
  • 1
    \$\begingroup\$ 5) Once again, pin 1 is an output connected to Vcc. If it tries to output 0 this will instantly fry the chip. There is no reason whatsoever to connect it to anything. 6) Normally, RS485 must be terminated at both ends. Terminator takes care of one end of a chain. But splitter sits at the other end. Yours is a special case. Sometimes resistor can be omitted on the transmitter side IF transmitter is at the beginning of a chain AND transmission is always in one direction. So, it is your choice. But it is not "instead". You need either both, or one at the end. \$\endgroup\$ – Maple Sep 14 '18 at 5:39
  • 1
    \$\begingroup\$ You can put 1 to output if the output is 1 itself. It's when output is at 0 (i.e. close to 0V) that bad things happen. Or the opposite - output at 1 (i.e. close to 5V) and you connect it to ground. This is a simple short circuit, no magic here. You don't connect hot and ground wires together after all. \$\endgroup\$ – Maple Sep 14 '18 at 9:55
  • 1
    \$\begingroup\$ And yes, thanks to isolation you pretty much have 5 separate RS485 networks. One begins with controller and ends with splitter. Four begin at splitter and end somewhere (potentially via many daisy-chained modules) with termination plug. BTW, if you remove input resistor and add "thru" jack to the splitter, you'd be able to place it anywhere in the controller's chain. Of course then you'd need a termination plug on "thru" output if splitter is the last device on that network. \$\endgroup\$ – Maple Sep 14 '18 at 10:07
  • 1
    \$\begingroup\$ What I described is not a "booster", because "thru" jack is wired directly to input jack, no boost here. It is a simple wiring option that makes physical placement of the splitter more flexible. For example this could be a single RS485 chain: Controller -> LED -> splitter -> LED -> Terminator. Then other 4 chains can be attached to the splitter in the middle. RS485 supports more than 30 nodes on the line. By connecting controller directly to splitter you achieve better isolation (i.e. controller protection), but you are under-utilizing the driving capacity of controller output. \$\endgroup\$ – Maple Sep 14 '18 at 18:38
1
\$\begingroup\$

A typical certification / commissioning test sequence for consumer electronics usually includes the following:

  1. Usage tests

    • Functional test (test each function / mode)
    • Performance test (test how the device performs under maximum load)
  2. Power supply tests

    • Power supply variations (test with minimum / maximum voltage)
    • Power supply failure (test where the power is repeatedly lost and restored)
    • Conducted transients (test reaction to bursts / surges)
    • Conducted interference (test reaction to high-frequency noise and base frequency harmonics)
  3. Environment tests

    • Temperature (cold / heat)
    • Humidity / moisture (e.g. damp heat / water mist / condensation)
    • Vibrations / shocks (test after a fall)
    • Inclinations (test the device upside down etc.)
    • Physical integrity (test if the case can withstand a certain weight or pressure)
  4. Safety tests

    • Insulation resistance (test at different temperature, humidity levels etc)
    • High voltage (test if the insulation withstands a certain voltage)
    • Flammability (typically a destructive test, skip)
  5. Electrical interference tests

    • ESD (apply electrostatic discharge to exposed connectors)
    • Low-frequency EM fields (stronger magnetic fields, capacitive coupling)
    • Radiated interference (test in presence of high-frequency EM fields)
  6. Electrical compliance tests

    • Conducted emissions (disturbance on power lines)
    • Radiated emissions
\$\endgroup\$
  • \$\begingroup\$ Thanks for this extensive number of tests. I will add some of them, I cannot do them all, since I don't have equipment for generating ESD, EM, radiation, not even a lab supply to change voltages (let alone around 220V except for a normal wall outlet socket). \$\endgroup\$ – Michel Keijzers Sep 13 '18 at 13:11
  • 1
    \$\begingroup\$ @MichelKeijzers Sure, I just wanted to give an overview. Your tests will obviously concentrate a lot on functional/performance part which you obviously know better than anyone since you designed the device. It's just incredible how many DIY devices stop working when you put them upside down or blow at them with a hair dryer. \$\endgroup\$ – Dmitry Grigoryev Sep 13 '18 at 13:33
  • \$\begingroup\$ I did not 'design' it, but I copied it and made some additions/improvements (mostly because of valuable members). I really intending to do a lot of 'physical' tests, because it will be carried around a lot (used on music stages). I'm a software engineer, so not used to ' hardware' testing :-) ... That hair dryer is a good point,, very easy way to test heating (which can be a problem on a warm stage). \$\endgroup\$ – Michel Keijzers Sep 13 '18 at 13:43

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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