screening

[duncan webster 1Participant @duncanwebster1]

I’m getting back onto a modification project that got stopped when the builders arrived. One Arduino controlled project (clock) has output via an open collector transistor, feeding the input of another Arduino (chiming mechanism). The second Arduino has 1k pull up resistor to 5V. The signal is 30 seconds on, 30 seconds off. The connecting wire is about 2ft long and has never actually given any problem in previous version, but I have the opportunity to make improvements if I knew how. I should mention that it’s all quite near a gas hob, and the igniter sent another Arduino clock doolally. Currently I have single core screened cable (ex headphones) with ground on the screen, but it occurs to me that spikes on the ground line might be just as bad as on the signal line. Its all powered off wall warts, so I don’t have ready access to a real earth. What I’m considering is using 2 core screened cable but with the screen connected to 0v via a capacitor and resistor in parallel, so the cap absorbs the spike and then discharges. Can one of you erudite chaps tell me if this has any chance of working and suggest values for cap and resistor?

[peter1972Participant @peter1972]

Consider using an opto-isolator (and associated LED resistor) in the chiming Arduino’s box driven from the open-collector of the clock project box so grounds are not connected together.

Alternatively, program the chiming Arduino to sample the incoming signal several times (with short interval between) before acting on the state of the signal if successive samples are at the same logic level.

[duncan webster 1Participant @duncanwebster1]

I know about denbouncing routines, but surely better to eliminate potential problem than handle it

[duncan webster 1]

On duncan webster 1 Said:

I know about denbouncing routines, but surely better to eliminate potential problem than handle it

Yes the code would be the same, but in this case to virtually eliminate the possibility of the chiming Arduino being triggered by interference. You have the luxury of monitoring a signal that has very low frequency and can sample the digital input very many times per second. Pesumably it does not matter if the chiming starts slightly later than it should!

I really do not think you should have any problem with the circuit you have. The 1kΩ resistor is fairly low impedance by electronics standards and 2ft of screened cable should not pickup much interference. Also you have a good signal voltage / current swing.

In situations like you have, people tend to overlook what happens when one power supply is on and the other is off. In this respect, the use of an open-collector transistor is good. Nevertheless, have you considered powering up both circuits from the same plug-in power supply?

When using a plug-in power supply, I am aware that the output voltage is floating with respect to earth but also aware that there will be some coupling between the incoming mains power and the output. If I measure one output from a plug-in power supply with my earthed oscilloscope, I see about 320V peak-to-peak waveform. One wonders what would be measured if an overhead power line is hit by lightning. It’s worth being aware of this, but it should not affect the reliability of your circuit. Anyway your circuit is far from being safety-critical.

[Robert Atkinson 2Participant @robertatkinson2]

A common DC power connection between the two boards is a good idea.
It’s possible the screen is providing a path for currents picked up on the power cables.

As peter1972 says an optoisolator will cure the issue.

Another approch would be to divide the pull up between the boards say 4k7 on each and put a small capacitor, say 100pF to 1nF, to ground on the input end.
If using twin screened cable make sure the cores are twisted and ground the screen at one end only. Normally the source end.

Robert.

[duncan webster 1Participant @duncanwebster1]

Thanks for that, unfortunately there isn’t room for an opto-isolator at the chiming end board, and having just one power supply would mean major surgery, so I’ll divide the pullup as Robert suggests.

 

All this has been filed away for next time I attempt something like this, you’re never too old to learn

Thanks again

[duncan webster 1Participant @duncanwebster1]

And I’ll incorporate software anti bounce as Peter suggests

[DMRParticipant @dmr]

No mention in any of this as to the physical layout. You cannot run straight wires next to each other and not get interference which may never bother you if they are all signal cables. Screening will not save the day if AC power leads are involved.
Anything involving mains power or any high voltage switching device should not be laid next to (in parallel with) signal cables. All your signal wiring needs to be the order of 50mm away from any power wires though crossing at right angles is OK. One alternative that can be tried is to use twisted pair signal wiring.

The wall warts (particularly if they are the modern switching versions) need to be apart from your signal sources too.

[duncan webster 1]

On duncan webster 1 Said:

Thanks for that, unfortunately there isn’t room for an opto-isolator at the chiming end board, and having just one power supply would mean major surgery, so I’ll divide the pullup as Robert suggests.

 

All this has been filed away for next time I attempt something like this, you’re never too old to learn

Thanks again

Obviously I have not seen your physical setup but optoisolators can be miniscule in size, Galvanic isolation solves the problem full stop.

I inherited a project (commercial, already in production) that had plagued the manufacturer with random problems when users connected later versions of external devices (cameras). I found that the problem was caused by the lack of isolation between the two systems. I had very limited options to add isolation without a PCB redesign and in any case many units were already in the field (literally! some were even in Chernobyl). As deployed systems did need regular visits to exchange memory cards, batteries etc I designed a new (identical looking) wiring loom that along with detailed instructions, the users could install on their visits.

The optical isolation was installed within the existing loom connector, a 9 pin Sub-D. The isolator I chose was a SSR (Solid State Relay) which had the advantage of its output switch being polarity insensitive. The opto LED needed less than 2mA and was easily driven by the MPU output.

Whilst the device was small and its surface mount ‘pins’ were delicate once wires were soldered to it and then protected inside adhesive lined heatshrink sleeving it easily fitted within the connector backshell. I will see if I can find any pictures of the assembly and post them. For reference the CPC1019N SSR was from RS (906-1026).

Ian P

[peter1972Participant @peter1972]

It’s not worth spending any money on this unless you find you have an issue.

[Author]

Posts