Power Quality Factors / EMF

[Chris15]

So I'm hoping to pick the collective brains trust to deal with a non entertianment issue.

I have what is effectively a system of panic buttons deployed on a site using a commercial alarm panel. It is false alarming.
The inputs of that panel are a voltage ADC, so a change on the voltage on the sense wire will trigger an alarm condition.
So would any of these be likely to induce a full volt or thereabouts?
- Running in the same cable tray as multiple (probably half a dozen) 50A 1ph sub mains, separation of roughly 400mm, parallel for probably 40 metres
- Those sub mains feed inverter air conditioners along the way, so I expect harmonics
- Those sub mains also feed hot water heaters - do water heaters have inrush currents worth discussing?

Entire mains is fed offf a generator power plant - a series of 800kVA Aggrekos conjoined properly.

Alarm wiring is using Cat5 but is not balanced.

Anything that would be helpful to clarify, just ask...

 

[STEVETERRY]

So I'm hoping to pick the collective brains trust to deal with a non entertianment issue.

I have what is effectively a system of panic buttons deployed on a site using a commercial alarm panel. It is false alarming.
The inputs of that panel are a voltage ADC, so a change on the voltage on the sense wire will trigger an alarm condition.
So would any of these be likely to induce a full volt or thereabouts?
- Running in the same cable tray as multiple (probably half a dozen) 50A 1ph sub mains, separation of roughly 400mm, parallel for probably 40 metres
- Those sub mains feed inverter air conditioners along the way, so I expect harmonics
- Those sub mains also feed hot water heaters - do water heaters have inrush currents worth discussing?

Entire mains is fed offf a generator power plant - a series of 800kVA Aggrekos conjoined properly.

Alarm wiring is using Cat5 but is not balanced.

Anything that would be helpful to clarify, just ask...

1. What is the input impedance of the analog "sense lines"?
2. What is the range of voltage they accept?
3. Are the lines open-circuit or terminated in a normal condition?
4. What are the expected voltages for normal and alarm conditions?

The layout you describe seems ripe for considerable analog noise to be induced on the sense lines.

ST

 

[Chris15]

1. What is the input impedance of the analog "sense lines"?
2. What is the range of voltage they accept?
3. Are the lines open-circuit or terminated in a normal condition?
3. What are the expected voltages for normal and alarm conditions?

The layout you describe seems ripe for considerable analog noise to be induced on the sense lines.

ST

Can I attempt to cover those in a slightly different way?
Through a combination of published information and a bit of quick maths, the circuit goes like this:

5V source
4k7 resistor
ADC sample point / "input" connector
2k2 resistor
6k8 resistor in parallel with the normally closed alarm switch
Ground

This means a tamper condition is either 5V or 0V,
An alarm condition is 3.3V
A normal sealed state is 1.6V.

That is based on a standard alarm application of a relay as the switch device.
In my application I have an open collector transistor in place.
So based on the Vdrop sums on the ground return line (common with the power supplying this radio receiver), there's about 0.3V dropped there and I lose roughly 0.1V across the BC817 output transistor. So apply the voltage division factor across the now 4.6V and add 0.4V and I believe my inputs are sitting at approx 2.0V in a sealed state. The current on the alarm zone / sense line reaches a maximum of 1mA if it is in short ciruit tamper, so the voltage drop there is negligible.

I have an enquiry in to the manufacturer to find what the threshold voltage is for triggering that alarm condition, because we all know full well there will be a range not a single value.
So what my thinking is that maybe that 3.3v range comes as low as 3V or even less and when I am already sitting at 2V, a volt of noise induced on the line is putting it over the threshold and therefore registering as an alarm.

The inputs have got a 400ms debounce on them, but I am waiting to hear back on how that is implemented, if it is simply sample at time 0 and then sample again at time 400ms and if they are both the same do something then I would think there's a possibility that I have some form of periodic induced interference that when the clock cycles align plays havoc but when they are slightly out everything's fine again until they overlap again.

That or the induced noise is more steady state and is holding the voltage above the threshold for longer than that 400ms anyway...

Thanks Steve for your assistance thus far.

 

[David Ashton]

Is your "ground" sourced back to the panel or is it connected to local "grounds'? an earth loop situation would likely cause the problem, induced voltage would seem most unlikely.

 

[Chris15]

Return to panel.
Only connections to any form of local power are relay isolated to drive a siren and a strobe.

 

[mrb]

are the unused conductors in the cat5 floating or grounded? The issue I see (however im no EE) is the high value resistors. You only have a fraction of a milliamp flowing, it seems as though reducing the value of the resistors (while keeping proper proportions) would increase noise immunity.

 

[Chris15]

Unfortunately the 4k7 is an SMD device mounted on a circuit board, so changing that's a bit tricky...

If you decrease the resistances then the increased current will start to bringh voltage drop on the lines into the mix...
There is no free lunch...

There are no free lines in the Cat5

 

[mrb]

Unfortunately the 4k7 is an SMD device mounted on a circuit board, so changing that's a bit tricky...

If you decrease the resistances then the increased current will start to bringh voltage drop on the lines into the mix...
There is no free lunch...

There are no free lines in the Cat5

perhaps wire up a circuit on some perfboard with a 1:1 op amp and replicate the voltage divider circuit with lower value resistors on the input side. VD isnt going to be a problem, you would be in the low tenths of a volt at 20ma with 120ft of 24ga

this is based on my assumption we are NOT talking about any soft of life safety system

 

[FMEng]

This sounds like a recipe for noise pickup. The circuit has no margin for noise. I would suggest isolating the long cable with an optoisolator chip. If you use a floating DC supply (wall wart) to drive the line and the LED side of the opto, it will in effect be a balanced circuit and more immune to noise, since it uses twisted pair. The opto should be mounted close to the alarm panel terminals.

I have isolated control signals connected to thousands of feet of telephone cable in this way, and it works fine.

This is an exmple of why good practice dictates that circuits should only be in raceways with other wires of similar voltage and current. Mic level circuits with mic level circuits. Speaker lines with speaker lines. AC line voltage only with the same, etc. In fact, I believe the NEC requires low voltage be separated from high voltage wiring.

 

[Chris15]

Oh I forgot to mention... there's 48 channels worth of inputs

I don't feel like building 48 op amps.
At this stage I'm really just trying to flesh out the likelihood of this being the causal factor more so than the resolution...

 

[David Ashton]

could you be picking up rf from a radio/tv taxi rf source, how about a small capacitor on each line to eliminate that option.A cable of particular length can pick up quite a high rf voltage.

 

[Chris15]

Possible though unlikely in this instance.
The installation site is pretty much in the middle of nowhere, the only television comes in off satellite dishes of several metres diameter...
The client does have Motorolas so that is a possibility...

 

[Dionysus]

This is one of those instances that I do recommend looking at it on a oscilloscope to see what is going on. What voltage DO you have going on?

 

[Chris15]

Um yeah about that...
The site's 3500 km away