sending power through long wires

[browniebob]

Good morning all, great forum. Here's my question: I'd like to send power through a few 32 gauge wires for a distance of several thousand feet. How do I figure out the limit on how power-hungry my load can be (it'll be a sensor) and what is a good way to send this power? My background is bioengineering so feel free to dumb down your response! The load will need DC but from some reading I've been doing, that seems like a silly way to send power. Please advise. Also, are there any online references you can suggest which I can read in order to become better grounded in related things, as I'll be doing this more in the future. On the other end of the wires (source side) I'll be able to basically pick any reasonable power source.

Best,
K

 

[DRWeig]

Hi K,

You could have a thorny problem if you want to send lots of power. You'll be limited by the ampacity of the 32 AWG conductors (which varies based on their temperature rating, how installed, and other factors that might tend to dissipate heat) and also by the voltage rating of the insulation on the conductors. Voltage drop will be a third factor to consider since your run is long.

Basically, 32AWG copper has a typical resistance of about 175 ohms per 1000 feet. A handbook I checked gives typical ampacity of about 500 mA in air for chassis-type wiring and only about 90 mA for power transmission.

If you're transmitting power, there is a trade-off between AC and DC, based on how the conductors are strung (capacitance).

Tell us how much power you need to transmit. Maybe then we can give you some better hints.

Good on ya,

Old Dave

 

[CarlPugh]

Do some research on sensors. Some sensors are two wire and the current through them depends of temperature, pressure, flow or whatever.
These sensors have a power supply that is built in and operates on the current through the sensor.

 

[buzzp]

On the surface, it looks like your asking for trouble. If the device hooked to it at the other end is wanting voltage (not a current sensor like 4-20mA), then you will probably end up with problems. Will wait for more info.

 

[dpc]

I think you'll want to use dc to avoid a lot of shunt capacitance losses and inductive voltage drop.

The #32 gauge will have considerable dc resistance - about 0.165 ohms/foot, (at 25 deg C).

Any way to run larger gauge wire for some portion of this long run?

 

[tinfoil]

Remedial Ohms Law: V(drop)=IR

If a 4-20mA sensor is to be used, you need to allow for at least 20mA of current.

For each 1000' of wire, you will have 175ohm of resistance, or about 500 Ohm for 3000' (I'm guessing what 'several' means).
20 mA x 500 Ohm is 10,000 mV, or 10 Volts!

That's just on the 'out' wire. You will have a further 10V drop on the 'return' wire of the circuit, for a total drop of 20V round-trip.

If your sensor power source were only 24VDC, then you only have 4V left across your sensor terminals!

I suspect that you are going to have problems with this arrangement. We use a minimum gauge of #20 for our sensor loops, and not for a distance of 'several thousand feet'

 

[itsmoked]

A point missed being mentioned so far is just plain old durability. 32AWG is weak, weak, weak, it's also very weak. Did I mention it's weak? If a snail oozes over it it will probably fail. Generally you don't want wire that's out in the "wild" under about 16AWG just because the jacketing can then have a reasonable wire/jacket strength.

 

[DRWeig]

Good point, itsmoked.

We limit wire size to 22 AWG absolute minimum because the apes who tighten the screw terminals will invariably smush smaller wires into oblivion.

There are utility-grade transducers that operate on a 0 to 1 mA signal for use over very long distances. If you can find one that senses the property that you'd like to sense, you may be able to deal with the voltage drop a lot better.

Old Dave

 

[browniebob]

Thanks folks (I wish there was a similar forum for biochemists!)

I'll be using a 30 awg wire, sorry for the confusion. I have space for 5-6 of them so I was thinking of either: 3 dc or 3 ac pairs, or one 3-phase set. What design issues should I be looking at? I would like to think of the wire and whatever scheme I'll be using for power transfer as a blackbox feeding into the load; this will help me better pick a sensor which fits the constraints, instead of considering both at the same time which would get complicated. The more power I can deliver, the better. If I can find a high voltage source at my base station, then can I jack up the frequency and overcome some of the losses? Either way, what sorts of characterizations should I be doing on this wire while it's still in a bundle (I have 3 strands of 2000' layed out around the walls of our lab right now).

K

 

[advidana]

30 gage is not much better. The distance depends on the loads (watts or volts and amps) you are going to use. to give you and idea If its a speaker maybe 40 ft but a flast light maybe 10 ft. a 100 watt blub maybe two inches, Etc.

 

[alehman]

The lowest loss will be with DC. Capacitive coupling will cause significant currnet leakge with AC. The higher the frequency, the greater the loss. If you don't need to transform voltage, stay with DC.

Actually one option may be to use a high voltage DC power supply and an electronic DC-DC converter at the sensor. Obviously you would need to stay within the voltage rating of the wires' insulation. This way you can transmit more power with less current and less loss. Power is transmitted to amplifiers on undersea fiber optic cables at several thousand volts DC by this means.

 

[itsmoked]

Okay I'll do it...

browniebob WHAT IS THE SENSOR?!?!?!?

There is a LOT of talent here u are wasting by dribbling out tiny bits of information and making us all guess how to help you. Give us the info we need and we can maybe save ourselves and YOU a whole lot of trouble and get you on the fast track to your solution.

 

[alehman]

I'll second that.

 

[alehman]

...Or perhaps he is working on something that's government classified.

 

[itsmoked]

Hahaha,

There's vague and then there's pointlessly vague.

 

[mc5w]

Please tell us how much voltage and current your sensor needs. Depending on the current level an appropriate wire size could be 24 gauge telephone cable, 18 gauge thermostat wire for the power plus telephone calbe or cat 5 cable for the signal, or even something heftier.

If your sensor is 4-wire ( 2 wires power in 2 wires signal out ) you very likely will need a voltage regulator right at the sensor so as to cope wioth voltage drop better.

 

[JBender]

Who's giving away 30 guage wire?

 

[browniebob]

I can see how that got annoying....

I'm using a little piezo accelerometer for looking at jolts to people's heads (the part was given to me). It has an input resistance of ~300ohm an output resistance of ~1kohm. I suppose I'll have to calibrate it against something prior to using. That doesn't help much, I know. I have only the part, no datasheet. The max output Ive been able to 'shake' out of this thing is around 10V, so I guess it needs at least 20mA assuming 50% waste. I also need to power 5 small LEDs (something like a fairchild mv8w00, which has 120mW/30mA rating). They dont need to be on at the same time, which makes life easier. I suppose then, that I could jack up the source voltage (dc) like alehman suggests to just under the threashold for the wires, and then somehow bring it down on the other side to a reasonable level...? if I have 6 wires, then maybe I can double up the current this way too?

 

[alehman]

Does your accelerometer have a part # on it? If so, it should be possible to find a data sheet. Looking at accelerometers on the web, in appears some require a constant current input of a few mA.

http://www.sensotec.com/catpages.shtml#acceleration

These devices seem to operate at fairly low signal levels. I'm concerned that unless the wires are shielded, there will be too much noise.

Have you considered going wireless? You could use a battery for the power supply, connect sensor output to a voltage controlled oscillator. Feed the oscillator output to the input of a commercial wireless microphone transmitter. The frequency of the tone at the wireless mic receiver would correspond to the measured acceleration. A little calibration work would be needed. Not sure about the LED's. What do they do?

 

[itsmoked]

browniebob; Thanks for more info.

Are you stuck with that sensor? I have built pedal sensors that monitor the speed of pedalling that used Analog Devices's really cool MEMS accelerometers. In this case I used wireless.(see Popular Science June 2001)

http://www.analog.com/en/subCat/0,2879,764%5F800%5F0%5F%5F0%5F,00.html

These all use under 1mA. Many have 2-3 axes all are very sensitive and fast.

From what I understand of your task I would couple one of these to a an 8 pin PIC processor or equivalent this would all run on about 7mA. This prevents running the raw signal anywhere. This will help you avoid the noise problems. You could encode the info into a current modulated string or use standard signaling methods like RS485. You can actually even modulate the power lines. One other thing this method helps with is data reduction. Often you reduce the amount of data right at the source which helps your whole project. For example if you are really only interested in the maximum acceleration experienced you can set a threshhold and only transmit readings that exceed it, or just the peak readings.