Current Sensing using copper trace

I observed a scenario a while back where a field service tech was installing a new circuit board with an onboard current sense shunt. It appeared to be a trace that fanned out from a large single one to multiple smaller traces. He was severing a number of the smaller traces while applying a constant current, and monitoring the output across the shunt. He indicated the reason for not using a resister was one of cost, and one board being able to accommodate multiple applications. I suspect even with very small traces being severed it would result in large stepped outputs. Possibly close was good enough?

Louis--

********************************************* Remove the two fish in address to respond

On 1-oz copper, a stretch of 14 mil trace .5454 inches long has .020000 ohms resistance. If you miss the length by a mil, the resistance will change by about 0.2%. It would be difficult to predict exactly how the resistance would actually work out, once you consider the resistance of the conductors where you are tapping into the shunt, component leads, and so on. You might get close, and then be able to adjust the circuit around that.

Of course your resistance could change from board to board depending on varying copper thickness, if it was plated through, edge geometry for different etch solutions, etc. It's something you'll very likely want to be able to adjust for. Perhaps build an adjustable gain into the amp you're using to measure the voltage across your resistance.

Or just spend the lousy $0.50 and get a 1% guaranteed resistor that doesn't have a +4000ppm/K tempco and vary with the production controls of your PCB supplier. That's my advice.

Best regards, Spehro Pefhany

Which was kind of where I was trying to nudge him, while allowing him the dignity of choosing between two options instead of choosing between idiocy and shame. ;-)

Yes, I know. I thought it was a bit too subtle for his good.

Best regards, Spehro Pefhany

tHE traces on circuit boards can be used to check out boards .When I was warrentee repair for a computer company, I often had to check for shorted IC's on circuit boards to replace them. NOw multiple chips would be on the same ground line. It was not practical to cut the traces on a multi layer board. I would take a low voltage, high current current source and connect it to the board under test. I would connect one lead of a sensitive voltmeter, with scales down to about

1 millivolt. to the negitive pin on the card. I would then with the probe check all the lines that fed into the ground line. At each point in the circuit where there was a branch I would measure the voltage. I would then check further up the trace on each of the feeds into the common point. If the voltage was the same as at the node, that would mean there was no current flow in that line. and the IC was ok. When I got to the bad IC I would find that there was a further voltage drop across this ground line, and I had my short circuit and bad IC. I used this method to locate hundreds of shorted IC's. The same method worked when trying to locate shorted components in other electronic equipment..

The same can be done on the high side.

On Wed, 04 Feb 2004 02:38:29 GMT, Spehro Pefhany Gave us:

Jeez... even our cheap 1/4 w resistors are 25 ppm. Are you referring to carbon comp, or carbon film maybe?

Metal film resistors are a lot more stable than that. At least ours are.

The suggestion of trimming traces, Andy, to calibrate the resistance, means you'll need minutes of a service man's time, or of a tech's time. No matter what assumptions you make about wage and labor rates, you'll burn more than the cost of adding a discrete resistor to the board.

It's a reasonable idea, asking the question you did, but when you run the numbers, the answer gets to be obvious.

Dark matter is still shooting off his mouth.

25ppm is 25 parts per million. I believe that is 0.0025 % I do not know if you can even buy a resistor that is .0025% as even lead length and soldiering job could change the resistance more than that. Let alone buy it cheap. I think DM is still being stupid. I have seldom see him post any thing but curses and foul language.

+4000ppm, give or take, is the tempco of copper @ room temperature, depending on purity and annealing. That is a decent output for a temperature sensor, and they are used commercially as such in some low-accuracy applications (stator RTDs). Copper corrodes so it's not as good as, say, Pt for long-term stability.

Most (if not all) pure metals have about this tempco at room temperature (~PTAT), you need alloys to get a good resistor.

Depends on the metal, eh?

Best regards, Spehro Pefhany

Thanks everyone for all the feedback. I kind of had an feeling it was not a good idea. I realy need to be within about +-10% or so on current measurement. I suppose if it was that easy no one would bother making these low ohm resistors would they?

Andy

On Wed, 04 Feb 2004 10:43:05 GMT, "bushbadee" Gave us:

No. your retarded ass is showing how little you know about the industry AGAIN.

YEP... Sure is.

Nope. Thermal tolerance is NOT the same as accuracy in value, you retard!

You are an idiot. Try Mouser, or Digi-Key. Both sell standard metal film resistors at 25 ppm 1%.

For you to shoot off your mouth (as usual) before even researching these facts that you obviously know not a goddamned thing about, is par for your alzheimer's course.

I think you are a riot!

You have seldom comprehended anything you have seen in your entire pathetic life, guess as you go boy!

Try again, you idiot! Get rid of the "B" baby bullshit, retard!

On Wed, 4 Feb 2004 08:31:51 -0600, "Andy Pichotta" Gave us:

Two 0.01 presicion resistors in series, or the exact 0.02 value is readily available.

It is a common thing in the industry... shunts.

Oh really they make these? I am fully aware they are available. I have them in the design already. I just had a thought on my long drive home from work. Wondering if I could just use a trace. Only a thought. So I thought I would check the opinion of some others out there. News groups work nice for that don't they? Thanks again everyone for the feedback.

Andy