current sensing with 2 parallel resistors

[Simon]

Right, I'm doing this design (now becoming well known as i keep asking questions about it  ) that is passing 20-25 amps and I'm thinking of using two parallel resistors for the current sensing to reduce strain, however I can only sense the voltage on one resistor, is there a risk due to me using 4.5 milli ohm resistors that track resistances and solder resistance will alter the 2 nearly symetrical paths and give me false readings due to one passing slightly more current than the other.

Attached is the section concerned of my PCB R00 and R01 are the sensing resistors, each coming from a relay IC3 is the current sensing chip.

Have I just opened a can of worms ? what if I took the line from the current sense chips to both resistors maybe through a say 1K resistor each ? so as to sum/average the two readings and not put 10A down a 0.64mm trace ?

[scrat]

Have I just opened a can of worms ? what if I took the line from the current sense chips to both resistors maybe through a say 1K resistor each ? so as to sum/average the two readings and not put 10A down a 0.64mm trace ?

I can't understand the sentence "and not put 10A down a 0.64mm trace". Where do 10A pass on a 0.64mm trace, in your layout?

However, summing the two differential signals could work.

[Simon]

if I were silly enough to run the 0.64mm trace past R00 and onto R01 I would be be linking the two together, but then a significant amount of current would try and go down it too messing up the measurement as the trace would become part of the main current path, if you look I have made sure the tracks to the sensing chip are not in the main current path.

you think "summing" the two voltages with resistors in series would work ?

[jahonen]

I would just put the sense resistors as close as possible each other and take the sense lines in between of resistors. Sense lines need not to be so thick, you can easily make them much thinner, if that makes it easier.

Regards,
Janne

[Simon]

that's a good idea, I'll have to do some pushing and shoving and make room to bring them together, I think I placed each one between relay pins, if they are in the middle they need more free room as they will clash with the relay pins

using the resistor adder does not work anyhow at least not in simulation

[Simon]

There, that might work

[mikeselectricstuff]

If you want to do it properly you can get four-terminal current-sensing resistors, to eliminate errors due to trace resstance etc.

[Simon]

yea but I think the 1/2 terminal left will be too small to take 10-15 amps shoved through it I think providing i stay away from main current carrying track it will do, I can software calibrate

[CafeLogic]

You can't sense only one resistor if they are in parallel. If you have a 2 ohm resistor in parallel with a 3 ohm resistor, you could probe the voltage drop across either and it would look like a 1.2 ohm resistor (assuming ideal traces). Putting them in series defeats your original purpose of using two since each of them will dissipate the exact same power as if it were the only one in the circuit. Using two in parallel will cause each to only dissipate 1/4 of the heat because you get half the voltage drop and twice the dissipation area. The down side is you only get half the sense voltage, making it that much harder to get an accurate reading.

[Simon]

yes my query was if there is significant track between the 2 then there will be an error in the reading over one only, I have solved this by bringing the two very close together so this is negligeable

[CafeLogic]

OH, I see now. I think the resistor tolerance would easily dominate any trace error unless you were really sloppy. If your app calls for 5% tolerance you can skip the resistors and just use the trace itself. I like this handy calculator for those sorts of things: http://circuitcalculator.com/wordpress/2006/01/24/trace-resistance-calculator/

[Simon]

I've been down that train of thought already, 1% is desirable and owing to carrying 20-25 amps a PCB resistor would be too large, as I'm working with 1-3 mR the board tracks could possibly cause a small difference

[veryevil]

If your looking to measure 20-30Amps why not use one of these http://www.allegromicro.com/en/Products/Categories/Sensors/currentsensor.asp (I recommend the 713) as it uses the Hall-Effect to measure the current and give out a linear voltage of 500mv + ~150mv/Amp

It also has an incredibly small resistance as it is works via a conducting path not a resistor and it comes in a simple sop8 package.

[Simon]

hm haven't really got the room, how do they work anyhow ?

[veryevil]

Basically, wire it up like this:




Then you get out:



Check the datasheet for more info

http://www.allegromicro.com/en/Products/Part_Numbers/0713/0713.pdf

[Simon]

so it is an 8 pin part and whats that round symbol ?

[veryevil]

Its the input DC Current.

Its an 8 Pin SOIC (not SOP as I said earlier) part which is tiny!

You pass the current path you want to measure into pins 1 and 2 (joined together with a large pad on PCB board) and then it comes out of pins 3 and 4 (which are also joined together). The only external component needed is a small cap between pins 5 and 6.

Other than that it just needs 5v and GND. The output then feeds into an ADC pin or some other system.

There is a 20Amp version and a 30Amp version with the only difference being the mv/A increase on each.

For more info see http://www.allegromicro.com/en/Products/Design/current_sensors/ look most of the way down to the bit about the ASC712 (the AC current monitor chip equivalent)

Also see this page about the ACS71x parts http://www.allegromicro.com/en/Products/Design/managing-external-magnetic-field-interference/

[veryevil]

Even more info available on the FAQ page for the chips

http://www.allegromicro.com/en/Products/Part_Numbers/0712/0712-13_faq.asp

[Simon]

hm thats interesting, I'll probably look into those later then, with the space saved for the resistors I'll have room for any other conditioning circuitry, I'm looking at 20-25 amps although one day may take the project to 50 amps but that's not now so no need to worry

[veryevil]

They also do 50-100 amp parts http://www.allegromicro.com/en/Products/Categories/Sensors/currentsensor.asp

[Simon]

ah hang on do these things only go up to 85C ? I need 125C or thats what I'm aiming for, once the current design is put to the test I'll know the actual working temp of the unit

[veryevil]

The chips themselves should get very hot at all with only having a 1.2mOhm internal resistance for the current path.

Using the max of 30Amps across 1.2mOhms with P = I²R give 900x0.0012 = 1.08Watts

Taking the thermal characteristics of the package (Junction-to-Ambient Thermal Resistance) of 23^oC/Watt it will only get ~23^oC above ambient

However if there are lots of other high temps about and its enclosed then there may be a problem

[Simon]

err yea I'm planning this to work in an ambient of 40C (maybe 50 due to being under a bonnet of a car) with a possible 30 C temperature rise in the case, of course the sensing circuitry is all part of this heat generation but still I'm also looking for very durable and basically indestructible so prefer to base it all on 125c parts, if this fails it can blow all of the cars electrics while far from help

[RayJones]

Sounds like your project is a regulator for controlling the alternators output from vintage cars - I'd imagine the regulators of old are hard to come by.

Temperature, bonnet, high current give it all away 

[Simon]

hehe you gotta find a way to control all that power under those temps - good luck , my solution was the result of a number of atempts although I'm yet to try this final one.

no the regs are easily available they are made in india, they usually break down quickly and are mechanical, considering what they do mechanically it is amazing (unfortunately they're not made like they used to be), basically a couple of specialized relays and the supporting framework provide more functions than meets the eye and i was surprised at how much effort went into replacing each function in software / hardware, it has been done before in fact many times projects were often published in magazines and you can get replacement regs for the later cars of the 70's, considering it's a relay the regulation is not too bad.

My aim is to make a more efficient unit with better stability and possibly extend functions of it. (some one has been working on it for years and never got there  )

Sad thing is the indian regs are now not much more than I can make the thing for in the first place my friend now tells me, this idea all started a year ago when they were three times the price they are now. so it might end up being more academic than practical but for the couple of cars my mate runs, it will be nice to see it work at least, in any case much testing will be required in all weathers and temps before i dare stake anything on it being fit for purpose