Non-invasive measurement of current in individual PCB trace...?

[David_]

Hello.

I want this for a few reasons but one thing I have been fantasising about is being able to measure the current going through a small peace of a PCB non-invasive, in order to measure individual traces or checking current through different places of a 0V plane or power plane.

So I could track down return currents in some quantitative manner, it would not matter if it wasn't very accurate but within 10% would be just fine.

Do you know of any such a device which I can make my self or buy for <=100USD ?
Is there any hall effect sensors suitable for such a task?

Regards

 

[Mr.Cool]

magnetic field detecting IC, they exist. the trace runs under the chip.

 

[dick_freebird]

Hall current sense ICs / IC-form modules tend to be slow,
more so as you get more sensitive (high gain op amp in
the mix, more gain = less bandwidth). These can be good
for traces, but for a plane you need to straddle the whole
conductive width and normal to current flow - can you
get a magnetometer that wide, and do you know the
orientation of current (presuming it's even a single
direction over time)?

For debug purposes a sacrificial board and an X-acto
knife (if your conductor plane isn't buried inside other
important layers) can be useful' section the plane and
reconnect it with jumper loops that can go through a
CT-1 (for HF / spikes) or a Hall sensor (for DC / LF).
Bound to be some differences at the margins, but a
gross mapping you can do this way when other devices
may be unsuited.

 

[srizbf]

[

You require a PCB current probe.
Nowadays , it is available readymade.
check the site:
https://www.aimtti.com/go/iprober/

 

[FvM]

Yes the TTI I-prober is quite nice, I bought one some years ago. It's however unrealistic to get 10% current accuracy even with calibration, simply due to lack of a well defined conductor geometry. And a quantitative measurement of power plane currents is about to impossible with a simple magnetic field probe.

Finally you want everything for < 100 USD

You may want to consider this cheap alternatives:
1. Measuring DC voltage drop with a millivoltmeter. Can be roughly calibrated by putting in known trace geometry or very exactly by injecting additional current. My preferred method to locate PCB shorts.
2. Measuring AC current (pulsed DC/square wave, sine) with a small sense coil and an integrator amplifier.

 

[joao.caracas]

take a look at **broken link removed**

 

[David_]

I'll have to wait to buy a real probe a couple of years or so, in the mean time I'll try out some hall effekt.
I have been aware of them for some time but there are a few different kinds and most seem to be about position sensing and I meant to ask you here on edaboard about that for a while.

Do you know what kind of hall effect sensor would be suitable?
Do you have any info on how I could relate sensitivity measured in milliTesla to a amount of current if we would assume that the sensor is pressed against the current carrying conductor?

That about 10% was really just a stab in the dark for what I considered to be really in-accurate but apparently I have much to learn about what accuracy isn't.

I have found this:
**broken link removed**

they also have that sensor but with 12- or 16-bit digital output(SPC. Short PWM Code) but the range is -200mT to +200mT, is that a range that would enable me to use it in any current sensing application?

I have been reading about flux concentrators and I don't quite understand if this is something I should utelize in this situation, do you know or what do you think'?

If we take a look at Allegro's assortment of IC's and how they are categorizing there types we have:

Magnetic Linear and Angular Position Sensor ICs:
Magnetic Digital Position Sensor ICs:
Current Sensor ICs:
Magnetic Speed Sensor ICs:

All other sensors on the market seems to fit into these categorys as well, most of them anyway.
But to me it would seem that it is linear magnetic sensors which is relevant to me...?
Most if not all current sense sensors do require the current path to enter and exit through the IC package.

Regards

 

[FvM]

The relation between trace current and magnetic field is ruled by elementary electromagnetic laws, you can either make simple hand calculations to get an idea about the order of magnetitude or perform precise simulations with a field solver tool.

Presently, your objective is obviously getting an idea about expectable flux density to decide about useable sensors.

The field of a straight thin conductor refers simply to the definition of magnetic quantities, B = µ0*I/r, µ0 = 1.256e-6 Vs/(A*m)
e.g. a current of 1 A generates 1.3 mT in 1 mm distance or 0.13 mT in 10 mm distance.

- - - Updated - - -

Consider a small linear hall effect sensor in SIP package like A1389UA or similar products from other vendors. The hall effect chip can be brought e.g. 2 mm near the conductor. The noise limited resolution of the sensor might be e.g. 0.01 mT which translates to about 15 mA.

 

[David_]

I just happend to have one of each of these at home:
AD22151: https://www.analog.com/media/en/technical-documentation/data-sheets/AD22151.pdf
DRV5053: https://www.ti.com/lit/ds/symlink/drv5053.pdf

I don't remember the fine points about these two sensors but if i recall correctly they are(at least AD22151) very versatile in that the sensitivity can be adjusted very freely, I will read go through the datasheets but first a matter more fundamental.

I have for some time been aware that I lack certain knowledge and some things haunt me more than others, I have tried on multiple occasions to research and grasp Maxwell's, Gauss's and the other people's equations(I am drawing a blank here but I'm sure you know what other persons it is whom I can't remember the name of) which are as I understand what mankind knowledge of these things stands upon.
As such it feel important for me to learn and I have from the first hour of showing interest for inductors noticed that apparently I should know this and if I don't I should read up on it but I have and I have failed.

I have so far felt that the text I have by chance tried to follow have been leaving out a great many things so that while reading the first chapter I lost track of what they are trying to say, or I find documents that has a metaphorical part in which the use different analogy's to explain things but then they come to the math and this has scarred me quite a bit because they have made these equations to appear rather hardcore and I have felt so far of that I have just quite.

But I am feeling as if its time for another stab at it,

1, does anyone here have any resources suitable for a novice?
resources to learn magnetics(in context of Inductor with cores and DC-DC converters) and I do mean from the very beginning so that I can cover all the relevant equations.

If not then could someone suggest a book they have experience from and would recommend, the problem with that option is that I have hardly he economy to buy a real good quality educational book as it is right now.

I can find lots of documents about magnetics but never any really good or any that actually do contain what they are advertising.
I have applied to a mathematics class which I will hopefully attend(it couldn't be more necessary for to do that)

2, I don't know if this is much to ask but maybe someone with the appropriate knowledge could suggest a step by step ladder showing which or rather who's equations to start of with and then continue with(if that approach is even possible) For now I assume that Maxwell is a good place to start.

Regards

 

[David_]

Hmm, I could have sworn I had written this update day's ago.

Well in any case I have made some progress in finding ICs for this purpose that I like to share, I have found two copanys that offer this kind of IC and they both have multiple sensors using the hall effect in different ways but for me the relevant parts are:

MLX91206 from Melexis:
It is programmable but requiers a special programmer from Melexis so I have to be happy with the default settings, if programmed you can change a lot of stuff such as switching from analog voltage output to 12-bit PWM output and adjust the sensitivity.
Field range: 10mT(there a a 25mT version as well)
Default sensitivity: 580mV/mT(if programmed it has a range of 460-700mV/mT)
Bandwidth: DC to 90kHz
Output: analog ratiometric output(if programmed you could use PWM output with 12 bit resolution)
Linearity: ±0.5%
Accuracy: ±1%
Supply Voltage: 5V
Response Time: 8µs
Supply Current(Max): 12mA
Polarization: Bidirectional
Package: SOIC-8
At Digi-Key: MLX91206LDC-CAL-001-SP-ND

**broken link removed** from LEM:
This is also programmable but it would appear to be done by the manufacturer on request.
Field range: ±3,3mT
Default sensitivity: 600mV/mT
Frequency 105kHz
Output: Ratiometric Voltage
Linearity: ±0.4%
Accuracy: -
Supply Voltage: 4.75 V - 5.5 V
Response Time: 3µs
Supply Current(Max): 20mA
Polarization: Bidirectional
Package: SOIC-8
At Digi-Key: 398-1057-1-ND

I plan to acquire one of each of these and mount them on a pen like probe while having the SOIC pins gently bent upwards and then insulated with shellac or something similar and the pen probe will be mated with a BNC male connector so I can atached them to a scope or Analog Discovery(USB 14-bit Multi-device, scope amongst others)

These sensors are of two different types and the MLX91206 one is using there patented IMC-HALL technology where IMC stands for Integrated Magneto-Concentrator and quote "Conventional planar Hall technology is only sensitive to the flux density applied orthogonally to the IC surface. The IMC-Hall® current sensor is sensitive to the flux density applied parallel to the IC surface."

While the FHS 40 sensors datasheet doesn't say more than "). It measures the magnetic field generated by the current flowing in a conductor such as a PCB track. The output voltage is proportional to that magnetic field."

One thought though, these are both SOIC-8 packages and as such they do not enable me to position them to measure the current in a very small trace and and to avoid putting it over two traces at once. I may have misunderstood flux concentrators but could one imaging it could be functional to put a pen tip shaped flux concentrator underneath the ICs? so that the side of the concentrator which mates with the SOIC-8 is as wide as the SOIC-8 silicon and the side that goes against the PCB would be maybe 1/10 the size of the SOIC-8 side width?

Another thought, it appears as the currents I would want to measure is of a magnitude that the fields at the sensor will be below 1mT or not much higher than that.
And if you look at digikey for example and brows through there magnetic sensors you'll find thst there are sensors that is specified for such a range(~1mT) but they are 3-axis Compass/Magnetometers, do you think such a sensor could be adapted for use as a current sensor?

I would hope that I could simply rad its data and translate that to current in a mcu?

 

[Warpspeed]

It seems that the sensitivity may require track currents in the region of an amp, which is fairly high.

Currents that high can often be seen with thermal imaging, and by measuring voltage drops along tracks.

I want this for a few reasons but one thing I have been fantasising about is being able to measure the current going through a small peace of a PCB non-invasive, in order to measure individual traces or checking current through different places of a 0V plane or power plane.

These types of problems crop up all the time in component level board repair.
Often a dead short between ground plane and power plane, and a huge range of possible causes, including many very guilty looking tantalum capacitors.

Wherever current flows, there will be both heating and voltage drops. And either can be used to follow where the current is going.

With power planes, a Hall probe is not likely to be of much help because of the very low current density. But probing with a millivoltmeter can be quite revealing.

 

[David_]

Ah I'd love to get my hands on equipment for doing thermal imaging, the price tag is the problem for such solutions though.

I have been interested in a camera named FLIR ONE:
**broken link removed**
it is the second upgraded addition and it is mounted on ether an Iphone or an Android phone. It is only using the phone as a power supply and the camera itself is completely self contained within the add-on(apart from the fact that it sends the data to the phone for displaying of course), I haven't jet been in touch with the company but I wonder if it might be possible to use for PCB pictures. Given the price tag of 250USD it may be quite good(what do I know, I am just hoping) but it looks impressive and it do take two pictures for every single resulting picture(one thermal imaging picture and one ordinary picture, these two are then layered over each other to some degree so that even while its is mostly an IR image you still get edges so as to more easily make out what the heat sources are).

In any case some day I am determined to get my self some thermal imaging capability.

I have know received two of these current sensors/transducers, FHS 40/SP600 and MLX91206-CAL-100 and will begin building some kind of probe. I can't utilize the full potential of these as they require programming with a special device to get the full range of sensitivity but I have bought the most sensitive versions because as warpspeed just expressed the amount of currents that can be distinguished may be in the range of a Ampere and above.

But I wonder if a second flux concentrator can help(they are already using such a construction inside the ICs) but if I test that I will be operating in the blind and just testing and hope I'll get something right.

I do get that these things will probably not be able to do what I would like them to but I will find out.

It was not the intended use but it appears to me that these may be great for use for sensing the secondary currents from the primary side in a offline SMPS, IF the operating frequency is well below 200kHz as the best bandwidth given by these kind of sensors tend to be 200kHz(and I take that does not mean that waveforms of 200kHz can be distinguished but they have to be some distance below that)

Thanks for mentioning thermal imaging, I think that might actually be my best chance at accomplishing this, I have seen it used in an app note to illustrate the difference between DC and high frequency AC current return paths.

I am particularly interested in the SMPS idea now, isn't it a interesting idea to use it to solve the isolation problem while sensing currents in SMPS designs?
They do have analog outputs and so there should be many SMPS controller chips that can be used without any extra circuits... no?

Regards

 

[Warpspeed]

I have been interested in a camera named FLIR ONE:

I once worked at a place that advertised that they could repair just about any electronic device or circuit board without a circuit diagram or any technical details.
That was a pretty big claim, but by golly they were good at this !

They had some extremely innovative methods, which turned everything I have learned over the years about fault finding and repair completely on its head.

Thermal imaging was just one technique of many, and when you can plainly see temperature gradients of much less than one degree, and compare a fully functional circuit board running next to an identical one that has a fault, a thermal image can sometimes be extremely revealing, and lead you straight to the problem.

The price of infra-red imaging is still coming down, and if I was not now in retirement, and I was still doing component level repairs, it is something I would definitely now have.

 

[David_]

With power planes, a Hall probe is not likely to be of much help because of the very low current density. But probing with a millivoltmeter can be quite revealing.

Good idea, thanks.
Then I'll begin looking into designing a millivoltmeter

I'll guess a instrument amplifier could be a good start, or simply a differential amplifier.

Another thought for sensing currents in single traces.
Given the capabilities of today current sense amplifiers such as LTC6102, couldn't it be a viable idea to design a circuit which can be attached to a peace of PCB trace through two sharp probes in order to sense the current going through that peace of PCB trace?

I am thinking that maybe one could use a small mΩ resistor(perhaps simply another "precision" trace) that when the probes connect creates a parallel resistance so that the length and width of the trace on the inspected PCB is not as picky and to even out the resistance so one most of the time know to some degree what the resistance is.
And then use for example LTC6102 with high gain to measure the voltage drop of the parallel combination.

What do you think?

Maybe I am just making things weird and difficult, I could then as well use sharp probes to first measure the resistance of a peace of trace and then use a millivoltmeter to measure the drop and calculate the current.

I guess that was already hinted at in warpspeeds last post.

Regards

 

[Warpspeed]

What's wrong with an ordinary multimeter on the dc millivolt range ?
Most can display down to 0.1mV resolution which is plenty.

Suppose you have a solid ground plane, and a solid power plane and a dozen bypass capacitors, one of which has become a dead short.

Get your bench power supply, set it to current limit (at a safe current) and put maybe half an amp through the shorted board. If you are lucky the shorted capacitor may start to get a bit warm.

If not, the capacitor with the lowest measurable voltage across it will be the shorted one.

You can do the same with tracks, but at a much lower safe current.
If no current flows, there will be no voltage drop.
Follow the voltage drops along the track branches, and it will lead you straight to the short.

 

[FvM]

What's wrong with an ordinary multimeter on the dc millivolt range ?
Most can display down to 0.1mV resolution which is plenty.

Yes, the method had been already suggested, e.g. in post #5 and your post #11. I can report that it's a rather perfect method to localize most PCB or component shorts.

But obviously the OP is biased towards a magnetic measurement. The latest approach to resistive measurements in post #10 doesn't even notice the previous suggestions of utilizing PCB trace resistance for current measurement. Bearing the dog for hunting?

 

[David_]

my bad, we where both writing millivoltmeter while I was for some reason thinking microvoltmeter.

In any case I will see what those sensors can do and then I'll return to tell you about it.

Regards

 

[FvM]

Micro versus millivolt. An economic standard 4.5 digit multimeter has 10 µV resolution in the 200 mV range. It's sufficient for most of the discussed DC voltage drop measurements.

 

[chuckey]

HPs method of testing for shorts on digital boards is to use a pulser and a current probe. The pulser put out a 1A 10nS pulse at a low frequency (1KHZ?). This is so the power within the pulse is too low to destroy gates. The current tracer does just that, it has a variable "gain" control and I guess it benefits from synchronous detection as the PRF is known. The tip of this probe was about 2.5mm diam and about 5mm long.
You could try this technique.
Frank

 

[David_]

this endeavour is not for detecting shorts, rather I am trying to visualize which path the current takes on the returning to the source.