How to solder this SMD current sensor, and how in the world can this handle 30a?

[Red Squirrel]

Bought some current transducers for a project and figured I'd buy extras for future requirements so I went with 30a ones for extra flexibility.  I did not realize they would be THIS small!  I always try to get through hole as it's just easier to deal with when prototyping or soldering by hand and if ever I get serious and want to actually make a product for manufacture then I'd go surface mount obviously. I figured I would just "dead bug" solder these on a small piece of protoboard and run some small lead wires over to pins and basically convert it to through hole, but wow, this is small. Though thankfully some of those pins are "doubled" so I don't have to be that precise I guess.



I do have some solder paste that I bought with anticipation that I might need it, but without an actual circuit board with an existing foot print, how would I go about soldering a part this small?

Also, how can this transducer actually support 30 amps?  I can't imagine passing 30 amps through something that small.  Don't you need like 8-10 gauge wire for 30 amps?  I'm not actually planing on passing that much current through it though, maybe like 5 amps max for the particular project I have in mind though depending on the wiring I will probably rate it at like 10 amps via microprocessor (show a warning or something) then put a 15 amp fuse. 

This is the part in case you are currious: http://www.digikey.ca/product-detail/en/0/620-1482-6-ND

I obviously skipped over the foot print/dimensions part of the datasheet when I went through it as it does say 3mm by 3mm, I guess I just badly assumed a part that can handle 30 amps would be much larger. 

Overall just curious what kind of technique I'd use to solder something this small.  Unfortunately I can't seem to find a place in Canada to buy a small solder tip for my Hakko so the one that came with it will have to do.

[indole]

Solder paste, stencil and hot air reflow.  Looking at the datasheet there is a 8-SOIC version, but max 25A of current.  They'd be substantially easier to solder than a QFN.

You may want to check out Proto Advantage, they have tons of SMT to DIP adapters and stencils and you may find a suitable footprint for the QFN. 
http://www.proto-advantage.com/store/index.php?cPath=2200_2218

I suspect it can handle 30A because it's a hall effect sensor and is effectively sensing the magnetic field produced by the passing current.  In the datasheet it's stating a 1.2mOhm internal conductor resistance, resulting in approximately 36mV drop at 30A or ~1W.

[con-f-use]

I second the hot air + (good quality!) solder paste. Don't even bother to try dead-bugging it, it's just frustrating. I soldered a NOA1212, which has the same pin pitch but is over-all smaller. I made a custom breakout-board at home and used toothpicks to apply the paste. So it's doable even without a stencil. Failiure rate was about 10%.

The 30A don't surprise me too much. You don't see how the contacts continue inside the package. I imagine, there's a mm of copper above. The .6 mOhm resistance (for QFN package) gives about halve a watt dissipation at full current and it's not unrealistic.

[tggzzz]

To avoid overheating, ensure there is a sufficiently good thermal path from the IC. "Sufficiently good" includes the PCB layout, construction and size. The manufacturer will often give guidance.

[jdraughn]

Have you used Eagle or anything similar to design a PCB yet?  If it were me I would design a breakout board for it and then order the boards from Oshpark.com and solder it using solder paste/hot air. It would be kind of a pain to just dab the right amount of solder paste onto the small pads so I would probably end up ordering a stencil too from oshstencils.com.

[con-f-use]

It would be kind of a pain to just dab the right amount of solder paste onto the small pads so I would probably end up ordering a stencil too from oshstencils.com.

Depends on how many you want to solder onto your newly created breakouts. For prototyping one usually needs only one or two. Hardly worth a stencil. Also with such small parts, cheap manufacturers sometimes screw up with HASL pads. They tend to fuse them together when coating. I've had to wick up excess solder from some osh and dirtypcb boards. My process to get the solder paste onto the pads is:
1.) Get toothpicks and dip them into the paste (paste in a small pot)
2.) Look at the paste whiskers at the tip
3.) Adjust paste's viscosity with solder flux
4.) Repeat steps 1-3 until the whiskers are thin enough for the pads
5.) Apply thinned paste whiskers to pads
It's nothing you want to do for hours, but for two or three chips it's less of a hassle than a stencil and much cheaper.

To avoid overheating, ensure there is a sufficiently good thermal path from the IC. "Sufficiently good" includes the PCB layout, construction and size. The manufacturer will often give guidance.

On page 13 of the datasheet. I wonder what the via under the pad means to accomplish. Usually I'd say its bad style because it decreases contact area and might make problems when soldering. As I said, the package will have to dissipate 0.6 W tops if you're insane enough to pass 31A through it. It can take 125 deg C (85 for the K variant). So even at a six times crappier board design than they quote, its temperature won't exceed the 125 C (ambient 24 C + 6 * 24C/W * 0.6W ~ 114 C - 4* for the K). For most applications, thermals won't matter much for this chip. Of course you don't want to push it to its limits nor fry a steak with your board when the summer is hot. Also dependability, accuracy and longevity correlate with temperature. All I'm saying, you'll need to work hard to screw up thermals for this chip.

[jdraughn]

I never thought about mixing in flux to change the viscosity, that would make it much easier.

It would be kind of a pain to just dab the right amount of solder paste onto the small pads so I would probably end up ordering a stencil too from oshstencils.com.

Depends on how many you want to solder onto your newly created breakouts. For prototyping one usually needs only one or two. Hardly worth a stencil. Also with such small parts, cheap manufacturers sometimes screw up with HASL pads. They tend to fuse them together when coating. I've had to wick up excess solder from some osh and dirtypcb boards. My process to get the solder paste onto the pads is:
1.) Get toothpicks and dip them into the paste (paste in a small pot)
2.) Look at the paste whiskers at the tip
3.) Adjust paste's viscosity with solder flux
4.) Repeat steps 1-3 until the whiskers are thin enough for the pads
5.) Applied thinned paste whiskers to pads
It's nothing you want to do for hours, but for two or three chips it's less of a hassle than a stencil and much cheaper.

[con-f-use]

I never thought about mixing in flux to change the viscosity, that would make it much easier.

I never really thought about, what solder paste really is. One day I had to inspect a tiny part with magnification and saw little balls of solder with rosin flux in between them. That's when the idea was born. It's kinda obvious, when you see it, but still a neat trick, you don't think of. However, solder paste composition is optimized and thinning it might have side effects on production stuff. For hobby it's just great and never bit me. You don't need much and just so, life is a bit easier.

[tggzzz]

To avoid overheating, ensure there is a sufficiently good thermal path from the IC. "Sufficiently good" includes the PCB layout, construction and size. The manufacturer will often give guidance.

On page 13 of the datasheet. I wonder what the via under the pad means to accomplish.

A thermal path to a copper plane on the other side of the board will reduce thermal resistance and reduce the temperature rise.

Note the use of 2oz copper, not the more usual 1oz.

Usually I'd say its bad style because it decreases contact area and might make problems when soldering. As I said, the package will have to dissipate 0.6 W tops if you're insane enough to pass 31A through it. It can take 125 deg C (85 for the K variant). So even at a six times crappier board design than they quote, its temperature won't exceed the 125 C (ambient 24 C + 6 * 24C/W * 0.6W ~ 114 C - 4* for the K). For most applications, thermals won't matter much for this chip. Of course you don't want to push it to its limits nor fry a steak with your board when the summer is hot. Also dependability, accuracy and longevity correlate with temperature. All I'm saying, you'll need to work hard to screw up thermals for this chip.

The complete thermal calculation is necessary. Only you have sufficient information for that to be possible.

[Fat]

Be a bit careful with how much flux you add.  I thinned some that way once and got too much in it. Flux bubbled and blew the chip off the board in my reflow oven.  Was interesting to watch. The chip actually raised up on the board, kind of like tombstoning.  I thought it would settle down and then it blew right off the PCB. I was using some very cheap paste I had bought from China.

Fat

[Red Squirrel]

Hmm guess I'll try the paste method with a tooth pick.  I have a magnifier so I'll try to do it by hand very carefully and apply directly to pads.  I'm going to want to connect that to something obviously, probably a wire, what is the best way to do that?  Do I just glue wires down in correct pattern and lay the chip on top?    Getting PCBs done is probably not worth it for such a small volume.  I've been playing around with KIcad so eventually want to look into it.  I'd probably make some break out boards in bulk for various type of packages.

For the amperage capacity I guess since it's such a short path of lower gauge it would not be as bad then?  It just surprises me it can handle that as something as simple as a loose wire nut on a 15 amp circuit can be problematic but guess voltage may play a role too, these would not have 120vac passing through them.   I guess a large pad would also act as a heat sink.  In my case I'll probably break out to 18awg wire somehow but I'll limit current to like 10 amps at 12 volts.

[con-f-use]

I'm going to want to connect that to something obviously, probably a wire, what is the best way to do that?

What the others and I described was exclusively for use with a breakout board. As I said using wire and dead-bugging it is neither a pleasant nor a good option. I'm sure someone, who makes boards at home would be willing to roll a batch for you. Also dirtypcb or the others are not that expensive and you can cram a lot of brakouts into their smalles 5cm x 5cm board area.

these would not have 120vac passing through them.

"The Allegro TM ACS711 provides economical and precise solutions for AC or DC current sensing in <100 V"

[Red Squirrel]

The premade breakout boards look interesting but I just have trouble spending more money for that than the chip itself cost not to mention shipping and customs on top of it and having to wait a few weeks.  I guess I will experiment with a way to solder wires to it. If I can lay it on top of wires that were somehow secured with paste on it then bake it, maybe?  Trick is any glue I might use will probably melt before the solder so glue probably wont work to secure wires.

[con-f-use]

I think it won't work reliably and you will end up ruining more chips than you can use. But than again, people have done .

Breakouts are not that expensive and should be under the customs threshold. In the end it will save you money and time and nerves. If its till to expensive, etch the breakout yourself. If you have a laser printer, toner transfer is really easy and cheap.

[Red Squirrel]

Tried to dead bug it, not happening.  If I spent enough time I probably could manage but it's quite tedius and I have at least 2 to do if I use these for my project.

I definitely should have double checked the size to get a better idea of how small it would be.  I need this to work by the 14th so that's not going to happen if I order anything, so think I'll just cut my losses and not use this in the project.  In the future when I learn kicad more to the point where I can make boards I might just get some boards made, but I'll want to wait till I get to a point where I want to make an actual project so make the cost more worth while.  Basically I can just throw in some break outs in blank areas of the panel for whatever project I decide to do as it's probably not worth spinning a board for just this.

[indole]

If you really need to get it working, digikey has an evaluation board for around 33$CDN.  When I order from Digikey I usually get it the next day if ordered before 4pm EST.   Here's the product:
http://www.digikey.ca/product-detail/en/ASEK711KEX-31AB-T-DK/620-1595-ND/4867659

You could also get the 8-SOIC version, which will be much easier to solder.  $4.29
http://www.digikey.ca/product-detail/en/ACS711ELCTR-25AB-T/620-1371-1-ND/2470595

Or if you have op amps on hand and the appropriate resistors, could just make a current sense circuit: http://cds.linear.com/docs/en/application-note/an105fa.pdf

[Red Squirrel]

Yeah 33 seems a bit steep, I can live without it.   My original plan was actually to use a small shunt resistor and op amp but was told an op amp does not operate linearly so it would not work, but all the reading I've done on op amps seem to me it would work? So will that in fact actually work?  I have not played with op amps yet but do have some here (LM358N) that I got from my last Digikey order but if another op amp is better can order that too.  Since yeah, Digikey shipping is very fast so ordering something new from there is probably my best option. Worse case I may indeed just get the soic version of those tranducers and try to dead bug that, probably easier.  Really not sure why I did not do that from get go.

[indole]

Should be doable with an LM358, will most likely need to run in split supply since the LM358 isn't rail to rail.   Here's a quick simulation I threw together in LTspice:


Gives 1V/A so if you need to sense a larger current just adjust the resistors accordingly.  You will need to take into account heat dissipation into the sense resistor as well, if you don't have a 100mOhm resistor can parallel up a bunch of 1Ohm or 10Ohms and that will also help spread the heat out.  I believe Dave did that with his uSupply project.  Take a look at his schematic for the uSupply at http://www.eevblog.com/files/uSupplyBenchRevC.pdf

[janekm]

Allego do make through-hole current sensors too btw: http://www.allegromicro.com/en/Products/Current-Sensor-ICs/Fifty-To-Two-Hundred-Amp-Integrated-Conductor-Sensor-ICs.aspx (though for 50-200A the hole is not that small )

They are much preferable to a sense resistor in high-current applications as you won't have a significant voltage drop and power dissipation in comparison (100uOhm conductor resistance...).

[Red Squirrel]

Yeah was looking at those for a separate project that requires 100a reading but since I'd only be using a fraction of the range it would end up requiring an op amp or something anyway.

Split supply will add a lot of complexity so I want to try to avoid it.  Is there a parameter in op amps to look for to not need that?  Is rail to rail and instrumentation all I need to look for?

[indole]

Rail to Rail single supply should do it.  When choosing you'll want to take into account the GPB, and offset voltage.  I'm assuming you won't need it for a high speed applicaton so a <1MHz GPB should be sufficient.
You could also look at specialized op amps for current sense applications:  http://www.digikey.ca/product-search/en?pv161=58&FV=fff40027%2Cfff800dd&mnonly=0&newproducts=0&ColumnSort=0&page=1&stock=1&quantity=0&ptm=0&fid=0&pageSize=500
The LT6106 might do the job: http://www.linear.com/product/LT6106.

[Red Squirrel]

Is voltage input offset what I'm looking at to know the minimum voltage it will take before it starts to react?  Lower the better right?  Thinking something like this: http://www.digikey.ca/product-detail/en/ALD1722PAL/1014-1095-ND/2414371

[indole]

It will depend on your application, but if you're not expecting to measure very low currents it will most likely not make a difference.  The offset voltage is added to the output so if the offset voltage is 1mV and you're op amp is set for a gain of 1V/A then it will account for an error of +/- 1mA.  *I could be mistaken, just my understanding of it*
Dave has a good series on op amps where he goes over offset voltage, bias current etc, should check it out.
https://www.youtube.com/playlist?list=PLvOlSehNtuHu2FviAaZaiyXwN41G4b1Lf