Author Topic: Silly way to put an smd, but how well do you think it may work?  (Read 3695 times)

0 Members and 1 Guest are viewing this topic.

Offline Rick Law

  • Super Contributor
  • ***
  • Posts: 2509
  • Country: us
Silly way to put an smd, but how well do you think it may work?
« on: January 23, 2015, 04:28:08 pm »
This is kind of silly thought but... I like to hear your view.

I want to improve the cooling of an SMD 0.05ohm current sensing shunt.  An idea come to mind:  how about instead of the SMD laying flat on the PCB, have it lay on side.    Like an inverted T.  The top of the T (now bottom) is the PCB, and the SMD's lay on side-edge is the vertical part of the inverted T.  The solder needs to be a bigger lump to give it mechanical support.  Once soldered back on the same spot, the SMD should sticking up like a wall.

With both top and bottom side of the SMD exposed,  I doubled the surface - both top and bottom of the SMD exposed.  With it sticking up like a wall should also improve air flow.  It should improve cooling at least some, but I am not sure if such improvement will make a meaningful improvement in cooling.

Anyone done something like that before to keep an SMD cooler?  How well did it work out?

Thanks
Rick
« Last Edit: January 23, 2015, 04:30:09 pm by Rick Law »
 

Offline Skimask

  • Super Contributor
  • ***
  • Posts: 1425
  • Country: us
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #1 on: January 23, 2015, 04:30:31 pm »
What about the reduced land contact area...
I didn't take it apart.
I turned it on.

The only stupid question is, well, most of them...

Save a fuse...Blow an electrician.
 

Offline Rick Law

  • Super Contributor
  • ***
  • Posts: 2509
  • Country: us
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #2 on: January 23, 2015, 04:34:56 pm »
What about the reduced land contact area...

I am not sure what you mean my reduced land contact area.  I think you mean the original area where the bottom of the SMD touches the PCB.

If so, the original land contact area is bare PCB with no copper (on either side) other than the soldering pads for the SMD.  I think PCB without copper is not a good heat conductor, so heat would not have a way to dissipate from the bottom.
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 12345
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #3 on: January 23, 2015, 04:37:34 pm »
Especially for small chips, you can substantially increase the dissipation rating by adding solid copper pour to the terminals.

You're still limited by rated temperature, which means you must qualify the result.  No time and money for testing the prototype = don't do it.  You also don't save much space, because you need at least as much copper area as a larger part would occupy anyway.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life? We can help.
 

Offline Rick Law

  • Super Contributor
  • ***
  • Posts: 2509
  • Country: us
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #4 on: January 23, 2015, 06:31:04 pm »
Especially for small chips, you can substantially increase the dissipation rating by adding solid copper pour to the terminals.

You're still limited by rated temperature, which means you must qualify the result.  No time and money for testing the prototype = don't do it.  You also don't save much space, because you need at least as much copper area as a larger part would occupy anyway.

Tim

I don't think it is hot enough to reach rated temperature.  I am just passing <3A which is <0.15W.  What caused me to think of fixing it is the original shunt there has a bad temp-co.  It is giving me reading (mV drop) of +- 2% (perhaps 3%) between cold verses fully warm up at that current.

So if I have a 2A fluctuation at say in the low single digit seconds, current is changing faster than the temperature of the shunt can change, the reading becomes a bit wild.

This shunt is about 6mm long and about 3mm wide.  Hard to measure given limited access and solder already on the ends.  I have some 3W shunts (400ppm/1C) that hardly made a difference at that current with the mildest of air flow cooling.  This shunt is giving me 1-2% or 1-3% change.

I was thinking along these lines:
1. Try to find some way to fit my 3W shunts through-hole that I can cool better
2. Try to get a shunt with better tempco

I was leaning toward #1 - Solder the 3W shunt to the underside somehow.  I have the 3W shunts at hand and also because I am not sure of the size until I remove the SMD to measure, plus the order then wait.

This idea of mounting the original shunt on the side occurred to me in the mean time hence the post to solicit your views.  That is something I can also do without having to remove-measure-order and then wait.
 

Offline tautech

  • Super Contributor
  • ***
  • Posts: 13767
  • Country: nz
  • Taupaki Technologies Ltd. NZ Siglent Distributor
    • Taupaki Technologies Ltd.
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #5 on: January 23, 2015, 07:06:43 pm »
This is why we still use copper clad PCB's, nothing conducts heat like copper for the price.
Conduct the heat away....it is your enemy.  >:D

Lie it flat and let the traces do the work for you.

Use any, some or all of the following:
2 oz or heavier board
Thicker traces
Polygon pours as Tim has mentioned
Stitch to opposite side polygon pours.

No doubt others that have more experience have other tricks.
Avid Rabid Hobbyist
 

Offline Paul Price

  • Super Contributor
  • ***
  • Posts: 1419
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #6 on: January 23, 2015, 08:07:25 pm »
Try to get a shunt with better tempco.
 

Offline Rick Law

  • Super Contributor
  • ***
  • Posts: 2509
  • Country: us
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #7 on: January 24, 2015, 03:59:50 am »
This is why we still use copper clad PCB's, nothing conducts heat like copper for the price.
Conduct the heat away....it is your enemy.  >:D

Lie it flat and let the traces do the work for you.

Use any, some or all of the following:
2 oz or heavier board
Thicker traces
Polygon pours as Tim has mentioned
Stitch to opposite side polygon pours.

No doubt others that have more experience have other tricks.

Unfortunately, it is a purchased already made stuff that I am trying to make better.  So changing PCB is out.  Not that I needed it better, but this is a learning project.  I will wait a bit longer to see if there is more advice.  If not, I think this is what I am going to do:

0. While I wait, finish my other project.

1. Take out the original SMD, remove all remaining solder on it so I can measure it's size well.  I need that to order replacement SMD with better tempco anyhow.

2. Re-solder it on via a longer wire (so the SMD is exposed and can be measured better).

3. The longer exposed wire should also allow me to do further experimentation...  That should allow me to learn more.      I am quite sure that it is the changing temperature, but I could be wrong.

Thanks for any further input.
 

Offline ConKbot

  • Super Contributor
  • ***
  • Posts: 1094
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #8 on: January 24, 2015, 07:42:56 am »
Put it on edge, and solder resistance will probably hose your accuracy ;)
see http://www.analog.com/library/analogDialogue/archives/46-06/shunt_resistors.html
for a test on how the footprint can affect the accuracy of the measurement. 
 

Online T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 12345
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #9 on: January 24, 2015, 07:53:55 am »
Then better to find a higher power and/or lower tempco part.  Pay $$$ for it.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life? We can help.
 

Offline Rick Law

  • Super Contributor
  • ***
  • Posts: 2509
  • Country: us
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #10 on: January 24, 2015, 08:12:56 am »
Put it on edge, and solder resistance will probably hose your accuracy ;)
see http://www.analog.com/library/analogDialogue/archives/46-06/shunt_resistors.html
for a test on how the footprint can affect the accuracy of the measurement.

The board has a calibration.  It should be able to account for a change in shunt.  That is one of the thing I want to find out about this board.  I am however concerned that my "improvements" is negative.

Then better to find a higher power and/or lower tempco part.  Pay $$$ for it.

Tim

I found a few 1% shunt at tempco of merely 50ppm shunt on Mouser at around just $2.  I have to measure the size of the installed SMD first to see if the new part is of the right size.  That would require taking it out and remove the excess solder from the ends to measure more precisely.  The 0.1% shunts is almost $10.  That would be an overkill.  I just don't want the wild swing as the shunt get hot or cool down showing me phantom current changes.

I need to keep that PSU setup intact for now until I finish my current simple project.
 

Offline KJDS

  • Super Contributor
  • ***
  • Posts: 2427
  • Country: gb
    • my website holding page
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #11 on: January 24, 2015, 08:54:02 am »
Just solder a piece of copper shim on each end to act as a heatsink.


Offline Rick Law

  • Super Contributor
  • ***
  • Posts: 2509
  • Country: us
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #12 on: January 24, 2015, 11:39:38 am »
I will keep the MSC series in mind.  Thanks for pointing it out.
 

Offline ConKbot

  • Super Contributor
  • ***
  • Posts: 1094
Re: Silly way to put an smd, but how well do you think it may work?
« Reply #13 on: January 24, 2015, 01:02:04 pm »
Put it on edge, and solder resistance will probably hose your accuracy ;)
see http://www.analog.com/library/analogDialogue/archives/46-06/shunt_resistors.html
for a test on how the footprint can affect the accuracy of the measurement.

The board has a calibration.  It should be able to account for a change in shunt.  That is one of the thing I want to find out about this board.  I am however concerned that my "improvements" is negative.



 :-+ Calibration would definitely null that. But solder would have its own tempco too.  And then the question is does the improved surface area outweigh the poorer contact to the copper in the PCB, and if it does, is it more stable with the poor tempco solder in the mix. Should be interesting to see how the results turn out.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf