EEVblog Electronics Community Forum
General => General Technical Chat => Topic started by: Poe on July 18, 2012, 09:19:38 pm
-
Just watched the EEV blog #314 and noticed the tinned traces intended to increase a trace's current carrying capability.
Since solder has something like ~1/10th the conductivity of copper, how effective would this be? .
Has anyone looked into this technique? I've seen it used quite a bit in older boards, but it never made much sense to me.
-
Just watched the EEV blog #314 and noticed the tinned traces intended to increase a trace's current carrying capability.
Since solder has something like ~1/10th the conductivity of copper, how effective would this be? .
Has anyone looked into this technique? I've seen it used quite a bit in older boards, but it never made much sense to me.
I also raised my eyebrows when Dave mentioned it.
Maybe it will even make it worse .... but right now i cant remember there i did read/hear that.
-
If you open a pretty standard desktop PC PSU and take a look on the main PCB, you'll see this technique all over the whole board.
The main intention is to reduce cost. 2oz or more copper thickness gets really expensive, as Dave said in the video.
I also don't think, that tinning the traces makes things worse. At least there is more conductive material available, so the resistance will go down in any case, that's for sure. The only question that arises, is the effectiveness of such tinning, because you wouldn't be able to get a even thickness of tin along the whole trace.
-
I also don't think, that tinning the traces makes things worse. At least there is more conductive material available, so the resistance will go down in any case, that's for sure.
Yes the resistance will be a little bit lower. Anyone with a good low Z multimeter there want to make some test ?
But that if the track cant get rid off the heat. ? The Tin must act like an heat insulator or do you think it will be better to get rid off the heat ?
-
Yes the resistance will be a little bit lower. Anyone with a good low Z multimeter there want to make some test ?
1oz copper is 35um. Putting 0.5mm of tin on a 1oz track would more than halve its resistance.
or do you think it will be better to get rid off the heat ?
More surface area, no solder resist, slight protrusion into air flow. I would say dissipation is likely improved slightly.
-
1oz copper is 35um. Putting 0.5mm of tin on a 1oz track would more than halve its resistance.
More surface area, no solder resist, slight protrusion into air flow. I would say dissipation is likely improved slightly.
Correct.
It can't do anything but help reduce the resistance and improve the dissipation (which is now lower thanks to the lower resistance)
Dave.
-
Since solder has something like ~1/10th the conductivity of copper, how effective would this be? .
IIRC, copper is about 2-3 times better, not 10 times better.
Dave.
-
The specific resistance (1m*1mm^2) of copper is 0,0178 ohm, tin is 0,1150. Thats a factor of 6,46.
For historic comparison: lead with a specific resistance of 0,2080 is even less condutive (11,68 times compared to copper), so the old Sn60Pb40 solder has a specific resistance of 0,1522 or 8,55 times worse than copper.
-
I do not think that the unmasked strips in the power supply were there to allow the solder to reduce resistance.
I think the track's design margin is so low, that they thought they may need to solder a length of copper wire on top of the tracks. That is why they only left a thin strip unmasked, rather then the whole track width.
They obviously decided they could get away with no copper wire, but if I owned the supply, I would add them.
-
Soldering copper wired onto tracks actually works rather well but also has some potential drawbacks. (Say potential vibration problems over time.)
There are also options to embed such wires into or on top of the PCB, but that is rather expensive and thus only used in very high power stuff. (Copper Inlay PCBs)
-
Resistivity figures from Wikipedia, adjusted to have same exponent
Copper 1.68×10?8
Tin 10.9×10?8
Lead 22×10?8
I've just done some actual measurements - short video uploading now..
-
I've just done some actual measurements - short video uploading now..
Poop, I was going to do that! :P
Dave.
-
I've just done some actual measurements - short video uploading now..
Thank you for taking time to do that. It will be very interesting info for me at least.
-
Does putting solder on high current PCB tracks help? (https://www.youtube.com/watch?v=Gy1K3ayPfOk#ws)
-
I've just done some actual measurements - short video uploading now..
Poop, I was going to do that! :P
Dave.
Yesss!!! Battle of the tear down titans. ;)
-
I would have thought that because of its low melting point, solder's conductivity would rise fairly fast as more current passed though it and heated it. But actual measurement is better than talk... time to watch the video.
edit: and of course I'm wrong, resistivity increases with temperature, not conductivity.
-
Does putting solder on high current PCB tracks help? (https://www.youtube.com/watch?v=Gy1K3ayPfOk#ws)
Mike a great video as always.
40% improvement is not bad i must admit ;)
That about long term use. I think about bad soldering there can be seen at hot component in old pcb (TV). Just a thing to consider.
-
Why is the video unlisted? Even the discussion originated here, I imagine the video can be interesting to "anyone".
-
It would be interesting to add fresh solder back to that track and see how much a heavy (manually added) solder coating affects the resistance.
-
I would have thought that because of its low melting point, solder's conductivity would rise fairly fast as more current passed though it and heated it. But actual measurement is better than talk... time to watch the video.
edit: and of course I'm wrong, resistivity increases with temperature, not conductivity.
Yes the equation to calculate ohm is not as easy as with pure CU.
-
Why is the video unlisted? Even the discussion originated here, I imagine the video can be interesting to "anyone".
Because I forgot to change it - I always set unlisted during upload to check it's OK and select the thumbnail image before publishing.
-
Is the 'normal' track there going to struggle with 1A? I would have thought it would have done ok, what happens at higher current?
-
I now also realise that my solder wick is shithouse. What brand do you use?
-
I now also realise that my solder wick is shithouse. What brand do you use?
Equally importantly, ask what iron he used ;D
-
I now also realise that my solder wick is shithouse. What brand do you use?
I use Chemwick brand and find it very good. Some of the cheap ones are rubbish. The bigger issue is usually people putting their fingers all over the braid. This contaminates it and really reduces its effectiveness.
-
Chemtronics Soder-wick (dispite the misspelling!), Iron is a Metcal
-
Is the 'normal' track there going to struggle with 1A? I would have thought it would have done ok, what happens at higher current?
You can work it out - I^2*R = 26 milliwatts dissipated at 1 amp, so negligible. On a PSU it will be more about reducing voltage drop - there are lots more heat sources on a PSU to worry about before track heating becomes a major issue.
-
@mikeselectricstuff
I assume that is 1oz copper on that board, is that correct?
And it was pb/sn that you removed.
-
Thank you mike. It definitely looks effective.
Although I wonder what kind of solder that was. There appears to be a significant conductivity range between solder types based on this chart from Indium:
http://www.microwaves101.com/encyclopedia/SolderChart.cfm#chart (http://www.microwaves101.com/encyclopedia/SolderChart.cfm#chart)
2% to 23% = 4x to 50x the resistance of copper? Is that right?
If the 4x stuff was removed and the 50x stuff added, the gain would be negligible. Other way round and the gain substantial. Does anyone know of 'typical' solders used in production environments today (pb-free types) and their conductivity range?
You might need to to define a solder type in your assembly documentation if the range is large.
Mike,
Not to be a pedantic bother, but could you apply 60/40 so we have a base for comparison?
-
I just shot a video on this. Will upload in due course.
I was able to get better results than Mike.
Dave.
-
I just shot a video on this. Will upload in due course.
I was able to get better results than Mike.
Dave.
...ya, but he was first!
;D
Thank you.
-
...ya, but he was first!
But I did it the proper way! :P
Dave.
-
Solder on PCB tracks - thermal test (https://www.youtube.com/watch?v=DUk9lt-PeRU#ws)
-
I just found a length of non cored solder left behind by the plumber. 200mm X 3mm dia. measures 0.53 ohms on my meters using croc clips for connection.
-
Matweb data
60/40 solder Electrical resistivity 0.00001499 ohm-cm
http://www.matweb.com/search/datasheet.aspx?matguid=06a31d97bb734b509043d81cf131b280&ckck=1 (http://www.matweb.com/search/datasheet.aspx?matguid=06a31d97bb734b509043d81cf131b280&ckck=1)
Oxygen-free high conductivity Copper, Soft, UNS C10200 Electrical Resistivity 0.00000169 - 0.00000173 ohm-cm
http://www.matweb.com/search/DataSheet.aspx?MatGUID=9aebe83845c04c1db5126fada6f76f7e (http://www.matweb.com/search/DataSheet.aspx?MatGUID=9aebe83845c04c1db5126fada6f76f7e)
60/40 solder is 8.76 times more resistive
-
Mike,
In your thermal test, it looks like the track with the solder had the heat spread much more evenly. That should mean that the soldered track has a better ability to loose heat via air conduction and radiation, but the shiny solder is probably radiates less heat then the coated track.
So overall, the end result is probably very similar.
It would be interesting to know how the coated copper track compared to the tinned track if you wick up most of the solder. Does a thinly tinned track get noticeably hotter then the masked track?
The risk with relying on the wave soldering to deposit solder on the track is that to get a thick layer of solder, the track has to be parallel to the solder wave. If the assembler puts the board through the wave soldering with the track at right angles to the wave, the thickness of the tinning will be much less.
Richard.
-
For serious current capability, solder copper wire or precut copper foil or sheet to the unmasked traces on the PCB. You can achieve anywhere from a few oz to well over 100 oz copper traces on demand this way.
Sometimes having 400+ amps and fine pitch components on the same board is a nice thing to be able to do.
-
For serious current capability, solder copper wire or precut copper foil or sheet to the unmasked traces on the PCB. You can achieve anywhere from a few oz to well over 100 oz copper traces on demand this way.
Sometimes having 400+ amps and fine pitch components on the same board is a nice thing to be able to do.
I've seen this done a number of times on higher quality power supplies. They just take a section of 18awg solid wire and snake it around where they need it. Down side here is that it has to increase the production time and labor costs substantially. So I would think this would be best left to where you need some really high current capacity in a limited amount of space, or the price bump won't have any significant effect on the profit margin. Anywhere 5 bucks can be a deal breaker on a final product, I can see where just tinning a trace would be the preferred solution.
-
I have just found that the new set of croc clips I got have a bad crimp joint and that is where the high resitance came in my measurement of solder.
I soldered the solder to the meter leads and got 0.1 ohms per meter with 3mm dia solid solder, no flux cores.
-
I have just found that the new set of croc clips I got have a bad crimp joint
The hours i have lost over the years to bad test leads :-[
The first thing i do with new croc clips test leads is solder the "crimp" before use.
-
I have just found that the new set of croc clips I got have a bad crimp joint and that is where the high resitance came in my measurement of solder.
I soldered the solder to the meter leads and got 0.1 ohms per meter with 3mm dia solid solder, no flux cores.
That is better, but still not quite there. You need to do a 4 wire measurement to test this accurately. Incidentally if you were doing a 4-wire measurement the bad crimp wouldn't have corrupted your measurement. 3 mm of lead-free solder should be about 0.02 ohm/meter.
-
For serious current capability, solder copper wire or precut copper foil or sheet to the unmasked traces on the PCB. You can achieve anywhere from a few oz to well over 100 oz copper traces on demand this way.
Sometimes having 400+ amps and fine pitch components on the same board is a nice thing to be able to do.
I've seen this done a number of times on higher quality power supplies. They just take a section of 18awg solid wire and snake it around where they need it. Down side here is that it has to increase the production time and labor costs substantially. So I would think this would be best left to where you need some really high current capacity in a limited amount of space, or the price bump won't have any significant effect on the profit margin. Anywhere 5 bucks can be a deal breaker on a final product, I can see where just tinning a trace would be the preferred solution.
wouldn't it be possible during manufacturing to mask off the board and electroplate extra ounces of copper on traces that needed high current carrying capabilities.
sometimes the board is done with a tin resist and electroplating to build up a 1/2-oz copper blank into a 1-oz copper board.
Something like this...
1. Drill all of the holes in a 1/2 oz. copper coated board.
2. Electroless copper coat to activate the vias.
3. Electroplate enough to stabilize the vias.
4. Mask the board to expose only the areas to be saved.
5. Electroplate 1/2 oz. of copper.
6. Electroplate tin.
7. Strip resist mask.
8. Etch in ammonium persulphate.
9. Strip the tin resist
10. Apply solder mask
11. Dip in solder for HASL
new steps 4 and 5 (and extra steps in between)
4. Mask the board to expose only high current traces
4.1 Electroplate 3 oz of copper
4.2 Strip the resist mask
4.3 Mask the board to expose all the traces to be saved, including high current traces
4.4 Electroplate 1/2 oz of copper
5 Electroplate tin
At this point, the board has 1 oz of copper everywhere, and 4 oz of copper on high current traces, and TIN resist everywhere.
I don't know if this would work. There might be severe undercutting on the 4oz copper areas, or maybe manufacturing processes wouldn't allow for the extra steps. The PCB designer would need to create an extra mask.
This is not "free" like the exposed soldermask, but is also not extra work during assembly like snaking wire or soldering copper foil or straps.
-
I like the way you think codeboy2k.
I have no idea if that would work though since I've never done plating.
-
Can be done, but will add signifigant cost to the board for very little gain. You would be better off to have a laser cut copper foil and place it on the board with solder paste and then IR reflow it.
If you need the high current tracks then you will use the 2 or 4oz copper and add a mezzanine board with the fine pitch components.
Otherwise you will add a copper wire to the board, or use solderwick and solder it to the appropriate places.
-
Can be done, but will add signifigant cost to the board for very little gain. You would be better off to have a laser cut copper foil and place it on the board with solder paste and then IR reflow it.
If you need the high current tracks then you will use the 2 or 4oz copper and add a mezzanine board with the fine pitch components.
Otherwise you will add a copper wire to the board, or use solderwick and solder it to the appropriate places.
Yeah these ways are certainly way cheaper. I like the idea of a mezzanine board of 1oz copper on top of a or beside a 2oz or 4oz board., or reflow a copper foil trace on top of existing traces that need a current upgrade.
-
Now here's something interesting from an LG LCD TV power supply. Ignoring the rather scorched MOSFET for now the one thing that is apparent is that they've tinned the traces, but they've done diagonal stripes on larger planes. Also, the pitch seems to be different on the output side. Why?
-
I'm pretty sure it's because their "fattening" method is so poorly thought through, that it'd create HUGE blobs on larger
surface areas. I've seen it on earlier REALLY cheap P/Supplies. I've had cases of 1 in 5 failing, and upon opening them up
found huge dangly blobs just shy of the case. I think the method they use is melting the solder in a WOK and flinging it
on to the PCB. Having many narrow tracks limits this effect.
-
Thank you mike. It definitely looks effective.
Although I wonder what kind of solder that was. There appears to be a significant conductivity range between solder types based on this chart from Indium:
http://www.microwaves101.com/encyclopedia/SolderChart.cfm#chart (http://www.microwaves101.com/encyclopedia/SolderChart.cfm#chart)
2% to 23% = 4x to 50x the resistance of copper? Is that right?
If the 4x stuff was removed and the 50x stuff added, the gain would be negligible. Other way round and the gain substantial. Does anyone know of 'typical' solders used in production environments today (pb-free types) and their conductivity range?
You might need to to define a solder type in your assembly documentation if the range is large.
Mike,
Not to be a pedantic bother, but could you apply 60/40 so we have a base for comparison?
Just got word back from our factory. We commonly use Cookson make SAC-305 for Pb-free assemblies. It is 44% more conductive than 60/40. The other Pb-free alloy we run which has "better performance" has 52% less conductivity. On Pb assemblies we use 63/37 which obviously is just a tad more conductive than 60/40.
Based on alloy alone, the parallel conductivity could vary three fold!
Since thickness is a crap-shoot, maybe it's a good idea to verify the conductivity of your assembler's alloy when using this technique?
Oh ya...
The SAC-305 conductivity is 16.6% IACS
99Pb/1Cu 12.6% IACS (similar to Dave's latest video)
63/37 is 11.9% IACS
60/40 is 11.5% IACS
MX alloy is 5.5% IACS
Copper obviously >100% IACS
-
WRT the LG PCB I guess that these are panelised, and can go through the wave bath either way, so the diagonal is the best compromise. The design is probably rotated so that a maximum number of mixed boards for that model will fit on a single larger panel.