Going form 0603 to 0402

[shadewind]

I've been using 0603 as a minimum size for my resistors and caps now for a while and I find it very simple to work with, both with an iron and with hot air reflow. On my latest project, I'm going to use a TI CC2500 proprietary RF transceiver chip and the data sheet strongly recommends using as small caps as possible, preferrably 0402. Especially the balun part (I'm using the reference design) seem picky.

My question is first, will it matter that much if I use 0603 instead of 0402. If it does matter, how hard is it to go down to 0402 if I find working with 0603 pretty effortless? Is it easier to do with hot air than with an iron? I have a good (Weller WD1000) soldering station with a good selection of tips with different sizes and good eyesight.

If I have 0402 in some places, I'm thinking I might as well have them all across the board. Any reason to NOT use 0402 all the way (when the values allow, of course)?

[Simon]

well I've never done much SMD stuff myself but I'm guessing that if you can easily do the larger size you may well cope with the smaller one. Why not give it a go ? drop a few smaller size parts in and see how you cope until you do a full transition.

[Neilm]

The main reason they may be recommending 0402 components is for EMI performance. Generally, as far as caps are concerned the bigger the component, the more inductance they will have. The more inductance means that they will have lower EMC performance than the smaller component. (self resonance).

I generally use 0603 components - I don't recall ever having to hand assemble but I would have thought it was possible with a decent magnifier.

Neil

[Conrad Hoffman]

After using the smaller parts for a while, 0603 will seem huge! I haven't been allowed to use an 0603 for years.

[mikeselectricstuff]

0402s are significantly more fiddly, but doable if you're comfortable with 0603 - they have an increased tendancy to tombstone when using paste and and stick to the iron when soldering with wire -  the smaller size means it doesn't take long for heat to transfer from one end to the other.  The latter can be a  plus when desoldering though as heating one end will usually melt both.

And of course for resistors, 0402 is  the point at which they stop marking them.

[IanB]

When people say "0402" without qualification, do they mean 0.04" x 0.02" or 0.4 mm x 0.2 mm?

Because I've seen (or rather haven't seen) some of those teeny tiny parts, and I really can't tell them apart from a blob of solder without a microscope...

[mikeselectricstuff]

0402 means 0.04 x 0.02"
You do occasionally see  metric sizes but these are not commonly used, and only cause confusion.

[joelby]

If you're comfortable with 0603, 0402 shouldn't be much of a problem if your eyes are okay or you have magnification. They can be a little fiddlier and I dot use them unless I have to, but it's rare that you need to.

[EEVblog]

When people say "0402" without qualification, do they mean 0.04" x 0.02" or 0.4 mm x 0.2 mm?

It's almost always imperial unless otherwise stated.
0402 imperial is infinitely more commonly talked about than 0402 metric.

Dave.

[EEVblog]

I've been using 0603 as a minimum size for my resistors and caps now for a while and I find it very simple to work with, both with an iron and with hot air reflow. On my latest project, I'm going to use a TI CC2500 proprietary RF transceiver chip and the data sheet strongly recommends using as small caps as possible, preferrably 0402. Especially the balun part (I'm using the reference design) seem picky.

My question is first, will it matter that much if I use 0603 instead of 0402. If it does matter, how hard is it to go down to 0402 if I find working with 0603 pretty effortless? Is it easier to do with hot air than with an iron? I have a good (Weller WD1000) soldering station with a good selection of tips with different sizes and good eyesight.

If I have 0402 in some places, I'm thinking I might as well have them all across the board. Any reason to NOT use 0402 all the way (when the values allow, of course)?

No, it's most likely won't matter much if you use 0603 instead of 0402, unless you are on some bleeding edge performance wise.
If you are forced into 0402 for some reason, then it's common to then use 0402 for other things too, as you have already paid any potential penalty in terms of manufacturing issues and cost.
But if you stuck with 0603 for most other components instead of 0402 then your yield could potentially be higher, and that can pay dividends.

Dave.

[shadewind]

I realize that the decoupling caps may not be that critical. If I follow the reference design decoupling (which consists of 220 pF caps as well), I'll probably be better off than I usually am.

The RF balun on the other hand I'm not sure about. The components are in the area of 1 nH and 1 pF. Won't parasitic inductance and capacitance play a role here? The question is probably more if it plays enough role for it to make a noticable difference.

[EEVblog]

As always, "it depends".
Your board layout could swamp your component worries in this case.
Surround the parts with ground flood fill etc and you gets lots of stray capacitance for example.

Dave.

[JuKu]

My experience is that even with good tools and good microscope, there is a relatively sharp threshold in difficulty. Basically, your hand trembles a given amount; if that amount is smaller than the component, it is going to be easy. if it is larger, you'll have issues. I used to be able to do 0603 relatively easy, and it was possible to do 0402. Now, 0805 is easy, 0603 possible but I don't look forward to it, 0402 I can't do. (I hate getting old!)

[shadewind]

As always, "it depends".
Your board layout could swamp your component worries in this case.
Surround the parts with ground flood fill etc and you gets lots of stray capacitance for example.

Dave.

I think the reference design assumes flood fill since the data sheet recommends it.

I don't want crappy performance so if I have to use 0402, I will but I won't if it isn't necessary for good performance.

[mikeselectricstuff]

As always, "it depends".
Your board layout could swamp your component worries in this case.
Surround the parts with ground flood fill etc and you gets lots of stray capacitance for example.

Dave.

I think the reference design assumes flood fill since the data sheet recommends it.

I don't want crappy performance so if I have to use 0402, I will but I won't if it isn't necessary for good performance.

I don't think it's about performance as such, just that the reference design layout has been optimised for 0402s, and would probably need tweaking of values for 0603 due to different parasitic inductance and capacitances. Unfortunately you need many iterations and some seriously expensive test gear to accurately optimise this sort of  thing.

[EEVblog]

My experience is that even with good tools and good microscope, there is a relatively sharp threshold in difficulty. Basically, your hand trembles a given amount; if that amount is smaller than the component, it is going to be easy. if it is larger, you'll have issues. I used to be able to do 0603 relatively easy, and it was possible to do 0402. Now, 0805 is easy, 0603 possible but I don't look forward to it, 0402 I can't do. (I hate getting old!)

My hand assembler charge a lot more for 0402 hand assembly, because all work is performed under magnification.
0603 and above is all by eye and therefore significantly faster and hence cheaper.
Personally I can do 0402 by eye, but I much prefer to do them under a microscope.
And at the low to mid range end of pick and place machines, they seem to be generally sorted into a ones that can do 0402 and ones that can't.
And of you have 0603 and above then you might be able to get your boards machine assembled quicker and cheaper than 0402 depending upon machine availability etc.

Dave.

[shadewind]

As always, "it depends".
Your board layout could swamp your component worries in this case.
Surround the parts with ground flood fill etc and you gets lots of stray capacitance for example.

Dave.

I think the reference design assumes flood fill since the data sheet recommends it.

I don't want crappy performance so if I have to use 0402, I will but I won't if it isn't necessary for good performance.

I don't think it's about performance as such, just that the reference design layout has been optimised for 0402s, and would probably need tweaking of values for 0603 due to different parasitic inductance and capacitances. Unfortunately you need many iterations and some seriously expensive test gear to accurately optimise this sort of  thing.

Yes, that's what I was assuming. So what you're saying is that it's probably true that using anything other than 0402 will give too much error with regards to the reference design assumptions to yield good performance?

A post on the TI forum from a TI employee seems to confirm this:

"Although chip antennas will spec themselves as a 50 ohm input impedance, they generally don't explain that it is only true if you follow their design guide exactly.  What this means is the feed trace to the antenna is usually part of the input impedance transforming network.  Not also that these antennas are generally quarter wave in form so the position of the ground plane will also effect the impedance as well as the radiation pattern.

Following the schematic of figure 10 in SLAU227E is a good start but more importantly is to follow the layout design of figure 13 as well.  Small changes in the layout can create large impedance changes at 2.4GHz.  Also, you must use 0402 parts or smaller as 0603 parts will never be manufacturable with any worthwhile yield.  Not saying you cannot get a layout to work with 0603 parts but you will always be hand tuning it for each board built because the part to part variation along with the parasitics is just to much to control reasonable in manufacturing.

Jim Noxon"