I've always used 0603 (1608 metric) size passives as the minimum size in my designs to optimize manufacturing. I'm told the smaller the part, the more likely it is to tombstone, stand up with only one end soldered.
The common sizes below 0603 seem to be 0402 and 0201. 0402 would not seem to have a big size reduction, but if it's an advantage without penalty, why not take it. 0201 is much smaller, but I wonder about how well it would work.
What sizes do you use, smaller than 0603? How much trouble do you have with them?
0603 is a nice size if hand soldering is needed for prototypes. The smaller sizes are sometimes needed for closely packed decoupling capacitors around processors and fpgas. The smaller sizes have less inductance. If dc blocking capacitors are needed on high frequency transmission lines then 0402 or 0201 are useful because their width is closer to that of the transmission line and they therefore cause less of an impedance mismatch. The tradeoff is that smaller capacitors may need more exotic dielectrics which have worse voltage and/or temperature coefficients. For resistors, similar tradeoffs seem to happen. High precision resistors are easier to make in larger packages. High voltage components are easier to make in larger packages. Power dissipation in resistors is more complicated, generally the larger the better. Contract manufacturers often refer to anything smaller than 0603 as "dust parts". If you drop them they are gone forever!
They also treat $100, 20 pin TSSOPs as "dust" parts. Seems the machine sometimes doesn't like a part orientation and tries to drop it in a bin, but the part goes flying instead!
Not worried about any high frequency issues. I will have a BGA on the board and the 0603/1608 will take up all the space of two balls on the back side. So an 0402/1005 with the pads would be about two ball areas. An 0201/0603 would just be one ball area and even more mountable around the vias.
Any parts that can't be had in X5R, I'd bump up the size. It's not like I don't have any 0805 or 1206 parts on the board.
I'm looking at replacing some op amps that can't be bought with a class D speaker driver chip. Seems to quell the high frequency noise, they indicate significant inductors. They go up in value with higher impedance loads for some reason. They have 33 uH with an 8 ohm load, so I'm worried I'd need hundreds of uH for a 50 ohm load. One on each of four outputs would be a bit much. So I'm looking at shrinking everything else I can.
That sounds like the often seen step in instructions, to do something "carefully". They never say just how "carefully". I never know if I should use the NIST traceable "careful" or if the generic, low-ball Chinese "careful" is sufficient. Not to mention when they don't bother to distinguish the Imperial "careful" or the metric "carefûl". It might only be 1 part in 10 different, but how do I know if that's important or not?
A balanced filter with a pair of 33uH inductors driving an 8 ohm resistive load will have a low-pass corner frequency of about 19kHz. A higher load resistance will need more inductance to get the same corner frequency. John
A higher resistance will have a higher corner frequency, no? The switching noise is what needs to be eliminated. That's around 400 kHz, or 20 times higher. I guess the idea is to have significant suppression at 400 kHz, with minimal impact on the 20 kHz pass band. Funny, that this corner frequency being pushed close to 20 kHz would seem to be the design point, yet the different makes of these chips specify different values of inductors, although not wildly different. I guess some are trying not to crowd the pass band.
So it's not likely I'll be able to use a class D amp then. The class AB amps seem to be limited to 5V power as they are mostly intended for earphones.
The LM8272 was a perfect part for the job, with few alternates. I guess that's why it's still such a long lead time while other parts seem to be easing up and more available. I used the LM8262 for two sockets that aren't driving the outputs, but I still need these parts for the output drivers. Space is tight on the board, and the stereo differential drivers would have freed up some space.
I still don't get how the semiconductor market was in fine shape before 2020 and now, three years later, still has not settled. It's not like anyone retired production capacity. Is anyone planning to build more capacity off the bleeding edge?
We use 0603 as our smallest, too. Sometimes I would like to use smaller but we rarely do. If so, we place smaller parts, usually smaller that 19 mil pitch ICs by hand.
What I am hearing from vendor(s) is that 0603 and larger parts will not be increasing production for those sizes. Not getting rid of them but just not increasing production scaling.
I hope we can get newer machines before we have to eventually go to
I've used down to 0201, mainly when there were no other variants available such as with very high speed sampling diodes. Generally, like others here, I stick with 0603 but when board space is at a premium then
0402 is my standard.
None. Tombstoning isn't much of a problem in a well-run reflow soldering process. I also have not noticed an increase in issues with smaller sizes of discretes. Large QFP sometimes produced solder shorts but that was usually related to screeding on too much paste.
While on the plane, I realized what is wrong with your statement. First, the output filter is LC, not RL, which makes a huge difference. This filter is not the low pass for the audio as such. It is a low pass to filter the 400 kHz noise. So the exact corner is not important, other than being well below 400 kHz, and in fact, should not be too close to 20 kHz, so as to not over attenuate the anti-alias filter on the DAC.
I did some calculations for a 40 kHz corner frequency and got an LC product of 16E-12. What remains is to decide how to apportion that product to each component. Taking a square root gives 4 uF and 4 uH as a starting point. The reactance of each component at 20 kHz (the highest audio component of interest) is 0.5 ohms for the inductor and 2 ohms for the capacitor. So this circuit will not be compromised by the 50 ohm load. Even an 8 ohm load would have much impact.
My goal is to reduce the size of the inductor, since there are four of them. This lowers the impedance of both the cap and inductor further, making the load less significant. However, the load on the amp needs to be considered. The above 4 uF/4 uH filter, produces 1.5 ohm load (per side) on the amp output. That's not so good. So I will need to increase the inductance and reduce the capacitance to get a higher impedance load for the amp output. So maybe a 10 uH inductor and a 1.5 uF cap? It gives 4 ohms load to the amp output at 20 kHz. 15 uH and 1 uF gives 6 ohms.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.