A while back I caught on to the fact SMD chip resistors were available in arrays of 4 (this seems to be bty far the most useful and common style), and switched from placing 4 discrete resistors to a single array, saving board space in the process. Costs are in the same ballpark, depending on the extent to which you cheap out on the resistor networks (I always get the slightly more expensive ones with concave terminals, which seem to make good joints during reflow much more reliably, With those requirements, and since I'm a baller and just throw money around on cadillac components for my designs, I buy them western distributors, meaning I might spend as much as 1.6 cents each in qty500, or more than 2.5 cents by the hundred (yes - my tastes are just tjhat extravagant you see) or 4-element CAT16-style 1206 arrays (I use the 1206 for ease of placing by hand -anything smaller and I just can;t place them accurately enough.
My next thought was "Wow, I'll bet I can get capacitor arrays like that too! Maybe pairs for crystal loading caps, or 4 0.1uF decoupling caps for nearby components in a single part"
No.
Such.
Luck.
They're way more exotic than resistor networks - like it ain't even close, And they're 30 to 70 cents from suppliers in the west, if you can get them at all, As I said above, I don't hesitate to put my mouth where my money is, but I ain't Jeff Bezos or nothin - come on! I can't afford that! Though I was able to source some 4 element 1206 0.1uF arrays, there was only one AliExpress seller offering them at prices like that, and I know I can't rely on
So what's the deal? I know discrete capacitors aren't more popular than networks popular because people want to increase the number of parts their manufacturing partners need to place, or lengthen their BOMs. So why is it that resistor networks are the rule, while capacitor networks are the exception?
for two termination, parts, I can handle 0805 - and if I have to, 0603 (I avoid more because they're annoying relative to the board space saved, rather than smaller - I have a few designs that require a 0603 or two), so I only use the 1206 caps for inputs to LDOs where you care whether the capacitance evaporates under applied voltage and the voltage is high enough that it's a bigger deal ( good article on it https://www.maximintegrated.com/en/design/technical-documents/tutorials/5/5527.html )
The simple answer is that resistors are more often used in situations where they are close together than capacitors are. The main use for capacitors in digital circuitry is for bypass/decoupling where they need to placed as physically close as possible to the power rail that they're decoupling. So they are more useful as single items. Whereas resistors are (were?) more often used as protection elements where you might want to current limit a bunch of inputs right next to each other, or as bus pullup/pulldown resistors where, again, you need multiple identical resistors (note the 9-pin bussed version with a single common lead), etc.
TBH, I haven't used a resistor array in years and I doubt I'd bother. It's just one more part to stock or source for relatively limited benefit.
IMO
Isn't the location important as well as the specs, such as placing the MLCC near the VCC of the IC?
Therefore, it is often placed in any position one by one, and the demand for arrays may be few.
But the resistor is not that as the capacitor.
I think that there is often no problem even if the trace is extended a little.
Well, resistors are essentially single-layer thin-film devices, while capacitor are frequently multi-layer devices; that probably makes it a lot more difficult to fabricate multiple independent capacitors on a single substrate... (?)
Given the way we generally manufacture semiconductors, manufacturing massive numbers of multi-layer components doesn't seem to be prohibitive. Besides, I don't think SMD capacitors are manufactured one by one either; I expect they basically make a big layered cake and then cut it into pieces for the individual components.
I think the answer really lies in what @cedarlakeinstruments said - the way they're used generally doesn't give any advantages to having capacitor arrays. Having e.g. an array of 4 x 100nF for decoupling would only mean that component placement gets more tricky (decoupling caps need to be very close to their accompanying IC after all) and have implications/impose limits to trace routing.
Well, crystal loading caps are just two at a time and not always 22 pF, so it probably is not a big market really.
Why would you have "4 0.1µF decoupling caps for nearby components"? Would they not all be connected to the same supply rail? That just does not make sense.
well. consider the example of a CH340 series serial adapter and an AVR microcontroller with power pins on the side facing the CH340 on a compact layout. The CH340 needs one each on Vcc and 3v3 pins, the AVR needs one, and the autoresert circuit needs another. There's your 4 0.1's. chips that spec 0.1 uF on two or more near-adjacent pins are hardly rare...
And surely there would be some demand for a combination one, eg, 0.1uF and 0.01uF in one package - 2 chips like that next to eachother would make good use of such a combo network! (Heck, it should be possible to do 1 uF. 0.1uF an 0.01uF in a single 1206, volumetrically speaking - give the the 1uF cap half of the volume... they wouldn't even need a separate terminal for each one since the majority use would be in parallel.... Because like, read any MCU datasheet, they almost always recommend multiple parallel decoupling caps per supply pin, of different values, The new AVR Dx-series say they want 1uF, and recommend 0.1uF on every pin (in the preliminary datasheet, they said 1 tenth of those values).
in some use cases you'd expect all on same rail, but that's far from universal. And then there are those chips that also have a pair of pins that they say "put a 0.1uF cap between these two pins" presumably to decouple some internal voltage regulator.... which are often near the power pins that also need 0.1uF caps.
Certainly a smaller market. But IMO the fact that 2-capacitor networks in the pF range aren't ubiquitous amazes me (for crystal loading caps) more than the dearth of 4-element ones - they are certainly lower volume. In fact, 2 cap networks barely seem to exist, and many of the ones that were available are being discontinued.
A major problem with packaging multiple capacitors in a single package would be coupling between the capacitors, since there is only a small distance between adjacent capacitors.
Why would the AVR need a separate supply capacitor than the CH340? Are they not connected to the same voltage? The whole idea of separate capacitors on the same rail is to have them immediately proximate to each chip, which placing the capacitors together, fails.
Only because the larger value is presumed to be an aluminium electrolytic which has substantial internal inductance. If you are using multi-layer ceramics, this would not be the case. You just want a capacitor designed for the job with low inductance.
Hi,
Bypass 0.1uF caps need to be as close to their respective IC, just because all the bypass caps share the same supply rails, they do not share the same physical circuit.
There are tracks on PCBs, wires in rats nest construction that are susceptible to pick up stray EMF from other wires and tracks.
Wire and tracks have resistance and inductance and inter track/wire capacitance.
Electrolytic and non-electrolytic caps are paralleled because each has a different frequency response.
I think this is basically it - thick film resistors in particular are effectively printed. Its easy to print several resistance traces on the same substrate, in fact is a technology often used for whole circuits (particularly for microwave devices).
I think the first 4 resistor SMT arrays I came across were on SIMMS/DIMMs or associated motherboard busses - an application wanting lots of parallel equal value resistances to control impedances on the address and data lines.
Digikey shows 5 pages of capacitor arrays in stock.
Sorted by price, the lowest cost is for a 4 x 0.1uF cap array,
"0.1µF Isolated Capacitor 4 Array 16 V X7R 0508 (1220 Metric)"
I think that would meet your CH340 example. $0.092 each in 100 lot quantity.
I could see 3 of them being useful there.
The AVR needs 3 caps itself, for VCC, a 2nd VCC on an SMD part, and one for AVCC. The pinout of a DIP or 32 pin TQFP I don't think is particularly friendly for a 4-cap array.
Huh, I wonder why I didn't see those (though I can't place 0508 accurately enough for 8-pin networks).
It definitely looks like we aren't able to fabricate multiple ceramic caps with anything like the ease we can resistors. All the examples are super specialty and exorbitantly priced... and the datasheet illustrations depict what appears to be.... two capacitors glued together, for high end RF stuff - clearly at least one of the caps isn't garden variety since the combo is priced at more than what I pay for all the parts on most of my designs plus the PCB they're mounted on. I am still surprised that 2-capacitor arrays for loading caps aren't common; I wonder if they interact with eachother more than one would want....
