15

enter image description here

Why would you put multiple resistors in line instead of one resistor that is the same value? In this case, I have a PCB with three 1k ohm resistors. What could be the reason for this instead of just one 3k ohm resistor? Thanks in advance!

Mr L
  • 159
  • 1
  • 3
  • 1
    Usually to share voltage drop or power dissipation... In this case what is V and I? My guess <10mA and 24V. Subtract diode drop and Pd = V²/3k = 161 mW shared by 3 small chips. – Tony Stewart EE75 Dec 02 '17 at 01:07
  • 1
    If it is 24V then this is probably for power dissipation - 0603 resistors are 1/10W. – τεκ Dec 02 '17 at 01:13
  • 3
    Not this particular case, but sometimes we do it for redundancy - where a failure open puts the device in a safe condition but a failure short could result in an undesirable outcome. – Spehro Pefhany Dec 02 '17 at 02:32
  • 10
    Are you sure that they are physically connected "in line"? I often line resistors up on my PCB but they aren't series'd together. The traces come in from the sides. – Ron Beyer Dec 02 '17 at 03:44
  • 1
    Nice photo. I appreciate the clean background. – A. I. Breveleri Dec 02 '17 at 16:02
  • What kind of abuse has that thing suffered? – whatsisname Dec 03 '17 at 01:30
  • The board looks like it is manufactured in volume, so it seems stupid, but they might have had a lot of 1k in inventory and designed around that. – Ross Millikan Dec 03 '17 at 03:45
  • Related, for parallel https://electronics.stackexchange.com/questions/197839 – PlasmaHH Dec 03 '17 at 12:07

4 Answers4

42

Three reasons I can think of:

  1. to allow for a higher power dissipation than one resistor can handle.

  2. to tolerate a higher voltage than one resistor can safely handle.

(Higher power or higher voltage resistors would be larger, which may not be convenient for the auto-place machine)

  1. to minimize the number of different resistor values required on the board (if I use lots of 1K resistors elsewhere, it may be cheaper to use three here, than to buy a few 3K resistors to use in one place)("cheaper" may include loading and programming the auto-place machine as well as the actual cost of the resistors).
Peter Bennett
  • 57,014
  • 1
  • 48
  • 127
  • 8
    4. to get the average resistance (since every resistor has a tolerance) closer to the actual value – Ferrybig Dec 02 '17 at 11:18
  • 13
    @Ferrybig That's not a guarantee, even statistically, since resistor values aren't always nicely distributed. If you actually need the tighter tolerance, buy it. – Abe Karplus Dec 02 '17 at 12:12
  • @AbeKarplus why wouldn't they be nicely distributed? If there's a known bias, why not use it to correct the rated value and then have some nicely distributed errors? – Andrei Dec 02 '17 at 15:03
  • 9
    @Ferrybig It doesn't work that way, rather the opposite. – itsLex Dec 02 '17 at 16:10
  • 6
    @Andrei Because you will still produce a big lot of outliers. Averaging three Gaussian variables gives a little guarantee. Moreover, if all three come from the same lot, they can all be equally biased by some systematic margin, and you are lost anyway. – yo' Dec 02 '17 at 16:59
  • "nicely distributed" as oppose to "normally distributed". Ill take nicely distributed components please –  Dec 02 '17 at 21:30
  • 1
    Averaging 3, still better than averaging 1... – Lamar Latrell Dec 02 '17 at 22:15
  • @LamarLatrell Yes, except for component cost, board space, number of placements, number of failure points, and design complexity. – Abe Karplus Dec 02 '17 at 22:46
  • @Andrei The manufacturer doesn't guarantee particular distribution, it merely guarantees that the value will fall within specified parameters. Typically, values differ more between batches than between units from one batch. – Agent_L Dec 03 '17 at 10:03
  • 7
    @LamarLatrell No. It's not a lottery with fair and random results. It's a spec. If I have a batch of resistors all reading exactly 1030 ohms each, I sell them as "1k, 3%". You know that you are averaging but you have no idea what are you averaging, which makes the process useless. If you need more precision, you buy better tolerance, if you don't need the precision, why bother? – Agent_L Dec 03 '17 at 10:07
  • 5
    @Ferrybig that will not work well when you buy from a manufacturer that does measurement and binning. – PlasmaHH Dec 03 '17 at 12:10
15

It may be a reason of power dissipation. Each resistor package of that size can support only a certain amount of power on the resistor so putting multiple in series spreads the power across multiple packages.

Another reason may be voltage rating. An SMT device like that will have a MAX voltage rating that it can handle without insulation break down. Multiple resistors in series will split the voltage drop across the various devices.

A last reason that this is sometimes done is that the circuit board design is done to it is easy to swap out certain resistors to be able to select a combination of different valued resistors that total up to certain calibrated requirement that may not be available in a standard valued single resistor.

Michael Karas
  • 56,889
  • 3
  • 70
  • 138
  • I have seen cases where resistors are put along each lead, say, a the power line and the return one. Think about wanting to have a voltage source with a series resistance of 300kOhms. I've seen applications where one 150k goes in series with each lead. Why would be the use of this? Thanks – Big6 Dec 02 '17 at 01:17
  • 1
    @sixcab Same reasons. – user253751 Dec 02 '17 at 07:35
  • @immibis I guess my question is, instead of placing resistors along each lead, why don't you put them in just a single lead? Say you want to create 10k and for that you use two 5k's. What would be the difference between placing a 5k in series with each lead of the source vs placing the 5k's in series with a single lead? – Big6 Dec 03 '17 at 17:26
13

In this case, reduction of production cost. Using a common value three times (1k) instead of a less common value once (3k) reduces the total number of different parts required by your design. In the pick and place process used in manufacturing, you are looking to minimize the number of different parts used, within reason. Loading new reels into the feeders can be a significantly greater cost than using a couple more of the same part.

Other reasons to make a chain like this can include increasing the effective voltage rating of the part, such as in the high voltage frontend of a meter or probe, but that doesn't apply here.

  • My answer jumps to conclusions. It doesn't look at all like a voltage tolerance case, but may well be for power or swap-out reasons as mentioned in other answers. Power in particular looks likely based on what is nearby. –  Dec 02 '17 at 01:11
  • 4
    Yeah. Those big-ass screw heads surrounded by a field of plated thru holes suggests the presence of a heat sink on the other side. – WhatRoughBeast Dec 02 '17 at 14:38
7

I wonder if the board has three resistors in a row because that's what the circuit needs: they aren't just substituting for a single resistor with three times the resistance.

Look to the left between each of the triple resistors, e.g. just above the "3" in the R23 label. Aren't those vias? It looks like there's a via next to each of the junctions in all of the triple-resistor groups. (Someone please let me know if I'm misinterpreting the photo.)

So my guess is the three resistors in each group could be wired in series, thus the placement, but there's also a tap at each of those junctions running to some other part of the circuit.

Or maybe the vias are there for some other reason of convenient layout - but it does seem like an interesting coincidence.

(Also, thanks everyone for the other answers! Always fun to learn some things about PCB manufacturing.)

Michael Geary
  • 171
  • 4