[sdiy] Resistor Networks....

[Steve Lenham]

On 30/08/2020 04:32, Pete Hartman wrote:
> I'm working on laying out a classic through hole module design with SMD 
> parts.  The original uses a fair number of resistor networks...
> 
> So I'm looking for seasoned opinions about whether there is any 
> important reason to keep resistor networks, when I can sub in 1% SMD 
> parts that will take up less space and be more flexible for placement.  
> Are the networks either 1) better matched even than 1% parts, or 2) have 
> some other characteristic which I'm not thinking of that might be important?

Although sometimes they save space over discrete parts, the main 
attraction of resnets in analogue designs is that the matching of the 
individual resistors is way better than the absolute tolerance specified 
for the whole network (and stays largely that way as temperature changes).

So they are a good choice in any position where the ratio of two or more 
resistances is more important than their absolute value. e.g. if you 
need to switch the gain of an amplifier in 6dB steps, or you want to 
create an inverted version of a signal with exactly the same amplitude.

In those places, a resnet will be cheaper than discrete resistors whose 
absolute tolerance - perhaps 0.01% - achieve the same tight ratio.

Old-school thin-film resnets offer e.g. eight individual resistors in a 
SO16/DIP16 package and can have really good absolute tolerance (0.1%) as 
well as close matching, but are jolly expensive.

Thick-film parts offering four individual resistors in a 1206 package 
(from the likes of Phycomp and Yageo) are cheap as dirt, though. Their 
absolute tolerance is nowhere near as good, 5% perhaps, but the segments 
are still pretty well matched. I've used loads of them in pro audio 
products.

I do agree that they are a bit of a cow to solder by hand and that they 
skew PCB layout a bit, but they are great for production when a machine 
takes the strain.

Cheers,

Steve L.
Benden Sound Technology