Manufacture and placement of SMT components is fully automated, only the P&P machine camera ever has to see them... Anyway 0805 is far larger than a full-stop, unlike the 01005!!
I've bought a new solder station which should keep the tip temperature within bounds, plus I've got some old 60/40 solder I'll use which only need about 180C to melt. A tiny tip, some Stannol 0.5mm solder and a magnifier and I should be good to go 🙂
Boy, it's difficult but it is achievable. I've managed, using a lot of flux, tinning the board first then holding the resistor down with tiny tweezers whilst applying a dab of heat. Using lead solder and a lower temperature seems to help too. Then cleaning the board with solvent, it doesn't look too bad but very difficult to get it straight. I've got some off cuts of carbon fibre and I'll see if these can be used as a guide, maybe a sliver of cf against the the adjacent component will keep it aligned.
There are 2512's if you want something larger. I've used them occasionally for special purposes. More expensive though.
Generally larger is better (up to 1206). Lower temperature coefficient (TCR) is important if you want low distortion and you plan on dissipating some power in the resistors. Low voltage coefficient (VCR) is also important if you have high voltage across the resistor and want low distortion.
Getting low TCR is easy, though I doubt there's much point in going below ±25 ppm/ºC in most audio circuits. VCR is rarely specified.
Definitely go with thin film types. I have measured the difference in distortion for a Vishay TNPW in 1206 vs 0603 and found the 0603 to be considerably worse (= higher distortion).
I've had good results with Susumu RG and RR series.
There might be some options in MELF but they can be a pain to work with because of their shape.
Tom
Getting low TCR is easy, though I doubt there's much point in going below ±25 ppm/ºC in most audio circuits. VCR is rarely specified.
Definitely go with thin film types. I have measured the difference in distortion for a Vishay TNPW in 1206 vs 0603 and found the 0603 to be considerably worse (= higher distortion).
I've had good results with Susumu RG and RR series.
There might be some options in MELF but they can be a pain to work with because of their shape.
Tom
Mouser has 60 (sixty) resistor types in stock in 2512 with a TCR of ±25 ppm/ºC or better versus 3354 types for 1206 with the same TCR. I get a pretty clear sense that the larger resistors are going away because there isn't a need for them anymore. You can get 1206 that can dissipate quite a bit of power now. Or 0612 (1206 with terminals on the long sides) if you really want to burn power.
Tom
+1
I used to solder SMD by hand. I still do for component swaps and such, but for prototype board builds I just showed the board in a toaster oven. It's easy, fast, and I get excellent results with it.
Tom
Probably should mention that if you use a toaster oven for reflow, do not also use it for food.
Is paralleling two resitors to get the same value than a standalone, usefull to get less heat or reduced temp co?
Not unless you want your food to taste like flux and have a strong desire for lead poisoning, cancer, and whatnot... 🙂
Sure. You'll get half the power dissipated in each so they'll heat up half as much. You can also do tricks like (R || R) + (R || R) = R. That'll work well in the cases where you can't find 2R and need R to be pretty exact.
Tom
Thanks, I use 5° PPM by superstition the feedback of transimpedance stage for dac I/V to reduce the error and 1206 size, while I don't hear difference with a 25PPM 0805 size or even lower 0603. But I have read the heating was an enemy here and could raise conversion errors.
okay so better to raise the size and lowisch tempco, then ? @Havoc
okay so better to raise the size and lowisch tempco, then ? @Havoc
Last edited:
Great if you want more power and lower noise. I am not sure about the temperature coefficient (TCR), but I’m pretty sure it will give you better tolerance.
Thermal noise will be exactly the same for one resistor with value R and two parallel resistors where each has a value 2R. The excess noise (1/f noise) that you get when they are biased may differ, though.
For two identical resistors in parallel the TCR of the parallel combination will be the same as the TCR of each individual resistor.
On average, yes. But worst case, no.
Tom
Doubt it, the number of lines is simply larger for higher volume products. Above 2512 you are starting to transition to heat-sink mounted resistors too (although you can argue that with IMS any SMT resistor can be heatsink mounted).
Maybe a few decades ago the transition from analog to switch-mode power circuitry would have re-shaped the resistor market, but I think things are stable now.
Sure. Demand plays in too. But it used to be that if you wanted a 1 W resistor you needed a 2512 sized one. Now you can get 1 W in 1206 and 0612. For that reason I expect 2512 to fall out of favour. You'll still need that size for the 2-3 W types.
It also seems 0805 is falling out of favour, so basically it's 1206 for power and 0603 and smaller for everything else.
All my new circuits use 0603 where I used to use 0805. I get the same or better performance and have many more parts to choose from. I've come across a couple of situations where 0402 would make the layout flow a bit nicer. Decoupling capacitors on ICs with tight pin pitch would be an example. But since I still build prototypes by hand (with a microscope and toaster oven) I've yet to go to 0402.
Tom
