juggling parts

Sometimes it's better to not give the tech access. It's not unusual for an adjustment to be needed one time at the factory to account for tolerances. Field repair is often at the board level, so no field replacement of anything that would impact that trim setting. People may whine about "the cost" and "the convenience", but the manufacturer sets the procedures to facilitate repair to a known state with factory tolerances.

Regardless of the topic, the headline has just reminded me of something the film critic Barry Norman once said on TV. He was commenting on a mention in a film of "pocket billiards". He just said that he didn't realise it was competitive!

AoE X-chapters, p 26, has some great data about exploding various resistors. The carbon comps win for short pulses, below about 1 millisecond.

I always assumed that 50 ohm cc's were best for high frequency termination, but the cheap carbon film resistors are about as good. Surface-mount thick films are better than either... less lead inductance.

It's preventing him from mucking with something he doesn't understand and doesn't have the equipment to calibrate. Besides, some products have a hundred cal factors.

If something in a signal chain is broken, twiddling pots will break more things. We rarely use trimpots, and then only to tweak gain in GHz-range amplifiers.

If it's not meant to be adjusted, then sot fixed parts are more a suitable analog comparison. A trimpot would be an unjustified vulnerability, simply from a reliability and cost standpoint.

I'm finding that that programmable part are increasingly dominant in non-repairability situations, so It rankles when the equation is made to an impractical analog equivalent.

RL

We do polynomial calibrations all the time. Try that with trimpots!

I've tried several low-value resistors for high-frequency matching pads and terminations. SMD 1206 resistors tend to be too capacitive, and the similar-sized MELFs are too inductive. It's easier to gimmick the MELFs, so that's what I tend to choose when it matters.

Jeroen Belleman

Lucky you! They were employed as trimmers by none other than Seymour Cray, back in the dark ages of RTL, before computers could Spice faster than techs could wirewrap.

On a sunny day (Tue, 16 Nov 2021 19:07:59 +0100) it happened Jeroen Belleman snipped-for-privacy@nospam.please wrote in <sn0s1v$1dfr$ snipped-for-privacy@gioia.aioe.org>:

The biggest factor often ends up being the PCB stackup. Assuming you clear the copper underneath, a pair of 100-ohm 0603 resistors in parallel is hard to beat.

-- john, KE5FX

Yow! Look at the inductance in that shield connecting wire!

On a sunny day (Tue, 16 Nov 2021 15:01:47 -0600) it happened John S snipped-for-privacy@invalid.org wrote in <sn167v$851$ snipped-for-privacy@dont-email.me:

Na, SWR OK on 2.4 GHz, wavelength is about 12.5 cm could solder it a bit shorter, OK. Spectrian 75W linear amp:

Exact used to make an arbitrary waveform generator with about 25 dial pots on the front panel, one per sample point. ;)

Cheers

Phil Hobbs

That wouldn't fly with me. :-)

Jeroen Belleman

In TIAs, it's often helpful to use an RC lead-lag network to compensate for the parallel capacitance of a high-value feedback resistor. A trimpot is good medicine for that.

Cheers

Phil Hobbs

Again, not a suitable comparison.

Polynomial calibration, for systems that drift under ALL influences . . . better measure them influences!

RL

And I think they are extinct!

Several people make thick-film resistors that drop into a DPAK footprint. They are good for 5 watts or so if you heatsink the paddle to a ground plane. They are actually pretty good at sub-nanosecond speeds.

I recently did a signal pickoff:

sig------1K 0603----+----amp---50r-----out | 20r 0603 || cap | gnd

It was horrible. I'll try something else.

Carbon comps have been used as poor-man's cryo temperature sensors. They are relatively immune to magnetic fields that trash diode sensors.