Precise voltages from imprecise parts..

For an oscilloscope? I don't think so.

Multiturn pots are at least as shock sensitive as singles, usually more.

Probably decided by some physicist or manager. If it's more expensive and harder to adjust, it must be better.

ROTFL. I've made asynchronous controllers that way. One of 'em eventually worked. That's why everyone prefers microprocessor-based controllers, of course; asynchronous means very susceptible to glitches and with trims located everywhere... very hard to reprogram.

In an actual manufacture situation, on the final-test bench, you never want an adjustment that doesn't go with an immediate feedback indication, A Lissajous pattern that looks right is SO much easier to adjust to than a frequency counter with a one-second update cycle. For the original poster, a 25.000V reference and a null indicator make for easy adjustment.

And you can make a 1k ohm divider with fixed precision resistors, then link a 1k pot in parallel to it, with a 1M resistor link, and a mismatch in the tempco makes almost no difference in the result. Heck, you could use a three-stage version, and get both coarse and fine trims. The full blown generalization of this, of course, is a Kelvin-Varley divider.

That's a better scheme, for sure, provided you can stand the dissipation in the low-ohm pot.

Loaded pots are neat--there are all sorts of cute nonlinear curves you can make with them. Finding new ones makes a good parlour game, like with 555s.

Cheers

Phil Hobbs

I once designed a PWM sort of DAC (all TTL) that had two trimpots, in theory one offset and one gain. The interaction was such that nobody could trim it; it agressively diverged. We had to work out a quantative procedure, with equations, to adjust it. I eventually spun the design to make the adjustments mostly orthogonal.

If you can parallel the element with a precision resistor you can get quite good stability. CRV on cermets is high too, another reason to use them as a pot.

Most of the modern multiturn trimpots are just single turn rotary pots with a mechanism on them anyway.

The really bad ones were the long ones with a linear leadscrew.

One of the secrets to getting good long term stability in the setting is to turn the trimpot from one end to the other several times before setting it to the final value. Naturally that's a major PITA with a 25 turn trimmer.

--sp

Motorize it!

There once was a DVM that used a motorized 10-turn pot with a mechanical Veeder-Root sort of numeric display, run in a self-nulling Wheatstone bridge config. It really wasn't bad.

Digitek I think it was.

---

John Fields

My usual ROT is that if the controls interact less than 10% (i.e. 1 turn in X requires a Y correction of 0.1 turns or less, and vice versa), anyone can do it; at 30% it requires an expert; and at 50% it's impossible except using a numerical algorithm.

Cheers

Phil Hobbs

Usually sold as "3/8 inch multi-turn pots".

Not in my experience. That's a 19mm trimpot, though you used to be able to get even long ones - 1.25 inch long. The quality of that that kind of trim- pot does depend on the manufacturer, the nature of the linear resistive ele ment and the construction of the moving contact - Vishay's top-of-the-line trim-pots had a multi-leaf contact, which gave very smooth adjustment, and no no-contact positions along the track. But even cheap and nasty 19mm trim pots were easier to set than 3/8" multi-turn trim-pots.

e.

Not something I've ever found necessary with a 19mm "long" trimpot.

Unnecessary trimpots are a sign of a novice designer, but any time you've g ot to get better than 0.1% precision, you've got to find yourself some way of adjusting the circuit to give precisely the output you want a the points where you need precision. A DAC can do the same job, and PWM can often do it better, but in a lot of situations at trimpot is the cheapest and simple st solution.

Auto-calibration can be fun - and can let you use temperature sensitive com ponents in precision application - but a trimpot takes much less design eff ort.

Odd that nobody has mentioned capacitance trimmers. Capacitors are a rarely available with better than +/-1% tolerances, so the world should be full o f trimming capacitors ...

If one uses a 'state variable' filter design, all C components are paired with a resistor; you can tune all RC time constants by trimming the resistors. It used to be that any hard drive would have a bandwidth-limiting filter module, ceramic wafer with obvious laser-trimmed resistance and surface mount capacitors.

Of the big three (R, L, C), the most expensive to trim is ... C.

Bill, you just have to take one apart and look at the way the parts fit together to see why they're bad. Differential expansion, in general, moves the wiper.

The circular types do not have that problem.

It's not "necessary" at all, just a manual procedure (seating the wiper) found to be worthwhile with burn-in and testing of thousands of units (high-quality sealed cermet single turn trimpots). The Japanese engineers I learned from were masters at getting superior performance and reliability out of cheap mass-produced parts- trimpots were something they didn't scrimp on by using the cheap multiturns or open types. It's quite easy to use all kinds of expensive parts and still get mediocre performance and reliability, and our competitors often managed to do so.

Calculations were also done for every range (of hundreds) to reduce the necessary range of adjustment and to nominally eliminate interaction (at least one way) so that (nominally) they could be adjusted very quickly, and even in the worst case of tolerances it wasn't too bad.

--Spehro Pefhany

Best regards, Spehro Pefhany

They are occasionally useful, such tasks as 'pulling' crystals, if you don't want do a DAC + varactor and use firmware/EEPROM or whatever.

--Spehro Pefhany

Best regards, Spehro Pefhany

There are lots of nice trimmer caps around.

On a sunny day (Mon, 20 Apr 2015 11:14:18 -0700) it happened John Larkin wrote in :

Mmm since I worked there and did it, who are you?

Yes phycisists, all over the place, and very highly qualified electronic designers. A large acellerator.

It is not 'harder to adjust', it is a LOT EASIER to adjust say a VFO or some reference. That is why I use the 25 turn types, for example here:

None of them good for more than a few picofarads of adjustment.

to get even long ones - 1.25 inch long. The quality of that that kind of tr im-pot does depend on the manufacturer, the nature of the linear resistive element and the construction of the moving contact - Vishay's top-of-the-li ne trim-pots had a multi-leaf contact, which gave very smooth adjustment, a nd no no-contact positions along the track. But even cheap and nasty 19mm t rimpots were easier to set than 3/8" multi-turn trim-pots.

Perhaps not. But my experience is that 19mm trim-pots with a linear resisti ve element work a whole lot better than 3/8" multi-turn trimmers.

Stability wasn't a problem we ran into, and the 19mm parts were a lot easie r to set.

Working out the relationship between the trim-pot's adjustment range, the c omponent variation you want it to compensate, and the observable the techni cian is following to set the trim-pot should be part of the design process.

I was once peeved that I had to increase the range of one of my trim-pots s o that the technicians could always wind the observable past the desired st ate, but that was the way they liked to do the adjustment.