AoE x-Chapters - 1x.2 Resistors

OTOH most stuff is cost cutting now, and less skilled engineers are also cheaper. Yeah, old audio mainly

NT

Grin, Ebers-Moll is cool now that I'm older, Transistor man get's you going when you don't know as much.

GH

I suppose, in a purely elementary treatment. But in a combination textbook + professional reference, I think it's out of place. ('Tain't my call, of course.)

If JL ever gets round to writing "Electronics From Scratch", he could get away with something like TM just fine.

Cheers

Phil Hobbs

They don't list the MD1248 series now. :(

Cheers

Phil Hobbs

Are there drawbacks other than resistor and voltage tolerances? I wouldn't do that because the small output voltage would inherit a lot of noise/ inaccuracy/drift from the power rails, but that's the only gotcha I can think of as long as 0.1% resistors are used.

Not a big fan of pulling DC current out of the wiper, as done in some other implementations.

Not a big fan of trimpots in general for that matter. They are a hassle in production, a liability in calibration, and impractical to automate. They are one of those components whose quality can only be expected to get worse as the volume drops. I've rambled about that before.

Don't use trimpots, use DACs.

-- john

What's the fastest MDAC around? Fastest DPOT?

On a sunny day (Tue, 6 Aug 2019 21:20:56 -0700 (PDT)) it happened "John Miles, KE5FX" wrote in :

If you have ever been through the calibration procedure of the old Tek scopes, it involved step by step adjusting of many trimpots. and people still sing the praises of that stuff today. They did not even use 10 turns.

When I worked in the TV studio in the seventies THOUSANDS of trimpots were one way or the other in signal and sync chain. Those never caused problems.

I use ten turns a lot. Some are better than others, Bourns was OK, using cheap Chinese ones now. But still have to see a problem with any trimpot.

DACs or digital potmeters when you need to dynamically change settings. Or maybe in strong vibration situations.

Dunno, good question. If I want a digitally-controlled attenuator for faster signals, I'll use something labeled for that purpose. Some very nice CMOS parts are available up to (mumble) GHz from outfits like Peregrine. They will work down to the audio range but the power has to be derated as you go below a MHz or so.

I don't think there are any good one-size-fits all trimpot substitutes, at least nothing affordable. VGAs bridge the gap between RF and DC, but most are noisier than I'd like. VGA designers don't seem to understand the whole Friis noise equation thing.

There are all kinds of corner cases where you have to use very different parts in order to avoid trimpots. I wouldn't try to claim otherwise... just saying that when I can avoid using a physical pot, I'll try pretty hard to do so.

-- john, KE5FX

I've seen countless that were bad. They were a major cause of failures.

NT

On a sunny day (Wed, 7 Aug 2019 02:37:15 -0700 (PDT)) it happened snipped-for-privacy@gmail.com wrote in :

It needed precision resistors, but this works OK:

Good luck at 2 GHz. Not every trimpot is used for DC offsets, by a lot.

Cheers

Phil Hobbs

Sure, AoE has changed much with the 3rd ed. I saw the 1st ed as a book to teach physics students enough to get things done in the lab.

It's much more than that now.

George H.

I almost always use the voltage-divider-on-the-wiper approach for DC adjustments.

I've never seen a current-related problem with a trimpot wiper at < 10 mA. Trimmers that are getting worn out (their lifetime is only 100 cycles or so) will get momentary opens on the wiper, and the voltage divider on the wiper prevents that from causing havoc, e.g. blowing up your diode laser.

Dpots and MDACs crap out below 1 MHz, making them essentially useless for most of what I do. Their bandwidth and isolation are both strong functions of the code selected, so from an electrical standpoint they're mostly crap. I do use dpots for current limits, which occasionally means programming them via a bus pirate and USB isolator because they're hanging off some higher-voltage rail.

The nice thing about a dpot is that it's hard to casually mis-adjust and then lie about it.

Cheers

Phil Hobbs

A screwdriver is a pretty good human interface.

There are RF digital step attenuators, and some (possibly) work down to DC, but they work in -shock- steps. And they need a digital input from somewhere. I've faked a digital attenuator with a half-pitch dipswitch and a few resistors, at the 4-bit level, but it's only 16 steps.

I don't see anything wrong with using a trimpot to tweak, say, a photodiode gain in a 300 MHz circuit, from a 2:1 variation down to 1%.

To me, there's nothing to avoid. Pots are cheap, easy, small, and reliable.

Most of our products are closed-box calibrated, with a cal table saved in flash inside. Testing is automated.

I once designed a PWM DAC circuit that used two trimpots to cal offset and gain. The interaction was so diabolical that nobody could cal it, not even the brilliant designer.

But to tweak the gain of a small low-noise DC-x00 MHz amp to 1%, a trimpot is perfect.

This is all analog, and has one pot:

The analog output is trimmed to 1 volt/mW at 850 nm. The photodiode gains are all over the place.

I don't know how Phil deals with parts whose specs regularly vary over a 3:1 range, and jump around every production batch, and are only vaguely suggested on data sheets.

I made a personal deal with one laser supplier that they would alert me to any major changes. Of course they didn't. I think they buy their lots of pds and lasers in dark alleys in Shanghai. Their fiber connector housings must be made from melted-down beer cans. An unmate/mate cycle is good for 3 dB gain change.

I never met Transistor Man. But I can (and do) explain transistors pretty well in about a half hour at a whiteboard with just a few concepts and a few numbers. Ebers-Moll is not the whole story at all.

Last few instruments that I designed, I don't think I've used a single bipolar transistor. They are going the way of tubes.

I still like relays!

What are some good trimpots for use at 2 GHz?

Seriously, I could use some of those for the variable-gain CFB amps I like to keep around the bench.

-- john, KE5FX

Photodiodes are generally pretty good, assuming you control the etalon fringes. The responsivity of BPW34s varies only a few percent unit-to-unit unless you're out in the tails of the response.

Diode lasers, now that's another issue. It's not uncommon for the threshold current to vary +-40%, and the major damage mechanism is facet overheating from too-high output power. The best way to handle that is a constant-power feedback loop and a current limit set by a dpot.

If I needed to modulate it rapidly with some known AM depth and/or FM deviation, I'd probably use a built-up R-2R attenuator and program it with hot tweezers.

Everybody seems to make their lasers in pots and pans, except for the very highest-volume ones. I'm using a $700 Nichia 488-nm diode today, which has a slow-axis beam divergence of 7 to 13 degrees, typical 10, and an aiming tolerance of +-5 degrees! Maintaining a consistent collimated beam diameter requires either a selection of different collimating lenses or some sort of varifocal or zoom arrangement. Blech.

As my Dad used to say when I complained about some chore, "If it were easy, I'd have done it myself." ;)

Cheers

Phil Hobbs

I've seen plenty of bad ones, too, in older gear. I don't see many in newer equipment at all, whether good or bad.

I've only used one trimpot in a commercial product. Brand new parts from Bourns, straight outta Thief River Falls and priced accordingly. A bunch of them failed after soldering. I took the hint.

-- john, KE5FX