DIGATAL CONTROLLED DECADE RESITOR

Uses witches to connect any one (or more) of a range of resistors between the input terminals.

It could be latching reed switches, or it might be semi-conductor switches - most likely big MOS-FETs.

Spooky!

Cheers

What's spooky about switches? It's an obvious typo - though not obvious enough for me to have caught it.

I got it right twice in the line you snipped

"It could be latching reed switches, or it might be semi-conductor switches - most likely big MOS-FETs. "

"Bill Sloman"

Uses witches to connect any one (or more) of a range of resistors between the input terminals.

At all ?

It could be latching reed switches, or it might be semi-conductor switches - most likely big MOS-FETs.

I suspect one or two, digitally controlled MOSFETs are involved.

The specs leave far too much to the imagination.

Don't even say what the voltage ratings are.

... Phil

Imagine a 4-quadrant MDAC with the (buffered) voltage on the terminals as the reference. Say the DAC controls a bipolar current source.

Load the DAC with a code representing G= 1/R(simulated) and you're done (conceptually, anyway), since I = V*G = V/R(simulated).

Range switching probably uses small relays.

There will be an interesting array of possible error sources etc. in this sort of thing, but most instrumentation uses 10uA-10mA currents and reasonable voltages.

We make a thing something like that.

Our control mechanism is a 16-bit MDAC.

Here is one simple way to use an MDAC to simulate a resistance:

The MDAC could be a real ladder-network thing, or could be PWM or delta-sigma.

We don't do it exactly that way, because we wanted our resistance to be fully bipolar and be a good resistor at AC up into the KHz, but the concept is similar.

It's not a trivial thing to do, at least to do accurately. This was one of the toughest analog designs that we've ever done. It was crucial that resistance transitions be smooth and glitch-free: if we're simulating an RTD to a jet engine control computer, and transiently open or short or bump while changing values, the FADEC may see an overtemp or bad sensor and shut down. That was a serious problem with competitive relay-based simulators.

That IET unit has impressive specs. It's not obvious if it is bipolar, or if it can simulate a resistance to an AC source, or if it glitches when you change value. Specs are sketchy.

DC offsets are killers. Ditto noise and loop stability. A simulated resistor has to be stable when connected to almost anything: voltages, currents, cables, multiplexed RTD scanners, AC, DC, millivolts, tens of volts. It's horrible.

has

Which is why I fancy latching relays or MOS-FET switches and an array of real resistors. There aren't any significant DC off-sets in that - give or take some thermocouple voltages, and with latching relays or MOS-FETs you aren't dissipating enough power to generate much in the way of temperature differences.

There's nothing particularly horrible in that kind of solution. Your

The unit is powered by four AAA cells - despite Phil Alison's scepticism I did read the specs. That's compatible with switching using latching relays, but I don't think you'd get much battery life if they were being used to keep regular relays closed.

has

"real resistors" is also more their speciality - they manufacture precision resistance standards.

Yes, Bill. It was what's known as a 'joke'. Not a very good one, admittedly, but also not a slight on your integrity, ability or virility.

Cheers

That makes a change. Some of our regulars don't seem to feel that a post is complete until they have managed to impugn all three.

has

It would be silly to make such a gadget out of resistors and switches. The numbers just don't work.

resistor has

Dunno, maybe. But it is strange there is no specific voltage limit as far as I can see. There is a simple 1W dissipation limit, like you might get if you used 1W resistors. And there are "30 calibration resistance values" to set during calibration. Which is about what you need for a binary stepped resistance chain isn't it? 0.1ohm to 24M, ~30 bits.

But there are other explanations for all the above.

resistor has

cables,

horrible.

It could be done with 30 resistors and 30 switches, But each resistor would have to be good for 1 watt. The resistors would probably have to be stepped in less than 2:1 steps to assure monoticity. The programming algorithm is messy, probably slow, but do-able. Some of the bidirectional switches would need milliohm ON resistances and low TCs. 30 latching relays would not fit in that enclosure, so it would be two mosfets per switch.

It's very messy, and would have nasty transition glitches. We decided to do it with MDACs, which takes fewer parts and makes smooth transitions.

This one looks "electronic" but has a 3 mV DC offset spec, lethal for RTD sims, looks noisy, and uses ISA bus!

Here's another electronic one:

It's an interesting problem.

Aren't all three implied impugned by a single descriptor... "weenie" ?>:-} ...Jim Thompson