Designing a "discrete" pot

[ADWSystems]

The switch setting on In- will be a constant but V1 will be changing. Therefore the current through the load and thus the voltage at the FET-resistor junction will change. If the FET-resistor voltage changes, then the resistance provided by the FET will change and therefore the total load. Which is not what I'm looking for. This works more like a constant current than a constant resistance.

 

[crutschow]

The switch setting on In- will be a constant but V1 will be changing. Therefore the current through the load and thus the voltage at the FET-resistor junction will change. If the FET-resistor voltage changes, then the resistance provided by the FET will change and therefore the total load. Which is not what I'm looking for. This works more like a constant current than a constant resistance.

You are misunderstanding what it does. The circuit does varies the load current with voltage, the same as a pure resistor would. It is acting exactly as a constant resistance not a constant current.

Remember that the resistor string is powered by the load voltage, thus its output varies with voltage. This causes the op amp to vary the MOSFET voltage to linearly vary the current with voltage, That's the same way a resistor behaves. If you don't believe, do the simulation.

 

[KeepItSimpleStupid]

I want to add something to this 21 question problem or is it "cloud problem solving"

1, Multi-tapped resistors are available. I once bought a 1000 W 0.5 ohm tapped resistor. Aside: I also needed a 200 W, 1 Meg Ohm resistor

2. Electronic loads are available: They can be constant R, V, I, and P. e.g. http://www.amrelpower.com/AMRELPowerProducts/eload_programmable_load.html

3. Battery testors are available: e.g. http://www.keithley.com/products/dcac/highspeedpower/battery/?mn=2308

 

[ADWSystems]

It's only 21 questions because we have broached a dozen or so different solutions. Some applicable, some not. The ones that weren't were because of things that a) I hadn't considered, b) I didn't state (such as a single input ruling out the dip switch option), c) that didn't apply, or c) someone was smart enough to think of. And I appreciate all of them, even others don't like the rambling that appears to stray from the main point. I find the discussions valuable because even if the whole solution doesn't fit, part of it may lead someone in a new direction, possibly toward an eligant solution. And I would consider this "cloud problem solving" with lots of people from all over chiming with ideas, suggestions, and critique on each others designs.

1. I thought of them but worried that setting the variety I found, would lead to inconsistancies from build to build. Not to mention something for someone to fiddle with on the production floorwhen they think they can "make it a little better".

2. I was vaguely aware of them. The ones I found were way overkill, these are closer but still appear to be overkill in many ways.

3. Though applicable to my other home project, these are a little out of my pocket's price range.

 

[carbonzit]

The op-amp solution

Questions to Dougy, since the op-amp idea was his:

I went ahead and simulated it (.asc file attached below for those who want to follow along), but I don't know how to evaluate it. When I change the resistor tap I get different currents (through the power supply). Couple questions:

1. How do I measure the resistance (between the two points indicated) with LTspice? The tools for measuring voltage and current are obvious, but I don't see any "ohmmeter" function. Or is this really a "virtual" resistance that can't be measured this way?

2. In your circuit, I ASS-U-ME that since the total resistance of the resistor string is relatively high (80KΩ), it's pretty much swamped by the lower resistance of the FET. Is this correct? or should one compensate for the parallel resistance?

3. I guess I too am having a hard time wrapping my head around the concept of a current source acting as a resistance. How exactly does that work? Wouldn't this circuit actually be injecting current into the circuit it's connected to, rather than acting as a resistance within that circuit?

In any case, if this did work, it would indeed seem to be a simple (and elegant) solution to the O.P.s problem (which, it seems, we still don't really have a good handle on; maybe 21 questions more?).

 

[dougy83]

1. How do I measure the resistance (between the two points indicated) with LTspice? The tools for measuring voltage and current are obvious, but I don't see any "ohmmeter" function. Or is this really a "virtual" resistance that can't be measured this way?

You can't measure resistance of an object without applying a current or voltage to it to get its response.

In LTSpice, I used "-V(resTop)/I(V1)" as my trace to show resistance.

2. In your circuit, I ASS-U-ME that since the total resistance of the resistor string is relatively high (80KΩ), it's pretty much swamped by the lower resistance of the FET. Is this correct? or should one compensate for the parallel resistance?

The resistor string is high because it is effectively in parallel with the 12.5/25/33/50... ohm resistance of the FET and 12.5R reference resistor. A larger string resistance will cause less error - but as it stands, the maximum error is when 100k is paralleled with the effective 100R - which is 0.1% error.

3. I guess I too am having a hard time wrapping my head around the concept of a current source acting as a resistance. How exactly does that work? Wouldn't this circuit actually be injecting current into the circuit it's connected to, rather than acting as a resistance within that circuit?

It's a voltage controlled current source - which is exactly what a resistor is.

 

[ADWSystems]

dougy has the idea, so there is enough information given to date. It is quite an elegant soultion (if I can wrap my head all the way around, I get pieces of it). I almost had it on my own at one point, but I didn't have the resistor string tied to the MOSFET drain which resulted in the opamp output varying and changing the total load.

 

[Misterbenn]

Sorry i haven't had time to read the whole thred but...

Could you not just use a digital pot or similar with a high power driver on the output to drive the load? Or use a standard pot / multipoint switch connected to a PIC which would control your driver.

 

[crutschow]

If you do the LTspice (free download) simulation of the circuit that CZ attached in post #45 you can see how the circuit works.

I highly recommend doing a Spice simulation of any circuit you design. It allows you to see and tweak the circuit operation before you build the circuit, and saves a lot of troubleshooting and redesign.

 

[ADWSystems]

crutschow, for some reason I can't figure out what is post #45 each page is number #1 to x. Strange. According to my comuter I have 5 posts #6 (your most recent is the 5th post #6)

I'll assume you are referring to me "wrapping my head around it". I have done the simulations with a dozen configuration and FETs but they don't help with evaluating the integration considerations and other variables not discussed here to keep the discussion focused (and me out of a courthouse, NDA and all). But thank you for the reminder.

 

[crutschow]

The post I was referring to was that from carbonzit, six posts previous to this one labeled THE OP AMP SOLUTION.

 

[ADWSystems]

Oops. My bad.

 

[carbonzit]

Oops. My bad.

Maybe not, if you're seeing the strange numbering behavior you reported (maybe you ought to tell the board-keepers about that: there may be a bug somewhere).

Anyhow, this is the post to which I attached a Spice simulation. Hope it helps.