Minimum detectable change in current or voltage

[Just_Plain_Bob]

Is there an interactive simulation setting that controls the minimum detectable change in voltage or current that the simulation will update on or display on an instrument (eg multimeter, scope etc).  Here's the situation: Use a source of 0.818181V applied across a network of a 1Meg in series with 9.9M in series with a 200k pot. Measure across the 9.9M/pot combo to return (gnd). Vary the pot. That represents a 1 percent change in a 10M resistor (9.9M - 10.1M). Neither an ammeter nor a voltmeter will detect the change at the junction of the 1M and 10M.  It represents a 300pA change and about a 7mV change neither of which is detected....

(0.818181 ++) ---- 1M ----ammeter---9.9M-----200kpot----gnd

                                        |

                                         --voltmeter-------------------------gnd

The simulation will not update circuit values changing as a result of varying the 200k pot.  V10.1.1 version.

[Just_Plain_Bob]

Anyone run into this? DeltaI actually 680pA not 300pA and is not sensed by Msim instruments....

[GarretF]

Hi,

I don't know exactly what your schematic looks like, but I think you may be connection the potentiometer incorrectly. You need to connect to the wiper so that the resistance changes.

 

[Just_Plain_Bob]

Garret,

Give me a break. Of course it's not the pot hookup. The point being made here was the instruments were not registering a low level change. What you're not seeing here is a switch was also being used to switch between two different values of resistance each variable by 1%. NI developers say the switch was the problem due to how the spice engine emulates a switch and suggested a work around of using an ABM to switch voltage onto one network or the other or to simply remove the switch and manually connect/disconnect. Not much of a workaround in my opinion. Defeats the purpose of having a switch. Tells me that msim doesn't handle the switch model very well and I consider it a bug.

The funny thing about it was the switch was only incidental for testing and not part of the real circuit. Install switch, circuit ceases to function. Remove switch and circuit functions (kind of) only showing 20% of the change that should be measured. How much simpler can you get than a few series resistors and measuring a current and voltage? Duh.

That's not all. A multimeter and a probe both read drastically different currents for the same series leg and both register a voltage change of only 1.5 mV when they should be seeing near 7mV instead. Another bug in my opinion.

I didn't think I needed to provide a schematic for a simple series network measuring a voltage and a current.

[GarretF]

Hi,

I apologize for making my post seem so simple. This forum often has students learning electronics, and questions about how to connect a pot do come up.

I looked at this again and I spoke with the developer, and the problem in this example seems to be the difference between the largest and smallest resistors in the schematic. Increasing the resistance on the ammeter seems to give correct results. Also, the default resistance of the voltmeter is 1GOhm, and the largest resistance in your circuit is 10MOhm. If I increase the voltmeter's resistance to 100GOhm, the results are accurate to 4 significant digits. I think this may be the cause of your seeing a change of 1.5 mV instead of 7 mV.

I think that we might be able to detect this problem, so I've entered a feature request to see if we can generate warnings in this case.

[Just_Plain_Bob]

Garret

Here is a portion of a test circuit where this all first surfaced. You will find that instruments do not respond well at all. Beyond that, if it were only an instrument issue, the buffering jfet should have permitted a meter to read the full change across the source resistor. The voltage change there is no different than unbuffered and that tells me msim is having difficulties in HiZ low current calculations. That's why I was wondering if there was some constant/setting controlling the smallest error value (or change) msim would detect and update circuit values on. In other words, the error value may be too large (convergence value too large) to accept the low current change in the network as a valid change in network current.

Again, switches were there only for testing. That's when it was discovered that switches rendered circuit completely nonfunctional. Then the measurement anomalies with both test instruments and DC Operating point analysis were noticed. This all leads me to the conclusion that perhaps a convergence value needs to be controlable by the user or simply made smaller depending on circuit Z. I know you have to draw the line somewhere or you'll never converge, but smaller or user settable would be good.

Robert

[GarretF]

There are some settings that control when the simulator accepts that the solution for a particular point in time is correct (ABSTOL, VNTOL, CHGTOL, RELTOL), and perhaps others that I don't know about. Unfortunately, I think this is an issue of relative resistances, so changing these would have no impact.

You can of course access and set these convergence parameters and others in Multisim. In case you haven't seen them yet, you can set them as follows:

Interactive Simulation

1. Click Simulate > Interactive Simulation Settings
2. Select the Analysis options tab
3. Select Use custom settings, and click Customize...

Analyses 

1. Open the analysis dialog (e.g. click Simulate > Analyses > Transient Analysis)
2.  Select the Analysis options tab
3. Select Use custom settings, and click Customize...

In terms of the way SPICE or Multisim works, there is no concept of a minimum detectable change in current/voltage, unless you are talking about the interface to the digital world, but even then it isn't a minimum change. The simulator will always use the last calculated current/voltage, and Multisim will faithfully display this result, whatever it is. Nevertheless, we do recognize the issue and I've entered a request for our Simulation and Modelling team to investigate, but I don't know if or when we might have a solution.

[Just_Plain_Bob]

Jarret

Thanx for staying with this. I did a slightly different workaround. (simulation tolerances unfortunately didn't help). What I did was take a current controlled voltage source and set the resistance to the test value in the circuit, ie, 10M so the 7mV would be generated when the 10M R was varied 1%. Still couldn't reliably use switches but it's better. I can go on with ckt design.

Thank you

[Just_Plain_Bob]

Garret

SO SORRY. I just realized I've been calling you Jarret.

RLP