Simulation of a three phase voltage meter: MOV error (Analysis: Step time too small) LTSPICE

Thread Starter

MarreneM

Joined May 2, 2020
8
Hi, I was hoping someone can help me to solve my error on LTspice.
I am designing a three phase voltage sensor using voltage divider resitors and a isolated differential amplifier. The voltage range I need to measure is max 440V_rms. I want to add protection by adding a thermistor and MOVs.

I first tried to add movs (VDRH05E250 x2 in series), but it gave me an error: "Anaylsis: Timestep too small; initial timepoint:trouble with d_lim1-instance d:u:1:filter: output:d4:1". Any ideas what the problem might be?
 

Attachments

Thread Starter

MarreneM

Joined May 2, 2020
8
Ohhh sorry, this is the grounds I created in this circuit as well. So Vdd1 is isolated from Vdd2. But it still gives the same error
1600337911480.png
 
Last edited:

ci139

Joined Jul 11, 2016
1,696
(for the 240VRMS) you could try 1...10MΩ in series with
the (100 ... 1000 pF or greater (... the capacitor perhaps in parallel with 4GΩ . . . 2TΩ) )
that -- if the 10µF causes some problems
 

ci139

Joined Jul 11, 2016
1,696
maybe set some series resistance to Vdd1 and Vdd2

solving the best GND referencing scheme for any particular circuit is a tedious work of guessing (for the Spice to be willing to eat it)
. . . sometimes it prefers the GND to be at ~AC side sometimes at –DC side sometimes at the neutral or summed 0 of ~AC sometimes elsewhere

there is no universal rules for such /// i suspect the COM is just the node name and you might want to tie it to GND as you do for GND1 as https://electronics.stackexchange.com/a/402222

since your detector circuit models' behaviour is unknown (U1 to U3 -- may require actual GND to "compute" properly) you likely want to keep this as a reference point
 
Last edited:

eetech00

Joined Jun 8, 2013
1,992
Hi, I was hoping someone can help me to solve my error on LTspice.
I am designing a three phase voltage sensor using voltage divider resitors and a isolated differential amplifier. The voltage range I need to measure is max 440V_rms. I want to add protection by adding a thermistor and MOVs.

I first tried to add movs (VDRH05E250 x2 in series), but it gave me an error: "Anaylsis: Timestep too small; initial timepoint:trouble with d_lim1-instance d:u:1:filter: output:d4:1". Any ideas what the problem might be?
These are the correct values for Three phase 230RMS source (you don't need the AC statements)

1600358871896.png
 

eetech00

Joined Jun 8, 2013
1,992
Hi, I was hoping someone can help me to solve my error on LTspice.
I am designing a three phase voltage sensor using voltage divider resitors and a isolated differential amplifier. The voltage range I need to measure is max 440V_rms. I want to add protection by adding a thermistor and MOVs.

I first tried to add movs (VDRH05E250 x2 in series), but it gave me an error: "Anaylsis: Timestep too small; initial timepoint:trouble with d_lim1-instance d:u:1:filter: output:d4:1". Any ideas what the problem might be?
This will help troubleshoot.

1. First, save your work.
2. Remove all blocks (window delete), then add them in one at time (undo one block, run sim, repeat).
Run sim after each add until error is produced.
 

MrAl

Joined Jun 17, 2014
7,811
Hi,

You do realize that in a Wye system the neutral is the junction of all three phases right? I assume you are working with a Wye system because of the way you draw the parts. If you are working with a delta system then you dont use a neutral.

An error such as the one you are seeing occurs when the resulting network equations are too 'stiff' and so the differential equation solver can not find a time step that satisfies both the short time constants of the circuit and the long time constants of the circuit. When it finds a time step for the short time constants it is so short that effective numerical derivative numbers appear to be the same for any local step in time and this can produce a numerical overflow. Problems like this can also occur if the equations result in something known as a 'shock' but that is harder to deal with.

The solution to the time constant difference problem is to look for all the parts that are responsible for the time constants of the circuit and try to adjust them so that they are closer together.
For example, if you found a 1 nanosecond time constant and a 20 hour time constant you would change the part values so that the 1 nanosecond increases to maybe 10 nanoseconds and/or the 20 hour time constant down to 2 hours. To do this look at the capacitors and inductors and associated resistor values and increase resistances if you have a very short time constant and decrease if you find a very long one.

One thing that looks suspicious is that 1gigohm resistor. Try reducing that.

Also note that a ground from a three phase Wye system can be the system ground and then measure the other nodes by using subtraction. For example, in a single phase full wave rectifier if you make the input AC source ground the system ground, you can measure the output by subtracting the voltage at the negative output of the bridge from the positive output of the bridge. One way to do this is to simply subtract using an output equation like
Vout=v(Vop)-v(Von)
and then you are again looking at the output of the bridge rectifier. Many people like to ground the negative output of the bridge, and then a large value resistor has to be used between the negative lead of the AC input supply to system ground (which is now the negative output of the bridge).

But check all the time constants first and compare.

Another way we might see numerical problems is with a capacitor that has to charge very very quickly. The solution here is to add some ESR to the capacitor which also happens to be more realistic anyway.
For an inductor, add some ESR to the inductor or perhaps a large value resistor in parallel (simulates core loss).
 
Last edited:

Thread Starter

MarreneM

Joined May 2, 2020
8
Okay so I did a bit of all the solutions.
I deleted my voltage sensors, and moved my ground to the neutral of the supply.
1600366163096.png
And then this happened...
1600366265331.png
What am I doing wrong? It is the MOV's. Can I place MOVs in delta connecting like that?
 
Top