Convergence issues with switched capacitor dc-dc converter

Hello,

I've created a dc-dc converter using a switched capacitor set up. When I run the circuit at an input voltage of 1V, I have no issues. However, for an input voltage of 2 V or greater, I get the following:

*************************************************
Transient Analysis `tran': time = (0 s -> 100 us)
*************************************************

Notice from spectre during IC analysis, during transient analysis `tran'.
There are 16 IC nodes defined.

Top 10 Solution Convergence failure counts accumulated from the beginning of simulation:
1 I2.M10:int_si
1 I24.M3:int_di
1 I24.M3:int_si
1 I2.M8:int_si
1 I1.M8:int_di
1 I1.M8:int_si
1 I24.M2:int_di
1 net9
1 cap1
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
1 net9
1 1
1 2
1 2by3
1 net4

Trying `homotopy = gmin' for initial conditions.
Top 10 Solution Convergence failure counts accumulated from the beginning of simulation:
6 I26.M3:int_si
6 I25.M3:int_si
4 I24.M3:int_di
4 I1.M8:int_di
4 I24.M2:int_di
4 net9
3 I26.M3:int_di
3 I25.M3:int_di
2 net30
2 net12
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
7 net4
4 2
4 2by3

Trying `homotopy = source' for initial conditions.
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
4 net4
3 2
3 2by3

Trying `homotopy = dptran' for initial conditions..

Trying `homotopy = ptran' for initial conditions..

Trying `homotopy = arclength' for initial conditions.
None of the instantiated devices support arclength homotopy. Skipping.

Error found by spectre during IC analysis, during transient analysis `tran'.
ERROR (SPECTRE-16385): There were 7 attempts to find the DC solution. In some of those attempts, a signal exceeded the blowup limit of its quantity. The last signal that failed is I(V2:p) = 1.99673 GA, for which the quantity is `I' and the blowup limit is (1 GA). It is possible that the circuit has no DC solution. If you really want signals this large, set the `blowup' parameter of this quantity to a larger value.
ERROR (SPECTRE-16080): No DC solution found (no convergence).

The following set of suggestions might help you avoid convergence difficulties. After you have a solution, write it to a nodeset file by using the `write' parameter, and read it back in on subsequent simulations by using the `readns' parameter.

1. Evaluate and resolve any notice, warning, or error messages.
2. Perform sanity check on the parameter values by using the parameter range checker (use ``+param param-limits-file'' as a command line argument) and heed any warnings. Print the minimum and maximum parameter value by using `info' analysis. Ensure that the bounds given for instance, model, output, temperature-dependent, and operating-point (if possible) parameters are reasonable.

3. Check the direction of both independent and dependent current sources. Convergence problems might result if current sources are connected such that they force current backward through diodes.
4. Small floating resistors connected to high impedance nodes can cause convergence difficulties. Avoid very small floating resistors, particularly small parasitic resistors in semiconductors. Instead, use voltage sources or iprobes to measure current.
5. If you have an estimate of what the solution should be, use nodeset statements or a nodeset file, and set as many nodes as possible.
6. Use realistic device models. Check all component parameters, particularly nonlinear device model parameters, to ensure that they are reasonable.
7. If simulating a bipolar analog circuit, ensure that the region parameter on all transistors and diodes is set correctly.
8. Loosen tolerances, particularly absolute tolerances like `iabstol' (on options statement). If tolerances are set too tight, they might preclude convergence.
9. Increase the value of gmin (on options statement).
10. Use numeric pivoting in the sparse matrix factorization by setting `pivotdc=yes' (on options statement). Sometimes, it is also necessary to increase the pivot threshold to a value in the range of 0.1 to 0.5 by using `pivrel' (on options statement).
11. Try to simplify the nonlinear component models to avoid regions that might contribute to convergence problems in the model.
12. Divide the circuit into smaller pieces and simulate them individually. However, ensure that the results are close to what they would be if you had simulated the whole circuit. Use the results to generate nodesets for the whole circuit.
13. If all else fails, replace the DC analysis with a transient analysis and modify all the independent sources to start at zero and ramp to their DC values. Run transient analysis well beyond the time when all the sources have reached their final value (remember that transient analysis is very cheap when none of the signals in the circuit are changing) and write the final point to a nodeset file. To make transient analysis more efficient, set the integration method to backward Euler (`method=euler') and loosen the local truncation error criteria by increasing `lteratio', say to 50. Occasionally, this approach fails or is very slow because the circuit contains an oscillator. Often, for finding the dc solution, the oscillation can be eliminated for by setting the minimum capacitance from each node to ground (`cmin') to a large value.

Analysis `tran' was terminated prematurely due to an error.
finalTimeOP: writing operating point information to rawfile.
Top 10 Solution Convergence failure counts accumulated from the beginning of simulation:
1 net17
1 I23.M10:int_di
1 I23.M10:int_si
1 I22.M10:int_di
1 I22.M10:int_si
1 I21.M10:int_di
1 I21.M10:int_si
1 I20.M10:int_di
1 I20.M10:int_si
1 I19.M10:int_di
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
1 net9
1 1
1 2
1 2by3
1 net4

Trying `homotopy = gmin'.
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
5 2
5 2by3

Trying `homotopy = source'.
Top 10 Solution Convergence failure counts accumulated from the beginning of simulation:
4 I24.M3:int_di
3 I24.M3:int_si
3 I24.M2:int_di
3 net9
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
3 2
3 2by3
2 1

Trying `homotopy = dptran'..
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
6 1
6 2
6 2by3

Trying `homotopy = ptran'..
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
5 2
5 2by3

Trying `homotopy = arclength'.
None of the instantiated devices support arclength homotopy. Skipping.
Top 10 Residue Convergence failure counts accumulated from the beginning of simulation:
5 2
5 2by3

Error found by spectre during DC analysis, during info `finalTimeOP'.
ERROR (SPECTRE-16385): There were 7 attempts to find the DC solution. In some of those attempts, a signal exceeded the blowup limit of its quantity. The last signal that failed is I(V1:p) = -3.01741 GA, for which the quantity is `I' and the blowup limit is (1 GA). It is possible that the circuit has no DC solution. If you really want signals this large, set the `blowup' parameter of this quantity to a larger value.
ERROR (SPECTRE-16080): No DC solution found (no convergence).

The values for every node on the last Newton iteration are given below. For those nodes that did not converge, the manner in which the convergence criteria were not satisfied is also given.
Failed test: | Value | > RelTol*Ref + AbsTol

Top 10 Residue too large Convergence failure:
V(2) = 0 V
residue too large: | -8.50663 TA | > 42.543 GA + 1 TA
V(2by3) = 0 V
residue too large: | -1.41694 TA | > 7.0946 GA + 1 TA

The following set of suggestions might help you avoid convergence difficulties. After you have a solution, write it to a nodeset file by using the `write' parameter, and read it back in on subsequent simulations by using the `readns' parameter.

1. Evaluate and resolve any notice, warning, or error messages.
2. Perform sanity check on the parameter values by using the parameter range checker (use ``+param param-limits-file'' as a command line argument) and heed any warnings. Print the minimum and maximum parameter value by using `info' analysis. Ensure that the bounds given for instance, model, output, temperature-dependent, and operating-point (if possible) parameters are reasonable.

3. Check the direction of both independent and dependent current sources. Convergence problems might result if current sources are connected such that they force current backward through diodes.
4. Small floating resistors connected to high impedance nodes can cause convergence difficulties. Avoid very small floating resistors, particularly small parasitic resistors in semiconductors. Instead, use voltage sources or iprobes to measure current.
5. If you have an estimate of what the solution should be, use nodeset statements or a nodeset file, and set as many nodes as possible.
6. Use realistic device models. Check all component parameters, particularly nonlinear device model parameters, to ensure that they are reasonable.
7. If simulating a bipolar analog circuit, ensure that the region parameter on all transistors and diodes is set correctly.
8. Loosen tolerances, particularly absolute tolerances like `iabstol' (on options statement). If tolerances are set too tight, they might preclude convergence.
9. Increase the value of gmin (on options statement).
10. Use numeric pivoting in the sparse matrix factorization by setting `pivotdc=yes' (on options statement). Sometimes, it is also necessary to increase the pivot threshold to a value in the range of 0.1 to 0.5 by using `pivrel' (on options statement).
11. Try to simplify the nonlinear component models to avoid regions that might contribute to convergence problems in the model.
12. Divide the circuit into smaller pieces and simulate them individually. However, ensure that the results are close to what they would be if you had simulated the whole circuit. Use the results to generate nodesets for the whole circuit.
13. If all else fails, replace the DC analysis with a transient analysis and modify all the independent sources to start at zero and ramp to their DC values. Run transient analysis well beyond the time when all the sources have reached their final value (remember that transient analysis is very cheap when none of the signals in the circuit are changing) and write the final point to a nodeset file. To make transient analysis more efficient, set the integration method to backward Euler (`method=euler') and loosen the local truncation error criteria by increasing `lteratio', say to 50. Occasionally, this approach fails or is very slow because the circuit contains an oscillator. Often, for finding the dc solution, the oscillation can be eliminated for by setting the minimum capacitance from each node to ground (`cmin') to a large value.

Analysis `finalTimeOP' was terminated prematurely due to an error.
modelParameter: writing model parameter values to rawfile.
element: writing instance parameter values to rawfile.
outputParameter: writing output parameter values to rawfile.
designParamVals: writing netlist parameters to rawfile.
primitives: writing primitives to rawfile.
subckts: writing subcircuits to rawfile.

Any ideas what is causing this?