Andrea Clary

What version of software is on your 4200A?

I am using 1.13.

I do not duplicate any error unless I do not set list sweep size of all SMU to same size.
It looks to me like you are doing that but is a topic to double check.

Initially you are doing 10 point list sweeps.
Then SMU3 does a larger number of points list sweep while all other SMU are a bias condition.
And it is when you attempt to again have 10 point list sweep that the illegal setup error occurs for SMU4 and SMU5.
It behaves like the "SS VL3,1,0.100000,1.1,0,0,0,0,0,0,0,0,0" command has not taken effect and it still thinks you have the more than 10 points list sweep defined for SMU3 when trying to define 10 point list for SMU4 and SMU5.

Link to service manual:
Service Manual on w140.com

Advise you to contact the mfg of the Nortec device.  Quick web search:  That seems to be Evident Scientific.

When using a fixed measure range, the value of that range imposes limits on valid values for the vlim setting.
Limit must be within 0.1 and 105% of the range value.

If the limit is encountered, the SMU will transition from CC to CV mode.
You provide the R;  SMU controls I or V but not both at same time.

The 2461 in digitizer mode has dual A/D.  So you can know both the actual I and the V even though your DUT is no longer being excited with the desired current if in CV mode.  Does the valve get hot enough before the limiting occurs?
 

Interesting topic.
A few years ago I had a customer use two SMU channels:  one for sensitive current measure and one to generate a voltage signal for a fiber alignment tool.
If simulating a changing current with few hundred Hz sinewave + resistors, we saw about 200usec of dwell time on each step of the following voltage signal.

In that case, the two SMU channels were within the same physical box, e.g., dual channel 2602B.

Can you use the two channels in the 2634B?
Since 2634B does not have TSP-LINk feature, it is limited for coordination with the 2657A.
If strongly determined to eliminate a PC from the mix, you could try the use of tsp-net and have the 2657A send commands over LAN to the 2634B.

Sounds like you are hoping the 1MHz digitizer in 2657A will give you fast loop rate.  But you are still taking single shot measurements, computing the voltage level and needing to source it.  I’d expect other overhead, specifically the voltage sourcing rate of the other SMU, will be your rate limiting factor.

Try:  older post - same topic

Q:  Is it possible to load a .csv to 2450 using pyVISA to generate the current pulses?

A:  Yes but will require some coding to pass the values of the csv to a source configuration list.
And building a trigger model to step through the list at a desired rate to achieve the waveform timing.

In your uploaded image of the pulse, they are about 2 seconds in duration.  Agree?
How many points would you want to use to approximate that pulse?

Is SMU sourcing only?  No measure?

Since your source function is current, the voltage compliance limit command would look like this:
:SOUR:CURR:VLIM 5

Separate from this is a protection level.  The instrument is capable of up to 200V.  The protection level lets you set an upper limit on this.
Suppose you meant to set the VLIM to 5.0 but passed 50 by mistake.  Having a PROTection level at 10 would restrict the output to 10V.

SOUR:VOLT:PROT PROT10

NOTE the enumerated values for protection level,  12 discrete values supported.

I note in the code, there is use of the TLINK connector digital triggering of the 2400 (TRIG:ILIN or TRIG:OLIN).
Unlike the 2602B, the 2604B does not have any digital IO.
The 2400 emulation is trying to map the TLINK triggering to digital IO lines in the 2600B.
Unfortunately, the model you picked does not have any digital IO.

And just checking: you mention you got the 2600B to be able to go to higher currents than what the 2400 can provide.
Using the emulation of 2400, the same 1Amp restriction will be applied.  You will not be able to access the new features of the 2604B when addressing it as a 2400.

The source meter can be configured to output a list of values from a source configuration list.
By controlling the speed of outputting each value from the list, the waveform can be generated.
Keep in mind, the max speed of 2450 for source + measure operations is about 1700 operations per second.

Github Sample Code for Current Sine Wave with 2450

What sort of timing in your arbitrary current were you hoping for?

Our model 6221 has an ARB mode with faster current sourcing than 2450.