Confusion Over Ranges and Limits

I'm having trouble understanding how measurement ranges, source limits, and source ranges all relate. I'm trying to figure out how to best to set up our unit for an application, which is basically this:

We are trying to observe a solenoid's self-heating behavior under various currents. We set up the 2461 to source current, then read the resistance as the part heats up. Once it reaches a trigger point (either a max heat or the resistance stabilizes), we then allow it to cool until repeating with another current level. We're using a LabVIEW program sending SCPI commands to the SMU.

What we seem to be experiencing is that even though we explicitly set SOUR:CURR:VLIM to some reasonably high value, we appear to be hitting a voltage limit equal to the number we set for DIG:VOLT:RANG. So in hotter tests, the resistance rises to a point where the voltage exceeds the DIG:VOLT:RANG, the unit can no longer hold the constant current, and it seems to switch to constant voltage. The manual describes some interplay between measurement ranges and source limits and ranges, but I'm having a hard time understanding exactly how. I don't think we want to just set DIG:VOLT:RANG to 24V or something and lose resolution at lower voltages, but I don't know if I can predict what a good value would be. We're currently setting DIG:VOLT:RANG to the calculated starting voltage + 20%, i.e. we expect the voltage to rise no more than 20%.

So I guess my questions are:

• Can someone confirm that DIG:VOLT:RANG can become the V limit while sourcing current, and perhaps even why or how?
• Is there a smarter way to handle this other than to just try and determine a better measurement range? We don't want to use auto range due to the time it takes.
• Is the problem that we are not actually specifying SOUR:CURR:RANG? Or even a SOUR:VOLT:RANG?

I can provide a full list of the commands being communicated, but here is an example with what I believe are the relevant ones in order:

:SOUR:VOLT:READ:BACK ON
:SOUR:CURR:READ:BACK ON
DIG:FUNC "VOLT"
SOUR:FUNC CURR
DIG:VOLT:RANG 9.035928
DIG:VOLT:SRATE 10000
DIG:VOLT:RSEN
DIG:VOLT:UNIT OHM
SOUR:CURR 0.464811
SOUR:CURR:VLIM 12.000000

Voltage quickly rose to about 9.036 where it stopped, and current began falling as if switching from CC to CV.

I hope that's all clear and thorough enough. Apologies if this is obvious, but the manual isn't quite clear. It mentions that these various settings are related, but sometimes stops there, declining to describe how or when one affects another.

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?
 

The valve does not actually reach the max temp before hitting the current limit, though it does rise and continues to rise after hitting the current limit, though not at the rate that it would if it were still under constant current.

I meant to update this sooner, but I have managed to make this tester work by simply better calculating the maximum expected voltage given the maximum desired temperature, and then still padding it with an extra 20%. Sending this more accurate value for DIG:VOLT:RANG allows all tests to run as desired. No new commands were added. After thinking about it for mere seconds, it makes perfect sense to do this because there's little point in setting an insufficient read range even if the SMU would source beyond that range. The whole point is monitoring output, so why would we want to output more than we can read back?! 

This seems like the best way to do this. Since the read is the most important, just set the best read range and expect the source range and limit to follow accordingly, rather than explicitly set them all and hope they don't disagree. Is that a good practice? Set what matters and let the SMU handle the rest?

Thank you once again for your help!