One gets frustrated when either a feature does not exist or does not work as it should but we have to be patient.
All beta programs always have room for improvement. Let’s all be patient and report bugs and enhancement requests directly by email to Mike Engelhardt.
The seven wonders of the world were not built in a day.
Qspice current source has instance parameters VSAT and VSAT2 to force a current source not to source power, which give such different. The solution is very simple, just add a 0V voltage source to sense current source current for B-source to use.
I used this method (ammeter as a zero voltage source) back in Spice2G. In LTspice it is possible to take the current of the current source for expression. You get used to good things quickly. I used this method to derive the volt-ampere characteristics of several bipolar transistors. For example, to compare their characteristics. This is not explained in detail in the Qspice HELP. Instead, there is an abstruse mention of Thévenin equivalent and Norton equivalent. That’s why I have a question. If it was mentioned that control sources cannot use the current of the current source, there would be no question.
I realized that fulfilling my wish could be costly (increased math). I can do without it. It would just be proper to mention this limitation in the HELP.
Please excuse me for hitting on the Qspice program. I see great promise in this program and have started to slowly make a collection of models as I did in LTspice.
Here’s my solution: Since the current coming from I1 is the same as the current entering R1, substitute I(I1) with I(R1). This eliminates the need for the V source in the top circuit.
I don’t need to model this simple circuit. It’s just a simple way to show the problem. Believe me - I’ve seen the resistor method. In LTspice program sometimes using current through resistor in dependent source expressions gives wrong result. Especially in the circuit where I tried to use the value of the source current in the expression there are no resistors.
I have reported operational issues to the author that I have found. Some have gotten addressed quickly. Others may take time needed to do them correctly.
In either case, Mike has been very understanding of people documenting potential issues and even suggestions as to improvements. I believe his goal is to provide a world-class sim tool that matches if not exceeds others like it.
The problem is that SPICE programs that support sensing the current of a current source don’t usually do it correctly. Instead, the current is from the previous time step and the small signal AC current is always zero.
Berkeley SPICE supported current sensing for anything internally represented as a Thévenin equivalent: V-devices, E-devices, L-devices, H-devices, and B-devices that are voltages or resistors specified as Thévenin.
QSPICE extends Berkeley SPICE to support correctly sensing the current in these these Norton(or trans-Norton) devices: F-devices, G-devices, R-devices, S-devices, and W-devices, which I’ve not seen elsewhere done correctly.