I have a solar power system (SMA Sunny Boy with 2 strings of panels). When the battery is nearing full charge or the inverter maximum output is reached and excess solar power is available the system throttles the amount of power coming from the solar panels. Observing the panel string voltages and currents it appears that it does this by reducing the string voltages while allowing the currents to be whatever the panel strings can supply. Presumably the voltage measurement is taken after some component that regulates the voltage. What type of circuit would be used to do this?
As described in this answer the output current of solar panels is directly related to the output voltage. If you short circuit the panels you will get a very high current - but basically no power, because the voltage of the panales drops towards 0 V. The same is true for a open circuit: In this case the voltage is pretty high, but there is no current flowing. Again, with P = I * U the power output of the panels is very low.
A solar inverter is using a switching DC/DC regulator at the input to controll voltage and current. In the typical use case this circuit would be controlled to find the point of maximum power (the MPP). But when there is a derating, because the inverter is not able to supply as much power as would be possible, it can increase its current draw from the panels and by that automatically decresing the voltage up to the point, where the desired power output is achieved.
In this diagram you can also see the power over panel voltage and current: Image from Analog.com
It's not complex BUT it may seem confusing at first :-).
PV panels are close to current sources with Iout loaded proportional to insolation.
I do not know exactly how the (extremely competent) Sunnyboy makers achieve their result BUT it could be done by using either raw on/off PWM or (more likely) a buck converter** = PWM with an inductive energy store to provide output energy during the off cycle.
Your description of the system providing full current and variable voltage makes sense if measured at the panels and not the output.
An intelligently controlled buck converter** can 'compel' the panels to assume any V or I which are less than the currently maximum available as long as the converter can dispose of the energy*. A panel loaded to more than its Vmp / Imp values at a given moment will provide LESS power, a current somewhere between Imp and Isc (ie slightly above Imp) and a voltage of whatever the system causes it to have.
eg a panel with Vmp=30V, Imp = 10A, Isc = 11A, Voc = 36V,
so Wmp = 300 W, under current insolation could be loaded down to 10V. It would then provide slightly more than 10A (as closer to S/C) at 10V = 100W.
If the panel was hard shorted by the controller it would provide about 11A at zero voltage so no power.
If raw PWM was used, at say 25% duty cycle into an optimum load, the panel would provide 30V, 10A when loaded and Voc, 0A when unloaded. Wout = 300W x 25% = 75W. Vpanel with a well filtered meter would be
(30 x 25% + 36 x 75% ) = 34.5V APPARENT. I apparent = 10A x 25% = 2.5A. Power apparent = 34.5 x 2.5A (too high) unless an RMS power meter was used that measured V & I simultaneously.
*As long as the VI product is less than the wattage available at the selected voltage. That's a quibble added for completeness. eg if Vmp = 30V and Imp = 10A so Wmp = 300A, you could get say 33V but not at 10A.
** Buck or boost or buck-boost depending on system design.