
Set up on 2022-04-22.
Confirmed that it worked reasonably: 100 pF air capacitor giving similar FDS results with/without series battery.

Did most of the work on 2022-04-24 evening. 
 up to series 3*

Then some revision on 2022-05-12 
 series 4* on the tall HV compressed-air capacitor 100pF, as the 100pF reference capacitor showed )
 series 5* on batteries in series with ~66uF capacitors, to infer further frequency range of 'EIS' of the batteries. 

See  plotting/plot__doplot.m  
for final plotting of results. 


Battery packs are 11 series-connected PP3 9V batteries.
  One is alkaline batteries (Panasonic Powerline Industrial Alkaline); this is the default, assumed in the absence of specification.
  The other is lithium (RND power Lithium Ultra Power), denoted 'Li' in the filenames. 
Dielectric spectroscopy instrument is IDAX300 (Pax Diagnostics, later Megger). 

Test objects for dielectric spectroscopy.
 (There were 4 different objects used: see the '*' markers in the following.)

The 3 objects in the first set were each done with the following situations: no battery, then all four combinations of the two batteries and two polarities. 
These objects:
  * 100 pF air capacitor (old 'normalkondensator', PTB-approved type), with guarding of frame, 
  * 2.2 nF yellow cylindrical (ICEL MPL 630V) (polypropylene?), 
  * the same 2.2 nF but with a parallel 820 kohm for a super-lossy example
In each case the FDS AC voltage was 100 V peak from the 'Internal 200V Peak' source. 
Some repeats were also made. 
The longest sweeps were 10 kHz down to 0.01 Hz at 3 points per decade. 
Many later sweeps were cut shorter, stopped usually before the final point (so the last recorded was 0.0215 Hz).

Then an even bigger capacitor was used:
  * a similar looking 0.22 uF yellow cylindrical capacitor (ICEL MPL 250V) 
For this, the internal 10V peak source was used, at 10 V peak amplitude. 
Purposes:  
	an example with voltages being DC >> AC instead of DC ~= AC. 
	even more effect of battery resistance due to the large capacitance,
	this large capacitance is quite realistic of large field-objects like transformer, machine winding, cable,
	good to try the 10V internal source that drives the output without buffer amplifier and that measures the output withiout a divider ... can we make anything strange happen by applying our DC?
Peculiarly, the *second* point of each sweep (4.6 kHz) got missed as 'current too high for instrument' but the first didn't ... bad feedback choice?
After the first sweep with no battery, the next two sweeps were with the 100 V battery in series (alkaline), but although it was supposed to be, and probably was, in the + direction; I became unsure (after disconnecting, probably so as to change + to -, but then getting distracted by thoughts of a possibly better sequence interleaving Li with Alk...). So these two measurements were named as 'x100Vdc' instead of '+100Vdc' ('x' for don't know).  Then a definite + sweep was done, followed by a repeat then a negative. The + repeat was to compare with the coming - sweep to see if polarization current due to the DC had noticeable effect on the FDS (answer: no; the +, repeat +, and -, were basically the same); note again that the capacitor was not swapped in direction, so a swap of the battery means a repolarization of the capacitor.  Then repeat without battery, then quick (1 each way + and - with the lithium battery. 

There is significant effect of the batteries on the uncompensated measurement with 'large' capacitance. 

Tables of fundamental-frequency data (C', C") from the interface program were saved for each object. 
Raw 'idf' output files were also saved, which could be interesting for e.g. checking harmonic levels: do the batteries behave asymmetrically enough that second harmonics could be clearly much more in the with-battery cases compared to the no-battery cases? 



In the later tests (12/May) a different 100 pF object was used, with lower loss at LF.
This is the venerable ~100 pF (96-ish?) compressed-air high-voltage capacitor. 

The 12/May tests also included batteries in series with ~66uF capacitance, which was intended more to tell us about the batteries than the capacitors!  I.e. even bigger capacitance would have been nice if easily available without significant loss or relaxation phenomena, since the sole purpose of the capacitor was to block DC current. 

For these measurements the battery was connected with its + pole towards the IDAX 'high' lead.
The circuit was:  IDAX output terminal, capacitor (3 parallel of ~22uF ICEL 'MPL' model), battery+, then battery- to the IDAX measure terminal.
I.e. parallel-looking coupling of IDAX and battery: + to +.
This direction is what would have been called '-' in the earlier FDS measurements, which were
based on whether the battery added or subtracted from the IDAX source voltage, to give the voltage across the capacitor.


