Digital Inductance/Capacitance Meter June 2018 NAN uF Display, Calibration, LCD Issue

Hello EPE fans,Has anyone built the Digital Inductance/ Capacitance Meter from the June 2018 issue?  I had a few issues (bought the wrong serial LCD and couldn't get the one I bought to give me anything other than it's slave address) and an inadvertent bridge of the oscillator output to ground but I've fixed both of those now.  So I now get the unit to calibrate as follows:C1 = 1022.33pFC2 =
91.07pHThen it displays:Cx = NAN uF 
F3 = 521570 HzWhen I add a 1 nF capacitor to the terminals, I can see the frequency change but the software still doesn't update the display with a value or has an issue calculating the value:

Cx = NAN uF
F3 = 369852 HzThe frequency value then falls by a couple of Hz every reading before settling to around 369831 Hz Has anyone else had an issue like this?  I'd appreciate any pointers as to what the issue could be and where to start looking for the problem.

I asked Jim Rowe the designer for comments and Jim has replied to us as follows:His reported problems are a bit puzzling, I must confess. Especially his reporting that his unit gives the first two calibration readings as:    C1 = 1022.33pF (which seems OK), but then    C2 = 91.07pH (which is quite weird, since it should probably display ‘L1 = 91.07uH’
then he reports it as displaying ‘Cx = NAN uF’, and continues to display this weird ‘value’ even when he adds a 1nF capacitor to the terminals and the frequency changes accordingly. It sounds as if the oscillator circuitry is probably working correctly, but either the Arduino program is going ‘a bit off the rails’ or the serial LCD is malfunctioning � perhaps due to it not being set to the I2C address that the Arduino program thinks it’s set to.

The main source of problems with this project has been concerning the addressing of the serial LCD display, so I suggest that this would be the first area for your reader to look. By the way, I gave quite a bit of information on addressing the I2C serial modules (which come in two ‘flavours’, with different address ranges) in episode 5 of the Cheap Asian modules series, which was published in the March 2017 issue of Silicon Chip (pages 82-85). This might help him in fixing the problem. 
One final thought. If he can’t find anything wrong with the serial LCD addressing, he would probably be advised to download the Arduino sketch again and reprogram the Arduino. 
I hope these comments help. 
Best wishes,Jim Rowe

Thank you for your reply.Actually, I made a typo in my original post.  The calibration was indeed 91.07 uH not pH as written.  So it is calibrating C1 and L1 properly and the LCD module addressing seems to be correct (I changed to the module suggested in the magazine).I have reloaded the sketch but I get the same behaviour.  I suspect that either there is an issue with the calculation or string conversion in the software or an error in the connection to the terminal posts.I will investigate further and let you know.

So, the frequency measured (370kHz) ties in with a capacitance of 1nF between the terminals.  This leads me to believe that the measurement portion of the software is working correctly but that the capacitance calculation or the conversion of the number to a string for display is causing the issue.  When I next get some time, I'll have to do some printf debugging to see the values of the variables being passed to the display.

It looks like the code is having an issue with squaring the number [the code is: F1sqrd = pow(F1, 2.0); and all the "Fxsqrd" global variables are defined as float].  I can't understand why because the result should be well within the range of a floating-point number.  If anyone can shed light on why the software is doing this, I'd be grateful.Freq1 = 520635Freq2 = 370375F1 = 520635.00F2 = 370375.00F1sqrd = ovfF2sqrd = ovfF3 = 520630.00F3sqrd = ovfCXval = nanCXuF = nanCdisp = Cx =   NAN uFF3 = 369380.00F3sqrd = ovfCXval = nanCXuF = nanCdisp = Cx =   NAN uF

OK, so I managed to get to the bottom of this now.  The following topic on the Arduino website was very informative: https://forum.arduino.cc/index.php?topic=206718.0The squared calculations were working fine, but the Serial.print() function cannot not print the full range of float numbers, so this gave the ovf rather than a real overflow.So I did some more digging and found the the CF variable was not defined, leading to an error in the CXval and LXval calculations.  There was some issue with the way the EEPROM stored the value of CF.  During debugging, I forced it to write the initial value (1.0) to the EEPROM by setting the flag 'MTorNot' to 0.  Now it seems to have stored the CF value in EEPROM and the calculations are working well.