THD testing is something that I have been eyeing for a while….
After I finished my Computer Engineering degree, I came back to this project and I have started to design the PCB in Eagle CAD. The problem is I only have a “Light” copy of Eagle and I can’t use more than 80cm2 of PCB area. However, I wanted to show the updated schematic design. What I showed originally was not the complete circuit, as it requires two full decade resistance boxes (which are set based on the capacitance), rather than what is shown/mentioned below.
Due to size constraints of Eagle Light, I had to slim this down and break it up into two PCBs, rather than one.
As you can see below, it is literally impossible to make the full schematic fit. Your options at this point are:
- Get a full copy of Eagle CAD ($500 per year)
- Break the circuit up into two boards, putting the decade resistance boxes on another PCB. (Extra Cost of x2 PCB)
- Buy two decade resistance boxes (of high quality) and plug them into the appropriate pad locations. ($30 x2)
For example, if you broke it up, you might do something like this:
Now, on adding Total Harmonic Distortion measurements, I want to first mention a few things. You can achieve this in many ways, although many of them are very expensive (dedicated testers). Ones like what you see below, can be achieved for $30-$60 (depending on the capacitance range/s needed and THD measurement resolution). It can be higher, much higher, especially if you need to make a reference bridge for 10-1000uF and you pick x20 Black Gate™ NX, you will have hundreds of dollars invested. I’d be interested to try that range, but I’d probably use Panasonic FM.
To start out, I will likely do 0.01uF to 0.68uF, I’ll use 20-30 capacitors, wired in a parallel configuration (capacitances add in parallel) with SPDT or SPST switches, as shown below.
MultiSim is the software you are seeing above,
MultiSim (By NI) is what most of professional engineers are using, or PSpice. I prefer MultiSim, because I have years of experience using it. I am still learning, don’t get me wrong, but I know how to do nearly all types of simulation testing/analysis types.
MultiSim is limited to -100dB or 0.001% THD and due to this, if I just wire up this circuit as intended, we’ll just see 0.000%. This isn’t very useful, as many things could happen and you’d still see that reading.
To ensure things were functioning as I intended them to, I added a function generator. You need to put this directly on the output of the DUT arm, not before the capacitor (or else it will filter it and bypass the point of doing it). I set mine for 3kHz and about 500mV. Our source signal is +- 5V and 1kHz, so this was exactly on the 3rd Harmonic, normally, the most problematic in audio circuits (2nd is also common).
Quote of my FB post
Essentially, you are creating a near ideal capacitor (using many low THD, like C0G or PP Film) by using many caps in parallel. Along with a 100ppm (or less) resistors in series. The resistor arms use multiple resistors in series, to create the lowest THD as possible. Groner says the voltage coefficients of the resistor arms are negligible for all 100ppm types he tried.
“As there is some correlation between the power coefficient of R1 (ref arm) and R2 (DUT arm), the net distortion is even lower due to common-mode rejection.” – S Groner
Then you can isolate the difference between your reference arm (which is near ideal) and the DUT arm, by looking at the differential of the two signalK. Hallman and S. Groner
Stay tuned as this design is constantly updated, until we have something that performs up to my desired quality level of -170 to -180dB. I will likely just update this post, so you’ll have to check back if interested in this. I’ll make a new post/page, when it’s completed.
Note: If all goes well, I will sell these boards (unpopulated). It will require different designs, for different capacitance ranges. This is due to the different capacitor sizes, lead spacings and lead diameters.