Boxing Day was put to good use assembling the latest incarnation of the Valve Analyser hardware and the results have been extremely positive. The board now incorporates a MAX1044 (ICL7660S) which creates a -ve rail by inverting Vin (set to around 7.5V) thus providing good rails for a TL074/084 as the anode current sensing amplifier in place of the expensive LM6144 rail-to-rail.
A first test using an ECC83 immediately demonstrated the benefits of the new design, with the first obvious improvement being the better current resolution at low anode currents by using an additional amplifier to create a more sensitive range. I opted for a gain of 4 for the second amplifier which yields low-end sensing ranges of approximately 60mA and 15mA (the Arduino ADCs are 10 bit so the resolution on the most sensitive range is roughly 15$\mu$A). The resulting anode plot for the ECC83 is shown below:
The curves are now noticeably smoother and closely match the published characteristics of an ECC83 – but I was puzzled by the fact that the curves suggested that the device was always conducting to a small degree (the curves appear asymptotic to a slight slope rather than the X axis). This measured phenomenon was not present with the previous circuit using the single rail LM6144. Going back to the circuit, it became apparent that the anode voltage sense network is a constant load across the capacitor banks in parallel with the device. A quick calculation (300V / 1419400$\Omega$) suggested that the current drawn by this network should be about 0.2mA with a voltage of 300V – this is about spot on for the measurements above. A quick change to the software to compensate for the effects of this network yielded the curves below:
The curves now are a very good match indeed for the published curves and show much better resolution than the Nano-based board. Furthermore, the fact that the circuitry is capable of accurately measuring the effect of the current flowing through the voltage sense network gives me good confidence that the device is now performing well for low anode currents.
I have now updated the Git repository with both the code updates and the new hardware – the software will continue to evolve against this hardware platform.