Valve THD measurement

Let’s take it out straight away: THD measurement will not give you an indication of how the valve “sounds” in a particular circuit. However, still s a good metric of how linear the valve  is at a particular point. Also can help you identifying electrode misalignment in old globe valves particularly if you don’t have a curve tracer. I usually look at the THD for driver valves. Finding the sweet spot for drivers has more to do with it’s driving capabilities  in my opinion (i.e. how many volts can swing for a particular application). Either way, I always start by looking at the THD for an initial operating point, then I refine it where possible to minimise distortion and then proceed to a listening test to do final adjustments. It may be that it “sounds” better at a different point which could well be higher distortion anyway.

So the next question you probably are asking yourself is: can I do this myself at home? Yes, you can and you don’t need to invest in expensive lab equipment for DIY audio purposes. Just look at Pete Millett’s interface which is an affordable solution to measure all sorts of voltages (DC and AC) in your circuits using a PC which can give you a resolution of around 100dB with a good quality soundcard.
I built a simple test box from components I had around. You can optimise this, so I won’t bother in justifying the selection of each component. I use my curve tracer to provide the bias section of the circuit shown below, various valve sockets, voltage panel meter, grid stoppers and some other bits.

Main reason why I built this externally was that I needed a screen supply, I had no space on the curve tracer chassis and also that in order to keep interference to a minimum of the tracer circuit, I needed to add some switch to turn the noisy parts of the tracer off when testing THD. Way too complicated in practice given space challenges I stated before, so I decided to go for a simple aluminium box holding the basic THD circuit parts.

In the diagram above,  a simple CCS load is used based on DN2540s with two potentiometers to allow coarse and fine adjustment of the anode current. The reason why  a CCS load is used and not a simple resistor is that the CCS provides the DUT with a very high impedance load which will help us to measure the THD of the valve itself rather than the THD of the valve circuit. A separate screen bias MOSFET source-follower (M3) was added with a coarse adjustment pot for the screen bias voltage using P4. C2 helps stabilising the bias voltage which it’s filtering aid is multiplied by M3 (i.e. a capacitor multiplier). Size of R7 helps protecting M3 and the rest of the circuit in case of an unexpected screen short. Haven’t added fuses or further protection as I generally use the valve tester to check the state of the DUT. Input signal and grid bias circuit are not included in the box as they are already inside my curve tracer, so no need to replicate this.

The screen bias arrangement is not optimal, but it was design to be used with only one HT variable supply. I have provided an external power socket for the screen bias in case in future a second HT supply is available for independent operation.

With this arrangement when testing tetrodes/pentodes, you will need to set the screen voltage pot down to 0V first. Then adjust the HT supply to the right output voltage which will be anode quiescent voltage plus the expected output swing plus some headroom, and then dial up the screen pot to set the desired screen voltage. A bit more clunky to operate, but this simple can help you testing tetrodes and pentodes very quickly.

M3 is a plastic TO220 so no need to isolate it when bolting it to the box.

Although Pete’s interface has an AC input capability, I decided to add an output cap to protect it when testing transmitting valves at voltages higher than 400V.

The circuit is driven by an external oscillator or the PC. You can set the input to obtain the right output swing based on the valve’s mu. I generally measure THD at 10 Vrms (+22.22dBu / +20dBV) unless I need a particular test. This allows me to compare valve’s linearity across different valve drivers, etc.

I have taken many THD plots using this small test box. You will see many of these THD plots in my website. So far I’m very happy with this little thingy which helps me in the workshop to do tests I need.

Here are some examples:

1. THD benchmark shows a sample testing of various triodes or triode-strapped pentodes drivers
2. 26 THD analysis shows THD across various brands of the famous 26 DHT
3. 955/VT-121
4. 24a Tetrode driver THD analysis