Notes on Building the Thing and Measuring the Results

Point-to-point wiring is FUN!  Done carefully it can be every bit as successful, traceable and do-able as PCB construction - and a damn sight quicker too!



Performance Measurements


Gain Structure

The open loop voltage gain of the overall amp, from input to loudspeaker, I measured at roughly +42dB (x 126). The closed loop gain I ended up with was +25.7dB (x 19.2). The difference between those open and closed loop gain figures is the amount of loop negative feedback; 42 - 25.7 = 16.3dB.

I was surprised at how close the closed loop figure measured (+25.7dB, x 19.2) was to the "op-amp" formula involving the feedback resistor (22 kOhm) and the 1 kOhm from 6AN8 pentode cathode to ground; (22K + 1K)/1K = 23 (+27.2dB). This formula assumes an infinitely high open loop gain for the amplifier, something far, far from the case in a valve amplifier of this nature. I recall seeing a modified formula for use in this actual situation, but it is understandably more complex, involving additional terms.

Frequency Response

I only ever made the most simplistic of frequency response measurements. With my odd output transformers I felt that achieving a flat +/-3dB response 40Hz~20kHz would be a minor miracle, so when I saw I'd easily got past that, actually quantifying the results seemed like gilding the lily! So I simply applied a 0.5Vpp sinewave to the input, and measured the output across an 8 Ohm resistor in place of the speaker, using my oscilloscope ("CRO" for short). The input signal level is chosen such as to produce a mid-level output - neither very low nor too high and consequently near overload. Frequency points to measure are pretty arbitrary but I chose a few that would give a reasonable spacing if plotted on logarithmic graph-paper, as frequency response measurements usually are; - linearly spaced measurements intuitively make sense but end up with big gaps between, when plotted logarithmically. If anything, it's sensible to make a few additional measurements toward the frequency extremes; where it is expected that response is changing rapidly - i.e. rolling off. That helps, to get a good graph, if you want to go that far. It's wise to pay particular attention to checking and rechecking the 0.5Vpp level of the signal generator at each measurement - the output of the signal generator isn't necessarily constant for all frequencies, and of course variation there makes a real joke out of subsequent measurements . . .! Not that that's ever wasted an hour or two for me . . .oh, NO . . .

There are quite a number of things that could be done to make such testing harder; adding capacitance across the resistive load, for example, better simulates real life complex-impedance loads, and provokes potential amplifier instability to a much greater degree a good thing to find out about during testing rather than use . . .) Testing nearer the amplifier's overload level is more indicative of "full power bandwidth" rather than a more favourable small signal bandwidth, is another example of more strenuous testing that typically might be done. Applying a 10kHz squarewave is another more critical test of upper frequency (and phase) response. A 10kHz squarewave is the summation of a series of sinewaves at multiples, (harmonics), of 10kHz. The more higher harmonics of that 10kHz fundamental that the amplifier will faithfully reproduce, the more faithful the squarewave will be, as it appears across the load. With experience, deficiencies in that output squarewave can be "read" in detail, pointing toward specific amplifier response-related problems.

All of which was a moot point in my case, as I didn't do any of it! Near enough was good enough. Bad, Paul, bad, bad, bad. I'm taking myself off for a sound whipping . . . er . . . now where was I?

Anyway, 0.5Vpp input produced a moderately even set of output levels, which with a bit of math (dB = -20log10(Vo/Vin)) gave a +/-0.2dB response from 40Hz~100kHz. This upper limit I thought was quite amazing, irrespective of the total lack of quality of my measurements! Since the dB scale is a relative one, it is necessary to chose a reference value and make all the other measurements relative to that; the value at either 400Hz or more often 1kHz is typically chosen for this purpose. I chose the 1kHz value because it gave a nice midpoint between a +0.2dB maximum figure at 400Hz and a -0.2dB minimum figure, between 4kHz and 40kHz.

It would have been a good idea to actually measure the -3dB points, high and low, as these are frequently used to quote and discuss frequency response of an amplifier. This would be done by firstly deciding what that datum level frequency was going to be - in my case I'd decided 0dB was at 1kHz, meaning 9.8Vpp output - and back-calculating what -3dB of that would be; multiplying the reference output level by 0.707 (the square root of 2) gives that - about 7Vpp in my case. Then sig-gen frequency is adjusted at upper and lower response extremes to find the points at which that amplifier output voltage level occurs.

Frequency response graph; click here for larger image.


In my case, as I said, having gotten past 20kHz I didn't bother with the upper limit - I noticed the response dropping fairly rapidly after 100kHz but didn't quantify it (I must go back and do that at some stage, just for interest's sake!). My sig-gen gets a bit dodgy below about 30Hz, so I'd have trouble measuring the lower -3dB point in this case anyway.

Frequency 40Hz 100Hz 400Hz 1kHz 4kHz 10kHz 40kHz 100kHz
Open Loop
for Vin = 100mV
10.6Vpp 12.2Vpp 12.6Vpp 12.6Vpp 12.8Vpp 12.2Vpp 9.2Vpp -
dB -1.5dB -0.3dB 0dB 0dB +0.1dB -0.3dB -2.7dB -
Closed Loop
for Vin = 500mV
9.9Vpp 9.9Vpp 10Vpp 9.8Vpp 9.6Vpp 9.6Vpp 9.6Vpp 10Vpp
dB +0.1dB +0.1dB +0.2dB 0dB -0.2dB -0.2dB -0.2dB +0.2dB


Hum and Noise

With the input pot at vol = min. and no input signal, I had about 11mV 100Hz hum across one channel 8 Ohm resistive load, (the channel physically laid out nearest the mains transformer and power supply section odly enough), and about 25mV across the other. These levels would be way too high if I was using headphones anywhere near the amplifier, but in practice sub-audible using my normal "K-Mart-Special" loudspeakers. Supply ripple at the voltage doubler was 17Vpp (on +365Vdc at the time of measurement), 0.5Vpp on the B+ (+318V) supply to the output stage and 5mVpp on the (+308Vdc) supply to the 6AN8.