Amplifier Output Frequency Responses

2019 Addition: I finally found out that the characteristic dip in these curves corresponds to the resonant frequency of the circuit, the main contributor of which seems to be the first input stage of the 12AX7 valves as I saw this alot in nearly all valve amps. Explained here:

Amplifier Output Frequency Responses
I thought it would be helpful while doing the Output Section Posts to log some data for my different amps output stages, either across the 6V6 anode, the 8/16 Ohm speaker output or both, depending on accessibility to the leads etc. I’m not taking them apart to do it!
In the case of a sealed cab, I can use a headphone or line out if available.
This will be a rough overview of general behaviour, but what I have seen so far is surprising to me, like the degree a speaker alters the frequency response of the whole amp, compared to the anode output.
For a detailed technical overview of the differences between valve and solid state amplifiers see these important parallel articles:
As the amps have different power outputs, I thought the best way to generate the most “fair” curves for all amps would be to take the voltages at 1kHz test tone, with a 200mV input, at about half the voltage that distortion starts to appear as the gain/volume knob is increased (or just ½ output voltage if it is a clean amp).
This gives fair swing either way for the different frequencies to boost or cut, depending on the amps characteristics, and would also show any distortion due to frequency related gains.
Aston Kit Amp
A+ Volts DC = 343V, Distortion starts at 40V AC Anode
References: 20V AC @ 1kHz, 200mV input
Aston Kit Amp FR
B&W Jubilee Transistor Amp
This is where I’m expecting a more linear response from an amp.
B&W Jubilee Transistor Amp FR.jpg
I still have a big dip from 100-500Hz, so I wonder how much of that is down to the FLS signal via the M Audio headphone socket. As expected otherwise from a transistor amp, it is flat from 1k Hz – 20k Hz. This is a clean only 26V A+ practice amp with no distortion channel, just a good tremolo. Its design may make a good pre-amp pedal, or just using the tremolo section as an addition to a valve amp maybe. I still can’t find a circuit diagram for this. It’s based on 2 Darlington pairs.
Maggie Amp – initial build – uncorrected bias – ½ distortion volume – 16 ohm load
Maggie Amp – initial build.jpg
Marshall Mercury Hi and Lo Inputs – 12” Greenback (note an EL84 output valve)
Marshall Mercury Hi and Lo Inputs.jpg
Fallen Angel 16 ohm – Channel 2 no boost/boost, and Channel 1 clean (note the orange and red lines are swapped incorrectly in the graph key).
Fallen Angel 16 ohm.jpg
There is little point doing the Vox AD30VT as it is software modelled, so the output tube characteristics is changed for each preset anyway – it’s the basis of its clever technology.
 Summary Conclusion
Ok, only a rough overview, and not accurately or best chosen parameters perhaps, but as the test kit is the same for all, with the best approximation to ½ volume of distortion for all, it still gives a conclusion in that the valve outputs voltages, and so current generally rises markedly with increasing frequency, all else being equal, compared to the transistor only amp. As the Lenard Audio site states, this is the basis of the perceived “valve power” loudness in comparisons with transistor amps – because it IS putting out more power at higher speaker frequency impedance! This is just the nature of valves, and does not diminish the technical characteristics of transistor amps – most people just prefer the sound of valves boosting speakers the way they do. That is why it is an unfair comparison of an X Watt transistor amp to an X Watt valve amp, depending how the figures were taken in the first place. If both are read as 100W at 1kHz say, then the valve amp may be putting out 4 times that at 10kHz compared to the transistor amp – of course it’s going to sound louder!
The Marshall gives mixed signals as it is a hybrid – 2 transistor pre amp stages, and a valve output (EL84).
I would like to do the Fallen Angel again at louder and distorted signals, as there is anything from 0 to 24dB boost between the Channel 2 results. They would not be a parallel curve as the boost works by changing the NFB with a capacitor value, as seen in the Maggie Boost section, and this NFB varies with frequency.
On the whole – an interesting exercise. It’s the shape of the curves that matters when comparing the different amps not the output levels, as this is the indicator of overall tone differences between them. It gives the general idea where they will be toppy or bassy etc.
What is also interesting is the difference between Hi and Lo inputs on the Marshall, with only the input impedance being the difference between the two curves – massive change in treble and bass frequency components of the signal, and probably overall strength, once passed the input.