A variety of amps and measured output power

[xtian]

I find myself with ten amps in the house--eight of my builds plus a JCM 800 2203 and a Mitchell PRO-100. I took the opportunity to measure each for output power before clipping. (1kHz sine at 1v p-p at input, observe output voltage p-p on scope, multiply by 0.3535, square, divide by output load.)

Here are my findings:

Code: Select all

amp            watts  power section     plate voltage
Twenty          3.2   2x 6V6 cathode   
Element 79      7     2x EL84 cathode   366
Hammond Stout   7     2x EL84 cathode   340
Shasta         12     2x 6V6 fixed      380
Rocket Dog     15     4x EL84 cathode   
AceTone        17     2x 6V6 fixed   
Fiddy TFL      35     2x EL34 fixed   
Fiddy V3       46     2x EL34 fixed   
PRO-100        63     4x 6L6 fixed   
JCM 800 2203   72     4x EL34 fixed   

First, I'm surprised the cathode biased amps are so much lower in power than fixed. I have no comparisons to make from this list (I don't have any fixed-biased EL84 amps). Would you expect the Rocket to be 15 watts? Would you expect cathode biased amps to be HALF the power of fixed?

Second, I obviously have issues with the "Twenty".

[katopan]

The problem with measuring this way is that everything is under significant sag with a steady sine wave more so than when driven with a real guitar. This will lead to significantly lower output power numbers than what the amp will actually deliver at max clean output. The fixed bias numbers will be low, and the cathode bias even lower (because of the cathode bias cooling off more under steady sine wave than with a dynamic guitar signal). That's why the cathode bias numbers are lower than you expect.

I have moved to measuring output power with a guitar, strumming until you can see the onset of clipping on the CRO, and then applying the calculation as per what you have. This gives higher peak voltages and the calculated power lines up with more generally accepted numbers.

The only amps I have tested I can compare to your list are:
Marshall 18 Watt clone vs. Hammond Stout (which is an 18 Watt variant) - With the guitar I measured 12.25W clean output power (Marshall advertised spec from decades ago is 13W), where under steady sine wave it was from memory around 9W.
2204 clone vs. 2203 clone (which you'd expect to be double) - 60W clean with guitar, and I can't remember the number for sine wave but it was much lower.
Deluxe Reverb vs. AceTone (is this a comparable plate V?) - 27.6W clean with guitar, and much lower with sine wave.

[teemuk]

I don't know if there's any other as indicative power rating than the continuous one. Yes, undoubtedly it will load the amp and the "burst power" rating may be significantly higher...

...but do we really want to take the route of car audio and their stupid power ratings where 50-watt amps turn kilowatt monsters when power is derived from burst power multiplied by someone's shoe size. Sigh.

Human hearing isn't really sensitive to transients but it is sensitive for "average" signal changes. For average loudness and cleannes we can't rely on burst, peak or transient power ratings, which can be all over the map. Continuos average power rating reveals real performance of the amp.

That aside, how comparable are those results? Are the measurements performed on same THD levels or are they relying on "visual calibration" of % of distortion?

[Jana]

I've always measured with a steady sine wave.

Some results (from memory)

"50 watt" Marshall with EL34 and 380 B+ = 35 watts
Four EL34 with 480 B+ and 2500 ohm OT = 98 watts
Two EL34 with 470 B+ and 3700 ohm OT = 60 watts
Two 5881 with 380 B+ = 26 watts
One 6L6 and 395 B+ = 11 watts

Sunn Model T with four 6550: With original GE or Tungsol tubes it would put out 150 watts clean. With newer tubes it would struggle to get 120 watts. There may be other factors involved, I don't know--don't have the amp any more. I just remember that once I put new production tubes in it, the original power levels were reduced.

Oh, I remember this amp at a place I used to work. It was a power amp for a vibration test shaker. The "voice coil" or shaker looked like a portable cement mixer. The power amp was this rack of tubes--rows and rows of 6550 tubes. The amp was 5000 watts. Someone decided it didn't have enough power so they got an even bigger amp with more tubes. This one was 10,000 watts! I have no idea the plate voltages. The "speaker cable" going from the amp to the shaker was about an inch in diameter for each lead.

[Stevem]

From memory, and this is with top notch output tubes testing in at 90% or better.
sorry about no B+ numbers.
Marshall 50 watt el34, 45 rms
Early Marshall plexi, 50 watt, 52 rms
Highwatt 50 el34, 60 rms
Highwatt custom 100 4-el34, 115 rms
Marshall major 4-kt88, 180 rms.
Marshall 2205 4-el34,88 rms
Fender 1967 bandmaster, 7581 tubes,43 rms.
Sunn Sonaro 6550, 62 rms
Oliver 1971 B200, 6l6gc 56 rms.rewired from 7027 tubes.
Fender 1976 pro reverb,6l6gc, 62 rms.
Highwatt 200, 4-kt88, 720 B+, 224 rms
Fender hot rod deluxe, 6l6gc,32 rms

[potatofarmer]

This is also dependent on the quality of the scoped waveform - both the resolution of the scope and at what point the operator decides is "just before clipping."

Most guitar amp power ratings I've seen say "XX watts at 5% (or 10%) distortion".

Plus, I mean, what's a few watts here or there between friends?

But to add to the list: I recently measured my "135W" Twin Reverb at 118W clean, continuous sine.

I've also got a homebrew amp with a cathode-biased pair of EH 7591s with 480V at the plate into an 8k4 load putting out a whopping 23W clean, continuous sine.

[Milkmansound]

Wattage is just a ballpark number for me. It's a range. There's too much variation from amp to amp, especially with the new production tubes and line voltages being what they are from house to house.

A 20W amp can be 15-22w. Are you gonna hear a difference anyway? Usually not with that small a range. 40W can be 35-45 and so on.

My 20W amp can smoke my 85W amp - depending on the source. Midrangey guitar is effective at filling a room whereas a clean steel guitar takes a lot of poop to get out there.

[xtian]

I agree that this testing method is inexact. But it has proven to be a reliable ballpark for two reasons. First, the power has been what I expected for the fixed biased amps. (And I'm learning what to expect for cathode biased ones.)

Second, regarding the comment about harmonic distortion, I find that the difference between where the wave is clearly sinusoidal, and then when it begins to clip and/or change shape, is only a couple of volts. In fact, the amplitude quickly levels off, becoming square at the limits of excursion. In retrospect, it would have been just as accurate to crank the amps, observe where the max p-p voltage appears as a square wave, and subtract a few percent.

I'm fine with this being a very conservative estimate. It's just a learning tool for me, and a way to compare amps, and to judge their general health.

---

As a related question, how do I calculate the power of a square wave at output?

[potatofarmer]

As a related question, how do I calculate the power of a square wave at output?

With a true RMS meter.

If you're getting a pure square wave, then the RMS voltage is equal to the peak voltage. With a mangled square wave, you need a 'true RMS' meter to measure the equivalent DC voltage as you can't (easily) calculate it.

There's a handy explanation here (skip to the bottom): http://www.allaboutcircuits.com/vol_2/chpt_1/3.html

[teemuk]

ond, regarding the comment about harmonic distortion, I find that the difference between where the wave is clearly sinusoidal, and then when it begins to clip and/or change shape, is only a couple of volts. In fact, the amplitude quickly levels off, becoming square at the limits of excursion.

Yep. Kinda debunks those myths about tubes clipping softly and generating endless resource of soft clipped power. In the end the truth is that they just hard clip to power supply rails like any other generic amp.

[sepulchre]

1kHz sine at 1v p-p at input, observe output voltage p-p on scope, multiply by 0.3535, square, divide by output load.

Pardon my ignorance but wouldn't you multiply by 0.707 of the peak to peak?

[teemuk]

Nope. 0.707 is the multiplier for calculating RMS (of sinusoidal wave) from the peak value. 0.3535 (which is 0.707 divided by 2) is used because he is using peak-to-peak value in the equation, which with sinusoidal wave is Vpeak*2.

[sepulchre]

Ah, I see. Thanks!

[martin manning]

ond, regarding the comment about harmonic distortion, I find that the difference between where the wave is clearly sinusoidal, and then when it begins to clip and/or change shape, is only a couple of volts. In fact, the amplitude quickly levels off, becoming square at the limits of excursion.

Yep. Kinda debunks those myths about tubes clipping softly and generating endless resource of soft clipped power. In the end the truth is that they just hard clip to power supply rails like any other generic amp.

NO TUBES FOR YOU!

Softer clipping in tubes vs. SS is not a myth, but I've never heard anybody claim "endless resource of soft-clipped power." Either way, once you reach the limits of the power supply, there you are.