power meter fun

[woodyc]

A few months back I got two cheap power meters, a WANF and KILL A WATT. More recently I also got a Murata ACM20.

They actually turned out to be useful. I normally plug in any amp and see if it powers up and what its doing audio-wise, but plugging into a power meter can also tell you where its biased or if it has a major issue. This has been a time saver at least twice in the past month. I think it will be even more useful as my database grows i.e. you have to know the STBY and ON watts for normal operation. [Of course for simple tube amps, you can just count tubes and do the math.]

Another curious tube amp thing is that the power factor always drops when you go from standby to on. (In Marshall 100W amps it goes from .95 to .85.) I think this may be because once the HV is on, more turns are engaged and the effective PT primary inductance increases? Thats just a guess. Maybe somebody here knows?

[Steve Conner]

I agree! I have what I think is the European equivalent of the Kill-A-Watt, and an ancient analog wattmeter that I found on Ebay. They're useful for all sorts of things. The Murata ACM20 looks interesting.

The power factor drops because the rectifier draws a distorted current waveform with a low power factor from its winding. That reflects to the PT primary and lowers the overall power factor. In standby only the filaments are on, and those are a pure resistive load with a power factor of 1.

[Prairie Dawg]

I've got a Kill a watt and use it often. I get live voltage and frequency as well as a good take on how much current an amp is pulling. Well worth the money.

[Jazz P Bass]

Kill-A-Watt is my main tool.
I find it amazing how much the line voltage fluctuates.
I am in SE Pennsylvania & it will go anywhere from 118 to 124 volts anytime it wants.
Try setting bias with that going on

[woodyc]

The power factor drops because the rectifier draws a distorted current waveform with a low power factor from its winding. That reflects to the PT primary and lowers the overall power factor. In standby only the filaments are on, and those are a pure resistive load with a power factor of 1.

Steve's right about this. I figured it was an inductance thing because the hv winding has more than 100x as many turns as the 6.3V and 5V. But without the recifier, pf~1 on the hv winding. (With the rectifier pf~.72.)

But that brings up a couple of curious questions. It seems like the current pulses will lead the voltage. Which means a tube amp is not an inductive load at all. Its capacitive. Is that right?

Also... I thought these power meters were only designed to work for sine waves, but if you look at the actual dissipation in the load and winding resistance, it seems like they are getting the right phase shift in spite of the current pulses. The watts are always slightly high but only by a few % -- and that may be just iron losses.

[Steve Conner]

The Kill-a watt type power meters sample the voltage and current digitally and do the math with a microprocessor.

So, the voltage and current readings should be true RMS, and the PF should take both harmonic distortion and phase displacement into account.

(How does harmonic distortion affect power factor? Best understood by Fourier analysis. If the line current has lots of 7th harmonic, say, but the line voltage has none, then the 7th harmonic current performs no useful work. Ditto if the 7th harmonic voltage were present, but 90 degrees out of phase with the current.)

I tested my cheap digital wattmeter against my mechanical one on a nasty load: a SCR controlled rectifier feeding a huge filter capacitor bank. They both agreed on the real power within about 5%.

When a rectifier is fed by a transformer, the leakage inductance of the transformer makes the current carry on a bit longer. The fundamental of the current lags the voltage slightly. But the low PF is mainly caused by waveform distortion, not phase displacement.

[woodyc]

So a tube amp is essentially purely a resistive load, just a weird one.

The Kill-a watt type power meters sample the voltage and current digitally and do the math with a microprocessor...

The question is what math they are doing. I was assuming they did some kind of peak detect and counting to get the phase shift. But in order to get a correct result in this case, I gather it must be doing a digital integration of i(t)*v(t).

[woodyc]

I thought I'd share a pic of the absurd way I plug in guitar amps now: wall - variac - breaker - acm20 - amp. (I usually keep a 5A fuse in the variac and the black box has 1, 2, 3, and 4 Amp breakers where the first three can be switched out.)

[Steve Conner]

No, it's not a resistive load because it generates harmonics in the line current, which resistors can't. I'd call it a non-linear load to avoid having to say if it were capacitive or inductive.

Yes, the power meter multiplies the voltage and current samples together and averages the result.

In a previous job I used a DSO with power quality analysis software that gave you three power factor results. A displacement power factor, a harmonic power factor and an overall one. The cheap power meters just display the overall one, which as far as I know is just the calculated wattage divided by (Vrms*Irms).

[woodyc]

No, it's not a resistive load because it generates harmonics...

Yeah, but as you said the harmonics don't matter and at 60Hz the V and I are basically in phase (see below.)

If anybody's interested, here's some typical waveforms for a guitar amp. Ch1 is current measured with a diff probe across a 1 ohm resistor, and Ch2 is voltage. (And yes, the voltage in my shop is that distorted.)

[Steve Conner]

The harmonics do matter. From the point of view of the power company, they're "wattless" power just as much as reactive power is: they inflate the RMS line current and increase losses in the wiring without doing any useful work in the load.

That's why they are lumped into the power factor, and it just seems wrong to call a load "resistive" when it doesn't have a PF of unity.

[woodyc]

The harmonics do matter. From the point of view of the power company, they're "wattless" power just as much as reactive power is: they inflate the RMS line current and increase losses in the wiring without doing any useful work in the load.

That's why they are lumped into the power factor, and it just seems wrong to call a load "resistive" when it doesn't have a PF of unity.

I'm familiar with the point of view of the power company since people in my neck of the woods pay for their harmonics. (see pic) But you do get my point right? I find it interesting that as loads tube amps have essentially no phase shift. Until a couple days ago, I hadn't even thought about what the current draw might look like. I'm kind of surprised nobody else finds this interesting. Posting stuff here often makes me feel like I farted in church.

[Steve Conner]

Well, in this case you're farting at the converted. I didn't comment on the waveforms because they're the same as what I saw when I tried it.

The charging pulses in the Marshall are quite modest because it has a sensibly sized transformer and filter caps. In a solid state hi-fi amp they can get really ugly.

[woodyc]

...In a solid state hi-fi amp they can get really ugly.

I don't have a lot of data yet, but I've noticed amps with switching supplies seem to have the lowest power factors. The worst was .44 for a Crate Power Block and the highest was a Markbass Little Mark III at .58. For solid state amps with linear supplies the lowest I've seen is a Fender Princeton Chorus at .59 and a Marshall MF350 at .61. In contrast a Crate GX20M is about .82, even cranked. Thats close to tube amp range. The waveforms I've checked have all been basically the same as what I posted above.

I wonder if you can improve these numbers by adding supply capacitance?