I would like to find a good way of implementing a true watt meter too. I dont wanna go with the complication of using two coils but if anybody knows a simple work around please do tell. I have an ancient true watt meter i picked up off the car boot. It's huge and must be worth a fortune. That has seperate coils in it but unfortunatly the scale is calibrated to 600 Watts. Not much use for me unless i'm using it to verify total power consumption from the mains. There must be a cheap way of sorting this out so we can display true watts on a cheap moving coil voltmeter.
Leigh
Leigh
Geek said:
Right !
Not to tell about multi ways enclosures
Be does'it really matters ?
Do you wanna know how many watts are effectively transfered to the loudspakers or how many watts the amp would be able to push in a know load ?
Indeed, I just need to check if the amp is working correctly and the voltage developed across various resistors is all what I care off
Matching the right loudspeaker with the right amplifier (and vice et versa) is a completly differrent story
Yves.
As mentioned, you could use an analogue multiplier, but you'd need more than just that, because real power is
P = V*I*Cosine(phi)
If you just multiplied the current and voltage, (no cosine term) you'd get the apparent power - which is not the real power because it doesn't take into account the ratio of speaker inductance to resistance. (or capacitance where/if that exists)
SOMEWHERE I have a circuit that does it - I'll have to dig it out.
The cheapest (and smallest component count) way is to use an embedded microcontroller (eg, Pic or atmel) but that assumes you can write code & have a programmer.
P = V*I*Cosine(phi)
If you just multiplied the current and voltage, (no cosine term) you'd get the apparent power - which is not the real power because it doesn't take into account the ratio of speaker inductance to resistance. (or capacitance where/if that exists)
SOMEWHERE I have a circuit that does it - I'll have to dig it out.
The cheapest (and smallest component count) way is to use an embedded microcontroller (eg, Pic or atmel) but that assumes you can write code & have a programmer.
I have not looked carefully at any commercial power measuring chips, but are they not primarily intended for 50Hz/60hz AC mains. They may not have the bandwidth to cope with audio. (just a guess)
Certainly the circuit I have - as it stands - uses a PWM conversion at around 1khz, so it would need to be modified to cope with audio.
Thinking more about it, a useful audio power meter would have to have a bandwidth in excess of 20kHz, since when amplifiers go into clipping, they generate a lot of ultrasonic harmonics (the tweeters are always the first to burn at parties).
Certainly the circuit I have - as it stands - uses a PWM conversion at around 1khz, so it would need to be modified to cope with audio.
Thinking more about it, a useful audio power meter would have to have a bandwidth in excess of 20kHz, since when amplifiers go into clipping, they generate a lot of ultrasonic harmonics (the tweeters are always the first to burn at parties).
I found the circuit I was thinking of... just got to copy the relevant part neatly - the original is undecipherable!
Unfortunately it has no practical component values - it just shows the operational theory.
But it got me thinking, as did the fact I'm currently also working on a log converter for a moving coil voltmeter (*): what is a suitable time constant for an amplifier watt meter? Does one want a peak-hold system, or to follow the time-average?
(*) Normal listening levels hardly show up as visible twitches on a moving coil (or bargraph) meter that has its full scale marked at something like 200W, so I need the low end (the -60dB to -15dB area) much magnified, and the -6dB to +3dB area rather compressed.
Unfortunately it has no practical component values - it just shows the operational theory.
But it got me thinking, as did the fact I'm currently also working on a log converter for a moving coil voltmeter (*): what is a suitable time constant for an amplifier watt meter? Does one want a peak-hold system, or to follow the time-average?
(*) Normal listening levels hardly show up as visible twitches on a moving coil (or bargraph) meter that has its full scale marked at something like 200W, so I need the low end (the -60dB to -15dB area) much magnified, and the -6dB to +3dB area rather compressed.
Steerpike said:
oouuuuuhhhhh (sound of engineering pain) -- don't use a log converter for this ---- too many expensive components. Take a look at the Linear Tech LTC1966 or LTC1968. You can get accuracy to a fraction of a dB and very good bandwidth. The datasheets for these components lift the kimona past the hootchie if you know what I mean.
I agree - I'm avoiding that route - they are too prone to drift and temperature dependence.
I'm doing it with a PIC (embedded microcontroller), which will let me do all kinds of extra sexy things at the push of a button, like conversion/calibration to phons!
And its *cheap*.
Those LTC chips though seem to be RMS-to-DC converters.
[which is conventionally done by Log/antilog processing (because squaring is equivalent to doubling a logarithm, and square rooting equivalent to halving a logarithm]
I'm not doing that - what I want is change the scaling factor of the meter, so its very sensitive at low signal levels, and insensitive a high levels.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.