[Amps] power meters for HF/VHF

[Dr. David Kirkby]

John Lyles wrote:
> The Cantenna in question (70 ohms) is no good. As others said, replace 
> the element, someone cooked it too hard.
> 
> Dr. Dave K. is spot on, Bird 43 is not a laboratory instrument. Just a 
> rough indicator. Certainly useful when trimming antennas to watch 
> reflected power, or to tune the input of an amplifier, or the output for 
> maximum. When measuring efficiency, gain, anything where you do need 
> better than 5%, use directional couplers and attenuators, and bench 
> power meters like the old HP 436/437/438 series. Other manufacturers 
> such as Boonton also make decent meters. These can be gotten on epay for
> reasonable cost. If you have access to a network analyzer, even the 
> TenTec kit, or a good gain/loss set, you can check and calibrate your 
> own couplers and pads at the exact frequency being tested.

I would argue you need to do this if you want better than 10-15%, not 5%. Also 
remember the Bird is (or at at least was) rated at +/- 5% of FSD. So if you are 
measuring at the half scale, Bird would claim your measurement would be within 
10%, not 5%. But given Birds specs are at least a factor of 2 optimistic, I'd 
reckon you have a 95% certainty of being within 20% if measuring at half the 
full scale.

> Good power metering is obviously needed if you are making commercial 
> rigs and amplifiers. Calorimetric dummy loads can be fashioned from 
> standard plumbing, if you can find an accurate flow meter and a pair of 
> thermometers in wells. But working with only a few deg C rise in water 
> temperature requires that all systematic errors be removed first. Like 
> offsets between thermometers, recording this value with no RF applied.

 From the point of view of a laboratory standard, any attempt to feed this 
directly from a mains water supply and measure the flow rate would be 
inaccurate, as the flow rate would depend on who has has just flushed the loo, 
or turned a tap on. It would never be a constant, so could not be used for a 
standard.

The one in the standards lab used a constant head of water, so the pressure was 
always the same. With the pressure a constant, the flow rate should never change.

The whole thing was in a temperature controlled room of course.

I suspect information about all this are published in papers by NIST, NPL etc.

I'm looking at this from the point of view of a primary standard. As I've made 
clear, the Bird 43 is fine for normal amateur use. Between a primary standard 
and a Bird 43, there are of course a whole range of other options. I believe the 
directional coupler and lab power meter is the most practical for hams that want 
something better than a Bird.

> I'd guess that the cooling air measurement would be fraught with errors. 
> We used small temperature sensors in the exhaust of the Broadcast 
> Electronics cavity amplifiers for FM, that would readout on the screen 
> and give you a warm feeling about how you were tuning the PA, for best 
> efficiency or far off of normal. But nothing was accurate enough to 
> calculate power.

I suspect, but do not know, that if you designed the amplifier with this method 
of power measurement in mind (rather than try to do it on a random amplifier), 
then it could be made quite accurate.

> Using calorimetric water loads, meters on the plate DC power, and 
> directional coupler measurements at work, I can usually get within 5%
> on power balance, which is considered good. I trust the couplers and 
> power meters the most, as I can measure the coupling of a line section 
> to better than 0.02 dB using a network analyzer. And the power meters 
> can be checked, send out for calibration. This is acceptable to the gov't.

If a calorimeter is designed as a standard (i.e. your *only* concern was best 
accuracy), and things like cost, size, response time were irrelevant), it would 
be better than the power meter/coupler. Those sensors on the HP power meters 
ultimately depend for their accuracy on a water calorimeter. Or at least they 
did, 25 years ago!

Dave