[NLRS] USB Power Meter

[Dr. Gerald N. Johnson]

The thermistor heads tend to not need calibration IF they can be zeroed. 
Fundamentally they are very close to primary power standards. The meter 
applies DC or AC bias (depending on model) with a bridge circuit 
measuring the resulting resistance of the thermistors and adjusting the 
bias for 50 ohms total for coax as seen at the RF port and lower for 
waveguide. Then when power is applied that resistance changes and the 
meter reduces the bias and indicates directly the power replaced from 
outside in the change in the bias. The only technique more predicatable 
is caliometric. The results are quite independent of frequency limited 
more by connector on the high end and the RF coupling capacitor on the 
low end. The thermistors are tiny and the most modern of heads include a 
second pair of thermistors to detect ambient temperature changes like 
holding the head in your hand.

Waveguide thermistor heads like the one with WR90 interface have been 
used as relative indicators at millimeter waves and at light. Light can 
penetrate to the thermistors and heat them.

Diode detectors have thermal drift as the barrier voltage is temperature 
sensitive as is back leakage and the forward bias current. Their 
junction capacitance limits high frequency response and the very low 
barrier voltage devices are nearly as fragile as thermistors for power 
overload. Then the meter is just gain with no feedback around the 
detector unlike the thermistor heads. Its easy to control that gain but 
not so easy to control the detector changes, so the 435 and later 
detector head meters include a front panel power reference for checking 
the head's calibration. Meters are cheap, while heads are outrageously 
priced.

Paul Wade, W1GHZ, has published a power calibrator circuit. Basically 
its a packaged 50 MHz crystal oscillator, a couple silicon switching 
diodes as clipper, a low pass filter, and an attenuator. $10 in parts if 
you have to buy the parts and the box. Its on his web site collection of 
projects.

Power meters tend to have a slow response and going digital tends to 
make that more difficult for alignment. For production pass/fail testing 
digital is great even for trouble shooting its great but a royal pain 
for alignment. And for digital the update rate depends a lot on how 
often the data is sampled and how much data is averaged to take out 
circuit noise. DC measurements at -70 dBm can't be very much voltage. 10 
^-10 watts on 50 ohms 70.7 microvolts rms, 100 microvolts peak, 100 
microvolts DC with the perfect half wave detector. Bound to be 
accompanied by noise. But the sensitivity is 30 or 40 dB better than a 
thermistor meter.

Absolute accuracy of the USB detectors would be a bit suspect without 
checking one regularly with a reference RF source such as Paul's or one 
in a 435 or 436 meter. I'd gain confidence if I could trust a switchable 
attenuator to test at lower power levels too.

I don't know what HP used for that power reference, I've never looked up 
that circuit. I expect Paul did. His reference will be a little 
temperature sensitive as the barrier voltage in his clipping diodes will 
affect the clipped voltage. I've not tested the one I built using a 100 
MHz oscillator with a scaled low pass filter. The finite switching time 
of the 1N4148 diodes may cause mine to be out of calibration.

73, Jerry, K0CQ



On 10/25/2013 9:44 AM, Doug Reed wrote:
>
>
> A quick look didn't find any similar accuracy specs for a HP
> 8481-series power sensor so I don't know if the USB meter is better or
> worse. I can say that a HP 435 analog power meter has a rated spec of
> up to 1% error in the full-scale meter reading, but that isn't the
> same as error in the power sensor. But the HP spec means the USB meter
> is NO WORSE than a HP 435 with the power sensor of your choice at
> frequencies below 4GHz. How they compare above 4GHz depends on the
> accuracy of the HP power sensor itself.
>
> Bottom line is that the USB sensor is probably no worse than a HP 436
> digital meter for accuracy. Depending on the computer, it may be a
> little clumsier to use. Another way to look at it is that upgrading
> from a HP 436 single channel meter to a HP 438 dual channel meter is a
> matter of buying a second USB power sensor and writing software to
> compare the readings from two USB ports. And if you write the software
> correctly you can calibrate out any(?) test setup errors.
>
> Personally, I'm usually more interested in relative power than in
> absolute power. If I can determine relative differences in power
> readings I can measure gain or adjust for peak or null. If I want to
> know absolute power I need a calibrated standard of some sort and you
> always have to wonder if the cal standard is correct, or more likely
> how far off it is. Since none of my test equipment is within any cal
> date, it is a moot point.
>
> Over all, I think I'm favorably impressed by the USB power meter
> specs. The +-1dB and +-2dB specs don't seem too impressive, but after
> looking at the HP specs I could find, I'm willing to bet that spec is
> probably pretty reasonable for a piece of test equipment under $200.
> If I already had a HP 436 with a good power head or could buy one for
> that price, I probably would. I probably would not sell my HP 435 to
> go out and buy the USB power sensor. But if I was considering buying a
> HP 431/432 and power sensor from a used source, versus buying this USB
> sensor, I'd give serious consideration to the USB sensor, assuming I
> can live within the 11GHz frequency limit and didn't need the wider
> selection of power heads available for the HP.
>
> 73, Doug Reed, N0NAS.
> ______________________________________________________________
> NLRS mailing list
> Home: http://mailman.qth.net/mailman/listinfo/nlrs
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:NLRS at mailman.qth.net
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>
>