[NLRS] power measurement (getting pretty L O N G)
Dr. Gerald N. Johnson
geraldj at ispwest.com
Mon Feb 13 12:10:39 EST 2006
On Sun, 2006-02-12 at 23:39 -0600, John P. Toscano wrote:
>
> Gerald wrote:
> > I guess I've not been memorizing HP catalogs lately (prices way beyond
> > my willingness to spend) so I missed the 8481B.
>
> Since I've been digging out information on the HP power meters, in case
> anyone is interested, here's what I found. There are also some 75 ohm
> coaxial and some 50 ohm waveguide sensors, but I am omitting those in
> this discussion because I consider them either of little ham interest
> (75 ohm) or probably very hard to find (waveguide sensors)
>
> HP 432A, has 7-position range switch (5 dB increments)
> -----------------------------------------------------------
> 478A sensor 0.01 GHz to 10.0 GHz 10mW FS to 0.01mW FS
> 8478B sensor 0.01 GHz to 18.0 GHz 10mW FS to 0.01mW FS
>
> Newer models of meter, which also have a built-in 1 mW calibration
> standard at 50 MHz.
Some of the 3 digit sensors aren't temperature compensated and so drift
a lot.
>
> HP 435A, 435B, 10-position range switch (5 dB increments)
> ---------------------------------------------------------
> 100 kHz to 4.2 GHz sensors
> 8482A sensor 3 microwatts to 0.1W FS
> 8482H sensor 30 microwatts to 3 W FS
> 8482B sensor 1000 microwatts to 25 W FS
>
> 10 MHz to 18 GHz sensors
> 8484A sensor 300 picowatts to 10 microwatts FS
> 8481A sensor 3 microwatts to 100 mW FS
> 8481H sensor 30 microwatts to 3 W FS
> 8481B sensor 1000 microwatts to 25 W FS
>
> HP 437A (digital model)
> ---------------------------------------------------------
> 100 kHz to 4.2 GHz sensors
> 8482A sensor 1 microwatt to 0.1W
> 8482H sensor 100 microwatts to 3 W
> 8482B sensor 1000 microwatts to 25 W
>
> 10 MHz to 18 GHz sensors
> 8484A sensor 100 picowatts to 10 microwatts
> 8481A sensor 1 microwatt to 100 mW
> 8481H sensor 100 microwatts to 3 W
> 8481B sensor 1000 microwatts to 25 W FS
>
> All of the above have been seen on eBay lately, and SOME are even
> guaranteed to be in working condition, many at reasonable prices.
I hadn't noticed what I thought were reasonable prices for anything
newer than 432A. Maybe I'm cheap.
>
> > their power dissipation capabilities, they were toast, burnt to a crisp,
> > and so "overdone."
>
> Sorry, I guess I was a bit slow on that one. Got it now!
>
> > Having multiple transverters with different input levels will lead to
> > the sensitive ones needing IF interface parts from overdrive in the heat
> > of a contest. Conversion to that common level will be good for their
> > longevity.
>
> Agreed, that is the goal. Right now I am "safe" because the 902 and
> 1296 transverters are on a dedicated IF radio that can only output a
> signal in the "well under 100 mW range" (too low for me to currently
> measure), and the 2304 and 3456 transverters are on a different IF radio
> that will eventually drive them all when everything is set up as I want.
> (The low-power IF rig is a modified IC-251a that I find a bit hard to
> use, which is another reason to move everything to a higher drive level.)
>
> > So one of the Analogue Devices log signal level chips that works to 8
> > GHz takes care of much of you're needs. Adding a down converter for
> > higher bands seems like it would be simple (and with a 100 MHz to 6 GHz
> > IF bandwidth, the LO might not need to be crystal controlled ;<)) and a
> > bandpass filter for an elementary spectrum analysis could be a more
> > useful instrument than the bare broadband wattmeter.
>
> W1GHZ has a project along those lines. See:
> http://www.w1ghz.org/new/portable_powermeter.pdf
>
> He uses an AD8307 logarithmic power detector chip for 0-150 MHz with
> good linearity from -70 dBm to +10 dBm, and an LT5508 chip, rated up to
> 7 GHz, which he finds usable from 220 MHz up to 10 GHz, although the
> response is not very linear and with a much smaller useful power range
> of -20 dBm to +13 dBm. In his kit, he uses a simple 10-step LED bar
> display to show approximate power, but the actual output voltage could
> be measured by another device (such as an analog or digital voltmeter),
> and with a bit of intelligence in an added controller chip, could yield
> a (reasonably) calibrated digital display.
>
> It could make a neat project, I suppose, to add the microcontroller
> intelligence and digital display to augment the crude 10-step bar graph.
> One would still need some external attenuators to measure more than 10
> or 20 mW, as would the older HP432/435/437 meters with most of the
> available sensor probes.
I see charts of error vs temperature for those chips in the latest
magazines. There is a wiggle each stage, about a decade and them a shift
with chip temperature. Not linear/log enough for a mirrored linear scale
(in dB) meter.
>
> I figured a nice attenuator set would be:
> -40 dB at 100W to make 100 W into 10 mW
> -30 dB at 10W to make 10 W into 10 mW
> -20 dB at 1W to make 1 W into 10 mW
> -10 dB at 1W to make 100mW into 10 mW
>
> I was looking around eBay for pieces to make such a collection, and did
> not find anything suitable unless I was willing to be limited to a
> maximum of 1.5 GHz, 2.2 GHz, or 4 GHz, and some of the prices were still
> high.
>
> I found a retail dealer on the net that sells lots of microwave parts,
> including attenuators that come pretty close to what I wanted:
> www.smelectronics.usa
> I never heard of them before tonight, but I was impressed with how close
> they could come to what I wanted:
>
> -40 dB at 50W to make 100 W into 10 mW (ok, 50 W into 5 mW)
> -30 dB at 10W to make 10 W into 10 mW
> -20 dB at 2W to make 1 W into 10 mW
> -10 dB at 2W to make 100mW into 10 mW
>
> These are all SMA male to SMA female bidirectional attenuators rated to
> cover 0 to 18 GHz! I found them on the "firesale" page, and they are
> special prices. In order from top to bottom, $200, $122, $23, and $25.
> These are brand new and guaranteed. I can't vouch for the vendor
> (yet), but I did buy a few items from them tonight and will report back
> if I have any trouble with them. But I have no other financial ties
> with them, just reporting what I found.
Some years ago I worked out logic to design attenuators automatically
using a choice of 1, 5, or 10% resistors with standard values and with
quarter, 1/2, 1, 2, 5, or 10 watt resistors. The program breaks each
stage of attenuator into one with maximum dissipation in the four
resistors the chosen resistor size and parallels the shunt resistor in
the T (starts with two in parallel for better RF results, the first
approximation of a disk resistor) sometimes with a third resistor to
improve the match. The program produces an asymmetrical attenuator. I
may have to resurrect it someday and modernize it from CBASIC to C and
to prove the designs have merit. Using leaded resistors might not work
well above half a GHz, but maybe using chips and stripline techniques
might work through 10 GHz.
>
> Oh yeah, just to "finish" the story...
>
> I bought a used, working HP432A meter with good 478A sensor from a ham
> in the state of MN today, for under $160 with shipping included. (I
> didn't feel like driving to Madison Lake and back.)
>
> I'm watching a couple of the 8478B sensors on eBay that are rated to 18
> GHz instead of only 10 GHz, even though the 478A sensor is "probably
> good enough" at 10.368 GHz. If the price of either (or both) stays low
> enough, I may bite. (Besides, as Jerry and others pointed out, they ARE
> fragile and spares can't hurt. The ones I'm looking at are guaranteed
> working.)
>
> I'm watching a couple of directional couplers on eBay. This would allow
> me to sample output power (at somewhere between -20 dB and -40 dB of the
> input power) with a dummy load or antenna on the output end. Some have
> one sample port, so you'd have to manually flip the connections to
> measure forward versus reflected power, others have both ports so you
> could sample both at the same time. And you wouldn't need a 50 watt
> attenuator, just a 50 watt load (or an antenna) if testing 50 watts...
Couplers will have more frequency sensitivity than attenuators, but if
they are calibrated either at the factory or with your wide range
milliwatt meter (with good attenuator) they can be very repeatable and
be calibrated at exactly your frequencies of interest.
>
> And I'm seriously considering buying a cheap cheap HP435A or 435B just
> to get the 1mW calibration output, even if the rest of the meter is
> shot, if the price is low enough. Who knows, maybe it could be gutted,
> keeping the calibration output and then using the W1GHZ circuit to drive
> the nice analog meter with its dB/Watt scales (with appropriate signal
> massaging). Another future project to add to the list, as if I don't
> have enough irons in the fire already!
>
> 73 (and good night!) from W0JT
> _______________________________________________
73, Jerry, K0CQ
--
Dr. Gerald N. Johnson <geraldj at ispwest.com>
More information about the NLRS
mailing list