Clifton Laboratories 7236 Clifton Road  Clifton VA 20124 tel: (703) 830 0368 fax: (703) 830 0711

E-mail: Jack.Smith@cliftonlaboratories.com
 

Revision History
04 June 2009 Original version

Table of Contents
Performance_Specifications
How_the_803A_Operates
Behind_the_Front_Panel
How_the_Bridge_is_Operated
50_Ohm_Load_Measurement_Results
Impedance_of_RG-223_Coaxial_Cable

Before the vector network analyzer, dinosaurs ruled the world and real men used bridges or slotted lines for VHF and UHF impedance measurements. Or so it seems.

In undergraduate electrical engineering school's transmission line laboratory in 1967, we made impedance measurements with both slotted lines and a much superior device—as it directly read impedance magnitude and phase—a Hewlett Packard 803A bridge. The transmission line lab had, of course, an accompanying HP 417A VHF/UHF super-regenerative receiver and a HP 608E signal generator.

A detailed analysis of the 803A's operating principles is provided in the HP Journal's April 1950 edition, with the following being a brief summary extracted from HP's 1960 catalog:

And, of course, whether it was wise to put the latest and  greatest test equipment into the hands of undergraduate EE students is another question. It is pretty difficult to damage the 803A bridge and associated equipment, short of dropping it out the lab's second story window, which I am also sure entered into the equation of which gear to outfit an introductory laboratory with.

It's been some years since I used the 803A and I found the phase adjustment knob rather stiff. The 803A manual says "no lubrication required. The instrument is lifetime lubricated at the factory." I suppose HP didn't think the lifetime of the 803A would be 50 to 60 years, so I decided to remove the front enclosure shell and lubricate the bearings.

I'm glad I did, as it let me see the mechanical care used by HP in the instrument.

The image below shows the 803A's main casting. It's aluminum, around 3/8th inch thick. I've reinstalled the phase adjustment knob so I can turn the control after oiling the bearings.

The main controls, I believe, are transferred into the real bridge part of the instrument located in the rear half of the box and casting via shafts connected to the gears.
 

One reason the building stood out is that it has a small antenna range on the roof, which is something you don't see every day.

Like most bridges, the controls are manipulated  to null the output signal. Since the 417A is a super-regenerative receiver, the signal generator should be AM modulated with a suitable tone, usually the standard 400 or 1000 Hz frequencies built into the 608C. Using headphones, the controls are adjusted for minimum received signal.

It's usually easier to find the balance point by reducing the signal generator output and quickly turning the impedance and phase controls through their range for a broad null. Then increase the signal level and adjust for a more accurate null condition.

The 803A's dial reads directly in ohms impedance. The phase angle is calibrated in degrees positive and negative, but is directly reading only at 100 MHz. (Or, megacycles as  the instruments say.) At other frequencies, the phase angle must be corrected by multiplying the dial reading by f/100 where f is the frequency.

The manual supplied with each 803A has several pages of unique correction curves, particular to that instrument, as identified by serial number. If the bridge were recalibrated by HP, a new set of correction curves would be generated.

As might be suspected, I don't have the correction curves for my particular bridge. This means the overall accuracy is closer to ±5% instead of the 2% range achievable with the magnitude and phase corrections applied.
 

To assess the 803A"s accuracy, I measured a Mini-Circuits KARN50-18 termination. The KARN50-18 is an inexpensive, but reasonable quality 50 ohm load in an N connector, with a return loss of 35 dB or more from DC to 1 GHz. Mini-Circuits "typical" data says the KARN50-18 will have a return loss of at least 44 dB up to 1 GHz. With a 44 dB return loss, a pure resistive load will be between 49.4 and 50.6 ohms.

The figure below shows the results over the range 55 - 480 MHz. (I've corrected the phase values for frequency.) It shows a periodic effect, as seen in the calibration data. The phase is reasonably close to the expected near 0 value, being 2.5 degrees in the worse case.

The impedance magnitude, however, shows a clear bias downward, with an average error of -6.4%. This is not too far from the ±5% (plus an additional term for frequency) error quoted by HP for the 803A without error correction.

As another test of the 803A, I measured the impedance of a short length of RG-223 coaxial cable equipped with male BNC connectors at both ends. As tested, the assembly includes an N-BNC adapter installed at the 803A's UNKNOWN port, and a BNC F-F adapter at the far end. The overall cable length, including the adapters is 32 inches. RG-223 has a characteristic impedance of 50 ohms.

I measured the cables impedance with the BNC F-F adapter open and with a BNC shorting cap. The cable characteristic impedance magnitude Z₀ is thus Z₀ = Sqrt(Z_SC*Z_OC) where Z_SC and Z_OC are the short circuit and open circuit magnitudes.

The results below show reasonably consistency, but with the same bias low as seen with the KARN50-18 data.
 

The result is Z_O = 50.1 ohms at 55 MHz and Z_O = 47.5 ohms at 100 MHz, representing considerable improvement over the uncorrected values.

If the cable's characteristic impedance has no reactive component,  then the open and short circuit impedance angles should be symmetrical about 0, i.e., the two angles should sum to 0 degrees. At 55 MHz, the two angles sum to 1.76 and to 1.70 at 100 MHz. This non-zero value reflects both a small instrument phase error and the fact that a real cable has a non-zero imaginary impedance component.