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

E-mail: Jack.Smith@cliftonlaboratories.com

Mike, W4XN, loaned me his Watkins-Johnson WJ-8617B HF/VHF/UHF receiver last week and asked me to use it and give him my opinion of it. I've decided to memorialize my impressions of  the WJ-8617B  receiver at this site as well.

If you are interested in reading more about Watkins Johnson, check Terry O'Laughlin web site http://watkins-johnson.terryo.org/ for a copy of the 8617B brochure, configuration guide and WJ price lists.

If you're not familiar with WJ products, they have been widely used by the US military and by three letter agencies. This particular 8617B tunes from 2 MHz to 1100 MHz and receives AM, FM, CW, SSB and pulse modulation, with a wide selection of IF bandwidths. It includes a panadapter module.

Oh, yes, I should mention that when new, this receiver cost about as much a new Cadillac automobile. WJ's 1992 list price (without options) shows the receiver starting at $24K, and Mike's receiver has at least $12K in "options."

In reading my observations, it's helpful to keep in mind what the 8617B and its relatives, were designed to accomplish. The 8617B is commonly called a "surveillance receiver" whose normal function is to detect and allow the interception of radio signals. This places a high premium on detecting signals, even ones that are only momentarily active, allowing the operator to quickly tune to those frequency and to demodulate the intelligence they may be carrying. Hence, features such as a wide range panadapter display are considered essential. Likewise, the ability to quickly tune to a new frequency and to demodulate both wide band and narrowband signals are essential. Multi-channel signals, or digital signals, can be demodulated with external equipment, connected at the receiver's IF output, or at baseband output.

When I worked in the FCC's Detroit Field Office in the late 1960's and early 1970's, we used surveillance receivers for many tasks, including tracking down interfering spurious signals from commercial broadcast and two-way transmitters, and, in conjunction with direction finding equipment, to located unlicensed FM broadcasters. At that time, the state-of-the-art available to us was analog-tuned receivers neither particularly stable nor accurate. Compared with that gear, the 8617B is an order of magnitude improved.

I'll leave to your imagination other uses of surveillance receivers, but one might envision an airplane full of WJ receivers, operators and tape recorders vacuuming up all sorts of interesting signals when flying in the right place. Or when attached to an antenna on the roof of an embassy in certain countries. Or ...

For a mundane example of how one might use the 8617B, see my page Radio Intelligence Example
 

WJ's product sheet describes the 8617B's features as:

The receiver under review includes the following options:

• 500-1100 MHz frequency extension
• 2-20 MHz frequency extension
• Installed IF bandwidths: 6.2 KHz, 20 KHz, 100 KHz, 1 MHz
• Extended memory (96 frequency/mode memories)
• BITE (built-in test equipment does a self-check on functionality)
• Signal monitor or panadapter, displaying ±2 MHz from the current tuned frequency
• Digitally-refreshed display, i.e.,  the panadapter uses digital memory to hold and display the signals
• Cursor  tuning (more on this later)
• Audio scan output (an audio frequency proportional to the tuned frequency during scan mode. I suspect this was used as a frequency cue when tape recording audio signals to analog  tape.)
• SSB option
• Variable BFO (synthesized BFO, tunable in 10 Hz steps)
• IEEE 488 digital remote control interface
• Wideband output (probably to connect to frequency diversity demodulators for multi-channel signals.)
• Noise riding threshold, a squelch system that mutes/un-mutes the receiver on actual RF signal to noise ratio

First, the panadapter. The cathode ray tube provides a frequency (horizontal axis) versus signal strength (vertical axis) display. The vertical range is selectable via a DIP switch inside the receiver for either 40 or 60 dB vertical range. The front panel controls also provide vertical  gain adjustment, horizontal span or sweep width, sweep speed, centering and the usual intensity and focus controls. In addition, the vertical axis may be switched between log and linear display and a "center frequency" pip or marker may be injected when desired.

Just below the panadapter is a headphone jack (no speaker level audio output from this receiver, although the headphone output can drive a small speaker to reasonable output levels), audio gain control and RF gain control, along with the power switch.

The 8617B has a switch for AGC (automatic gain control) on/off. Unlike almost all amateur radio receivers, there is no selectable attack or release time. Rather, the AGC is either on or off. The 8617B's service manual says that the AGC speed is automatically selected between fast and slow, but the selection criteria is not clear. It seems to be related to video versus audio output. Perhaps further study of the service manual will clarifiy this point. There is, however, no manual AGC attack or release adjustment.

When AGC is selected, the receiver displays the signal strength in dBm in  the LED display area, as illustrated below. I checked the signal strength display accuracy against a known good signal generator and found the display to be significantly in error, ranging from 15 dB to 9 dB over the range -120 to -40 dBm.

In addition to AGC, the receiver incorporates AFC or automatic frequency control. AFC, when activated, tunes the receiver to the strongest signal within the AFC  range. AFC in the 8617B is interesting. In a commercial FM broadcast receiver, for example, before the days of synthesized tuners, AFC was a common feature, and it would correct for perhaps 20 or 30 or even 50 KHz of drift or tuning error. The standard channel spacing in commercial FM broadcast is 200 KHz, so the designers limited the AFC range to prevent a strong adjacent channel station from capturing the AFC. (Adjacent and second adjacent channels were not assigned within the same service area, of course, but AFC capture could come up in an automobile FM receiver as you kept listening to a particular station as you drove out of its protected service area.

In keeping with the purpose of a surveillance receiver, the 8617B's AFC range is ±10x  the bandwidth. If a 100 KHz bandwidth is selected, the AFC range is ±1 MHz. At first, this seems like a grossly excessive AFC range. In the FM broadcast paradigm, for example, it would be unworkable as  stronger signals hundreds of KHz away would be continually capturing the AFC. Were the WJ engineers wrong? Not at all. Place yourself in that hypothetical aircraft, flying outside the 12 mile limit of some un-named country and you see a signal pop up on the panadapter. You quickly tune towards it, but it's gone before you can tune it in. With a wide range AFC, all you have to do is get close and the AFC will reach out and capture the signal, tuning the receiver to its frequency in a matter of milliseconds. (WJ added another nifty feature aiding in this type of interception as well - cursor tuning, which I'll discuss later.) If there are a  dozen signals of interest, closely spaced, then AFC isn't going to be helpful. In other cases it might, literally, be a life saver.
 

Bandwidth is selected via five push-button switches. As with all WJ switches, when selected an LED illuminates indicating the switch is selected.

The switches are labeled with two bandwidths per switch. In some 8617B receivers an option was installed for 10 bandwidth positions instead of the normal five. This receiver has four filter and demodulators installed, 6.2 KHz, 20 KHz, 100 KHz and 1000 KHz. These filters have shape factors that are nothing to write home about if compared with amateur radio receivers. The shape factor (3:60 dB, not the amateur radio 6:60 dB standard) of these four filters is 3:1. Again, there's a reason for this. These filters are optimized for pulse, digital and multi-channel signals and hence use filters designs with relatively  gentle roll off in the stop band. WJ didn't intend for  the 8617B to be used in a DX contest or during Sweepstakes to pull out weak signals from strong adjacent signals. Rather, it was designed to permit collecting clean signal samples from a variety of modulation techniques, perhaps to be tape recorded for later analysis using more specialized equipment.

The particular bandwidths  this receiver is equipped with have a Goldilocks problem. 6.2 KHz is useful for AM, but it's too wide for SSB in many instances. 20 KHz is a  good selection for two-way voice FM. 100 KHz, however, is optimum for television audio and that's about it. It's way too narrow for commercial FM broadcast. 1 MHz is too narrow for television video and it's about four or five times the optimum bandwidth for FM broadcast. When dealing with surplus equipment, of course, you take 'em as you get 'em, and I'm sure the filter bandwidth installed was carefully selected for a much different use than I've put the receiver to.

In another departure from amateur practice, each filter has a separate associated FM detectors, optimized for the filter bandwidth. SSB and AM use a common detector.

The detector or mode is selected by five push-button switches, with AM, FM, CW, PLS (pulse) and SSB detectors being installed in this particular receiver. USB and LSB are selected by cycling the SSB mode button.

Now is as good a place as any to discuss WJ's human interface philosophy. WJ is not big on multi-function switches, or switches that select different functions by tapping, pressing, holding longer and the like, as can be found in some amateur gear and much consumer electronics. For one thing, there's no need for it. With a target market that's large price-insensitive, saving the price of a few dozen switches isn't critical. And, as seen in the interior photographs of the 8617B, the front panel can't be made smaller so there's no reason to save room.

Still, some switches have multiple functions, For example, the down memory selection switch's second function, NRT selection, is engaged by first pressing the "Function ↑" switch. When the function switch is engaged, the function switch LED illuminates and the standard switches LEDs become dark. The shifted function is indicated by reverse lettering, black type on a silver background, immediately above any dual-purpose switch.

Finally, place yourself in that airplane flying 12 miles off the coast of a hostile country. You're strapped in, bouncing around from turbulence, perhaps from an armed interceptor flying close enough to cause wake turbulence. Maybe you are wearing gloves. It's a tense encounter and your heart is pumping. It's a lot easier to hit the large push button to select the desired function than to try and tap or tap and hold or whatever, is required in a multi-function switch.
 

Listening to noise in the absence of a signal is fatiguing. And, a set of relay contacts or a transistor switch that operates when a signal is present can turn on a tape recorder or activate an indicator. The 8617B has two devices to accomplish this task.

The first is a more-or-less standard "COR" or carrier operated relay.  Its threshold is set using the up and down arrow keys from 0 (not active) to 40 dB, in 1 dB steps. The COR also controls the front panel audio and mutes the receiver in the absence of a signal above the threshold, i.e., the classic receiver squelch function.

"NRT" or "noise riding threshold" is a slightly different anti-noise device installed as an option on this receiver. NRT works by comparing the received signal with the RF noise, un-muting the receiver and activating the COR relay when the set minimum signal-to-noise level is achieved. Like the standard COR, the NRT has a range of disabled to 40 dB.

The selected COR or NRT level is displayed in the LED panel. NRT selection requires first pressing the shift or "function ↑" switch.

In my brief exposure to the 8617B, I haven't found much difference between the NRT and the conventional COR. The NRT should provide superior performance where the noise level changes, such as thunderstorms or intermittent line noise. Updated ... I added an external broadband preamp ahead of the 8617B and found the standard COR function to be difficult to use—as the noise level changed with frequency the COR level required adjustment every MHz or two. The NRT squelch, however, was rock solid and showed no signs of variation with changes in the noise floor.

The 8617B has two antenna input ports and the user may manually select antenna 1 or antenna 2 via a push button. When selected, antenna 2 is active. In addition, antenna 1 and 2 selection may be automatically made, based upon the tuning frequency. This is accomplished by setting the receiver to the desired crossover point and invoking a sequence to establish this as the switching point. So far, I've only used manual antenna switching.
 

The tuning steps are selected with two options. The front panel switches select 1 MHz, 10 KHz or 100 Hz tuning steps.

Alternatively, by removing the top cover and flipping a DIP switch,  "variable tuning" may be selected. Variable  tuning in this  regard means that tuning steps are 100 Hz, 1 KHz, 10 KHz, 100 KHz, 1 MHz, 10 MHz and 100 MHz. The step selected is indicated by the displayed frequency digit blinking. Pressing the 10 KHz and 100 Hz switches move the selected tuning digit (step size) left and right. For example, if set for 10 MHz  tuning steps, the leading 2would slowly blink in the image above.

The receiver was set with variable tuning disabled and I've now activated it. It's better than the three 100x steps available with the front panel, but in my opinion, WJ didn't quite get this right. If you are working in, for example, the 450-470 MHz band, you would like the receiver  to  tune in steps  corresponding to how the band is channelized, 25 KHz in this case, or 12.5 KHz if one considers the "split" channels. Almost all VHF communications systems are channelized and channels don't always fall in 1 or 10 or 100 KHz steps. As it is now, you would tune in, say, 10 KHz steps to get close and then in 1 KHz steps to the channel frequency. (500 Hz error isn't going to be an issue in most cases). The "get close and use the AFC" isn't always useful because (a) the frequency you wish may not have a signal on it at the moment and (b) in a busy frequency range, the capture range is just too great. A variable AFC capture range would be useful in this situation.

While I'm at it, WJ also has another human systems problem in my view. I own a WJ HF-1000 HF receiver and it has the same problem—the tuning knob is located at the far right hand side of the receiver, within an inch or so of the front right rack handle. I'm left handed and it's  awkward to use the tuning knob, between reaching across the receiver and the limited clearance between the knob and the rack handle. A  right-handed person might find the tuning knob easier to use, although I suspect the rack handle clearance would bother them as well.

Frequency Memory is featured in the 8617B.

The receiver has 96 memory positions, with each memory position saving frequency, bandwidth, mode, COR level, AFC and AGC status, antenna selection as well as RF/IF gain setting. (96 Memories requires the expanded memory option.)

Saving operating parameters requires first selecting the memory ID, done by memory up/down buttons that cycle through memory positions 00..95. The current memory ID is displayed in the main LED display section.

After the desired memory address is selected, pressing the STO button saves the current parameters.

Recalling a saved memory position requires two switch operations after the desired memory address has been set. First, pressing the RCL (recall) switch loads the saved parameters into the display. Pressing the EXC (execute) switch switches the receiver to the saved parameters. I suspect requiring two separate switch activations is an operator error reduction measure, as it requires a deliberate action to actually change frequency. It also permits the operator to view the parameters saved in memory without actually making the frequency change. If the operator decides not to activate  the recalled memory parameters, pressing the MAN (manual) switch returns the receiver to the last used parameters.

Scanning in the 8617B is also provided, with three scan modes.

The first mode is "single band scanning" where the receiver starts at frequency F1 and stops at frequency F2, pausing whenever it encounters a signal above the COR or NRT threshold. F1 and F2 are entered in two consecutive memory locations, with F1, the lower frequency, being saved to an even numbered memory address, e.g., save F1, the start frequency, at memory location 10, and F2, the stop frequency, in memory location 11. The receiver is set to memory position 10, in this example. Pressing the SCAN switch starts scanning from F1 to F2, with the scan time adjustable through the DWELL control. The scan frequency steps are at 50% of the selected bandwidth, up to a maximum of 65536 steps. Since scan mode requires two memory positions, 48 F1/F2 combinations may be saved.

The second scan mode is called "step" scanning. Step scanning sequentially steps through the 96 memory positions, stopping when a signal is detected. (Fewer than all  96 positions can be stepped, of course. In addition, individual channels may be locked out from scanning.)

The third scan mode, "multiple band scanning" is initiated similar to the straight frequency scanning mode, except that the user starts the scan from an odd memory address. In this mode, scanning starts at the F1 represented by memory channel 00, stopping at the F2 entered in channel 01, then starts scanning again at F1 saved in memory channel 02, stopping at F2 in channel 03, etc. recycling to memory channel 00 when it reaches the odd address from which the multiple band scan was started.
 

• Frequency standard (time base) can be selected as internal or external via a switch with a BNC connector providing a sample output (internal mode) or serving as the reference input (external mode). The reference in/out is 1 MHz.
• BITE (built-in test equipment) jack. I can't find a reference to the output parameters of the BITE output (BNC). As a guess, it provides a pass/fail output based on the BITE results.
• X,Y and Z output (BNC) parallel the internal panadapter horizontal (X), vertical (Y) and retrace blanking (Z) axis so that an external display may be used with the receiver.
• Remote control in this receiver is an IEEE-488 or HPIB or GPIB as it is also known. This permits remote controlling the receiver from a computer using the interface originally developed by Hewlett Packard for test equipment. Some 8617B receivers have an RS232 serial remote control port instead of an IEEE-488 port.
• Scan output (BNC) is a variable frequency audio tone proportional to the scan frequency. This was likely used with a stereo or multi-channel audio tape recorder as a simple method of associating captured audio with a particular frequency, without requiring computer control and data capture.
• Antenna input ports are two Type N connectors. The operator may manually select between the two antenna ports, or selection may be automatic depending upon the frequency setting.
• COR is a current sink transistor to ground, active when a signal stronger than the COR threshold is received. When NRT mode is selected, this port is controlled by the NRT function. (BNC)
• Audio out is a fixed level (not changed by the front panel audio level control) 600 ohm audio output, 10 milliwatts typical maximum level. (A screwdriver potentiometer on the rear panel sets the maximum audio output level.)
• SW IF Out is the switched IF output, i.e., a 21.4 MHz IF sample with a bandwidth equal to the currently selected IF filter.
• FM MON provides a DC coupled FM output with the current selected bandwidth.
• SW Video is a DC coupled, wideband video signal (AM or FM, depending upon the mode selected) Of course, the term "video" does not necessarily mean a television signal. Rather, it is a wideband demodulated output that may be used for multiple signal types.
• WB IF Out is a fixed level (-30 dBm) 4 MHz bandwidth IF sample signal centered at 21.4 MHz. The level is automatically adjusted to -30 dBm by a separate AGC system. It may be used for external demodulators, or external panadapter or the like.
• AC Input. is through a standard three-blade IEC connector. Power is adjustable via a switch for 100, 120, 220 or 240 volts, 50/60 Hz.

Again, the difference between amateur radio equipment and WJ is apparent. Except for the standard IEEE-488 connector, each output connector has a single purpose and is a shielded BNC connector—there's not an RCA jack nor a multi-pin DIN connector to be seen.

In contrast, it's only a slight exaggeration to say that WJ never built two identical receivers. Between user option selections and changed and updated designs, there are major differences between different 8617B versions. The open chassis approach WJ uses is more amenable to customization.
 

There's a reason for emphasis on shielding. There's the obvious one, of course, to prevent one receiver stage from interfering with another, or from one receiver leaking signal into the next one in the rack.

There's a more important reason. I don't know for sure, but it's reasonable to assume the 8617B (or a derivative of it) meets the stringent TEMPEST signal leakage criteria. Two stories from the open literature demonstrate why one might be concerned with signal leakage from electronic equipment operated in sensitive locations. Both examples are from Peter Wright's book Spycatcher

First, in the 1950's the French Embassy in London used a mechanical teleprinter and an on-line encryption device so that the 5-level Baudot signals sent outside the embassy were protected by high level encryption. However, careful examination of the encrypted data pulses showed small ghost images or glitches from the source un-encrypted signals. It was thus possible to read the signals as if  they were completely unencrypted. (DeGaulle was President of France at the time and relations between the UK and France were frosty.)

Second, the UK's MI5 counter-intelligence operation uncovered the operating frequencies used by the Soviet Union in the UK. By using a sensitive receiver, they were able to receive the local oscillator leakage and then by adding or subtracting the associated IF frequency, they could determine the frequency to which the embassy receiver was tuned.

Similar techniques are said to be used in determining whether a household has an unlicensed TV set in the UK and with computer monitors it's possible to reconstruct an almost perfect pixel-by-pixel image of a monitor from some distance.

The 8617B's signal leakage specification is less than -100 dBm, which is quite effective at making these sorts of recoveries difficult to impossible.

I should add that the 2nd LO synthesizer actually generates the 100 Hz tuning steps in the range 4.4 - 5.4 MHz, and is translated to 530 MHz via a mixer.

The 2-20 MHz HF range extension works in a similar fashion, being up-converted to 551 MHz.

I've ignored the microcontroller and power supply, display, panadapter and the like to emphasize the 8617B's frequency conversion scheme.

A complete verification of the 8617B's performance specifications would take a week or two of steady lab work, so I've only looked at a few data points.

In general, I like the 8617B. As might be suspected from its intended purpose, it does a better job at VHF and UHF than listening to SSB or CW signals. In no particular order, my observations are:

• Although the "variable tuning step" option makes tuning easier  than with the 100 Hz, 10 KHz and 1 MHz steps, it falls short of optimum. User interaction would be considerably improved if specific tuning steps, e.g., 12.5 KHz for the 450-470 MHz band, were implemented.
• Frequency scanning does not stop on  the exact center frequency. This is a function of the scan step, e.g., if the bandwidth is 20 KHz, the scan is executed in 10 KHz steps, 50% of the bandwidth. A signal centered, for example, on a 5 KHz raster will be locked to 5 KHz high or 5 KHz low. Even if the frequency raster is an even 10 KHz (146.790 MHz, for example) I found the receiver stopped at 146.780 MHz. The tuning algorithm likely stops scanning when the COR threshold is exceeded and this can easily happen when the receiver is a few KHz off an exact center tune.
• Along the same lines, I miss a built-in center tuning indicator. It should be possible to build an outboard center tuning indicator using the DC-coupled FM output available on the rear panel, but for a receiver with so many bells and whistles, omitting a zero center indicator is odd.
• The digital panadapter is quite nicely implemented, particularly when scanning.
• I really like the cursor tuning mode.
• The first local oscillator synthesizer tunes in 1 MHz increments and there's an obvious (but brief) glitch on the panadapter as you tune across 1 MHz boundaries. The synthesizer is fast, on the order of a few milliseconds, but even a brief glitch is noticeable.
• SSB and CW modes are usable but suffer from two problems. First, the receiver's 6.2 KHz bandwidth is excessive for either mode. Installed bandwidths are, of course, a function of how the receiver was equipped by its original purchaser. The second note I made is that the AGC is not really optimized for SSB where a fast attack, slow release is considered optimum. The result is background noise captures the AGC during long speech pauses. Disabling the AGC and running manual RF gain is another option.
• The AFC range is much greater than one might wish for frequency ranges with multiple close spaced signals. As discussed earlier, this is an intentional design decision  to aid the 8617B's normal function.
• For FM broadcast demodulation, the IF bandwidth selection of 100 KHz is much too narrow and 1 MHz is much too wide. A 300 KHz bandwidth filter and detector would be a better choice, but this again reflects the decisions of the receiver's original purchaser.
• I don't like the far right hand side tuning knob's position and its proximity to the front right rack handle. A right handed operator might find the design acceptable. Tuning is smooth, although not equal to the best amateur radio receivers.
• I like the front panel control arrangement, with a switch for every function, more or less. Where a second function is implemented, the operation is clearly indicated by the switch LEDs and the front panel legend.
• Sensitivity is more than adequate for a surveillance receiver. With a 20 dB FM quieting sensitivity of around 1.5 microvolts, it isn't going to win any prizes as the most sensitive receiver on the block. However, an external preamplifier can always be installed to improve sensitivity. Depending on the RF environment, however, increased sensitivity may require additional filtering between the antenna and the preamplifier. The 8617B's designers likely intentionally traded off sensitivity for improved intermodulation and out-of-band signal  rejection.
• There's no built-in speaker audio amplifier, although the 600 ohm headphone output can produce usable volume in a quiet room when operated into a loudspeaker. Again, this is a design decision as these receivers normally are operated into a tape recorder or used by an operator with headphones.  The headphone audio level is more than adequate.
• The main display is relatively uncluttered and provides the required information.
• The 8617B's 3rd order intermodulation intercept, IP3, is +8 dBm (typical) for 20-500 MHz and 0 dBm (typical) for 500-1100 MHz. By current HF amateur radio standards, these are rather poor numbers.  Likewise, the phase noise specification of -105 dBc at 20 KHz offset is relatively poor. Some of this can  be attributed to the  receiver's age—first sold in the early 1980's it must have started design around 1980. The state-of-the-art has advanced considerably since then in  both IP3 and synthesizer phase noise. It's also not clear that in the normal surveillance environment the 8617B's performance figures are unacceptable. A lot depends on the density of signals in the local environment and their relative strengths.