Skip to content

Heathkit SB-620 Scanalyzer

May 31, 2017

SB-620 FrontDespite my interest in vintage electronics now taking a back seat to other things, I have been interested in obtaining one of these for a while and recently acquired one, a dead one. These devices are usually known as Panoramic Adaptors and were used by the Signal Corps. There was another well known commercial one suitable for radio hams but I cannot recall who made it and a quick Google search hasn’t helped.

The purpose of the device is to allow a radio operator to see what’s going on close by the frequency he is working or considering working, out to +/- 25kHz or so. Functionally, it is a narrow range spectrum analyzer. The fixed ranges are 10kHz and 50kHz, a variable range allows the unit to scan out to +/- 250kHz or slightly more, depending on the IF frequency of the receiver it is looking at. There were different coils available for the local oscillator and a builder/ user would have to set the LO up for the IF frequency of interest. In my case, I have a Heathkit SB-102 that has an IF of 3395kHz and fortunately the unit I have was set up for 3395kHz though apparently with the wrong oscillator coil, more on that below.

It has turned out to be one of the more interesting pieces of kit that I have encountered. The operating principal is the use of a frequency changer (mixer) and a sweeping local oscillator. The oscillator is swept by a voltage that is derived from the X axis sweep voltage and applied to a varactor diode. As the oscillator sweeps, when a signal is present that causes a difference frequency of 350kHz that signal is passed by the IF strip in the analyzer, is then rectified and presented to the Y axis. The signal being synchronous with the X axis results in a stationary peak or a “pip”. There is a crystal filter before the IF strip that sharpens the resolution to 1kHz. The sweep is generated using a neon relaxation circuit.

The first job was to bring the thing back to life. There was a poorly patched in replacement CRT negative supply capacitor that did not bode well and sure enough there was no HV-. I soon established that the corresponding transformer winding was open. This is actually helpful in an odd kind of way because such failures are usually (in my experience) due to insulation failure and that means the transformer is rendered completely useless. I took the transformer apart far enough to discover the remains of a burnt lead out from one end of the winding. I surmise the winding burnt when the aforementioned capacitor failed. So the transformer could still supply the B+ and heaters, all I had to do was to sort out a work-around for the HV-. I wound up (pun intended) making a voltage quadrupler that would be powered by the B+ winding. Even with current draw multiplication due to the quadrupler, since the HV- chain only draws around 200μA the additional load on the B+ winding is negligible. However there was still an issue. Because the B+ is also rectified by a full wave doubler, the B+ winding is elevated some +250V and if I connected directly, that elevation would subtract from the negative HV. So I capacitor coupled the system to the B+ winding. Further it is necessary to connect the correct side of the chain – marked “A” below – to the top end of the B+ winding otherwise it will not work. Here’s a picture of each side of the quadrupler with the diagram:Heathkit SB-620 HV Repair V Quadrupler

I found it necessary to increase the value of the caps for each stage to maintain the successive voltage gain, you can see from the measured voltages noted on the diagram that the chain steps fairly uniformly to the desired voltage. It worked out perfectly! It also fitted perfectly:SB-620 Bottom

I went through the calibration protocol and encountered a problem that took a while for me to comprehend (my knowledge of RF is sketchy at best). On setting the local oscillator (LO), I found that it seemed to tune well however, the instructions go on to test whether the LO is set at the correct frequency, i.e. 350kHz under the incoming IF frequency or 350kHz over. The method is to raise the incoming frequency by precisely 700kHz and an image peak should appear to the centre of the screen. Well, it didn’t. Said another way, the LO should be 350 above the incoming IF such that if the incoming signal is raised 700, it will first pass through the LO frequency and finally be 350 above so that an image signal will pass the analyzer IF strip. I later revisited Stephen Lafferty’s notes on constructing different LO frequency arrangements (the original parts are as rare as rocking horse poo) and he noted that for an incoming IF of 3395, the LO should be at 3745 whereas mine was at 3045. I queried this claiming ignorance of matters RF and Stephen confirmed that his figure is correct. Stephen’s article is here. We went back and forth a little, both being puzzled. The problem is that the LO coil (L3) is not marked with a part number. We concluded, I think, that the coil in my unit is not correct for an incoming IF of 3395kHz. Stephen had used a trimmer for the LO tank capacitor, and I followed his lead. The coil in my unit was already adjusted close to minimum inductance and so the fixed 56pF cap had to be reduced in value to raise the LO frequency. I replaced the cap with a 7 to 47pF trimmer and was easily able to obtain the correct LO frequency and the 700kHz image per the instructions. HOWEVER, the law of unintended consequences intervened: Before making this change I had taken a picture showing combined 1 and 4 kHz modulation with the peaks occuring in the correct locations thus:SD-620 1 & 4kHz Modulation Scan

(The small pips are harmonics. Note, the pip centre control is excessively touchy on that range hence the poor positioning in the picture above. The flash is due the the scan being very slow compared with the shutter speed.) After the modification, I was unable to reproduce this result with the modulation peaks occurring at approximately 1/2 scale. After much thought (well what passes for thought in my slow old brain), I realised that the reduction in fixed capacitance in the LO circuit meant that the swing of the varactor now had increased influence, approximately double, and that the bandwidths were now double the intended settings. Rather than screwing around with the varactor swing voltage settings, I chose to  increase the capacitance and settle for the LO being 350kHz below the IF frequency at 3045kHz. It just works better and is more satisfying to play with this way. Here’s what it looks like squashed up by the doubled bandwidth:SB620 1 and 4kHz modulation, wrong scalingI then decided to address the excessive pip centering sensitivity; The pip centre pot is 5k. I established the resistive setting with the pip centered then replaced the pot with a 2k part with appropriate resistors each side. This reduced the sensitivity usefully while still permitting the pip to be moved to the side of the screen when using the 50kHz sweep range. Again, it makes the device more fun to play with.

The next problem was that the sweep circuit then failed on the 10kHz range. This requires the slowest sweep rate and the charging resistance is more than 100Meg, the currents being in the few micro-amp range. I checked the values of the various resistances and they were surprisingly close. I also replaced the capacitor (the green cap in the picture above) but it still would not sweep on the 10kHz range. So, once again, a puzzler. I decided to see if perhaps grid current draw from the X amp was stalling the relaxation circuit. Changing tubes didn’t work, I tried several. However, I found that both the plate resistors were almost double the specified values despite being marked correctly. They are too small in size for the job and in my experience, carbon composition resistors develop cracks when run too hot over time. I replaced them with correct values and sure enough relaxation action resumed (if that isn’t a contradiction in terms). This might seem surprising however, it’s a Kirchoff thing, the currents entering a node must sum to zero. In this case the node is the triode and if the plate voltage is too low to cause sufficient conduction from the cathode, the current to balance the node must come from the grid. (I’ve encountered this before and at that time it stumped me for a while. I was developing a hybrid cascode for a phono stage with a fet in the bottom and a triode in the top. The triode was pulling the grid voltage reference down and I finally realised that I needed to increase the plate voltage.)

The rest of the calibration is pretty mundane though I encountered another issue when the instructions direct one to amplitude modulate the incoming IF at 50kHz to set the scan width for the 50k setting. Neither of my signal generators (GR 1001A and HP 8610A would accept AM at that high of a frequency. It’s not necessary, I found that if I modulated at 10kHz, the harmonics took care of the rest of the band. It’s not a laboratory quality instrument, so the deal is to mark the scan width control for the setting that gives correct placement of the modulation frequencies on the screen and (presumably) check it from time to time. Here’s what 5kHz modulation scanned on the 50kHz range looks like, showing that the harmonics can be used for sweep width setting certainly out to 20kHz.SB-620 10kHz and harmonics

Here is what 1kHz FM looks like on the 50kHz range showing the IF passband: I didn’t understand what I was seeing and again, Stephen came to my aid: “To explain it in detail, I would need to know a little more, but generally it seems typical to me. The key question is whether you mean that the modulating frequency is 1kHz or the FM deviation is 1kHz. I will assume it’s the modulating frequency. From that, I would estimate that the peak FM deviation is 7kHz and the span of the display is 40-50kHz. (I estimated the peak deviation by counting the 1kHz peaks from the center to the edge of the deviation.) Then the modulation index is m = Fpk/Fm = 7kHz/1kHz = 7.”

SB-620 1kHz FM

I managed to get it to see the IF from my SB-102. The sensitivity is just adequate and Stephen built a 17dB gain stage into his. The noise floor does show well on mine with good peak amplitudes on quite weak signals (my antenna basically isn’t). Still, here’s something from the 40M band, the primary signal is from France:SB-620 Scanning on 40M




From → All Posts

  1. Bruce Baur permalink


    Glad to see you can’t put the soldering iron down totally. Why didn’t you become and electrical engineer? Well, you are better than some EE’s. I like your HV solution.

    An interesting instrument. At the museum we have something similar, an FM deviation meter made by Radio Specialty, but it looks at the RF off the air rather than the IF. It has been on the web site for sale and no takers, so it will now go to recycle.


    • Thanks Bruce. Somehow, I got the notion that I was too thick to cope with the math to become an EE. As it turned out, the math required of a ME is probably no less testing, we have our version of the Maxwell equations in the Navier Stokes equations. Ugh. (Or not if one is good enough at math to appreciate the beauty.)

  2. Quite an impressive repair. I would probably have failed at multiple issues on the way..

    • Thanks Lars. Actually, I had more trouble getting an adequate level of 3395kHz IF from my Heathkit SB-102!
      Peace, Richard.

  3. I am very interested to get one of them bec. here is a Heathkit Line in restoration, missing this unit. (313, 401, 600, 610)

    • Hi Martin.
      Do you know what the IF frequency for your receiver is?
      I will keep you in mind. Would you be interested in the SB-102? Note, I’m just wondering, I’m not quite ready to let them all go.
      Is there a link to your Heathkit line?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: