Tektronix 7L14 Spectrum Analyzer

Robin Whittle rw@firstpr.com.au  Page established 2013-01-07
Last update 2013-03-14

../ To the Joy of Test Equipment page.
../../ To the main First Principles website.

This is a temporary form of the page which I intend to develop here - with some images of the 1st mixer of the 7L14:  http://www.barrytech.com/tektronix/tek7000/tek7l14.html

2013-01-07 Preliminary work

This is to support discussions on the Tekscopes mailing list starting 2013-01-06 with message 86565:


Fixing these diodes is not exactly a joyous project but I will be happy when it is done!

If I am successful with this I will document what I did properly.

The location of the 1st mixer is shown with 'x'.  This is looking at the inside of the left section once it has been opened up.  The cylindrical device is the YIG Local Oscillator.

Here it is labeled "A2".  (Fig 3 Chassis Left Side exploded diagram on page 9-19 of the Artek PDF manual: http://artekmanuals.com .)

The box is about 1 x 1 x 0.5 inches.

Note the "I" is used as the RF input.  On the Tekscopes list, Ed Breya explained (2013-01-06):
For a low-band Spectrum Analyzer, the input signal will necessarily be applied to the IF port, which is DC-coupled. The RF and LO ports are transformer-coupled, and typically operate at microwave frequencies.
The normal range of the 7L14 is from 10kHz to 1.8GHz.  The 1st Local Oscillator runs at 2.095 to 3.895GHz.  When running at 2.095GHz, frequencies around 0Hz are modulated by the switching of the diodes and the associated transformers to become 2.095GHz signals which are tuned in by the IF (Intermediate Frequency) system which is centred on this frequency.  As the 1st LO is stepped up in frequency, the lower sideband of the input signal, where 0 Hz is stepped up to the 1st LO frequency, is the one of interest.  For example if the 1st LO is running at 2.195GHz, then input signals in the 0.1GHz range, in the lower sideband, are now at the 2.095GHz frequency which is tuned in by the 1st IF system.

I didn't find a good way to dismantle this thing - the whole case is soldered together and melting that solder also melts connections inside and the anchoring of the hard coaxial cables.  I used a hot air gun.

The thin teflon fiberglass PCB which had the ring quad diode mounted on top of it, with two solder joints to the bottom left.  The middle connection is used as the RF input and the top connection is used as the IF output.

The notional diagram of the 1st mixer is the A4 section of the block diagram below.  The LO input (the 1st Local Oscillator of the 7L14) is used to drive the ring quad with either positive to the top and negative to the bottom, or the reverse.  In that diagram, with positive to the top and negative to the bottom, the left pair of diodes turn on and form a reasonably low-impedance ground point.  In the other half of the cycle, with the LO signal reversed, these are in a high impedance off state and the right pair of diodes are turned on, forming a reasonably low-impedance ground point.

This alternation of ground point steers the input signal to flow either leftwards in the top transformer of A4 or rightwards.  The output winding of that transformer is the IF output.  So the IF signal is the input signal modulated by the +/-/+/- switching of the LO signal. 

The actual output transformer arrangement is trickier than the A4 diagram.  The input signal travels along the stripline conductor from the middle terminal on the right.  See the reverse of the PCB below where this area is a ground plane.  The little square is evidently a capacitor, I guess to perform a low-pass filter function.  Then the input signal will travel downwards down the left of the right leg of the "transformer" on the lower left above (lower right below).  The signal goes through either the left or the right leg of the ring quad, through one pair of diodes or the other (actually only through one diode of the pair if the input signal is positive and the other diode if it is negative) to ground via the diode drive arrangement described below, to the local ground point which is the via just to the lower right of the transformer in the image above.

These two legs of the transformer have matching conductors on the rear side, as seen below.  The left leg (top side, above photo) has on its rear side (right below) a short conductor to ground.  I am not sure why it is shorter.  The right leg (above, for the topside) has on its rear side a full length track to ground, which goes towards the ground via which is used for the ring quad drive arrangement.  The upper end of this rear side right leg (top) connects to a curved ground wire of what is effectively a balanced transmission line or transformer (depending on how I think about it).  That rear side track terminates at ground on the left of the photo below.

On the topside this balanced transmission line or transformer is driven by a via from the shorter leg (rear side right below).  I think the inductive coupling of these top and rear tracks causes the balanced signal being driven into it by the rear-side secondary legs of the transformer to be transformed into a single-ended IF output signal, which appears at the top right terminal of the photo above.

Below the IF output (terminal "L") is the rear one on the right.  I think what is going on here is a high pass filter.  However, at this point (2013-03-03) I have reassembled everything but have still not got the spectrum analyzer going.

This is a green-red enamel wire pair.  Both green and red are grounded at the near end, where they are soldered to the SMA input terminal.  One of these at the far end (for the IF output coax connection) goes to ground and the other goes to the signal conductor of the IF output.  I might be wrong about this, but I think these are the connections and I think it functions as a high-pass filter.  We are only interested in ~2.095GHz at this point.

The lug in the middle grounds the centre of the rear side of the PCB.

In the photo above there are two large solder blobs on the left of the PCB.  These would be on the right of the photo below, on the rear side of the PCB.  I can't figure out how these solder joints were made.  When I reassembled the mixer I used some small wire wrap wires to achieve similar connections.

Here you can see the old ring quad still soldered to its drive arrangement, which is decidedly odd.

"A4" is the labeling for the 1st mixer.  In this 7L14 there is no limiter or 1.8GHz filter.

This is a traditional, easy to understand, notional schematic of a double-balanced mixer.  However the actual device is constructed in a rather different fashion.

In the notional schematic, the LO (Local Oscillator) signal, between about 2.095 and 3.895GHz, is fed to the bottom right of the mixer.  This drives the top and the bottom of the ring quad with out-of phase versions of the LO signal.  If the diodes are numbered:

    D4    D1
    D3    D2

Then when the top is positive and the bottom negative, D4 and D3 conduct, leaving their junction (the left terminal of the IF transformer primary) grounded.  This does not produce any signal, on its own, in the IF transformer.  However it steers the current from the RF input ("1" on the left) through these diodes so a positive RF signal will exit the IF transformer primary from its left terminal.  When the LO signal is negative at the top and positive at the bottom, the reverse happens: D1 and D2 are turned on and the RF signal leaves the IF transformer primary from the right side, and therefore drives the IF transformer with opposite polarity.

So the RF signal is multiplied by +1, -1, +1, -1 etc. at the LO frequency, creating in the IF transformer the sum and difference of the LO frequency (and also its harmonics, or at least its odd harmonics) and all the frequencies in the RF input. 

I am choosing between these diodes from RF Elettronica di Rota Franco in Italy.  Page 3 of the A-2.pdf Diodes catalogue:  http://www.rfmicrowave.it/eng/catalogue.html lists:

I chose the 5082-2277, which has light blue epoxy you can see in some photos below.

There is also a bonded pair of enameled wires taking the Local Oscillator signal from the bottom coaxial cable to the ring quad, which is shown here detached from the front of the PCB.

A close-up of the ring quad diode package.  The ceramic disk is about 3.2mm diameter (~1/8").

Here is my diagram of how I think the ring quad is driven.

The LO (1st Local Oscillator in the 7L14) drives the red wire of the first enamel wire pair.  The green is grounded at this end.  I think this forms a balanced line which delivers the signal as a differential pair of signals at the other end of the wire pair.  The green drives the X terminal of the ring quad (I don't know which way the quad conducts - it doesn't matter) and the red drives the Y terminal. 

If this was the only circuitry, the green drive to X would be little or nothing and most of the voltage would be positive and negative on the Y terminal.

However, the fiendish little loop of wire to the right of the ring quad forms a transformer with connections to ground, which would tend to cause the X voltage to be equal and opposite to the Y voltage.  The LO only runs between 2.095 and 3.895GHz (in the normal 7L14 - in mine it goes another 0.7GHz higher still because it has no 1.8GHz LPF internally and it can handle input signals up to 2.5GHz) so these tiny pieces of wire presumably form quite good transformers at these frequencies.

That's it for the 1st mixer, as best I understand it.  I haven't yet got the machine working, so perhaps my understanding is wrong, or my repair work is bad, or something else is wrong in the rest of the spectrum analyzer.

2013-03-03  I installed the new diode ring and re-assembled the 1st mixer

More explanation later when I get the machine going.  At present I am making up three Tektronix 7000 series flexible extender cables from kits designed and made by John Griessen of the TekScopes list. 

When I finish these I will be able to run the 7L14 from them, on top of an R7306 rack mount scope with its top cover removed.  Then, I should be able to debug it in various ways, including, if necessary, by removing the 1st mixer, desoldering its cover (I didn't completely solder it, so I should be able to get it off without the hot air gun) and running the 1st mixer in the open air.  Then I should be able to debug signal levels of the LO at both X and Y terminals of the ring quad.  The 1st LO is currently able to run down to 2.095GHz (it shouldn't, but it does, and I can measure this via its front panel output port) so if it is driving the ring quad then I should be able to  put a small DC current into the RF input and get a small but consistent 2.095GHz signal out of the IF port.  This should be tuned, amplified, frequency shifted etc. etc, and deflect the beam of the scope vertically.

By hook or by crook I should be able to get this thing working!

Here are some photos.

The photos above and below were taken through a Bausch and Lomb stereo zoom microscope.

2013-03-14 - still working on the machine . . .

[tek7000extenders] More information to follow, assuming I get it going properly.  For now, here is a photo of John Griessen's magnificent Tektronix 7000 series flexible extender cables.  (See the Tekscopes mailing list link above to contact John.)

The cables are long enough to plug into the scope without taking off a cover.  I took the right cover off a 7306 so I could work on the 7L14 with a greater length of flexible cable, and so the 7L14 could be to the side of the scope, rather than in front of it.


I got it working . . .  The rebuilt 1st mixer does work.  There may have been some other problems in other modules which prevented it working at first after I repaired the 1st mixer.

Beware of broken solder joints where the hard-tube coaxial cable joins the connectors.  This can happen when wrangling things in and out, including by twisting the coax.  Beware that some of these connectors, with or without such a break, have their pins somewhat short.  There's no obvious fix for this in itself, but the workaround is to file away the cylindrical part of the socket that connector screws into, so the male pin goes further into the socket.

I found it helpful to put 105MHz into the black coaxial cable which carries the 2nd IF from the left side to the right side.  This enabled me to ensure that the right side was working properly.  I didn't read the calibration instructions but figured out roughly how to set the "LOG CAL" (really a gain control on vertical deflection) and the AMPL CAL (gain for pre-resolution amplifier), along with the offset pots in the Function IF to get reasonably good alignment of Lin, 2dB/div and 10db/div.  There's no trimmer for the scale of the 10db/div, so I set the LOG CAL to get that matching 10dB/div.  Then, I used the offset for 10dB/div to get its -30dB level at vertical 8 (top of the screen) at the same level at which it was there in Lin mode as well.  There's a trimmer for 2dB/div gain but I found it best to leave it be.

These things are not for the faint hearted.  My adventures started with a bunch of smoking and burnt (dead short) tantalums. (This is a 1982 model . . .)  I won't be able to document all I did.

I was working with a beautifully photocopied manual. (Tucker Electronics is offering a genuine manual on eBay for USD$150 discounted to $97.50:  http://www.ebay.com/itm/300874409506 ).  Without that manual, which is now heavily annotated with pencil, red pen and Post-It notes, I would not have figured this out.  I pulled apart various other modules and took photos.  Don't try taking off the little pipe entry into A8.  The tip of the wire has a gold flexible bond to the LPF there.  I repaired mine with conductive epoxy.

I haven't attempted to get the digital display section going yet, and I may not bother.  It is not yet frequency calibrated but I figure this will be relatively easy.   I will post some more photos at some time in the future.

(All photos taken with a Sony F-707 camera, which is from 2001: 
http://www.dpreview.com/products/sony/compacts/sony_dscf707 )