Micro-Tel MSR-904A Microwave Receiver: a broken trace, a replaced pot, and an escape proof guarantee

First of all, mystery solved – the remote enable input for the IF attenuator. After quite painful tracing of wires and disassembly of the logic boards, a broken trace! Just a little crack, but big enough to block the electrons’ flow. A little bit of solder, and then, suddenly, the MSR-904A’s IF attenuator can be remotely controlled.

Today, a also the dual pot for the F2 adjustment arrived – fitted, also this, working again.

But most interesting, a little box, with a label not seen before:

msr-904a mains cable label

The part has a NSN number, original unit price was no less then 71.38 USD!

Enclosed – the most special mains power cable I have ever seen:

msr-904a mains cable
-it was well packaged in a multilayer heat-sealed bag.

And, a label, which will be kept – Federal Prison Industries: Escape Proof Guarantee
unicor federal prison industries escape proof guarantee

Micro-Tel MSR-904A Microwave Receiver: crosshair, imagesetting film printing, remote control input

Some more progress related to the MSR-904A:

(1) The crosshair. The old one is badly damaged.
msr-904a crosshair

Sure, I could just use a laser or inkjet printer, but with a laser, the resolution is not good-doesn’t look sharp enough. With inkjet, I doubt it would be permanent enough, and also there, the printout is never sharp enough. So I decided to go for the solution that also Micro-Tel used, so-called imagesetting film. Back home in Germany, not problem, but here – first I need to find a source. Turns out, not too difficult, about 20 USD for 8 pcs of crosshair (minimum order fee), a company located in New Jersey, not around the corner, but not far. 2400×2400 dpi, acetate/emulsion film.

With a film printing service identified, we need to get the digital data, of the crosshair. With the human eye being a pretty precise tool to determine even small differences, and aiming for perfection with the replica – first, determined the line width and distances with a measurement microscope. Such a little microscope is extremely handy, I use it all the time for inspecting circuit boards, etc.

msr-904a crosshair measurement microscope

msr-904a crosshair measurement
one of the big divisions: 0.275 mm (26 pt at 2400 dpi) wide. Line is about 0.125 (12 pt at 2400 dpi) wide.

Here – a reconstruction drawing:

The new prints should be in the mail tomorrow!

(2) The remote control input: a 37 pin connector! Fortunately, not all wires are connected (x in the draft).

msr-904a remote control input

Controlling the function, bands and IF bandwidth, and detector characteristics (log-lin), all fairly straightforward. All of these inputs seem to have pull-up resistors, so grounding them works fine to switch. This is quite hand for control via optocoupler – no external voltage required.

The IF attenuator – traced the lines to the control board, it is a BCD control input, 2×4 bit. Unfortunately, I can’t get it to switch… the ‘enable’ signal doesn’t seem to reach the control board – more effort will be required to trace this last line! – Solved: a broken trace on one of the logic boards!

Some of the pins, despite having wires attached to them, remain unidentified – or might be reserved for options not implemented. Except for the IF attenuator control enable signal, all functions needed have been identified anyway.

msr-904a remote control input pinout

(3) The remote frequency control input – analog voltage: quick check with a frequency counter connected to the LO sample, and a DC voltage supplied – it is a 0 to 10 V input.
Scaling of the input voltage can be adjusted on the A6B2 board, R56 is for the mid-range adjustement (offset – supply 5 V and adjust for mid-band frequency), R68 is the gain adjustment (set at 0 V, and adjust for lower band limit; check setting at 10 V – should be at the high end of the band – and it really is).

Micro-Tel MSR-904A Microwave Receiver: some progress

With the basics done (power supply, potentiometer), a few hours were spent to get everything tuned up.
And, quite amazingly, it is receiving:

msr-904a first sign of activity
– notice the dirt, and the sticker residues. Also the crosshair (which is printed on a piece of plastic foil), will be replaced.

Finally, the exterior. The front panel, easy enough, all brushed and cleaned with diluted isopropyl alcohol.

The top, bottom and side panels with the sticky green stuff – all the old junk (“paint”) has been removed, my soakin the panels in methylated sprits, sanding, solvent cleaning, sanding. Then, a layer of aluminum primer (self-etching automotive primer). Followed by a light sanding, and a layer of ‘Hunters Green” alkyd paint. After 10 hours of air drying, final curing at about 150-170 deg C, for a bit over an hour.

All in all, quite an effort. The result –
msr-904a panels - newly painted

Missing items – one fastener like this – no idea where to find, seems like a part from the aerospace industry.
micro-tel quick release fasteners
Should you have any of these around, even of somewhat different length, or if you know a source, please let me know!

msr-904a receiving at 8.1 ghz
Sweeping test around 8.1 GHz – with the refurbished panels installed.

Two more handles are still needed – either need to get spare handles from a parts units (which may be impossible to find), or ship the MSR-904A with 2 handles only, and provide the remaining two later, once I had a chance to fabricate them back home in Germany. At least, I have the exact dimensions measured, just a matter of some CNC milling.

msr-904a receiving AM modulated signal at 8.1 ghz
-this is a test using a 8.1 GHz AM modulated signal, with about 1 kHz modulation frequency. Carefully checked the IF chain (different chains are used, depending on filter setting) – the MSR-904A uses 250 MHz, 160 MHz (by mixing the 250 MHz IF with 410 MHz, from a low noise LO), and 21.4 MHz (for the 100 kHz BW setting).
All seems to be functional.

With the receiver now basically functioning – some weekness of the AFC circuit alignment, and the frequency control was noticed. Therefore, some more effort was spent on the frequency control and AFC circuits, and the tuning indicator circuits. Really tough without any instructions or schematic.
There are some nice indicators on the front panel, LED bar graph displays – one for signal strength, and one for tuning.

These displays, now, in working condition and properly adjusted, are great fun to use. They are extremely responsive – nothing to compare with the time lag and sluggish nature of a modern SDR.

After several hours – here, receiving at 6.1 GHz, with 1 MHz bandwidth, and the AFC keeping the frequency, counteracting artificial drift:
msr-904a receiving at 6.15 GHz with AFC active

Monitoring of the AFC control is by looking at the IF frequency, 250 MHz (on the EIP 545A counter), derived from the (non-phaselocked) MSR-904A LO frequency, minus the RF input frequency (from the Gigatronics 605 Microwave Synthesizer; the EIP 545A is locked to the 10 MHz signal from the 605).
This setup allows me to check for any drift of the MSR-904A IF chain (and AFC, if activated), to 1 Hz resolution.
Had it running now for several hours, no issue, signal stays perfectly tuned.

The only remaining item, internally, is the alignment of the cross-band assembly – still lacking one CD4051 multiplexer circuit – which is on its way. A quick check with a CD4051 taken from another board showed that there is no defect, the board just needs some alignment of the band-to-band transition points. The crossband assembly allows a full 0.5-18 GHz sweep, with automatic band selection.

Micro-Tel MSR-904A Microwave Receiver: the monitor output

The MSR-904A has a few outputs (and inputs), most of them, easy to identify, but two, are pretty much a mystery to me, with no schematic – their function is clear, but with all these wires TTL logic boards -difficult to guess the pinout.

msr-904a monitor output

The more easy thing first, the monitor output. Arguably, this was intended to be connected to storage displays, digitizer, chart recorders, or the like – to more permanently record the activity over the bands.

So, what do we have. Using a scope, and a multimeter, and activating more or less all the functions of the apparatus, that’s what I found out.

msr-904a monitor output - pinout

Sorry for the rough draft, but any questions, please ask – for the given purpose, good enough.

The only pin that doesn’t do much, is pin 9 – always stays on logic 5 V – maybe a +5 V supply line? Doesn’t seem to be an essential function, anyway.

Micro-Tel MSR-904A: RF tuner block diagram and component specs

Just to learn a bit about the way Micro-Tel was doing their engineering, let’s have a quick look at the way in which it functions, at least for the RF front end. After some study and tracing, here is the block diagram:

msr-904a rf tuner block diagram
Some parts don’t have model numbers – because these are hidden, and I don’t want to take the thing apart.

Luckily quite a few of the parts are still available, and datasheets are available for most. The YIGs are Avantek, have a customer part number, but I assume, essentially, slightly modified line item parts, with some specific specications. The only really uncommon part, is a “TREK MICROWAVE” 0.48-2.01 GHz 3-stage YTF. Didn’t know that such low frequency YIG filters existed, this has serial #00003.

YIG filter 4021-104

Even more, TREK doesn’t ring a bell for me, and the filter really looks very much like Systron Donner technology, with the characteristic color, and square-cube shape.

As it turns out, TREK acquired the YIG division of Systron, some time around 1984… so, this mystery solved.

The only part where no data seems available is the 2-18 GHz mixer, Avantek SX83 series, but can’t find a -1612 model anywhere.

Some datasheets:
5B120-2330_25-O_OP – 5915011438953 _ chebyschev 2330 MHz 25 MHz bw

k&l 5L120-300-0 low pass 300 mhz

8L120-2050-0 low pass – 5915012428744 _K&L filter

narda 4244-6 082-Couplers

a34 datasheet

qbh 101 amplifier datasheet

narda 4203-10 coupler

Micro-Tel MSR-904A: some basic repairs

Quick initial assessment, these are some of the items that will need attention:

(1) Exterior. Need to fabricate instrument feet, re-paint the panels, handles are missing – either need to get spares, or fabricate replacement handles (can only be done back at the main workshop in Germany, lacking machine tools here).

(2) The ground leakage -need to check the power supply. Hope it is not the transformer or other critical part.

(3) Power cable. Absolutely non-standard! Uses a BENDIX connector, 3 pin, type PT02E8-3P-027.
msr-904a mains connector bendix PT02E8-3P
Interestingly enough, found a suitable cable, especially made for the MSR-904, on xbay, Army surplus! PN: SC-D-627094-5FT NSN:5995-00-165-3806, the guy has more then 10 pieces – seems the Army was really worried to run out of cables for their MSR-904s.

(4) The frequency display works but doesn’t show the right frequencies.

(5) The F2 adjustment (upper sweep stop frequency in F1-F2 mode) doesn’t work.

(6) Figuring out the major adjustment pots – this is all documentation I have:
msr-904a documentation

(7) Figuring out the pinout of the “Monitor” port (intended to connect a storage scope, I might connect a digitizer), and of the “Remote” port – the remote control signals (TTL).

(8) Figuring out the external frequency control and phase lock voltage requirements.

(9) Drafting a block diagram of the RF deck and IF chain, just to better understand the inner workings, and to see, which parts-components Micro-Tel used.

Now, on item (2).

The power supply –
msr-904a triple shielded power supply
It’s held in place, and held together, by a cup full of screws. And, it has a layer of what is presumably Mu metal (high magnetically shielding sheet metal), to keep the 50-60 Hz in the transformer.

The filter, at the input, it is a sight in itself. Not sure how much it would cost to fabricate a custom aluminum case of this size, and to manually assembly it these days. Parts value alone, over 250 USD.
msr-904a mains filter
Well, and as it turns out, exactly these parts are leaky. SPI filters, 51-321-610, still available, after being around for 30+ years, at Mouser and elsewhere – 119.64 USD each, 18 pcs minimum order….

The spec data:
SCI 51-321-610 hermetic RFI filter

After a quick thought – I will give these parts a miss. With all the shielding, transformers and wires, we can do without hermetic feed-through filters – keeping in mind that also the bottom and top lid of the unit have ventillation holes.

So, filters removed, and wires re-connected… and, quite to my satisfaction, no ground leakage any more – not even a few microamps.

Before putting it back together – quick check of the power supply – all seems to be working fine now, and well adjusted.

Item (4) – complicated. Took me quite some hours. The frequency meter is acutally a voltage meter, and this is controlled by the tuning voltage, and a complex digital circuit spread over some hard to reach board. After searching around – it’s just a defective CMOS multiplexer switch, setting the gain of one of the voltage conditioning stages (which are needed to handle the various bands). It’s and CD4051, standard item, no problem. Put in a good one, from another part of the circuit that is not criticial at this point, and ordered a few spares, just USD 1.75 for 3 pcs, including shipment, from Macau.

msr-904a a few more broken parts

On item (5) – the sweep circuit is pretty similar to the Micro-Tel SG-811, and for the SG-811, I have the schematics around.
It uses a dual 10 turn potentiometer. 10k.
msr-904a helipot 8106 defective

Helipot 6108 series, a type common to high-grade analog-control instruments. Seems that one of the stages (the stage that controls the sweep range) is defective – the hybrid resistance material used for the pot (these don’t use wire, because they are made for high resolution applications) is open at the “cold” end – sweeper is always at full scale.

As these are all fully sealed units, no way to repair – found an exact replacement second-hand, for a reasonable charge. For the time being- changed the wires: the F2 display (controlled by the second stage of the pot, which is still working), has been disabled, and the wires changed so that the acutal sweep range is now controllable – so I can do all adjustments, just don’t get a display for the F2 frequency.

Well, and after all this, the unit is at least basically working, responding to controls, and not triggering any fuses. To move things further, setting it up with a few GHz range synthesizers, and an EIP 545A counter, for some first tests with RF.

msr-904a adjustments

To be continued…

Micro-Tel MSR-904A Microwave Receiver: the big box arrived!

Look what I found on the doorstep yesterday:
msr-904a big box

Wrapped in 20+ feet of bubble wrap, nothing less than a Micro-Tel MSR-904A Microwave Receiver, needing some TLC, later, to be added to a special equipment collection elsewhere.

The MSR-904A. Arguably, it is the last member of a series of 18 GHz+ receivers, build by Micro-Tel at Baltimore, MA, and intended for surveillance work, by governmental agencies. If you ask the right people, these receivers are pretty famous, and have been considered a strategic item for a long time.

They are build using all discrete parts, and hardwired CMOS and TTL logic. After all, many parts, but if you have seen other Micro-Tel instruments, not too unfamiliar. Some say, 80s technology, but actually, is is build in time-less style – from the best components available (not only at the time – these components, YIGs and filters haven’t really improved since).

Some performance data:
Frequency range: 0.5-18 GHz – fundamental mixing; fully YIG pre-selected over the full range (using 18 dB drop-off filter, i.e., three YIG spheres; one preselector for 0.5-2 GHz, the other, 2 to 18 GHz).
1st Image rejection, 70 dB, and 65 dB at above 12 GHz.
IF rejection: >70 dB
IF filters: 100 kHz, 1 MHz, 5 MHz, 30 MHz – quite handy.
IM3: about 5 dBm
LOG and LIN detectors
AM and FM demodulator
Spurious: 90 dBm at input equivalent over full range.
Noise figure is about 20 dB

Note: All in all, 3+3+1+1 = 9 YIG spheres are used, and an uncounted number of filter crystalls. The 100 kHz 21.4 IF filter, it’s quite impressive.

In the 2 to 18 GHz range, a 250 MHz-21.4 MHz IF chain is used, with LO 250 MHz above the signal.
For 0.5 to 2 GHz the signal is mixed with an additional 2.08 GHz from an auxilliary LO. I.e., LO frequency is 2330 MHz (2080+250 MHz) above signal.

The other great things about it:

(1) Fully fundamental mixing, using YIGs – lowest phase noise possible. Fully preselected.

(2) Unit has a 250 MHz IF output, with about 40 MHz bandwidth – this makes this unit ideally suitable as down-converter, if you want or need to receive at medium to high GHz frequencies. Can be directly fed to any SDR for demodulation. The MSR-904A has very small group delay, seems pretty suitable for handling of digital modulation schemes.

(3) It is fully remotely controllable, and has a phase lock input – will hook it up to a ADF41020, and/or a fractional-N PLL, same PLLs as already developed and tested for the Micro-Tel 1295 receiver. Such PLL unit will go along with the MSR-904A, once the repair and proper adjustment and testing is finished. Micro-Tel used to offer a frequency stabilizer (PLL) for the MSR-904A, but I have never seen one offered. If you have one, please let me know!

Two downsides – NO serice manual, no manual or documentation at all. If you have one, even if only for another MSR unit (MSR-901, MSR-902, MSR 903), please, let me know.

Second downside – the condition. Well, there don’t seem to be many of the MSR-904A around for sale any more. One unit I know off, but it doesn’t have the panoramic (scope) display. Other might be available, at outrageous cost. This unit was sold even blow the market value of a fraction of the components.

msr-904a as received - front

Note the tuning know – different from the typical Micro-Tel style. But nevertheless, seems to be the original, unmodified part.

msr-904a as received - top

msr-904a as received - bottom

msr-904a as received - back

The full repair, it will be a major job, because currently, it is a bit beat up – I wish, the earlier owners would have treated it a bit more carefully, and Micro-Tel should have never touched the green paint that just isn’t lasting and a sticky mess on a good number of their instruments – fortunately, only the panels are affected, and these are easy to remove – re-painting already in process!
Also, it doesn’t seem to work well, powers up, but seems to have a leaky supply – keeps tiggering the RCD. It needs a through inspection.

Cosmetically at least, the inner working are in much better shape than the exterior would suggest.

Center – edge-connector boards, mainly YIG driver and analog control. The metal box on the right – the RF box with the microwave stuff. The other items – IF converters, detectors and so on. Everything: very well shielded.
msr-904a top view

2.08 auxilliary LO. Mixers.
msr-904a aux LO filters mixer

A Narda 2-18 GHz broadband -10 dB coupler. Still available from Narda today!
msr-904a narda 4203-10 LO sample coupler

The preselectors: S082-1630 (2-18 GHz, might work up to 20-22 GHz), and a custom Systron Donner 0.48-2.05 GHz YTF.
msr-904a preselectors

3562A Dynamic Signal Analyzer: LCD retrofit NewScope-5

Yet, another job related to the 3562A, same machine that also had the ROM board defect. This unit also has a weak display, and I have been asked to check replacement-repair options.

After a brief search, there aren’t any spare CRTs around, for the 1345A display used in the 3562A. The last good ones might have been purchased-stockpiled some years ago, by corporation that need to keep equipment going.

Well, fair enough. Luckily, failed CRTs and their limitations are a common feature of dated test equipment. So other have already invested time and effort to provide a solution: the LCD retrofit.

Why not just replace the whole instrument, with something new, up-tp-date, and more manageable? Several reasons:

(1) Many clients have proprietary-custom software running certain automated tests, using certain types and specs of test equipment. The final product specs have often been agreed upon with the OEM, in contracts that are a big hazzle to change. Some of these products, in fact, most, have long service life, so the test rig needs to be kept alive, more or less, at any cost.

(2) Cost – well, new test gear of the proper kind is outrageously expensive. Not taking about plain stuff here, but powerful equipment, network analyzers, spectrum analyzers.

(3) Servicabilty: talking about smaller businesses, not big corporation, often it is quite handy to stick with somewhat dated analyzers etc, because they are perfectly up for the task, the operators are trained, both in using and serviceing them, and often, spare units-parts units are around and can be procured at a fraction of the cost of repair of new equipment.

(4) Quality. Arguably, and except for new digital signal-fast scope stuff, the instruments build in the late 80s and early 90s might be the best ever build. Most of them have specs and typical performance far above what most regular quality can provide nowadays. The reason is simple, in the 80s, these rigs were build for the military and related agencies as the key target market. Nowadays, for consumer electronics, consumer communications… One exemple: the HP 8566B spectrum analyzer. Not sure about the price of such equipment nowadays, if build new – certainly 6 digits. On xbay, they go for about few cents per USD list 1985 list price… and be assured, no big deal to get them working in-spec.

The LCD replacement kits have come down in price considerably – earlier on, still USD 1000+, now, check this out:
NewScope-5 offer
USD 400. Not bad. This includes display. LVDS driver. And certainly, the controller board, to adapt the display to the 3562A.

Here, from the inside:
3652a lcd conversion newscope-5

The display: it’s a rather dated SHARP LCD DISPLAY, LQ057V3DG02, TFT 5.7″, 640×480. But rather then stockpiling CRTs, I now have a few of these in stock now – found them for about 15 EUR each surplus – this will allow service of the LCD-retrofit units, for years to come, without any need for modification to fit another type of LCD display…

I can tell you, such retrofit is worth every dollar. If you have any of the CRT analyzer with the screen gettin dimmer. Don’t hurt your eyes.

3562a lcd conversions newscope-5
A quick glance at it – it’s great to work with it – color display, rather than dim green display.
The color settings work fine for the most part – there are some little bugs in the NewScope-5, related to the text color, in some menues (first character has different color). However, this has really no impact on the great advantage of such LCD retrofit.

3562A Dynamic Signal Analyzer: EROMS fixed, finally!

The last and only remaining item to get the 3562A with the defective A3 ROM Board (03562-66503) back into service, replacing the defective EPROM. Well, I thought this would be a 30 minute job, but it ended up taking a few hours longer. Why? Multiple reasons:

(1) A bug in the AVR eprom reader software, specifically written to read the 3562A rom board (and similar boards, or other instruments – software always needs some adaption). One of the address lines was not toggled-ended up with corrupted data read from the “good” rom board.
Learning for today: always check the EPROM data read for validity, by checking for repeat patters, and by reviewing all the byte values. There should be at least a few 0xff values, otherwise, one of the data bus lines may be broken.

(2) Turns out, there are two versions even of the Rev B rom board, same part number, but eproms U118 and U218 that have the checksums are different. So, needed to desolder these two as well, and replace with the updated version from the working board.

The good ones on the left, the bad guys on the right…
3562a re-programmable vs one-time-programmable eproms 27256

(3) With all this, my eprom programmer, the only one I have that can handle 27256 EPROM had a defective jumper! No contact on one of the address lines….

After all these efforts: that’s the board, after repair:
3562a rom board

The replaced EPROMs are now in sockets – just in case, should they ever fail again.

Just in case you have to do a similar job – here are the EPROM images. Keep in mind, Rev B, 36x 27256 EPROM (sure, you can also use 27C256).
hp 3562a A3 ROM BOARD rev B

And, finally, let’s switch the power on-
3562a rom repair - complete!

All tests passed!

Note – just discovered, someone is offering a spare 3562A ROM board for about USD 100 on xbay…. well, well, but in the end, better a thorough repair, with all EPROM images captured, than just switching some board!

Avantek AFT-4231-10F 2-4 GHz Amplifier: some characterization and modeling

The task for today – characterization of a bunch of microwave amplifiers, Avantek/HP AFT-4231-10F. These are quite rugged and affordable components, widely available surplus, and hermetically sealed – will last forever, if things are not messed up completely.

aft-4231-10f under test

The specification however, it’s not quite clear, and no detailled information could be found on the web. That’s why I have been asked to come up with measurements and a calculation model that allows to estimate the gain (and the actual maximum output power, and the necessary input power, to reach close to maximum output), at any given frequency and input power. Also, it needs to be checked how far above 4 GHz this device still works.
Last item is to measure the supply voltage sensitivity of the gain, to get a feeling on the required stabilization, to avoid incidental AM on the signal.

The datasheet –
aft series amplifier

The only equipment at hand at my temporary workshop here, a microwave source, EIP 928, and an HP 8565A spectrum analyzer was used to measure the gain at various input levels. Accuracy of this setup is about 1 dB.

Some of the results (0 dBm input: blue diamonds; 10 dBm input: green triangles):
aft-4231-10f pout at 0dbm and 10 dbm pin vs frq

To get a proper continuous description, these data were fit to a non-linear function, fractional polynomial term (fits are done using Tablecurve 2D, an excellent program, highly recommended, but doesn’t come cheap):
gain fit
The gain fit (0 dB input) can also be used to describe the maximum power, with some scaling factors – this considerably reduces the number of parameters needed, and the calculation effort later, when implemented in a microcontroller. Black lines in above diagram show the fit results.

For the gain compression, a 2nd order polynomial is used, and scaled for the 10 dBm input gain.
aft-4231-10f gain compression vs pin at 3 GHz

Once this is all established, no big deal to see the full picture.

Gain, at various input power levels, Pin:
aft-4231-10f gain vs frq at various pin

Output power, Pout, at various input power levels, Pin:
aft-4231-10f pout vs frq at various pin

Accordingly, no problem to get 18 dBm+ in the 1.8 to 4.5 GHz range, perfect for the application requirement.

The final item – supply voltage impact on gain: tested at 3 GHz, 0 dBm input power.
Using a Micro-Tel 1295 test receiver, the reference level was set to 0 dB at 15 V supply voltage, which is the nominal voltage.
Down to 9.0 V, the AFT stays within an excellent 0.01 dB variation. Output power slightly increases (0.15-0.25 dB) down to 6 V. At about 5 V, amplification cuts out. So the AFT can work with any voltage from 10 to 15 V, at about 80 mA, and seems to have pretty good internal regulation.

amp avantek aft-4231-10f