PLL

ADF41020 18 GHz PLL: universal divider and PLL board

2016/08/03 Simon Leave a comment

I cannot praise Analog Devices enough for the ingenious designs, and for providing parts like the ADF41020, a fully integrated 18 GHz PLL. This is actually part of a major design effort for a multi-channel frontend, here just a description of the small test board used to establish the general circuit layout and board design.

Probably interesting is also the hand-soldering of the LFCSP leadframe package, which is actually not as difficult as it seems. For soldering of the pad, there is a large via in the center, which does provide good heat-sinking and is easy to solder through the 1.2 mm board.

Above, the layout, below, 10 boards – 14 dollars and a few weeks later.

For soldering, best use 0.5 mm Ag-containing SMD solder, with Type 32 flux, which is halogen free resin flux.

To mount the LFCSP, first apply some solder to the chip pads, but not to the center/heat sink pad. Apply some flux to the board (which is already pre-tinned; use any good SMD flux pen). Then align with a good magnifier, using some Kapton tape to hold the chip in place – leave one side exposed. Then solder, in one stroke, using a medium hot soldering tip. Reflow another time – one side done. Remove the Kapton tape, and solder the remaining 3 sides. Then stick down the chip with Kapton tape again (to avoid any remote chance of movement, in case all the solder melts during the next step). Turn around the boards, and solder through the via, with a fine solder tip.

For a test, just apply a test signal to the input, and use the “MUX” output to check for any pulses. There we go:

These pulses aren’t quite long, so it is one of the few occasions where a scope more advanced than the 2215 Tektronix is really useful in the home shop… same pulses on a HP/Agilent/Keysight 54720A, 54713B plug-in, and 100 MHz 1:10 probe.

These fast risetime pulses, and the various prescalers, dividers and good input sensitivity make the ADF41020 quite useful for any PLL and frequency counting applications.

2 GHz in, 25 kHz out — confirmed.

About the input sensitivity: the ADF41020 is specified over a 4 to 18 GHz range – how about lower frequencies? A quick look at the input circuit shows a 3 pF capacitor – which equals a reactance of about 53 Ohms, at 1 GHz (i.e., the capacitor and termination resistor will cut the input power available to the buffer approximately in half).

… quite useful down to 1 GHz, no problem or instability at all. Also checked the the reproducibility, for 3 devices – not a lot of scatter.

EIP 545A Microwave Counter: power meter upgrade

EIP 545A Microwave Counter: another power meter upgrade

2016/08/03 Simon Leave a comment

This is just a brief update on an earlier post, related to the power meter option for EIP counters, EIP Power Meter Upgrade. The current 545A came without any options, even without GPIB, and I was not quite sure if the firmware power meter option firmware would work, even if no GPIB option is present. Well, easy enough, put it to test.

It didn’t take much, a few (rather old 2516, and 2532) EPROMs, a resistor, some spare ICs, and a few inches of solder…

Here, the result: all working just fine, counting away at several GHz, and power reading appears to be resonably accurate. I would not substitute it for a proper power meter, for calibration purposes, but the power meter option is a handy feature for any quick test and confirmation.

Various

Major relocation, and numerous 230 V conversions…

2016/08/03 Simon Leave a comment

Long time no post, not because there is nothing happening here, more to the opposite. Relocated from the US, East Coast, back to Germany, including the US section of my electronics shop, 40+ pieces of heavy test gear. All made it over the sea just fine, in a 20′ container. Now, changing all the fuses and converting everything to 230 VAC mains voltage. I will spare you the details, just a few impressions for some trusty HP power supplies. These actually require some re-wiring, you have to break to circuit traces, and install a wire bridge.

Installing new plugs… wires properly protected.

Now just install a bridge between the middle solder points. Great that there are schematics and manuals, even for 50 year old devices!

…don’t forget to replace the fuse with one of the proper size for 230 VAC operation!

Various

HP 3325B A26 Assembly: low byte SRAM, and an incorrect signature

2016/06/21 Simon 1 Comment

A quick comment on the 3325B service manual – Thanks to a kind contributor.

The kernel SA test is a bit flaky on A26U27 PIN 10. This is an inverter output, and can’t be same as the input… correct signature is A70F. Appears to be a misprint in the HP service manual!

In the present case – the culprit was one of the SRAMs, the lower byte. Fortunately, not really a rare part, and easily sourced and replaced, once you know that this is the circuit at fault. You might want to check this very carefully, because there are multiple versions of the 3325A/3325B control boards.

Various

YIG Oscillator/YTO Analyzer: linearity, output power, hysteresis

2016/06/19 Simon Leave a comment

With a good number of YIG oscillators (YTOs) around, time for a few tests. Rather than writing down all the numbers, a quick test setup, with two power supplies, an EIP 545A counter (with built-in power meter), an Agilent 66319D power supply (used as current source for the main tune and FM coils of the YTO), a 14.5 dB attenuator (just happens to be the value I had around) and a few cables.
When performing such test, make sure to put a 10 dB (or larger) attenuator of really good quality (low SWR) directly at the YTO output, without any adaptors or cables in between – some YTOs will show inaccuracies of not properly terminated in 50 Ohms.

This is a view of the setup:

One of the YTOs under test, a S081-0320 2 to 8 GHz Avantek part.

As you can see, the output frequency is pretty linear vs. main coil current.

The tests are carried out first with increasing coil current, then with decreasing coil current, and the hysteresis is calculated (difference of output frequency, when a approaching from higher vs. lower current).

One thing to watch out for are thermal effects, but let the YIG warm-up for 1 hour or so, with the main coil at half-range current. The effects aren’t all that big for the Avantek YTO discussed here, but you never now, for lower quality parts, other manufacturers, and so on.

Micro-Tel MSR-902C

Micro-Tel MSR-902C Microwave Surveillance Receiver: A3B5 and A3B6 analysis and repair

2016/05/21 Simon Leave a comment

Some progress with the MSR-902C, which is basically working, but not on all bands – see earlier post.
With no documentation on the MSR-902C available, except some data sheets, I first traced the band select signals – and they appear to be generated on the A3B5 board, which we may call, band control board. This board also converts the 0-4 V tuning voltage from the tune dial to a 0.5-9 V signal that corresponds to 1 to 18 GHz (0.5 V/GHz slope).

This is a top view of the A3B5 assembly. On the upper left, the tuning control voltage converter, left half, voltage comparators for the 1-18 GHz multi-band mode. These comparators assign the band number to the tuning voltage, when the 1-18 GHz range is used. It is disabled (even supply voltage cut) when the MSR-902C is in single-band mode (selected by band selector switch).

First observations – there are control LEDs that seem to indicate which band is currently active, but for some reason, they don’t light up on all bands. Very strange. Probed around the logic signals, and at least some signals are there, still no LED lighting up. Very suspicious. After some head-scratching, decided to probe the LEDs by supplying some external power, and poking around with a resistor. Turns out, some of the LEDs are dead. They just barely light up with 20 mA of current applied, and no sign of light at all at the current level supplied to them by the A3B5 assy.

After repair of the defective leads, you can clearly see the difference of the new LED, vs. the old LED – some of the old ones are still working, but not very bright any more, and I’m going to replace them all, once I have this back in the main workshop with a better supply of 3 mm red LEDs.

Another strange observation – two of the logic chips are rather hot. What is going on there? Furtunately, these are in sockets, a 7401, and a 7404, and a quick test revealed that one gate of each of these chips is sinking current, about 200 mA. So these need to be replaced.

Mugshots of the culprits so far….

Not sure if it is very clear, but here are the connector signals of the A3B5 assy, and the description of the adjustment pots. First adjust for appropriate tuning voltages, then adjust for proper band switching in multi-band mode, monitoring the LEDs.

The next thing, the A3B6 assy. No apparent defect, but still needed to find out what it does, and how to adjust. It appears to be the multi-band control assy, converting the 1-18 GHz full-range tuning voltages to tuning voltages for the individual bands, by applying offset and slope corrections. The offsets/slopes are selected by CMOS multiplexers as it is custom for most of the MSR series designs. The output tuning voltage is buffered, and forwarded to the other circuits.

Here, you can clearly see the order of the adjustment potentiometers. For adjustment, if may be best to first align A3B5, and then supply appropriate reference signals to the MSR-902C, or measure the LO frequency, and do the fine adjustment with the 1-18 GHz full range mode selected, and tuning through all the bands. The fine adjustments would need to be done both at the low end (for offset), and at the high end (for gain), for each band. No big deal, once you know which of the pots to turn.

Further repair will have to wait a bit, until a few spare 7401 have been received. But all is looking pretty good.

HP 8662A Synthesized Signal Generator

HP 8662A 8663A Power Supply A7A3 Assy: base transformer defect

2016/05/21 Simon 4 Comments

A little note, Thanks to a kind fellow sharing this repair info with me, related to the A7A3 assy of the 8662/8663A generators: the 9100-4018 base drive transformer. There are two of these on the 08862-60289 board, protected by a fuse and diodes, nothing should acutally happen to them, but things can go wrong.

Below, the two versions of the A7A3 boards, left, the older 08662-60289, and the more recent 08662-60604. This post only refers to the -60289, the base drive transformers of the newer units look the same, but have part number 9100-5291 – don’t know if the can be exchanged.

A snipet of the power supply circuit, there are two identical base drive transformers, center tapped on one side.

Cross reference – this is a NSN part, the most prominent manufacturer seems to have been Fil-Mag, which use to be a Sprague company long ago.

Note that these little beasts come at a quite hefty price! That’s well over USD 100, just for the two base drive transformers – I hope, HP did not pay the list price, or anywhere close to it.

Here is a rare view of the interals, after heavy work with sandpaper and other means (these transformers are potted, but as with all potting compounds it can be removed, if you have plenty of time, a good supply of tools and don’t mind the dust and dirt).

It uses an OJ41408 bobin/pot core, PC14/8 size. Wire is about 0.1 mm size. The pot cores are still available from Magnetics Corporation, mag-inc.com, and the material is just a regular mid-frequency ferrite. So you might be lucky with just using any average good pot core with about 5000 permittivity (e.g., N30 ferrite).

Fingers crossed that you will never need this information, because it is quite a laborious effort to reverse-engineer the internals, and to fabricate a new transformer manually. Tempted to say, I could manufactur them well below list price, if someone would need a 1000 pieces….

Micro-Tel MSR-902C

Micro-Tel MSR-902C Microwave Surveillance Receiver: power back on – first signs of (extraterrestrial?) life

2016/05/10 Simon Leave a comment

Today, a few spare MJ12002 transistors arrived. No time to lose, and put them into the power supply. Note that the new transistors are 1983 data code, whereas the Micro-Tel originals were 1988… fixing the power supply with old parts, but no reason to assume that these transistors have any issues with age. With such power supplies, I would always suggest to use a pair of transistors of the same manufacturer, rather than mixing up two very different devices. This is why both transistors were replaced, not just the defective part.

After this replacement, connected a 10 Ohms 25 Watts load resistor, and grounded the Interlock and ON/OFF lines. When powering up, the green AC ON light comes on, but not for too long. Look at the set of fuses sacrificed in the process:

Another set of tests – no issues found, all working fine. Something must be loading the power supply, and I can’t get any negative voltages out of it – but there must be at least one negative rail to provide -15 V to the various opamps in the receiver.

Not to long and the culprit was found – a shorted tantalum, a T310 series Kemet tantalum, directly at the – what turned out to be, -18 V output. Check out the date code. Why did Micro-Tel put a 1979, week 38 dated device, in such kind of expensive and specialized equipment (other parts suggest that this unit was made about 1989, at a price of about $40-50k – that’s about $70k in today’s dollars…).

Some tests show that there is a +18 V, -18 V, and a +12 V output. All are routed through feed-through capacitors. A fair bit of effort, and cost!

First test with the actual receiver connected –

– connected the 1-18 GHz tuner – a bit of a cable mess.

To test the basic functions, like, IF chain, detectors, etc, a 1.5 GHz test signal from a HP 8642B was routed to the tuner. And, to my greatest satisfaction, the MSR-902C is actually receiving!

1 kHz AM modulation…

… also tested the FM and AM detectors, both in sweep and fixed modes, the AFC, the IF gain, the marker – all working. Also the 8-12 GHz, and 12-18 GHz ranges, working fine. Clear signal down to -105 dBm input. So all working and pretty well tune.

Unfortunatly, this is not the case for the 2 to 8 GHz ranges – the frequency display is not showing a reasonable value – not sure what is going on here. Maybe something with the band logic, or the signal multiplexers (see the MSR-904A repair story – these instruments are notorious for defective CMOS multiplexers).

So far, so good – at least in some bands, we would receive satellites, or signals from other galaxies, given, there aren’t many strong sources out there, in space, and all the other solar systems, too far away!

HP 3562A Dynamic Signal Analyzer

HP 3562A Dynamic Signal Analyzer: a blackened transistor, and a dead S-RAM

2016/05/09 Simon Leave a comment

Here, Thanks to Michael from Zurich, Switzerland, and for the benefit of everyone with a 3562A showing similar signs of disrepair:

The 3562 A shows a fault code, is on the A2 (SYSTEM CPU/HPIB) error code 19, which means Monitor RAM Test Multiple Monitor RAM failures.
All voltages checked, and they are OK, the ripple is OK and the clocks are OK too. Everything was OK, no smoke, but I still had to solve two issues.

Issue (1)
Non-working Display Unit HP 1345A

First of all I guessed that this issue was coming from non-working A2 (CPU) but installing the test jumper to get the test screen did not work. Measured the voltages on top of the HP 1345A. These did not show any issues (no shortened tantalum capacitor) on +5V/+15V/-15V. Checked a few more voltages but not the +105V.
So I removed the HP 1345A unit to do a visual inspection and noticed the defect on the A3 board (Low Voltage Power Supply). Q1/Q2 did not look so good.

For sure without +105V we do not get any picture from the HP 1345A. I removed the faulty Q1/Q2 and solder some test cables for +5V/+15V/-15V/+105V to A3 to power the assembly from an external power supply. Lucky me, no other defect and the test picture came up.

I decide to order SG3524 (pulse-width-modulator), MJE180, and all capacitors of the A3 to replace them all. After the rework on A3 I carefully powered up the +15V supply which is used for the DC/DC converter to generate the +105V and measured the current.

See the re-worked A3 assy:

No issue seen anymore, the +105V is working. I added a 2.7k resistor to create a nominal load (approx. 37mA) on the +105V path to adjust the +105V. Just to know what was causing the burned Q1/Q2 I swapped the new SG3524 with the original one and I see that the current was increasing like hell when slowly increase the +15V voltage. So I guess the major problem was a defect SG3524 here.

Issue (2)
Faulty A2 board with hex error code 19

After installing the now working HP 1345A back into the HP 3562A I got a bit more detailed information about the A2 problem.

It says that there is a problem on the low byte SRAM on A2. To ensure that nothing else causing this problem (bus issues) I removed the A2 from the cabinet and it can be operated completely standalone. From the LED on A2 I still got the same error code as before (when A2 was installed) so at least there is no other
board causing this issue and I really can focus on the A2 board. First I checked all signals on the two 32kx8 SRAMS (U212/U211) with a scope but I did not see anything defect, everything looked so far good (no shorts, activity on all signals, etc).
So I attached my nice Philips logic analyzer.

Playing a bit with the logic analyzer, but did not get any more results so I believed what the monitor test logs said and replaced the low byte SRAM (U211) with a new one (ordered 70ns ones from Mouser).

After replace the SRAM the self test is passing on the A2 and it’s now time to install the A2 board back again in the
HP 3562A cabinet (and crossing fingers!!!!!).

With the changed SRAM, my HP3562A boots up without any other errors and issues and is ready to be used again!

Micro-Tel MSR-902C

Micro-Tel MSR-902C Microwave Surveillance Receiver: a metal box, microwave plumbing – 1 to 18 GHz tuner revealed

2016/05/08 Simon Leave a comment

With no manuals available, some investigations were carried out to better understand the workings of the MSR-902C microwave tuner, which has a 1 to 18 GHz range, good noise figure, fully-fundamental mixing with 3-stage preselection over the full band. IF output is 250 MHz, so the tuner can be combined with any resonable SDR or other modern receiver, as a down-converter, offering about 40~60 MHz bandwidth, and 60 to 70 dB image rejection, and huge capacity to deal with out-of-band overload signals.

This is the rough scheme, leaving out all ordinary electronics in the case, just the microwave parts (note that there is another SMA attenuator in the feed line of the splitter, coming from the 8-18 GHz YTO, not shown in the sketch).

Essentially, there are two inputs. One covering 1 to 12 GHz, and another one, covering 12 to 18 GHz. The 8 to 18 GHz YTO is used for both bands, and PIN switches are used throughout to route the signals.
The IF goes through a 300 MHz low-pass and a +13 dB monolithic amplifier.

Note that there are some different/earlier versions of the MSR-902 and maybe also MSR-902C which use a slightly different configuration, with a LO doubler. Maybe the could not get proper 8-18 GHz YTOs at the time, at any resonable cost, and had to resort to another topology (using a doubler) for this reason. However, I have never seen any of the earlier tuners, and can only report what I heared about them, with documentation on these units being almost completely absent.

For some of the key devices, see references below. Glad not to show list prices, as these would quickly add up to USD 10 or 20k, for all these microwave parts. Not to mention that these are all US made, most advanced and highest grade components of their kind. Datecodes are from the late 80s, mostly 1988, but still today, there aren’t much other options around to build a tuner of this kind. Maybe there just aren’t enough entities around that can afford such device nowadays, and software and digital signal processing certainly have contributed that todays devices can achieve perfect results even with less expensive, heavy, and energy-consuming parts. Still it is very instructive to study the design of this tuner. It even has a LO sample output, and with some effort, all the YTOs could be phase locked with relative ease (using GeSi dividers, etc).