Category Archives: HP 8662A Synthesized Signal Generator

HP Attenuator 33321SB AKA 33321-60026 ATTENUATOR: fixing a defective spare part

After some consideration, I decided on the fix of the the 33321SB attenuator. The defect – the plastic of the grey holders broke off, probably, a matter of age and a matter of stress in the special construction of the 33321SB with the side connector.

It is not the first attenuator that has such defect, so it seems to be a certain weakness of the HP design, albeit, a weakness showing up much after the design life.

Originally, I planned to fix it back home in Germany, with some precision machine tools – drill two holes, fit/glue two plasic rods, and the mount it in the original fashion. But currently, it is uncertain when I can go to Germany again, so I decided on an alternative fix. There are several unused HP attenuators around here, so I checked their internals for a contact of the same design (length). Take care, there are many different configurations of these attenuators that all differ in the length of the contacts, the distances between contact, and so on.

Found a good donor – now, cutting the piece with a razor blade.

The part is obviously fragile, better don’t touch the contact.

After assembly, it seems to work well and fits well. To avoid further risk of breakage, added some 2K epoxy glue underneath the metal spring – it is not visible on the picture. So I hope it will be a permanent repair. Otherwise, we will fix it again using better tools and methods.

With the spare part repaired, and installed in the 8662A, the generator is working well again, at all power levels.

HP 33321SB aka 33321-60026 Attenuator: HP 8662A Synthesized Signal Generator fix

Finally, I received the attenuator set for the HP 8662A repair. The very special 5+40+40 db attenuator with a side connector (rather than the usual design of these attenuators with two top SMA connectors).
However, a quick test showed that the 5 dB step is not working right. The “through” path is fine, but when I engage the 5 dB attenuator, is is something like 45 dB, unstable. Maybe a blown segment? That would be an easy fix.

So, I opened it up carefully, and found a broken contact. It seems the side connection is pushing a bit on the contact, and over the year, this design caused the plastic to fail. Tough to fix without precision drills and machine tools. So it will need to wait for the virus to clear, and for me to go to the German workshop again, to attempt an repair.

There are two small plastic studs, and they broke off, maybe the plastic got brittle over time with the pressure of the side contact pushing.
Let’s think about how to fix it, if at all possible.

Note – the seller was kind enough to refund half of the price, without having me to ship back the part, fair enough. At least some spare coils and segments should I need to fix other attenuators in the future.

HP 8662A Synthesized Signal Generator: almost fixed!

Still working on the 8662A, a unit that is in very good shape but someone tampered with the power supply in a non-expert attempt to fix it.

After a little bit of waiting, CMOS ICs worth a few dollars arrived, and a single 4049 CMOS fixed the power supply. Probably, the defective caps on the power converter board destroyed one of the transistors, and then lead to a voltage spike that destroyed the CMOS (which has a connection to the sense line) – only that gate got destroyed, the other gates still work.

There was no need for any adjustments, the voltages are still accurate to the resolution of the meters I have around here.

Back up and running, the machine is working fine and giving good output. At least after installing a coax line to bridge the missing attenuator.

There are some small issues that were easily fixed. A few non reactive keys that were traced to a bad contact of the connector related to one of the lines of the key matrix.

… and some issues of the modulation display. It is stuck at “20”, but the 1st digit (+-1) is working. Turned out to be a defective trace of the multilayer motherboard. Maybe someone damaged it by pulling on the ribbon cable, or similar bad handling – it is one of the outermost pins.

Easily fixed with a wire and some silicon glue.

Some study on the attenuator – to my great surprise, the 8662A uses a retrofit set, to replace the obsolete attenuator at the time of construction of this very late production unit. And, it does not(!) use the attenuators of the 8663A. These don’t fit, neither in values nor in size, because there is an interference of one of the SMA connectors with the bandpass filter that is also mounted to the inner front plate. The missing attenuator, a 33321-60026 is a 85 dB unit, one segment 5 dB, two segments 40 dB each – these are switched simultaneously. A really special part, and difficult to assemble even if you have several other step attenuators because of the length of the 40 dB chip…

I don’t actually need to attenuator that much, but luckily found this set on ebay from a seller in Israel. The kind guy offered it for a very reasonable price, great! So we can restore the unit to full function.

Meanwhile, the attenuator set has arrived in Germany, where I will pick it up during the next business trip – coming up soon. In any case, I will also need to pick up numerous screws, because the unit arrived with several screws missing, even some quite essential ones.

Anyway, there are plenty of HP screws around.

HP 8662A Synthesized Signal Generator: An almost new generator, with a pretty old power supply, and a serial number mystery

This is another one of the marvelous HP 8662A generators, I got it eventually for free because it had parts missing but otherwise, a good unit.

The missing parts were the mains filter, and the transformer for the OCXO. The filter, I just replaced it with a fairly standard commercial unit, should be good enough for now. The transformer it is a bit more tricky, because of the limited space and rate type. By luck and coincidence I found a spare in Japan. 25 Euros, fair enough.

Be aware that there are many versions of the input assemblies – and my unit has a 1982 vintage power supply, with a 1996 back plane. This is not going to work well, so I needed to study all the wires and connections. No wonder someone had given up on the repair job before.

The later version:

Mounted these parts – yet, I didn’t turn the unit on, because of a blown fuse. After some inspection of the power supply, the common issues. A few bad caps on the driver assembly, replaced with a pair of new 22 uF axial caps.

Other issues – one of the main switching transistors is dead – ordered some spare transistors, new old stock, from Greece, and took them to Japan during the last business trip to Germany… these are MJ16012 power transistors. 800 Volt, 15 Amp 175 Watts. I don’t trust any copies of these, but rather bought good old Motorola parts. 1986 vintage.

Good practice with the 8662A power supply repairs – checked the control assembly.

The oscillator is working, 40 kHz present, but no drive output – how come? Some quick checks clearly show issues around the U1 inverter, a defective logic gate. Someone must have tried to fix it before, but seems he didn’t have the proper tools and solder – all sticky and bad, so I cleaned it with alcohol.

Desoldered the U1 4049 CMOS, with quite some difficulty because the drill used was quite small, and difficult to desolder.

Some strange observation – all the versions and date codes. The power supply control assembly, maybe 1997 vintage. The power supply base assembly and low voltage circuit, 1982 vintage. The A7A3 power converter assembly, 1993 with some parts dating back to 1986…

The serial – 34xx would suggest 1960+34=1994, but this can’t be, because the frame was only made past 12/1995.

Checked two random assemblies – also these were made in 1997.

Probably fair to say, a 1997 unit, and someone tried to fix it some time after that with a 1982 supply. Why he changed the power supply baseboard, and the removed the oven transformer, no clue.

The defective gate of the control assembly – order a few new CMOS circuits. Plenty of them back home in Germany, but here, it is faster and cheaper to order them new from Thailand… in addition, ordered a few 4013 and so on just in case.

Lastly, this unit also has an attenuator missing. Not uncommon for the 8662A. It is a 33321-60028 65 dB 5 V attenuator. 5/40/20 dB elements. Maybe I will find one on eBay, but actually, I have not much use for it because this 8662A will be running at +10 dBm all the time as a reference LO for my VCO test rig and phase noise analyzer.

The 08662-60001 low noise VCO: the heart of the HP 8662A

This is certainly one of the electronic antiques and marvels, which had the most hidden and non-spectacular impact on mankind and development of mobile communications in the 80s and 90s – the VCO of the 8662A/8663A, the defacto standard for low close-in phase noise signal generators.

The concept, it is a switch reactance oscillator, i.e., the inductances will be switched, to cover the 320 to 640 MHz range, rather than doing this all by varactor diode tuning.

For low phase noise, you need an oscillator with high Q – this is achieved by strongly biasing the varactor diode over the full tuning range, and by low resistance PIN diodes (two in parallel, see below!), that switch the thin-film inductances embedded between lexane disks.

Note the small indent on the golden patch on the PCB? This is where the copper bolt reaches in to adjust the frequency offset.

All this is housed in extruded aluminum, end a layer of special steel which absorbs magnetic fields. So far, I never had the chance to look inside of these – but a kind reader this block, Martin, shared this picture and I put it up here for those interested.

The designer of this marvelous and magic device, his name, Dieter Scherer, a German fellow of HP, unfortunately, I have never met him and don’t know if he is still alive. Sure he left behind great achievement and a legacy of high frequency engineering.

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

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.

8662 8663 pwr supply

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

08662-60289 base drive schematic

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.

9100-4018 master cross ref

5950-01-267-1279 aka 9100-4018 transformer

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).

8662 8663 pwr fil-mag 42z994 internals

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,, 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).

8662 8663 pwr oj41408

mag-inc 41408ug pot core

mag-inc j material

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….

8662A Synthesized Signal Generator: another rather straightforward repair

These 8662A generators keep coming in, seems that after so many years (like, 30 years) of 24/7 service, some little things are coming up. But not really sure if the frequency of failure/repair really increases, because with all these 1000s of parts, most of the units have seen some repair already, and it is just in the ordinary course of things that there can be defects every once in a while.

This unit add issues with the reference sum loop (A6A5), showing up in intervals of about 10-20 MHz. Knowing that a 10/20 MHz reference is used for the ref sum loop assy, this was the first and most likely cause of the defect. So, I substituted the 10/20 MHz signal, but, to no effect. Therefore, the defect is most likely located on the A6A5 assy.

8662a a6a5

One option would be to switch the assembly with one of my 8662A/8663A units, but these are currently in use and I don’t want to pull out assemblies that are difficult to fix and need adjustments, etc.

So, next, checked the pre-tune circuit, and, issue found. TP14 is at a constant -2 V, rather than variable voltage as per requirement given in the service manual. Digital input is OK – maybe the fault is in the semiconductors – one transistor, and 4 FETs switching a current source.
With spares at hand, these suspicious parts were quickly replaced. Effect: no effect. Big mystery.

Now, connected an external resistor, to measure the pretune DAC (4 FETs), and, it works! Even more mysterious.

Finally, also checked the traces and the resistors, and found the ‘2K’ Trimpot to have about 40k!! With the wiper contact having some effect, but not much.

8662a a6a5 assy schematic

8662 defective trimpot

Desoldering the Trimpot – all seems to be OK with it. Maybe some kind of aged solder joint? Whatever it was, the Trimpot is now fixed, the A6A5 assy re-adjusted, and the 8662A working again!!

HPAK (HP Agilent Keysight) 8662A Synthesized Signal Generator: ref sum loop adjustment, output amp/peak detector repair

Introduced in about 1980, the 8662A is a marvelous machine, it is an ultra-clean signal source, with very low close-in noise. This is the kind of oscillator used as a reference for phase noise tests, narrow channel receiver testing, and so on. Just a quick glance at it inner working, and it is clear that only the most brilliant engineers must have been working on this apparatus in those days. Sure enough, this did not come cheap, about 30 k$ in 1981, something like 75+ k$ nowadays…

Even more interesting, these machines are still in use today, and are still valued for the same reason – hardly any synthesized generator exists that has similar close-in noise.

The block diagram – essentially, there are two ultra low noise switched inductor osciallators that form the reference and output sum loops, together with with some fractional and n divider circuitry, to allow for the fine resolution. The output sections has a heterodyne band, and other bands derived by doubling or dividing the 320-640 MHz fundamental output of the output sum loop oscillator.
8662a block diagram

As with all wonders and complex machine, they sometimes stop working. This one had two issues reported, inaccurate output power (and error #10B – ALC error for some frequencies/settings), and a ref sum loop unlock condition at certain frequencies. The latter issue had been persisting for some time, but the ownder didn’t use it at the critical frequencies, the ALC loop error just recently seems to have come up, for now external reasons, apparently, during a measurement.

(1) The output sum loop A6A5, error #06. This turned out to be a drift issue, of the low noise REF oscillator. This had drifted out of pretune span, and needed just very slight adjustment of the tuning screw (under a heavy metal cover, rear panel of the instrument). Such adjustment is not mentioned in the manual, but easy enough, set the generator to 640 MHz, check that the pretune is properly adjusted (adjustment is done for -3.75 V at 320 MHz), introduce a -34 V tuning voltage signal, and adjust for a frequency a bit above 640 MHz, say, 642.5 MHz, to give the phase loop a bit of room to operate.

08662-60417 oscillator

Great care must be exercised, because the adjustment uses a copper (!) screw – red frame in the above picture, most likely, acting as a capacitor vs. a metal surface inside of the unit (don’t open up these oscillators – they are wraped in Mu metal or similar exotic material). Don’t overtighten the retaining nut, copper is pretty soft.

After that, the ref loop is stable again, even when sweeping through the full band, for some hours.

(2) The more tricky part, as it turns out, the ALC loop. This is implemented on assemblies A4A1 (which has the peak detector), and A4A7, the ALC control, this has a DAC, to set the target voltage, and a regulator, to keep the output signal at the level set by the DAC output.
8662a a4a7 probing
After some time of probing around, and substituting the input of the A4A1 assembly by a calibrated 100 MHz signal, it is clear that the A4A1 peak detector delivers insufficient voltage, and that the voltage is frequency dependent.

A quick glance at the schematic:

8662a output amp

Blue frame – the bandwidth limiting circuit, enganged at below 10 MHz – first guess: something is at fault, in this circuit. Well, unfortunately,after even more probing, and even swapping some transistor (they tested good but you never know) – the focus shifted to the part in the red box. The peak detector diode! This is really bad news, because it is part of the output amp microcircuit, part number 08662-67008. Carefully desoldered, and checked – the detector diode has 0.3 V forward, which is fine, and 0.7 V backwards – this is a far too low reverse breakdown voltage. Something must have happened with this diode, maybe, it is just age. So, I openend up the circuit (the lid can be lifted off with a knife, glued on with silver epoxy), and inspected with a microscope – no obvious defect, all kind of nicely wire-bonded parts on sapphire(?) substrate – nothing I can fix with the tools at hand.


But, as luck comes along, found two of these circuits on ebay. Very mysterious. They look like old and used and ripped out of the boards (with a screwdriver, not by desoldering), and I suspected that the two might just be damaged parts – but 14 days right of return were offered, so not an issue. Even then, it doesn’t really make sense to rip the out of the board, every reasonably skilled and knowledgeable person would rather desolder them carefully, and fit a spare. However, these spare still had the through hold plated vias on the connectors, from the board. Glad the feed-throughs at the bottom of these microcircuits are so sturdy! Well, the only explanation I have – someone saw the golden parts, on some odd circuit boards, and only wanted to keep the gold, but not the boards. Fair enough.

And, this is almost the end of the story, the spares arrived within a few days, and I cleaned them up, and fitted the most “used” looking 08662-67008, and, quite to my surprise – working just fine.

The only thing that remained was the amp bias adjustment, the offset adjustement (both on A4A1) and the CW power adjustment (A4A7) – also checked the other alignments of A4A7, but they were all still fine.

In the end, still one output amp in the box, for the next 8662A, and the current one, back alive.