HP 8640B signal generator repair

As sold, not working. An oldy but a goody.

I seem to have gone down a rabbit hole that a few others I read about have travelled. It goes something like this. You have the remote need for some esoteric test equipment to work on radios and stuff and you are overcome by the shear beauty and horrifying complexity of HP’s equipment from the 70’s, 80’s and 90’s.

Truth be known, I had a frequency counter and a signal generator already but they were not what I wanted and so, during a car boot sale in August I purchased a HP 5328B for an amazingly reasonable amount given what they cost when new (this too seems to be a point many in my boat make). I couldn’t wait to get home and take a look inside. It has the crystal oscillator oven and seemed to me to be a lot larger and make a lot more noise than would be expected to perform this task but it was gorgeous inside with it’s 1970’s vintage gold trace PCB’s and red 7 segment display. This immediately led me to start searching for more HP equipment, the next being a signal generator.

I was lucky to find one for sale on eBay not working from a chap here in Perth and so put and offer in for this HP 8640B and picked it up. Seem these units are used as a bit of a standard when it comes to quality of signals produced.

After getting it home and plugging it in, sure enough it did not work – sure the fan came on but there was nothing else working, no display, no needle movement – nothing.

POWER SUPPLY

First things first I ripped the top and bottom covers off and took a look at the power supply boards, all but 1 of the red LED’s were on, the one off was (from memory) the +5.2V rail. I noticed that the 0.022uF “Rifa” capacitor across the 240V in line had at some stage blown up and et out the magic smoke – this seems to be it’s normal mode of operation at this age and so I noted I should change that at some point but it isn’t going to stop the thing from running.

Puling the A12 rectifier assembly I tested the multitude of diodes being used in the rectifiers. All these seemed to test fine so that wasn’t the problem. I removed the 3 voltage rail cards and re seated them, and I also removed the 4 power supply pass transistors and re seated those as well. Switching on after this finds all voltages are now present and the front numerical displays started working, however, despite my knob twiddling only had a fixed range of frequencies, ie, changing the range selector had no effect, also, the FM modulation didn;t seem to work and the was no power output displaying on the needle.

One of the many times I had this on the bench

Later, a bit more fiddling with pass transistors finds one of the contacts on the power supply board that connects to one of the legs of a pass transistor had failed (broken) so I I’ll need to get a new connection for that at some stage as it is flaky.

Power supply pass transistors

ANALOGUE SWITCH LOGIC CONTACTS – RANGE & DEVIATION

After a bit of reading I find that a common problem now days is the little contacts that rotate when you move the range switch, deviation and output level are getting to the age where the plastic welding holding them to the plastic switch disc is starting to fail. Sure enough, mine had the same problem and so it was time to pull these out and undertake a repair.

The first problem I encountered was the Allen key grub screws holding the knobs on these controls to the shafts on the switch mechanism. I loosened all but one of the grub screws (It always just one isn’t it) in the knobs, the last one ended up stripping internally so the hex key just spun and so, despite my trying to glue the key in there with super glue and Devcon I couldn’t get it undone. These grub screws are hardened of course so no normal drill can drill them out – I had to order a special 1.5mm solid carbide drill bit from RS Components to have any chance of drilling it out – this tiny bloody grub screw put a stop to the project for 2 weeks while I waited for the drill bit to arrive. after arrival of the drill bit I was able to drill out the grub screw – I’m sure this must happens all the time as the grub screws are very small – why they didn’t go larger I have no idea – seems plenty of room to go up a size or 2 on the range switch knob.

Range and Deviation selectors in the instrument

After finally getting the combined Range and Deviation switch mechanism out I found I was missing 5 of these little contact figures across the 3 separate rotating plastic discs which had come loose and fallen off over the years into the bowels of the machine and despite my searching and shaking of the instrument, I could only find 2 rattling around in there so I had to make 3 new ones.

Oh dear – none left on this at all!

I ended up visiting my local electronics store and bought a range of things that i hopes could have internals that would work as replacement contacts, this range of thing included an edge connector block which, after pulling apart had an almost endless supply of contacts if you undertook some modifications (see below). These replacements are made of a thicker gauge metal which may be a problem and one must remember to remove any sharp edges that may exist on the contact fingers (I used a very fine Arkansas sharpening stone) so as to not damage to PCB that it contacts with.

After using a 2 pack epoxy to glue the old fingers on again and also the 3 new made fingers I had the range function working again and the modulation selector seemed to work as well.

CRACKED RANGE GEARS

This too seesm to be a common fault in these machines. Apaprently the plastic used was called Delron and that has a tendanmcy to shrink over teh years causing stress cracks as teh bub the gears are cast on is of brass which will not shrink and something has to give.

One of the bevel gears was badly cracked so I broek it completley and then glue it back again using epoxy. After drying one of teh teeth just wasn’t quit liinging up well enough to mesh priopoerly and was jaming thinsg up, a littel bit of a file on teh offending tooth to widen the gap was just enoiugh to gain clearance needed and it all roated as well as can be expected.

As a side note, remade solid brass gears are now available online which may be on the Christmas list this year.

Range and deviation assembly detail

OUTPUT POWER

Even after getting the range switch working the gauge showing output power was not registering anything (it worked when selecting AM or FM modulation % but not output level voltage). I tested the AGC board and its didn’t seem to be giving me the correct readings. A bit if hunting around and tracing resulted in me pulling out the output power selector/attenuator assembly which had another analogue switch rotor with the same little contact fingers on it. In place inspection teh contacts looked to be there but on removal I see that 3 of them were broken and wouldn’t have been making good contact. Seems these contacts feedback into the AGC and which also feeds to meter and so its an important part of the output circuit.

Gluing on these contacts again and after reinstallation it all seems to work again, needle moves with changes in the output power vernier, annunciator lights work as expected. YOU RIPPER!!

Working in all it’s glory. Yes I know the 8640B also has a counter built in but where is the fun in that.

CONCLUSION

This instrument is the first one I worked on and I learnt a lot. Just getting my head around the HP Service manual was a steep learning curve but I’m happy to have a fully functional signal generator. I sold my old signal generator and frequency counter and have these HP machines proudly in my lab. I plan to get a few more HP instruments, would be good to get a Spectrum Analyser so am on the look out for a broken one to try my luck again.

Amateur radio interference to NBN (VDSL2) connection in Australia

From time to time I transmit on the 80m and 40m band. The frequencies in these bands overlap with the frequencies utilised by the NBN here in Australia. The NBN in Australia utilises VDSL2 often using the existing copper phone line for the final stretch from the node to the premises – in my case the final 700m.

When I first started transmitting I noticed my internet connection would drop. After a bit of experimentation I found that I can see the interference in the admin screen of my modem (FRITZ!Box 7490).

Here is what the SNR graphs look like steady state

When I transmit on 80m or 40m there is plenty of interference introduced by the transmitter as seen in the following screenshots.

Test transmission on 80m
Test transmission on 40m

I have just finish studying for my Advanced licence and so have had the ear of a few people with the extensive knowledge on all things interference lately. They kindly suggested winding the phone line around ferrite rods which I tried (clump of 4 rods with 5m of phone line wound around it) but the internet still drops when I test transmit 100W CW, have also added various values of capacitor in shunt between red and green phone line. As can be seen from the following screen shots, the various capacitance have some fairly dramatic impacts on the bandwidth that is available to the connection and hence what speeds I get on my VDSL2 (NBN) connection.

20,000pF resulted in 11.05MBits Down and 1.45kBits Up

20,000pF in shunt – really knocks the internet connection around. notice only have 1 download and 1 upload band

10,000pF results in 14.MBits Down and 3MBits Up

1,000pF results in 26MBits Down and9.7MBIts up

33pF its getting back to normal now 41MBits Down and 11.5MBits Up.

I hope there are others out there that find this page and it can assist in identifying issues you may have and what doesn’t seem to work for me. If you have any suggestions on what will fix this persistent issue I’m all ears.