No Script, No Fear, All Opinion
RSS icon Home icon
  • EEVblog #261 – Marconi 2023 RF Signal Generator Teardown

    Posted on March 22nd, 2012 EEVblog 26 comments


    Teardown Photos HERE
    What’s inside the Marconi Instruments 2023 9KHz to 1.2GHz Signal Generator?

    Be Sociable, Share!
    • dave-o

      Dave, Aeroflex makes super high end communications analyzers. Really great kit, but crazy expensive. More than a nice car kind of prices. But totally worth it if you need the capabilities. Not a bad place for Marconi to land. Those guys really do grok RF.

    • wilfred

      Your videos are really getting to a high standard. I like the way you have introduced the block diagram overlay.

      I think you should add a few seconds at the end of a teardown to show the device still works. Not that I have doubts but it would be a good way to finish.

      Could you show the bandwidth of the Rigol too?

    • http://alternet.us.com hpux735

      That copper pipe going to the output is “hard pipe,” a kind of coax.

      That is just an awesome piece of kit. Those RF systems make me happy.

    • Fox

      The tubes on the output N-connector is actually a semirigid coaxcable, so no extra coaxcable inside the tube.

      Exellent video as ever, of truly pornografic RF test gear.

    • http://wardyprojects.blogspot.com Adam Ward

      Great video Dave, one of your best ever IMHO. Fascinating to see such well designed and engineered kit as this.

      Actually we have an Aeroflex/IFR testing unit where I work (a GSM RF analyser to be specific). Sadly it’s a nasty, obsolete and unsupported wreck and a real pig to work with. It has no buttons or dials and no display — everything is done over some evil, opaque, binary serial port protocol which is also undocumented!

      Your video made me jealous. After all, I’m not allowed to take our Aeroflex apart :(

      Maybe when nobody’s looking…….

      • Tony

        What is it with high-end gear, their performance is just stellar but the “user interface” always seems to be like an afterthough – either it is binary knobs, some arcane protocol or some horrible PC software that crashes if you happen to *think* something it doesn’t like.
        Damn, I still have nightmares of a really high-end Anritsu test system I used briefly some ten years ago … Hrr…

    • JPL

      As mentioned before, that “tubing” is probably RG402 or RG401 semirigid.

      I really dig the construction, but it’s fairly low frequency stuff. The central traces look like microstrip since the “roof” (so to speak) is too far away to make a proper stripline.

      The parts density is insane in that lower RF section. In an actual microwave device you’re likely to see a lot more of your passives made from copper structures.

    • Jay Ts

      Fantastic video, Dave! I agree that this is one of your best so far. I wish everything I owned were made as well as that product. Too bad it doesn’t go down to 1 Hz, because that means I have no use for it. (haha)

    • Bob

      Dave–

      Talk to some of your ham radio friends –

      As mentioned, the “pipe” to the output connector looks like .141 semirigid line. The interior is silver plated, with a special solid dielectric, and a silver plated centre conductor. (Also available in 0.085 and 0.25 diameters.) Good to many GHz. One of the fun parts of working with semirigid line is bending it without causing kinks.

      The long strip of circuitry feeding the output amp is the set of attenuator stages, with those stages shielded from each other — think about it — you need fancy shielding to be able to attenuate a signal reliably from +13 down to -130 dBm!

      You can do 1.2 GHz with FR4 — but the board has to be made by a quality house. Folks use FR4 all the time for Wi-Fi (on both 2.4 and 5 GHz). Yeah, a lot of test gear will use Duroid (Rogers PTFE substrate) material for the really critical sections, or fancier substrates such as ceramics.

      Even back in the early 90′s, 1.2 GHz was fairly low frequency stuff.

      Those conductive gaskets are critical to proper device operation — I’d be interested in knowing how close to spec the device is after reassembly!

      Thanks for the teardown! Keep ‘em coming!

    • Hendi (dg3hda)

      I guess that the output section is a bit different. The Beryllium device might be the final amplifier stage, RF output crossing the diecast part. The longish PCB with the sections looks like a output attenuator: each pair of relays (powered by the DC feedthroughs) selects the signal path to go either straight thru the microstripline or thru the “T” style resistor bunch. Those might do the big steps, and a PIN diode attenuator somewhere the smaller steps. Separating the attenuator from the rest and the separator steps between themselves reduces leakage of signal – which I guess is also the main reason for the semirigid output line.

    • allan

      Awesome teardown, mate.
      I luv it!
      Did you manage to get the signal generator reassembled and working again? If so,
      I’d be mighty impressed, with all those screws. Learned about RF design from watching this video–Thanks.

    • Joop

      Nice teardown. I actually own the same generator. It came in not working. That is, the keyboard was not responsive and an error was in the display. It turned out it was set to operate on the external 10MHz reference. So first I had to take apart the front panel keyboard. The keys are those rubber type things that press down on a PCB. The rubbers were cleaned as well as the PCB. I also used some rub-on silver compound on the contacts.
      After that it was possible to operate the generator again and disable the external reference setting. Most stuff seems to work now, although the LF output is not working. Modulation does work, so maybe somebody blasted some output buffer or something.

      As to why the keyboard went bad, I do not know. A nasty environment perhaps? Or maybe the keys were never used and the generator was always controlled externally? Hopefully your keys are all ok.

      One interesting thing was to see the reverse power protection had tripped 7 times in its life! A good thing the poor generator has such a feature.

      • JPL

        Most likely, those 7 times were when someone turned the output on without a load.

    • Bill Clay

      What is the purpose of the beryllium oxide?

      • Chrs

        Some RF components use beryllium oxide as ceramic substrate material because it has high low thermal conductivity.

        • Chrs

          Sorry, “high thermal conductivity”, of course.

    • huh

      Speaking of signal generators for mere mortals, have anyone tried those cheap AD9851 DDS modules on Ebay? They’re about US$20 shipped: less than the analog, and sadly obsolete, MAX038.

    • Murphy

      Porn indeed … love your videos. Thumbs up!

    • Tony Darby

      Great teardown with very clear defintion video.
      Now I know you are not into amateur radio, but it would be nice to see you put this fine instrument to practical use.

    • Hendi (dg3hda)

      Speaking of Joops external references…
      Rubidium standard cough cough

      • Joop

        These are 2nd-hand quite affordable nowadays. One might even buy two to have a spare one in case it brakes down. I think I used it (or some other 10MHz unit) in the process of identifying all the problems with the unit.
        Before repairing the front keyboard I also used a GPIB interface to see if the generator worked OK (signal wise). Note that a for the price of a computer GPIB interface you can easily buy 2 Rb clocks. RS232 was not an option at the time since one has to configure either GPIB or RS232 for the external control and GPIB was configured on mine. Now I have keyboard control I can select either interface.

    • Bill

      Did you get it put together and working? LOL
      Thumbs up!

    • Chris Jones

      The beryllium oxide transistor would be the output amplifier, and poking it with a screwdriver would be a bad idea. All of the relays (RLC – RLL) on the smaller board on the underside are individual stages of resistive attenuators (pi-attenuators) which can be individually bypassed by switching the relays, to get different attenuation values. The signals going through the feedthrough capacitors to the ribbon cable would just power the coils of the relays.

      What I would really like to see is the low-noise tray of a 2042 signal generator. That must have something very clever in it to have phase noise so much better than the other Marconi signal generators.

      oh, and:
      http://en.wikipedia.org/wiki/File:Semi_rigid.jpg

    • Sam Reaves

      Hi Dave,

      Just curious. For your Agilent scope mandwidth test I was wondering if you set the starting frequency reference amplitude to 50KHz and then run out to 500MHz what bandwidth to you get for your -3dB point? Normally the scope bandwidth tests that I have done normally use 50KHz as the reference point. You should get a little better bandwidth measurement that way.

      73′s

      Sam
      W3OHM

      P.S. I received the uCurrent. Thanks for a very useful piece of equipment.

      • Greg

        I do not remember Agilent procedure, but Tektronix recommend for i.e. TDS 5000 series (500 & 1000 MHz) BW reference frequency 10MHz or less, so not make a difference. In this case I would care about short, good quality, thick RF cable and/or power loss compensation with the power meter and divider just on the input of the scope. But it was not proper BW measurement, just demonstration.

    • risto

      I own a third hand Marconi 2019 /up to 1.04GHz/ sig gen. It is a year younger than me /1982/ and after a series of repairs, it still works.
      lots of standard components inside…BFR9x transistors, 74LS logic, 74S, CD4000,8085 CPU, some early canned Miniciruits ring diode mixers…all in all off the shelf easily replaceable parts.

      The previous owner has replaced all of the electrolytic caps on the power supply and processor boards with new ones, and, from what he told me, it seems that the instrument wiped out its firmware somehow, so he had to reprogram all those uv erasable memories in here. Go figure…

      There were some wild variations in the output power, highly frequency dependent, too. I cracked it open and after hours of meticulous probing with a RF voltmeter, it turned out that some of the surface mount bypass caps on the output variable gain amplifier board had deteriorated completely for no obvious reason… unreliable multilayer technology back then, i would guess.
      Also, one of the VCOs had particularly ”loud” uncorrelated random spikes of phase noise. Another cracked cap, i thought, but no, i replaced all of them and the noise was still there… Since it too has those printed inductors, i thought that maybe it is some structural inconsistency in the vco board stackup that produced that noise. So i took the entire vco board out and tossed it in the oven at 150C overnight. The mysterious noise has never returned ever since.
      Then, one of the step attenuator pads wasnt being activated correctly, i had to readjust the electromagnet switching it in or out. Another pad was way off at 1GHz, so i balanced that out as well…not to mention that FM modulation calibration was at some frequencies off by as much as 20%, but it’s bang on now. I used a spectrum analyzer with some built in analog demodulator goodnes to realign the 16kg beastie.

      it seems to work now…at least it has an OCXO reference inside.

      Apart from that,looking into Dave’s video, the rainbow ribbon cables, the berylla warning sticker and the Made in UK plate have made it way into the mid 90s…

      i now have a dead USSR 1.2GHz signal generator needing repair…