TSP #35 - Teardown, Analysis and Repair of an HP/Agilent 5347A 20GHz Frequency Counter & Power Meter

In this episode Shahriar takes a close look at an HP/Agilent 5347A 20GHz Frequency Counter and Power Meter. This defective unit does not provide any frequency information from Input 2 which is rated to operate up to 20GHz. Before the teardown and repair attempts, the principle operation of the instrument is reviewed.

The properties of a Step Recovery Diode (SRD) is presented along with the theory and practical aspects of generating a frequency comb. The heterodyne architecture of the frequency counter is explained in detail with the mechanism of detection and calculation of the input frequency.

During the teardown of the unit the synthesizer board, motherboard, power meter reference board and the main RF assembly are shown. The schematic of the synthesizer board and the RF board are also described. The defective component is identified and examined under the microscope. The slides for this episode can be found at The Signal Path website.

The Signal Path

[alexschoenheit]

On the sampler: the input at the right is connected by a ribbon to a black small capacitor chip to ground. Then, a long ribbon connects to a single horizontal line, which is a microstrip line above the ground. Then, a third ribbon is feeding the SRD diode, which is the blue square chip on the right of the microstrip line. On the leftmost side of the microstrip - a 50ohm termination to ground (dark square). Since we are dealing with a very short pulses here, every bit of ribbon interconnecting is an inductance. Therefore all 3 ribbons are inductors with a tapped off capacitor to ground between the first 2 inductors. Effectively - this is a matching circuit, the longer ribbon used here as a impulse forming choke. In the middle - a very thin vertical slotline is coupled to the microstrip. This 90 degrees microstrip-slotline arrangement forms an electromagnetic mutual coupling. The slotline, being a balanced line by its very nature serves here as a wideband BALUN, feeding diode (NOT FET!!!) sampler in the top middle. Usually, its enough to drive the SRD with +17-20dbm to get good harmonic rich output. But here, HP guys had a serious dilemma: To avoid spurious mixing, and possible frequency measurement errors, they needed a nice and clean pulse to drive the sampler and this forbids any reflections going back&forth on the feeding microstrip. Because of that, the microstrip is very well terminated at its end. In such a case, MOST of the energy is dissipated in the termination, so relatively low amount of energy is coupled to the slotline. But in order to minimize the mixing loss (sampling mixers usually provide -20-25db conversion loss which is already bad), the pulse must be strong enough to drive the schottky sampling diodes ON properly. Because of the low percentage of pulse energy coupled to the slotline (low efficiency coupling), the driving power was increased from 50-100mW to huge 300mW (+25dbm).

[rubusroo]

Off sick today & I can frankly think of nothing I'd rather do than to watch one of Bell Laboratories finest electronic engineers for an hour. I may not understand everything yet, but the way he explains everything in minutia detail is a massive help to me. Thanks for these videos +TheSignalPathBlog.

Guys like you +EEVBlog, +mjlorton & +mikeselectricstuff don't get the recognition you deserve considering how many budding EE students/hobbyists you help.

[michaelhernandez]

By far the best electronic repair videos on YouTube

[Richardincancale]

Fascinating architecture in that instrument - thanks for explaining it. Use of the comb frequency generator reminds me of the classic Barlow Wadley Triple Loop architecture used in the 1950s RA17 HF receiver to guarantee that all three IF frequencies stay locked even if the fundamental LO drifts.

[redliquid]

Amount of knowlegde, that you're sharing with us is outstanding. Honestly - you have the most valuable channel on YouTube in electronics subject. Period.

[bulltza]

The only thing I think I am missing is the initial symptom before attempting repair. Don't forget the next time!

Amazing video, keep it up! You can't imagine how much I am learning from your videos

[katelikesrectangles]

I had such an "OH!" moment from your FFT explanation. Amazing design. Thank you so much!

[wither]

Brilliant as usual. Always love the theory part as well as the practical. MORE!!!

[ElGatoLoco]

I don't know jack about electronics but I find this fascinating for some reason.

[capriracer]

Some interesting additional information on SRD's is included in Hewlett Packard application note 918 entitled "Pulse and Waveform Generation with Step Recovery Diodes". In the introduction to this app note, 928, 983, 984, and 989 are also referenced.

[jovianconflict]

I really like these in depth videos. Good work!

[vaualbus]

Arleady a new video! You have the power of upload a vide when i'm at university and cannot watch immidantly! I have to wait....

[capriracer]

Also, Since the class B amplifier only has one transistor, you are only amplifying one half of the cycle. If this were in the traditional "push pull" configuration, you would have amplification at both ends of the waveform. However, since this is an RF power amplifier being fed into a step recover diode, the distortion present is not necessarily a problem. The most important thing in this circuit is to drive the single transistor into full saturation so as to attain the power needed to trigger the IF sufficiently after leaving the SRD.

[WAX]

I can't find the Wiha pliers set that you showed. Please provide the part number and source if possible. They look like a great set to add to my workbench. I already have your same screwdriver set. 

I absolutely watch every video you produce. They are the single greatest source of continuing education available to me. Thank you for all you do for our community.

[glenslick]

It might have been interesting to show a comparison of probing the broken unit against the working unit at the same point under the same conditions.

[ThingsWhichArentWork]

What a superb video and excellent explanation of everything. I wish I'd had you as a lecturer back in my university days. Thanks for all the work you put into making these videos - it must take you ages. I'd suggest taking a look at a bit of software called ManyCam for doing screen capture (with audio, webcams, media clips etc alongside it) - it's somewhat buggy but very feature rich and there's a free trial (for up to two input sources) - You do need a PC with a hell of a lot of grunt to capture at 1080p from multiple sources however.

[robbowman]

Great video, thanks! Fingers crossed that you can find a donor part and complete the repair. Or even better, how about characterising your working one and making a new module? That would be a seriously good learning exercise, and I think it should be feasible given that 20Ghz is nearly DC these days :)

[SetMyLife]

My answer: The 14dBm signal coming into the BJT base has a huge swing. That induces large currents in the base pin. The transistor then operates over ranges much larger than it's linear region, so it's non-linearities come up.

[raymundhofmann]

What about making a replacement circuit out of available microwave parts? Even if it would be  not fully up to spec, it would give a great video.

[HungHoang-jcpo]

It is really good tutorial. I have exactly the same unit and have slight same problem with no output voltage and current. I did full self test and had pass. But no output. Do you have any clue for the issue? Thank you