Author Topic: Probing high frequency sections of an rf board with a spectrum analyser  (Read 3292 times)

0 Members and 1 Guest are viewing this topic.

Offline jeremy

  • Frequent Contributor
  • **
  • Posts: 909
  • Country: au
Hi all,

Bit of a newbie in this area so I'm looking for some advice. I'm trying to fix a spectrum analyser, and I've narrowed the problem down to the RF part. Is there a good way to probe these transmission lines using another spectrum analyser? In particular, I'm curious about how to tap into impedance controlled traces correctly.

I'm assuming the answer is "go get an active probe", but I'm kind of hoping that there is some other tricks which can get the job done.
 

Offline SaabFAN

  • Frequent Contributor
  • **
  • Posts: 665
  • Country: de
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #1 on: November 23, 2016, 04:14:14 pm »
I believe the easiest method is to use a piece of Coax with the outer shield removed but no metal showing. With that you can get near the traces and look at signals. You won't get accurate power measurements, but you should be able to see if the signal is there and if it is distorted (for example by a damaged component).

For building an active probe, you might want to check this thread: https://www.eevblog.com/forum/projects/homebrew-spectrum-analyser-high-impedance-probe/

Online tggzzz

  • Super Contributor
  • ***
  • Posts: 10272
  • Country: gb
    • Having fun doing more, with less
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #2 on: November 23, 2016, 04:34:56 pm »
The first thing to do is obtain a service manual, and see what that tells you to do.

I'm not sure how "probing the transmission lines" will help you. It would help if you outlined the circuit and exactly what you want to determine and why.

If the transmission lines are coax transmission lines, then disconnect them and connect them directly to your known good SA.

If the transmission lines are on a PCB, then you might use a *10 low impedance probe connected to your known good SA. That can be professional or homebrew (470ohm resistor in series with coax); in either case you will change the impedance and cause reflections, but that might not be important for these tests.

If it is "in the RF section", then there's a good chance the mixer diodes are blown, either due to much power or due to any DC.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline metrologist

  • Super Contributor
  • ***
  • Posts: 1802
  • Country: 00
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #3 on: November 23, 2016, 04:43:10 pm »
I recall making a passive low impedance RF probe using a few inches of semi-rigid coax.

Trim the end so you have a probe tip with the center conductor. About an inch up, strip half the outer shield on one side to expose the center conductor and cut out a section that can be bridged with a SMD cap, then solder a 50 ohm smd resistor from center (cap) to outer shield, on the analyzer side.

And here is another: http://www.emcesd.com/1ghzprob.htm
« Last Edit: November 23, 2016, 04:49:09 pm by metrologist »
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 8722
  • Country: gb
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #4 on: November 23, 2016, 05:19:21 pm »
As others have suggested, actually touching a tuned stripline is with a metal probe is not usually a great idea. Striplines and similar RF stages ar easily detuned by metal touching them. Non contact methods may be used and for that I have used capacitive coupled probes but even they can detune a stage and effect the reading on the test eqiuipment.

I would normally be checking the input and output stages of a tuned stripline at an appropriate point to determine the behaviour of the stripline and 'buffer' stages. That may be done via non contact probes, high impedance probes or a low impedance probe if the buffer stage can handle the additional load and its effects taken into account where measurements are made.  The cheapest non active high impedance contact probe just uses a coaxial cable with the end split out to braid (earth) and a non inductive 1K resistor as the probe from the centre conductor. Calibration will be non existent but you see the signals and it will work as a relative level indicator on similar impedance circuits. Keep all lead lengths as short as possible at microwave frequencies.

As you will know, your Spectrum Analyser has a 50 Ohm input. Whilst this is common for RF work, it is less than ideal when probing inside equipment as 50 Ohms is a low impedance for many RF stages and will adversely effect their behaviour by loading them. The additional capacitance of a coaxial cable can also severely detune circuits or load them. probing RF circuits at any point other than an active buffer output can be fraught with problems.

As for active probes, I own a few. Commercial offerings are expensive as the active electronics inside them needs to provide a pretty flat response over their operating frequency range. DIY active probes may be built relatively cheaply as has been mentioned. The poor er frequency response of simple active probes may be normalised if a tracking generator or broad band noise source is available. Compensations have to be made for the calibration source signal level of course !

Much depends upon why you want to image the spectrum in an RF stage. If it s for the presence f a signal, that is easy. If you actually want an accurate level measurement, that can be harder to achieve. Relative readings are the mid ground where you can decide if the gain of a stage is not as expected. If you have the same of lower signal level at the output of an MMIC gain block amplifier as that at its input, then there is likely something very wrong with that stage ! I should state that an MMIC is often a 50 Ohm impedance device so you can probe input and output quite easily with just a coaxial cable but remember the effect of the additional 50 Ohm loading and potential mismatch standing wave effects of adding the test lead coax to the circuit. 

Finally, the use of an oscilloscope probe in 1:1 or 10:1 mode is tempting but fraught with issues. The oscilloscope probe was designed for a termination impedance of 1 Meg Ohm, NOT 50 Ohms. As such its resistive coax and 10:1 divider is totally wrong for the spectrum analyser and the bandwidth tends to be very low anyway.You can find some active oscilloscope probes that have a switchable 50 Ohm / 1 Meg Ohm output. These will work with a Spectrum Analyser but the bandwidth may be quite limited.

Always remember that active probes are quite fragile and their input signal level limits must be respected or overload will occur resulting in strange outputs (intermodulation) and in some cases, probe amplifier destruction.

Hope this helps a little.

Fraser
 

Offline TheSteve

  • Supporter
  • ****
  • Posts: 3033
  • Country: ca
  • GHz
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #5 on: November 23, 2016, 11:21:25 pm »
Another spectrum analyzer should have a 50 ohm input so in a pinch you can try using some 50 ohm coax with a series 450 ohm resistor as the probe tip. It actually works quite well if you use a decent quality resistor and keep the leads as short as possible. This will also work with the scope that has a 50 ohm termination. This all assumes you're not trying to work into the GHz.
Be sure to watch input levels and DC offsets to protect your working analyzer.

Check out this thread to see how well coax with a 470 ohm resistor(all I had) actually worked compared to passive and active probes.
https://www.eevblog.com/forum/testgear/comparing-an-active-probe-to-passive-probes-for-fun/
VE7FM
 

Offline jeremy

  • Frequent Contributor
  • **
  • Posts: 909
  • Country: au
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #6 on: November 24, 2016, 12:29:07 am »
Thanks all. The SA in question is a low end FieldFox-ish one (don't have the model number on me at the moment). Managed to get hold of an engineer at keysight in their local repair depot who told me a service manual for this device doesn't exist even on his system. The method of "repair" is to just replace the entire board for a flat fee.

Input bandwidth is 9khz - 3Ghz, so I am expecting multi ghz signals. 9khz to about 2Ghz works fine, then the trace drops to the noise floor. So I don't think it is DC related. I was hoping that I could maybe probe around to find the area where the 2-3Ghz band gets stopped. Anyway, I haven't had a great deal of time to look at it lately, was just curious about this particular issue.

The idea of using a bare piece of coax/capacitive probe and holding it near is very interesting, never would have come up with that myself! And thanks for your link TheSteve, very interesting.
 

Offline VK5RC

  • Supporter
  • ****
  • Posts: 2292
  • Country: au
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #7 on: November 27, 2016, 11:07:37 am »
Can you post some photos,  I would imagine all the goodness happens in the first cm or so,  attenuator then mixer/ADC.
Whoah! Watch where that landed we might need it later.
 

Offline jeremy

  • Frequent Contributor
  • **
  • Posts: 909
  • Country: au
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #8 on: November 27, 2016, 11:27:00 am »
Yep was planning on doing that asap, but life is pretty busy at the moment so fun stuff has to wait. It's a bugger to get the thing apart, but I have looked at it before and I don't think it's so simple, certainly not just a few cm of rf. It also has a tracking gen on the same board, and is >2 layer (iirc, there were some patterns of vias which make it look like there is a buried waveguide underneath), so that helps to confuse the issue.

Although I'd certainly be happy to be wrong on this one!  ;)
« Last Edit: November 27, 2016, 11:29:58 am by jeremy »
 

Offline VK5RC

  • Supporter
  • ****
  • Posts: 2292
  • Country: au
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #9 on: November 27, 2016, 11:11:13 pm »
Totally understand re busy-ness of life, that board will be really interesting I suspect. Lots of 'microwave goodness' tucked well away I suspect.
I have 'snowflaked'  my 5 and 10GHz transverters, even a 1mm square bit of copper at those frequencies makes a big impact on transmission line characteristics. If not interpreted very carefully, 'probe' results could lead you down the wrong rabbit hole.
Whoah! Watch where that landed we might need it later.
 

Offline Fraser

  • Super Contributor
  • ***
  • Posts: 8722
  • Country: gb
Re: Probing high frequency sections of an rf board with a spectrum analyser
« Reply #10 on: November 27, 2016, 11:51:02 pm »
I am an RF Explorer owner and have had excellent experience discussing matters with the designer, Ariel.

Ariel has continued his development work and a recent release from him is a very neat EMC pick-up probe at a very competitive price.

See here:

https://www.seeedstudio.com/RF-Explorer-H-Loop-Near-Field-Antenna-(RFEAN25)-p-2720.html

More details here:

http://j3.rf-explorer.com/menu-the-news/174-near-field-antenna-for-rf-emi-emc-diagnosis

and the Draft datasheet is here:

https://www.yumpu.com/en/document/view/56260179/semiflexible/3

Coupling graph to 1GHz for two common PCB types is attached.

Worth $27 me thinks  :)

I have no commercial connection with this product but will be buying one for myself soon.
« Last Edit: November 27, 2016, 11:59:12 pm by Fraser »
 
The following users thanked this post: nugglix


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf