The test of the eBay frequency divider is probably more relevant for many viewers than the phase noise clock standard (still interesting of course). It would be great if you could test some of the most popular modules from Aliexpress or eBay (for example the ADF4351 or AD8317 boards) and analyze them with your nice test equipment. Because most hobbyists won't have the ability to verify the functionality of the RF boards from eBay over a basic "it seems to work" level.
You don't need much (if any) equipment to know that many of the chinese RF boards are complete junk. With little experience in RF design, just by looking at some of the modules available, you get the awfulness immediately.
In my opinion, Using a R&S FSWP the phase noise was quite good but starting at 80dbc a lot of wideband spurs are present as the filters in the design are not matched to kill the re-entrant nature of the filters along with the spurs being generated by the multipliers bleeding down the chain. As a lab source I had to add a lot of external filtering. Still for the price a nice reference.
Interesting! If I had to guess I'd say looks like they decided to market a device they had perhaps developed/used internally for R&D. I can think of several uses, a nice clean clock for a DDS is one. My guesses for the OCXO are either Wenzel or Rakon :)
Wonderful teardown and review, as per usual Shahriar, I have been looking at signal hound, very interesting company and very exciting products. I especially like the fact that it is made right here in the USofA. I would like to formally request that you make a teardown and review of the Rigol RSA 5032 Real Time Spectrum analyzer. I am very interested in this unit and would like you to compare it between the Sound Hound and the Tektronix PC based RSAs of the same bandwidth capability. I feel this would be very beneficial to those of us who are interested in the higher end of the Amature Radio bands. Thanks again Shahriar for all you do.
At 12.5 minutes, you show the spectrum on a silent spectrum anyzer, and state that you can see the phase noise from this plot. I believe this plot is shaped like this because of the applied window function that is selected. Those sidelobes are a typical thing for signal windowing. So you can't get the phase noise from this view. Please correct me if I'm wrong...
Johannes van der Vegt there is no windowing applied. This is the phase noise response of the Siglent analyzer. This is a swept spectrum, not a wideband FFT.
Are you sure about that? When RBW is < 30kHz on this analyser, sweep mode "FFT" is available. In your measurement, span is only 2MHz and RBW=1kHz. On the left of the plot, I see "FFT". Or am I misinterpreting the screen? (I don't know this specific instrument at all!)
WRT to the power supply of this kind of sensitive instruments, can you somehow correlate the power supply noise to the output clock with a feedback loop? My first guess is that it would still inject harmonics because of the square nature of the power supply waveform.
Powering it up with a 12V battery, would be interesting. It all depends the quality of the wall wart, I guess. On the other end noise created by the wall wart or switching power supply would still be very far from the fundamental.
Question:... I just got a GNSS Dis OSC Clock... its square wave output is terrible looking to me, but it measures correctly on the OSCOPE, and on my OWON multimeter in freq counter mode (not the best test equipment)... is that good enough to sync my equipment, or should I be looking for a SINEWAVE Clock instead?
As long as the waveform has a clear zero crossing with margin, it would work fine. But you can always create a simple LC bandpass filter and get yourself a much nicer sinusoidal signal.
@@Thesignalpath .. Thanks for the quick response... I just through up a tiny 1 min post to show ya what I'm seeing.... Again thanks for the response... th-cam.com/video/U2H8PlMXxx4/w-d-xo.htmlsi=pPPVNSthX0kEG-VM
@Oogobuk Thanks for making the video and nice setup. A few comments which you may find helpful: 1) Make sure that the signal you are looking at is terminated into 50-Ohms. I could not tell from the video, but if you terminate the coax into HiZ, in some cases, you can create additional ripple in the signal. The output driver from the OSC driver assumes 50-Ohms. If your scope does not have 50-Ohm, you can make a little jig and add a parallel 50-Ohm to the scope input, as close to the scope input as possible. 2) The signal shows overshoot/undershoot which can also be easily removed by some low-pass filtering of the signal. Again keep in mind that you have to make your low-pass be 50-Ohm matched and terminate it into 50-Ohm. 3) You could also just buy one of these: (sv1afn.com/en/products/10-mhz-band-pass-filter.html) Keep in mind that this too must be terminated into 50-Ohms.
@@Thesignalpath Yes, thanks alot, I'll be exploring thoes 3 steps... I think the 0 volt crossing may be fine (the counter does read it)... I'll try the lowpass filter first... I was thinking some caps would smooth it out... or runnint it through a fast transistor may regenerate it pretty well too?...
With fixed frequencies like here, it's easy. Distort the sine wave, e.g. make square wave out of it or use non linear components like diodes and then filter the wanted harmonic out of the frequency comb. Look at the schematics of HP-8656B, it has discretely built frequency multipliers inside to get 800MHz, IIRC.
Hi Shariar, You can actually validate the phase noise of this source with the Rohde & Schwarz FSWP8 + Option FSWP-B60. Phase Noise @1 GHz carrier: 1 Hz is -56 dBc 10 Hz is -88 dBc 100 Hz is -116 dBc 1 kHz is -143 dBc 10 kHz is -166 dBc 100 kHz is - 173 dBc Higher offsets are almost at thermal noise (limit) at 22 degrees Celsius.
HP 8662A/8663A has better phase noise ;-D Phase noise measurement with Sigilent @12:50: Marker is at -58dBm with RBW of 1000Hz, 10*log(1000Hz RBW)=-30dBm for 1000Hz RBW, gives -88dBm/Hz phase noise. Reference that to dBm of center peak, gives you dBc phase noise
With a clean source I believe you can improve the PN floor of your MXA by using NFE, would love to see that if you have time - literature.cdn.keysight.com/litweb/pdf/5990-5340EN.pdf I had a play with it here on data from a Signalhound SA124B www.eevblog.com/forum/testgear/phase-noise-head-to-head-siglent-ssa3032x-vs-hp-8566b-vs-signalhound-sa124b/msg1628197/#msg1628197
Although thats a lush device, It only runs at 1GHz.. This must still be akin to an ants technology to a high frequency God like you. The Signal Path, Ripping the universe 42THz at a time ;P
The test of the eBay frequency divider is probably more relevant for many viewers than the phase noise clock standard (still interesting of course). It would be great if you could test some of the most popular modules from Aliexpress or eBay (for example the ADF4351 or AD8317 boards) and analyze them with your nice test equipment. Because most hobbyists won't have the ability to verify the functionality of the RF boards from eBay over a basic "it seems to work" level.
You don't need much (if any) equipment to know that many of the chinese RF boards are complete junk. With little experience in RF design, just by looking at some of the modules available, you get the awfulness immediately.
In my opinion, Using a R&S FSWP the phase noise was quite good but starting at 80dbc a lot of wideband spurs are present as the filters in the design are not matched to kill the re-entrant nature of the filters along with the spurs being generated by the multipliers bleeding down the chain. As a lab source I had to add a lot of external filtering. Still for the price a nice reference.
Interesting! If I had to guess I'd say looks like they decided to market a device they had perhaps developed/used internally for R&D. I can think of several uses, a nice clean clock for a DDS is one. My guesses for the OCXO are either Wenzel or Rakon :)
Or Valpey Fisher?
The lack of warm up or ready status indicators is especially surprising given the OCXO.
Does it matter for phase noise?
No, just for frequency stability.
Wonderful teardown and review, as per usual Shahriar, I have been looking at signal hound, very interesting company and very exciting products. I especially like the fact that it is made right here in the USofA. I would like to formally request that you make a teardown and review of the Rigol RSA 5032 Real Time Spectrum analyzer. I am very interested in this unit and would like you to compare it between the Sound Hound and the Tektronix PC based RSAs of the same bandwidth capability. I feel this would be very beneficial to those of us who are interested in the higher end of the Amature Radio bands. Thanks again Shahriar for all you do.
Can please you share part numbers of ultra low noise regulator and frequency multipliers?. If they are visible 😀
At 12.5 minutes, you show the spectrum on a silent spectrum anyzer, and state that you can see the phase noise from this plot.
I believe this plot is shaped like this because of the applied window function that is selected. Those sidelobes are a typical thing for signal windowing. So you can't get the phase noise from this view.
Please correct me if I'm wrong...
Johannes van der Vegt there is no windowing applied. This is the phase noise response of the Siglent analyzer. This is a swept spectrum, not a wideband FFT.
Are you sure about that? When RBW is < 30kHz on this analyser, sweep mode "FFT" is available. In your measurement, span is only 2MHz and RBW=1kHz. On the left of the plot, I see "FFT". Or am I misinterpreting the screen? (I don't know this specific instrument at all!)
That FFT is on a narrow capture spectrum. Not the entire viewed span. You can see the sweep happen across the screen.
What sort of IM product was that between 4 and 5 GHz?
WRT to the power supply of this kind of sensitive instruments, can you somehow correlate the power supply noise to the output clock with a feedback loop? My first guess is that it would still inject harmonics because of the square nature of the power supply waveform.
Powering it up with a 12V battery, would be interesting. It all depends the quality of the wall wart, I guess. On the other end noise created by the wall wart or switching power supply would still be very far from the fundamental.
Fotofab is a company that makes off-the-shelf foldable solder-down EMI shields.
Wait... You're still around?! I naturally assumed you dropped off the face of the earth when the videos stopped on your channel. P
I'm not dead, I just mutated into a software developer (so.... close?).
Well that'll do it. haha Either way, love your channel - you're one of the reasons I started making videos. :D
At work: "Hey do you know linux?"
Me: "A bit"
And here we are.
Very interesting! You are the best!
Question:... I just got a GNSS Dis OSC Clock... its square wave output is terrible looking to me, but it measures correctly on the OSCOPE, and on my OWON multimeter in freq counter mode (not the best test equipment)... is that good enough to sync my equipment, or should I be looking for a SINEWAVE Clock instead?
As long as the waveform has a clear zero crossing with margin, it would work fine. But you can always create a simple LC bandpass filter and get yourself a much nicer sinusoidal signal.
@@Thesignalpath .. Thanks for the quick response... I just through up a tiny 1 min post to show ya what I'm seeing.... Again thanks for the response... th-cam.com/video/U2H8PlMXxx4/w-d-xo.htmlsi=pPPVNSthX0kEG-VM
@Oogobuk
Thanks for making the video and nice setup. A few comments which you may find helpful:
1) Make sure that the signal you are looking at is terminated into 50-Ohms. I could not tell from the video, but if you terminate the coax into HiZ, in some cases, you can create additional ripple in the signal. The output driver from the OSC driver assumes 50-Ohms. If your scope does not have 50-Ohm, you can make a little jig and add a parallel 50-Ohm to the scope input, as close to the scope input as possible.
2) The signal shows overshoot/undershoot which can also be easily removed by some low-pass filtering of the signal. Again keep in mind that you have to make your low-pass be 50-Ohm matched and terminate it into 50-Ohm.
3) You could also just buy one of these: (sv1afn.com/en/products/10-mhz-band-pass-filter.html) Keep in mind that this too must be terminated into 50-Ohms.
Did you see my last comment?
@@Thesignalpath Yes, thanks alot, I'll be exploring thoes 3 steps... I think the 0 volt crossing may be fine (the counter does read it)... I'll try the lowpass filter first... I was thinking some caps would smooth it out... or runnint it through a fast transistor may regenerate it pretty well too?...
Thanks agsin Shahriar!
It’d be nice if it had the option of phase locking the 100 MHz XO to a 10 MHz input, with say a 1 Hz loop BW.
How does an x5 frequency multiplier work? I.e., what would the circuit diagram for it look like?
With fixed frequencies like here, it's easy. Distort the sine wave, e.g. make square wave out of it or use non linear components like diodes and then filter the wanted harmonic out of the frequency comb. Look at the schematics of HP-8656B, it has discretely built frequency multipliers inside to get 800MHz, IIRC.
How does a x5 multipler work? "Works just fine!" :-)
Hi Shariar,
You can actually validate the phase noise of this source with the Rohde & Schwarz FSWP8 + Option FSWP-B60.
Phase Noise @1 GHz carrier:
1 Hz is -56 dBc
10 Hz is -88 dBc
100 Hz is -116 dBc
1 kHz is -143 dBc
10 kHz is -166 dBc
100 kHz is - 173 dBc
Higher offsets are almost at thermal noise (limit) at 22 degrees Celsius.
Hi, long time follower. Can you tear down a TWTA? Thanks for the great content!
HP 8662A/8663A has better phase noise ;-D
Phase noise measurement with Sigilent @12:50:
Marker is at -58dBm with RBW of 1000Hz, 10*log(1000Hz RBW)=-30dBm for 1000Hz RBW, gives -88dBm/Hz phase noise. Reference that to dBm of center peak, gives you dBc phase noise
Some of these cheap eBay RF boards have fake via stitching. Most of the vias are only on the copper layers (no drill holes).
I know Im gonna love this video. Thumbs up before watching.
👍👍
Interesting bit of gear.
I wouldn't say so. Just a few not so special components sold for thousands of dollars.
like your work, very exacting,
wow
With a clean source I believe you can improve the PN floor of your MXA by using NFE, would love to see that if you have time - literature.cdn.keysight.com/litweb/pdf/5990-5340EN.pdf
I had a play with it here on data from a Signalhound SA124B
www.eevblog.com/forum/testgear/phase-noise-head-to-head-siglent-ssa3032x-vs-hp-8566b-vs-signalhound-sa124b/msg1628197/#msg1628197
Although thats a lush device, It only runs at 1GHz.. This must still be akin to an ants technology to a high frequency God like you.
The Signal Path, Ripping the universe 42THz at a time ;P
He may as well be sometimes.
$1500 btw
Let me clear that up: this is 23 minutes on a fixed sinusoid whose main attribute is that it is perfectly encoding nothing? Still watching.
Welcome to metrology, it's fascinating. (no sarcasm)
10 dBm=10 mW
First :-D