Author Topic: Spectrum Analyzer Rigol TG-815  (Read 7184 times)

0 Members and 1 Guest are viewing this topic.

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Spectrum Analyzer Rigol TG-815
« on: March 03, 2015, 01:39:32 PM »
I have not had any luck figuring this out.  I have a Rigol TG-815.  When I have a direct connection from the output of the tracking generator to the input of the analyzer, everything works fine and I have a relatively smooth line.  When I take this same setup and add attenuation via coaxial attenuators, say 70db, to the point when the signal is close to the noise floor, ripples begin to appear (roughly 10db).  The frequency spread of the ripple changes as I change the coaxial connector length.  Also, it does not matter where I place the attenuators (e.g. at the output of the tracking generator or input to the spectrum analyzer. Any ideas as to what's causing the ripple?
 

Offline ion54

  • Contributor
  • Posts: 34
  • Country: ca
Re: Spectrum Analyzer Rigol TG-815
« Reply #1 on: March 03, 2015, 01:52:33 PM »
Looks to me like a lack of impedance matching. Are you sure your coaxial attenuator is pure resistive?
 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #2 on: March 03, 2015, 01:56:58 PM »
I thought of an impedance mismatch at first as well.  But I'm not sure where it would be coming from.  The attenuators are from mini circuits and rated at 50ohms.  Also, the ripples I'm describing only occur when the attenuation pushed the signal near the noise floor.  Maybe it's a minor mismatch that I'm only seeing when the attenuation is that high.  Not sure though if anyone else has encountered this and has a simple explanation?
 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #3 on: March 04, 2015, 11:23:19 AM »
Maybe an impedance mismatch this is amplified, somehow, as the signal approaches the noise floor?
 

Offline ion54

  • Contributor
  • Posts: 34
  • Country: ca
Re: Spectrum Analyzer Rigol TG-815
« Reply #4 on: March 04, 2015, 11:48:14 PM »
I don't have the capability to test your set-up and the picture is not very good. Could you explain all the steps from full signal to the level you start seeing that amplitude variation? How many attenuators are cascaded. Does the amplitude start oscillating with the first attenuator installed? The noise floor of the Rigol TG815 is about -120 to -130. Where is your signal starting to become affected? If you connect more attenuators are they attached back to back or are you using cables between attenuators? Some more pictures of the set-ups, with better resolution, might help too.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2173
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #5 on: March 05, 2015, 08:22:10 AM »
I've not used a Rigol 815 but I'd expect that this effect will be due to poor leakage/bypass performance inside the analyser when the TG is enabled. i.e. the TG signal can bypass your attenuators by another path within the analyser itself.

So as you add more and more attenuation the bypass/leakage signal will begin to become significant wrt your wanted (attenuated) signal. So, depending on the relative phase of the two signals where they eventually meet inside the analyser, there will be a summing or cancelling effect that will appear as ripple on the display. The ripple will get worse the more attenuation you add up until the point the two signals are at the same amplitude where they meet. Also, the  longer the cable you use the more peaks and troughs you will see as there will be a steeper phase shift wrt frequency in a longer cable resulting in more cases where the relative phase of the two signals sums or cancels.

« Last Edit: March 05, 2015, 08:52:39 AM by G0HZU »
 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #6 on: March 06, 2015, 11:02:30 AM »
Thanks!  I can see this occurring.  When a low level of attenuation is used, the TG primary signal path dominates and you don't see much of a ripple.  When the signal is attenuated by, let's say 70db, the primary signal path is weak, and the weak primary is adding with the weak leakage and you see a substantial ripple. Makes sense.
 

Offline Howardlong

  • Super Contributor
  • ***
  • Posts: 3678
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #7 on: March 06, 2015, 11:39:27 AM »
I just tried it on mine. This is the result. Attenuation was from four SMA pads and one N pad. I normalised with the same setup minus the pads. I tried to duplicate the OP's setup, but it showed up a little blurred so I'm not sure I have exactly the same setup.





 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #8 on: March 06, 2015, 11:45:10 AM »
I will say that the ripple I noticed was significantly more prominent in the 500+ MHZ range.  For me, it did not matter where the attenuators were located so I think your setup is probably fine.  What does it look like from 500Mhz to 1.5Ghz?
 

Offline markone

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: it
Re: Spectrum Analyzer Rigol TG-815
« Reply #9 on: March 06, 2015, 11:46:18 AM »
Thanks!  I can see this occurring.  When a low level of attenuation is used, the TG primary signal path dominates and you don't see much of a ripple.  When the signal is attenuated by, let's say 70db, the primary signal path is weak, and the weak primary is adding with the weak leakage and you see a substantial ripple. Makes sense.

I also do not have TG-815 but i'm considering to buy it and i have a couple of questions about this situation :

1) what happen if you disconnect the attenuator leaving the cable on place during the sweep ?   
2) what happen if you disconnect everything during the sweep ?

If the leakage is inside the instrument (i hope is not the case...) you should see a responce curve quite far from the noise floor (let's say -120dBm ) even at "open circuit" condition.

Am i wrong ?
 

Offline Howardlong

  • Super Contributor
  • ***
  • Posts: 3678
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #10 on: March 06, 2015, 11:59:03 AM »
I will say that the ripple I noticed was significantly more prominent in the 500+ MHZ range.  For me, it did not matter where the attenuators were located so I think your setup is probably fine.  What does it look like from 500Mhz to 1.5Ghz?

I'll take a look in the morning, it's 1am here and I Just powered down the shop and put everything away. I chose the span 1MHz-500MHz as that looked like what you had, but it was a bit blurry.

What cables are you using? Are you sure they're not 75 ohm? Some BNCs are 75 ohm, as indeed are some N types but that's much rarer.
 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #11 on: March 06, 2015, 12:08:26 PM »
FYI, here are some better pictures.  Everything in my system is 50 ohm. Top line in each photo is the TG signal with a direct connection from output to analyzer input (saved line).  The bottom line is the noise floor at a band width of 300 Khz over the full scale of the instrument (again, a saved line).  The middle line in the first picture is the the signal with 20db of attenuation.  As you can see, there is not much ripple.  The second picture is with 50db of attenuation.  As you can see, there is significant ripple. The ripple is close to the noise floor.

markone, I like the machine and it is great performance for the price.
 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #12 on: March 06, 2015, 12:10:19 PM »
Second picture
 

Offline markone

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: it
Re: Spectrum Analyzer Rigol TG-815
« Reply #13 on: March 06, 2015, 12:34:33 PM »
>The bottom line is the noise floor at a band width of 300 Khz over the full scale of the instrument (again, a saved line)

So the noise floor level is around -70 dBm ?
 

Offline Howardlong

  • Super Contributor
  • ***
  • Posts: 3678
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #14 on: March 06, 2015, 12:34:38 PM »
Right, couldn't sleep!

Full span, 30kHz RBW, 70dB attenuator:


Full span, 30kHz RBW, TG on, no cable:


Full span, 30kHz RBW, TG off, no cable:

So above about 800MHz, the TG leakage is about the same level as your attenuated signal, leading to the in phase and out of phase peaks and troughs you're seeing. What it doesn't explain is why in your OP, which looked like it was a span from 1MHz to 500MHz, why you were seeing the same peaks and troughs.
 

Offline georges80

  • Frequent Contributor
  • **
  • Posts: 664
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #15 on: March 06, 2015, 01:04:04 PM »
Instead of taking camera pictures, a screen capture provides better 'detail'....

http://peter.dreisiebner.at/rigol-bildschirmkopie-lan/download.htm

cheers,
george.
My 'job': www.taskled.com
 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #16 on: March 06, 2015, 01:06:16 PM »
Right, the first picture was something I actually noticed on a friend's machine.  Honestly, I don't know why his is acting up at that lower frequency.  He could have been using cheap coax,  but I'm not certain.  On my setup, it behaved much better.  Your setup even looks better than mine.  Maybe a better coax or the use of sma attenuators?

markone - the noise floor can get as low as -135 db, but to get there, like with any spectrum analyzer, you have to limit the input bandwidth.  To get to -135db, I have to 1) turn on the pre-amp 2) reduce the bandwidth to about 100 or so hertz, and possibly reduce input attenuation.  I did not do this for my demonstration as it would have taken too long to sweep the entire 1.5 Ghz and was not necessary.  Again, I think it's a good buy. It won't perform the same as a $10k+ unit, but then again, unless you want to spend 10k+, your options are limited.
 

Offline markone

  • Frequent Contributor
  • **
  • Posts: 367
  • Country: it
Re: Spectrum Analyzer Rigol TG-815
« Reply #17 on: March 06, 2015, 02:02:26 PM »
>markone - the noise floor can get as low as -135 db

My initial assessment was made thinking that the test condition was different, with a much lower noise floor.

I'm trying to understand which is the actual dynamic range of this instrument, it would be great to see what happen if you could manage RBW and preamp status to obtain a noise floor at least around -90 dBm and repeat the 50dB attenuator analysis.

Thanks for your attention.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2173
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #18 on: March 07, 2015, 01:11:23 AM »
Quote
What it doesn't explain is why in your OP, which looked like it was a span from 1MHz to 500MHz, why you were seeing the same peaks and troughs.

I've not used the 815 but can you alter the TG power level in the menus? The OP's plot looked to be a normalised measurement and I'd expect you could see this effect get worse if it had been normalised with a lower TG power level.

Also, depending on where the signals 'meet' inside the analyser you could also see changes wrt the internal attenuator setting.

There could also be some common mode effects here as well. However, The results for leakage don't look very good on any of the analyser plots above 1GHz.

 

Offline Howardlong

  • Super Contributor
  • ***
  • Posts: 3678
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #19 on: March 07, 2015, 01:21:25 AM »
Quote
What it doesn't explain is why in your OP, which looked like it was a span from 1MHz to 500MHz, why you were seeing the same peaks and troughs.

I've not used the 815 but can you alter the TG power level in the menus? The OP's plot looked to be a normalised measurement and I'd expect you could see this effect get worse if it had been normalised with a lower TG power level.

Also, depending on where the signals 'meet' inside the analyser you could also see changes wrt the internal attenuator setting.

There could also be some common mode effects here as well. However, The results for leakage don't look very good on any of the analyser plots above 1GHz.

In the test I wound it up to full smoke, 0dBm. Interestingly, the difference in an unconnected (leaked) sweep at 0dBm and -20dBm isn't much, I would imagine a lot of the leakage is pre-attenuator.
 

Offline G0HZU

  • Super Contributor
  • ***
  • Posts: 2173
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #20 on: March 07, 2015, 01:30:57 AM »
Quote
In the test I wound it up to full smoke, 0dBm. Interestingly, the difference in an unconnected (leaked) sweep at 0dBm and -20dBm isn't much, I would imagine a lot of the leakage is pre-attenuator.

My first impression when looking at your plot and Joe's second set of images was that you had (perhaps) each done your tests at different TG drive levels prior to normalisation. Was Joe's test normalised at -10dBm TG drive level? That might explain why yours looks to have lower leakage relative to the 0dB normalised reference (when compared to Joe with the -10dBm TG power level).


« Last Edit: March 07, 2015, 01:36:42 AM by G0HZU »
 

Offline Howardlong

  • Super Contributor
  • ***
  • Posts: 3678
  • Country: gb
Re: Spectrum Analyzer Rigol TG-815
« Reply #21 on: March 07, 2015, 01:59:10 AM »
It's a very valid question. Indeed we should be very clear on the parameters used beyond defaults when making these tests.
 

Offline JoeAtl30319

  • Contributor
  • Posts: 36
  • Country: us
Re: Spectrum Analyzer Rigol TG-815
« Reply #22 on: March 07, 2015, 11:10:21 AM »
FYI, here is a nice clear shot, for those who are interest, of an analysis I just performed.  On this analysis, I set the bandwidth to 3khz.  This had the effect of lowering the noise to around -95db.  I re-preformed the -50db test (top line).  The middle line is with 69db of attenuation.  The bottom line represents the noise at the 3khz bandwidth.  Here are some other setting I used with this test:

TG level set to -20dbm
Normalizing not used
Coax = 2ft RG58/U (50Ohm)
Internal attenuator set to 20dbm (30dbm was default)
Pre Amp = On
Attenuator = 50 Ohm

Let me know if I missed anything.
 


Offline fqahmad66

  • Regular Contributor
  • *
  • Posts: 58
  • Country: 00
Re: Spectrum Analyzer Rigol TG-815
« Reply #24 on: March 07, 2015, 04:30:25 PM »
FYI, here is a nice clear shot, for those who are interest, of an analysis I just performed.  On this analysis, I set the bandwidth to 3khz.  This had the effect of lowering the noise to around -95db.  I re-preformed the -50db test (top line).  The middle line is with 69db of attenuation.  The bottom line represents the noise at the 3khz bandwidth.  Here are some other setting I used with this test:

TG level set to -20dbm
Normalizing not used
Coax = 2ft RG58/U (50Ohm)
Internal attenuator set to 20dbm (30dbm was default)
Pre Amp = On
Attenuator = 50 Ohm

Let me know if I missed anything.

I think if you move the cable and those bumps change then it is cable...
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf

 

http://opalkelly.com/