RF spurii measurement - advice sought - UK Vintage Radio Repair and Restoration Discussion Forum

Skywave · 7th Feb 2008, 11:30 pm

Preamble: I've chosen this Section since I am predominately going to refer to 145 MHz - but the problem could be anywhere in the VHF spectrum.

The problem: I have a VHF transmitter capable of 30 watts of unmodulated carrier wave into a 50-ohm power attenuator. I need to measure the non-harmonic spurii within +/- 500 KHz of this carrier at this power level. I have a calibrated spectrum analyzer and a range of attenuators for the analyzer's input.
How do I reliably measure the spurii (if any) in the stated bandwidth?

Comments: A few thoughts by way of a "steer". . .
The basic problem is this: if I connect the O/P from the power attenuator directly to the analyzer, the high level of the fundamental at 145MHz will necessitate such a level of attenuation at the analyzer's input (so as to avoid overloading the analyzer and creating spurii within the analyzer) that the "to-be-measured" spurii from the transmitter will then be too weak to measure on the analyzer (I need to check for spurii levels in the region of 70 to 75 dB below the carrier).

If I was looking to assess harmonic-related spurii, it would be easy to construct a narrow-bandwidth series L/C notch filter to attenuate the fundamental, since the harmonic content is well-spaced from this fundamental, and the harmonic-level measurement will, therefore, be reliable. Unfortunately, using this technique for "close-in" spurii (such as my requirement here) isn't going to work, since the Q of this L/C filter would have to be impossibly high. Even if I build a multi-section stop-band filter (such as an Elliptic / Caeur), I still run the risk of some in-band spurii falling within this stop-band (again, a question of Q).

So, as you can see, I've had a good, long think about this - and I'm stuck on this one.

So I thought I'd put out a shout here to tap other people's brains!

All comments are appreciated.

Al / Skywave.

Ray Cooper · 7th Feb 2008, 11:53 pm

Just a few brief thoughts :

Don't know just what model of spectrum analyzer you're proposing to use, but I suspect that if you set carrier to the top line, then the analyzer's noise floor is going to be about -70dB, so you're a bit stuffed...

But all is not lost. Turn down the input attenuator 10dB (carrier spike will ascend off the screen) and note any spurii. Then turn it another 10dB down, and re-note the spurii. If they've only gone up 10dB, then the analyzer is not contributing to the spurii and you may reasonably accept the result. If they've gone up noticeably more than 10dB, then the analyzer is getting into the act and you'll have to think in terms of some notch filter as suggested.

GMB · 8th Feb 2008, 12:06 am

Are you requiring the carrier to be modulated while making this measurement?

If not then I was wondering if there might be the possibility of combining the signal with itself delayed by a precisely cut transmission line to cancel out much of the carrier thus reducing the signal to a more manageable level.

Skywave · 8th Feb 2008, 12:47 am

Ah - Ha! Yes - I should have added that I am referring to an unmodulated carrier wave.
The idea of a section of line to attenuate the fundamental is noted - but I expect that the B/W will not be narrow enough at these frequencies.

Spectrum analyzer is an Anritsu MS26xx, (where xx = something I can't remember off-hand): 10 Khz to several GHz.
And yes - the vertical scaling on the screen implies that the dynamic range is of the order of 70 dB. The "quick check" of the analyzer's dynamic range which you suggest, Ray, is indeed worthwhile - I'll do that - thanks.

Seems to me that the only real answer is an analyzer with a better dynamic range.

Thanks guys.

Al / Skywave.

MichaelR · 8th Feb 2008, 10:32 am

Hi Al,

Please remember that your spectrum analyser is is a broadband input and even though you may set the display to not show the carrier , it is still there as far as your analyser input is concerned.Most analysers are superheterodyne in design with a broadband input.

It is most important thet you set the input attenuator of your analyser to give an input to its input mixer of about -30dbm for spurious free operation. ( spurious generated in the analyser due to input mixer overload).

You will I suspect need to preselct somehow, either using a high quality notch filter on the front end to notch down the carrier. You should be able to make a filter with about 40 db of attenuation.

The other way is try a coupler whereby you could have a 40 or 50 db coupling value.

one of the previous posts does also raise a valid method for identifying internally generated spurri by knocking down the input attenuator and noting the drop level in the spurii levels

You mention Anristu I am very familiar with their range , I worked for them for 15 years.

Regards
Mike

GMB · 8th Feb 2008, 10:33 am

Quote:

but I expect that the B/W will not be narrow enough at these frequencies.

I haven't tried it yet, but theoretically and subject to frequency stability, if you split the signal and remix it again with one route massively delayed then according to the actual delay you can achieve whatever "bandwidth" you want.

I expect the snag is in the practicalities of course.

(I will be looking for a solution to this problem myself eventually, when restoration works are completed.)

MichaelR · 8th Feb 2008, 10:44 am

Just noted your measurement requirment in more detail

You will struggle to make measurements with a 2601 series analyser of on screen 75db , you need more on screen dynamic range.

Mike

MichaelR · 8th Feb 2008, 2:10 pm

Al,

you could couple down 30 dbto40, attenuate another 10 db or so and finish off going through a tuneable xtal notch filter, you may just be able to get as much as 75 db down from that signal with a very narrow resolution bandwidth, if you are using the 2601 it may have the 30Hz filter fitted. The noise floor achievable with narrow spans could be better than -120dbm

Mike

Skywave · 8th Feb 2008, 5:32 pm

Quote:

Originally Posted by MichaelR

Just noted your measurement requirement in more detail

You will struggle to make measurements with a 2601 series analyser of on-screen 75dB; you need more on-screen dynamic range.

Mike

Hi Mike -
Which is the conclusion that I found myself being drawn towards - but I wanted independent confirmation from others - just to be sure.

The analyzer is an MS2621A.
I ran a quick dynamic range assessment test (as per Ray's Post) and found that once the signal "spike" goes off the top of the screen, the internally-generated harmonics start to appear. At 10 dB "above" the screen top, it's well into non-linearity (surprise, surprise

)

Al / Skywave.

MichaelR · 8th Feb 2008, 6:28 pm

Quote:

Originally Posted by Skywave

Hi Mike -
Which is the conclusion that I found myself being drawn towards - but I wanted independent confirmation from others - just to be sure.

The analyzer is an MS2621A.
I ran a quick dynamic range assessment test (as per Ray's Post) and found that once the signal "spike" goes off the top of the screen, the internally-generated harmonics start to appear. At 10 dB "above" the screen top, it's well into non-linearity (surprise, surprise

)

Al / Skywave.

The 2621 is a good analyser, if you have a drive level of -35dbm into the mixer the internaly genrated spurii will be 80db down on that analyser.I would not put a 30dbm signal on the front end even though it is specd to do so , one mistake with the manual settings and you will blow the analyser front end.

use a coupler with 30db coupling and set manually 40db attenuation for the analyser.Spurii and 3rd order products shouldl be 80 db down from carrier.

I am assuming from your concern therefore that you are wanting to measure lower, if you are doing what I think you are doing you would need 100db dynamic range.

regards
Mike

Skywave · 8th Feb 2008, 11:31 pm

Mike - thanks for that Post immediately above.

A couple of points arising:

(a) As I recall, the front panel of the MS2621 states "do not exceed +25dBm". I don't have a manual for it - so I suppose that the spec. figure of +30 dBm (that you quote) includes a 'safety margin' of 5dBm ?

Incidentally, as 'routine default', I always hang a 30 dB external attenuator on the input to this analyzer before measuring anything; attenuators are cheap to replace compared to the 'front end' of an analyzer.

Same idea as always setting the d.c. volts range on the AVO to 1Kv. before measuring an unknown voltage - I learnt the importance of this a long, long time ago.

(b) Re: your suggestion above.
With a -30 dB coupler & -40dB attenuation setting in the analyzer, all input signals will be equally attenuated by 70dB: the 'unwanted' carrier and the 'looked for' spurii. Although I can see that this will prevent overload of the 1st. mixer in the analyzer, (

) won't this level of attenuation result in the 'looked for' spurii being undetectable, since they'll be buried in the noise, bearing in mind that they could be 50 dB (or more) below that of the carrier?

Or am I missing something obvious (again)

Al / Skywave.

MichaelR · 9th Feb 2008, 12:05 am

When you view the coupled signal , you can set it at the ref level at the top of the screen hence with 30db attenuatIon your carrier level is now , 0dbm.

Your on screen dynamice range should be upto 80db hence from reference value down to -80dbm approx noise floor.With a -20dbm to -30dbm approx mixer drive level this will give intermods and spurii generated in the analyser at levels near the noise floor of about -80dbm. You can identify these once you have the baisc settings correct by using "Ray's " method.

I am sure the analyser has a screen dynamic range of 100db. ( Graticule )

On screen dynamic range is what "window" you can display. If your noise floor is -100dbm then you can display upwards from -100dbm to -20dbm ref setting. Your noise floor is dependent upon RBW setting ( narrower the better ) together with the drive level to the mixer.

On screen dynamic range is normally restricted by the detector performance in the analyser.

regards
Mike

Skywave · 9th Feb 2008, 11:12 am

Quote:

Originally Posted by MichaelR

I am sure the analyser has a screen dynamic range of 100db. ( Graticule )
Mike

The screen has a total of 9 horizontal graticule lines, each 1 cm. apart. My understanding is that this gives a "screen dynamic range" of 8 divs. times 10 dB per div. = 80 dB.

Al / Skywave.

MichaelR · 9th Feb 2008, 11:53 am

Quote:

Originally Posted by Skywave

The screen has a total of 9 horizontal graticule lines, each 1 cm. apart. My understanding is that this gives a "screen dynamic range" of 8 divs. times 10 dB per div. = 80 dB.

Al / Skywave.

The top line is always regarded as the ref line , you may be thinking that is some "boxing or border". if you count down each line from the top you will get 100db if you are set to 10db/div

The best noise floor @ 30hz RBW will be -115dbm and as I say the best "on screen dynamic range" ( defined as the best carrier to noise measurement available on one screen ) will be about 80db.

I do not know what you want to measure but also bear in mind depending how close you want to measure to you carrier the phase noise and filter shape are important. This analyser is pretty good in that the phase noise is better than -100dbc from the filter peak @10khz. This is good for a microwave analyser but may not be good enough for a high grade HF measurement.

regards
Mike

MichaelR · 9th Feb 2008, 10:32 pm

Quote:

Originally Posted by Skywave

Mike - thanks for that Post immediately above.

(b) Re: your suggestion above.
With a -30 dB coupler & -40dB attenuation setting in the analyzer, all input signals will be equally attenuated by 70dB: the 'unwanted' carrier and the 'looked for' spurii. Although I can see that this will prevent overload of the 1st. mixer in the analyzer, (

) won't this level of attenuation result in the 'looked for' spurii being undetectable, since they'll be buried in the noise, bearing in mind that they could be 50 dB (or more) below that of the carrier?

Or am I missing something obvious (again)

Al / Skywave.

Speed reading can sometimes be no good

Al I just noted your comment here. When I mention the 40 db setting of the spectrum analyser attenuation remember this attenuator on most analysers is fitted in line between the input and the first mixer. It sets the drive level to the mixer it does not attenuate the main input signal. Settings around this level will give you the ability to see 80db down from carrier. The main input signal goes through the input with no further attenuation.

mike

Skywave · 9th Feb 2008, 10:49 pm

This Thread now seems to have gone quiet: it's time for me to move on.

First, I would now like to say 'thanks' to those members who have made the useful suggestions as per the above.

Second, I am aware that this Topic is of a highly-specialised nature and will probably be of interest to only a very narrow range of readers. With this is mind, I am now going to stop enquiring further, take the time out to slowly absorb all that has been said - a lot of this is really quite complex (to me, anyway

) - and use it to assist & further my own investigations.

Third, in the event of anything relevant then coming to light, I will return here to ask further or (hopefully & preferably) report on progress.

Finally, I will be occasionally viewing this Thread to check for any subsequent Posts. And again - my thanks goes to all those above.

TTFN

Al / Skywave.

7th Feb 2008, 11:30 pm	#1
Skywave Rest in Peace Join Date: Jun 2006 Location: Chard, South Somerset, UK. Posts: 7,457	RF spurii measurement - advice sought Preamble: I've chosen this Section since I am predominately going to refer to 145 MHz - but the problem could be anywhere in the VHF spectrum. The problem: I have a VHF transmitter capable of 30 watts of unmodulated carrier wave into a 50-ohm power attenuator. I need to measure the non-harmonic spurii within +/- 500 KHz of this carrier at this power level. I have a calibrated spectrum analyzer and a range of attenuators for the analyzer's input. How do I reliably measure the spurii (if any) in the stated bandwidth? Comments: A few thoughts by way of a "steer". . . The basic problem is this: if I connect the O/P from the power attenuator directly to the analyzer, the high level of the fundamental at 145MHz will necessitate such a level of attenuation at the analyzer's input (so as to avoid overloading the analyzer and creating spurii within the analyzer) that the "to-be-measured" spurii from the transmitter will then be too weak to measure on the analyzer (I need to check for spurii levels in the region of 70 to 75 dB below the carrier). If I was looking to assess harmonic-related spurii, it would be easy to construct a narrow-bandwidth series L/C notch filter to attenuate the fundamental, since the harmonic content is well-spaced from this fundamental, and the harmonic-level measurement will, therefore, be reliable. Unfortunately, using this technique for "close-in" spurii (such as my requirement here) isn't going to work, since the Q of this L/C filter would have to be impossibly high. Even if I build a multi-section stop-band filter (such as an Elliptic / Caeur), I still run the risk of some in-band spurii falling within this stop-band (again, a question of Q). So, as you can see, I've had a good, long think about this - and I'm stuck on this one. So I thought I'd put out a shout here to tap other people's brains! All comments are appreciated. Al / Skywave.

7th Feb 2008, 11:53 pm	#2
Ray Cooper Retired Dormant Member Join Date: Jan 2007 Location: Porthmadog, Gwynedd, UK. Posts: 199	Re: RF spurii measurement - advice sought Just a few brief thoughts : Don't know just what model of spectrum analyzer you're proposing to use, but I suspect that if you set carrier to the top line, then the analyzer's noise floor is going to be about -70dB, so you're a bit stuffed... But all is not lost. Turn down the input attenuator 10dB (carrier spike will ascend off the screen) and note any spurii. Then turn it another 10dB down, and re-note the spurii. If they've only gone up 10dB, then the analyzer is not contributing to the spurii and you may reasonably accept the result. If they've gone up noticeably more than 10dB, then the analyzer is getting into the act and you'll have to think in terms of some notch filter as suggested.

8th Feb 2008, 12:06 am	#3
GMB Dekatron Join Date: Aug 2003 Location: near Reading (and sometimes Torquay) Posts: 3,099	Re: RF spurii measurement - advice sought Are you requiring the carrier to be modulated while making this measurement? If not then I was wondering if there might be the possibility of combining the signal with itself delayed by a precisely cut transmission line to cancel out much of the carrier thus reducing the signal to a more manageable level. Last edited by GMB; 8th Feb 2008 at 12:17 am.

8th Feb 2008, 12:47 am	#4
Skywave Rest in Peace Join Date: Jun 2006 Location: Chard, South Somerset, UK. Posts: 7,457	Re: RF spurii measurement - advice sought Ah - Ha! Yes - I should have added that I am referring to an unmodulated carrier wave. The idea of a section of line to attenuate the fundamental is noted - but I expect that the B/W will not be narrow enough at these frequencies. Spectrum analyzer is an Anritsu MS26xx, (where xx = something I can't remember off-hand): 10 Khz to several GHz. And yes - the vertical scaling on the screen implies that the dynamic range is of the order of 70 dB. The "quick check" of the analyzer's dynamic range which you suggest, Ray, is indeed worthwhile - I'll do that - thanks. Seems to me that the only real answer is an analyzer with a better dynamic range. Thanks guys. Al / Skywave.

8th Feb 2008, 10:32 am	#5
MichaelR Retired Dormant Member Join Date: Mar 2006 Location: Sheffield, South Yorkshire, UK. Posts: 1,587	Re: RF spurii measurement - advice sought Hi Al, Please remember that your spectrum analyser is is a broadband input and even though you may set the display to not show the carrier , it is still there as far as your analyser input is concerned.Most analysers are superheterodyne in design with a broadband input. It is most important thet you set the input attenuator of your analyser to give an input to its input mixer of about -30dbm for spurious free operation. ( spurious generated in the analyser due to input mixer overload). You will I suspect need to preselct somehow, either using a high quality notch filter on the front end to notch down the carrier. You should be able to make a filter with about 40 db of attenuation. The other way is try a coupler whereby you could have a 40 or 50 db coupling value. one of the previous posts does also raise a valid method for identifying internally generated spurri by knocking down the input attenuator and noting the drop level in the spurii levels You mention Anristu I am very familiar with their range , I worked for them for 15 years. Regards Mike Last edited by MichaelR; 8th Feb 2008 at 10:39 am. Reason: adding comment

8th Feb 2008, 10:44 am	#7
MichaelR Retired Dormant Member Join Date: Mar 2006 Location: Sheffield, South Yorkshire, UK. Posts: 1,587	Re: RF spurii measurement - advice sought Just noted your measurement requirment in more detail You will struggle to make measurements with a 2601 series analyser of on screen 75db , you need more on screen dynamic range. Mike

8th Feb 2008, 2:10 pm	#8
MichaelR Retired Dormant Member Join Date: Mar 2006 Location: Sheffield, South Yorkshire, UK. Posts: 1,587	Re: RF spurii measurement - advice sought Al, you could couple down 30 dbto40, attenuate another 10 db or so and finish off going through a tuneable xtal notch filter, you may just be able to get as much as 75 db down from that signal with a very narrow resolution bandwidth, if you are using the 2601 it may have the 30Hz filter fitted. The noise floor achievable with narrow spans could be better than -120dbm Mike

8th Feb 2008, 11:31 pm	#11
Skywave Rest in Peace Join Date: Jun 2006 Location: Chard, South Somerset, UK. Posts: 7,457	Re: RF spurii measurement - advice sought Mike - thanks for that Post immediately above. A couple of points arising: (a) As I recall, the front panel of the MS2621 states "do not exceed +25dBm". I don't have a manual for it - so I suppose that the spec. figure of +30 dBm (that you quote) includes a 'safety margin' of 5dBm ? Incidentally, as 'routine default', I always hang a 30 dB external attenuator on the input to this analyzer before measuring anything; attenuators are cheap to replace compared to the 'front end' of an analyzer. Same idea as always setting the d.c. volts range on the AVO to 1Kv. before measuring an unknown voltage - I learnt the importance of this a long, long time ago. (b) Re: your suggestion above. With a -30 dB coupler & -40dB attenuation setting in the analyzer, all input signals will be equally attenuated by 70dB: the 'unwanted' carrier and the 'looked for' spurii. Although I can see that this will prevent overload of the 1st. mixer in the analyzer, () won't this level of attenuation result in the 'looked for' spurii being undetectable, since they'll be buried in the noise, bearing in mind that they could be 50 dB (or more) below that of the carrier? Or am I missing something obvious (again) Al / Skywave.

9th Feb 2008, 12:05 am	#12
MichaelR Retired Dormant Member Join Date: Mar 2006 Location: Sheffield, South Yorkshire, UK. Posts: 1,587	Re: RF spurii measurement - advice sought When you view the coupled signal , you can set it at the ref level at the top of the screen hence with 30db attenuatIon your carrier level is now , 0dbm. Your on screen dynamice range should be upto 80db hence from reference value down to -80dbm approx noise floor.With a -20dbm to -30dbm approx mixer drive level this will give intermods and spurii generated in the analyser at levels near the noise floor of about -80dbm. You can identify these once you have the baisc settings correct by using "Ray's " method. I am sure the analyser has a screen dynamic range of 100db. ( Graticule ) On screen dynamic range is what "window" you can display. If your noise floor is -100dbm then you can display upwards from -100dbm to -20dbm ref setting. Your noise floor is dependent upon RBW setting ( narrower the better ) together with the drive level to the mixer. On screen dynamic range is normally restricted by the detector performance in the analyser. regards Mike Last edited by MichaelR; 9th Feb 2008 at 12:07 am. Reason: spelling

9th Feb 2008, 10:49 pm	#16
Skywave Rest in Peace Join Date: Jun 2006 Location: Chard, South Somerset, UK. Posts: 7,457	Re: RF spurii measurement - advice sought This Thread now seems to have gone quiet: it's time for me to move on. First, I would now like to say 'thanks' to those members who have made the useful suggestions as per the above. Second, I am aware that this Topic is of a highly-specialised nature and will probably be of interest to only a very narrow range of readers. With this is mind, I am now going to stop enquiring further, take the time out to slowly absorb all that has been said - a lot of this is really quite complex (to me, anyway ) - and use it to assist & further my own investigations. Third, in the event of anything relevant then coming to light, I will return here to ask further or (hopefully & preferably) report on progress. Finally, I will be occasionally viewing this Thread to check for any subsequent Posts. And again - my thanks goes to all those above. TTFN Al / Skywave.