Spectrum Analyzer Probe & Power line communication

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
This is a comprehensive question and I don’t where it will end.

I am new to deal with spectrum analyzer but worked for 3 years with the oscilloscope.
I have a new Rigol DSA815 spectrum Analyzer and it didn’t come with any probe. I searched upon probing for the spectrum analyzer. It is definitely not the oscilloscope probe.

One my search I come with this conclusion:
1- Probing in spectrum analyzer requires the impedance of the probe to be equal to 50? to match the spectrum analyzer input impedance.
2- The RF input is N male, many adapters (like N male to BNC) have a 50? resistor too.
3- When probing to the spectrum analyzer you have to take care of the output resistance of circuit under test as not to load it.
4- Co-axial cables are used for probing purposes in Spectrum analyzers.


What I want to measure:
We have a power line communication energy meters that have suffered from poor communication. It works in a range of 122 KHz.
As a beginning, we have noticed that if there are many Fluorescent lamps have a PFC capacitor in the input which will attenuate the PLC signal for sure and it will affect the communication. But this has to be measured accurately.


I am intending to make a PLC decoupling and filter circuit to measure the amplitude of the transmitted signal with the lamp switched ON and OFF with the spectrum analyzer.

What is the consideration that should be taken to probe the signal under study?
Is my approach is reasonable or it is not applicable at all?!
 

Wuerstchenhund

Joined Aug 31, 2017
189
This isn't really a good job for a Spectrum Analyzer. The best tool for a task like that is a good DSO with decent FFT and an appropriate active HV differential probe. Piece of cake.

If you don't have an HV differential probe then you could use a galvanically separated transformer instead and measure with a High-Z probe on the secondary side, which should be adequate for a 122kHz signal. Just make sure there is galvanic separation as itherwise you'll put your scope and yourself in danger.

Then set the scope in FFT mode and observe the transmissions in the time and frequency domain.
 
Last edited:

crutschow

Joined Mar 14, 2008
34,284
Your spectrum analyzer goes from 9 kHz to 1.5 GHz, so it could measure the signals you want, but it does have a 50Ω input impedance, and you need to allow for that.
At those frequencies, you don't have to match the 50Ω impedance to the line, but you do need to block the large 50-60Hz main's voltage which you can do with a low-pass RC filter with a corner of about 100kHz.

A standard isolation transformer will likely attenuate the 122kHz signal significantly, so you may want to consider an audio isolation transformer with sufficient frequency response.
You can put the audio transformer after the low-pass filter to avoid saturating the transformer, but be aware that the filter circuit is at a dangerous voltage and treat it accordingly. :eek:
 

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
This isn't really a good job for a Spectrum Analyzer. The best tool for a task like that is a good DSO with decent FFT and an appropriate active HV differential probe. Piece of cake.

If you don't have an HV differential probe then you could use a galvanically separated transformer instead and measure with a High-Z probe on the secondary side, which should be adequate for a 122kHz signal. Just make sure there is galvanic separation as itherwise you'll put your scope and yourself in danger.

Then set the scope in FFT mode and observe the transmissions in the time and frequency domain.
I have a Tektronix TDS2014 but never thought it will be that accurate in math function. I will try it.
 

Wuerstchenhund

Joined Aug 31, 2017
189
I have a Tektronix TDS2014 but never thought it will be that accurate in math function.
Unfortunately it isn't. The Tek TDS2014 only has minuscule sample memory (I think 2.5k or something similar poor) which is insufficient for getting proper FFT resolution.

FFT only starts to become useful at 16kpts, and better modern entry-level scopes process at least 64kpts, with some (i.e. GW Instek GDS-1000B) going up to 1Mpts.

You're unlikely to get any decent results with that TDS2014.
 

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
Your spectrum analyzer goes from 9 kHz to 1.5 GHz, so it could measure the signals you want, but it does have a 50Ω input impedance, and you need to allow for that.
At those frequencies, you don't have to match the 50Ω impedance to the line, but you do need to block the large 50-60Hz main's voltage which you can do with a low-pass RC filter with a corner of about 100kHz.

A standard isolation transformer will likely attenuate the 122kHz signal significantly, so you may want to consider an audio isolation transformer with sufficient frequency response.
You can put the audio transformer after the low-pass filter to avoid saturating the transformer, but be aware that the filter circuit is at a dangerous voltage and treat it accordingly. :eek:
Ok, I will probably use a 100nF X2 capacitor with VAC T60403-K5024-X044 transformer.
Then I connect the output of the filter directly to the SA?
 

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
Unfortunately it isn't. The Tek TDS2014 only has minuscule sample memory (I think 2.5k or something similar poor) which is insufficient for getting proper FFT resolution.

FFT only starts to become useful at 16kpts, and better modern entry-level scopes process at least 64kpts, with some (i.e. GW Instek GDS-1000B) going up to 1Mpts.

You're unlikely to get any decent results with that TDS2014.
mmm, that means I have to stick with the SA since it is the only thing available.
Why do you think it is n't the perfect job for the SA?
 

Wuerstchenhund

Joined Aug 31, 2017
189
Why do you think it is n't the perfect job for the SA?
Although most SA's go down to 9kHz, LF isn't really what they are made for (unless you have an LF SA). Most SAs have very sensitive inputs that can't tolerate DC (which kills the input stage), they are meant to be connected to properly terminated transmission lines (which mains isn't), and don't support special probes like active HV differential probes.

That doesn't mean you can't make it work somehow, i.e. as crutschow said with an galvanic separated audio transformer plus a mains frequency filter and some impedance matching, which can give you some (more or less vague) measurement results. The best tool for such a job however is a proper DSO with active HV differential probe.

You didn't say if this was for hobby or business but if it's for the latter I'd probably consider renting a proper scope + HV probe for a day (or if these use cases appear more often, consider spending the money on a good DSO). Of course, if it's for hobby and all you have is the TDS2014 then I guess telling you what the best way is is of no use and you'll have to make do with the SA.
 

crutschow

Joined Mar 14, 2008
34,284
they are meant to be connected to properly terminated transmission lines (which mains isn't)
Matching is required to minimize high frequency reflections in the signal line, and that's not a problem at 122kHz.
The SA goes down to 9kHz with a 1Hz minimum resolution bandwidth and <-110dBm noise level, so it should give reasonably good results at 122kHz.
And I believe the TS is more concerned about relative signal levels than absolute levels.
 

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
Although most SA's go down to 9kHz, LF isn't really what they are made for (unless you have an LF SA). Most SAs have very sensitive inputs that can't tolerate DC (which kills the input stage), they are meant to be connected to properly terminated transmission lines (which mains isn't), and don't support special probes like active HV differential probes.

That doesn't mean you can't make it work somehow, i.e. as crutschow said with an galvanic separated audio transformer plus a mains frequency filter and some impedance matching, which can give you some (more or less vague) measurement results. The best tool for such a job however is a proper DSO with active HV differential probe.

You didn't say if this was for hobby or business but if it's for the latter I'd probably consider renting a proper scope + HV probe for a day (or if these use cases appear more often, consider spending the money on a good DSO). Of course, if it's for hobby and all you have is the TDS2014 then I guess telling you what the best way is is of no use and you'll have to make do with the SA.
Good advice thank you.
it is for business and we can buy the DSO but it will take time. The SA was bought primarily for conducted EMI measurement. And we thought it will do the same job in PLC case. We also might use LISN to measure the amplitude of the signal of the PLC.
 

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
Matching is required to minimize high frequency reflections in the signal line, and that's not a problem at 122kHz.
The SA goes down to 9kHz with a 1Hz minimum resolution bandwidth and <-110dBm noise level, so it should give reasonably good results at 122kHz.
And I believe the TS is more concerned about relative signal levels than absolute levels.
I will be interested in the two relative and absolute levels. I think I can obtain the absolute levels if I knew the exact attenuation that is made by the measuring circuit.
 

Wuerstchenhund

Joined Aug 31, 2017
189
Matching is required to minimize high frequency reflections in the signal line, and that's not a problem at 122kHz
The impact of mis-match on a 122kHz signal will be negligible (although matching is relevant whenever the signal isn't d.c.), but the problem here is that the mains line acts like a large antenna, potentially catching a wide range of noise and other high frequency stuff, and in a non-matched line this can cause reflections and intermodulation products which can very well affect your measurements.

To avoid that you'd need to use a LISN device.
 
Last edited:

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
The impact of mis-match on a 122kHz signal will be negligible (although matching is relevant whenever the signal isn't d.c.), but the problem here is that the mains line acts like a large antenna, potentially catching a wide range of noise and other high frequency stuff, and in a non-matched line this can cause reflections and intermodulation products which can very well affect your measurements.

To avoid that you'd need to use a LISN device.
Even If a bandpass filter, it will affect the measurements?
 

Wuerstchenhund

Joined Aug 31, 2017
189
Even If a bandpass filter, it will affect the measurements?
It potentially can, yes. The problem is that your test instrument has an low-Z input which expects a matched line, and if you can't provide a matched line then pretty much all bets are off. For conducted EMI measurements a LISN helps because it provides a specific impedance, and because it acts as a low-pass it filters out the crap that can be found on a typical mains line, which is what you need for EMI measurements because you only want to see the interference caused by your UUT (and not the crap that sits on a real mains line). And because a LISN provides a specific impedance for the mains line it's now pretty easy to provide some impedance matching for a low-Z instrument (most LISNs have an 50ohms output for a measurement device so you just connect your SA or EMI receiver and you're good to go).

However, if I understand your original posting right then you don't want to do conductive EMI testing, you want to measure a specific frequency on a real (unmatched) mains line, which is something completely different. The lack of proper impedance matching will lead to side effects which might very well impact the 122kHz signal which you're interested in. Your "probing" (transformer) itself will impact the signal, because of its resistance and high inductance. Without matching you'll also have a hard time getting realistic amplitude measurements (and comparing an unmatched line with an unknown Z with a generator signal with a different Z won't necessarily give you realistic results). Also, while the frequency range of most general purpose SAs goes down to 9kHz, they often perform pretty badly so low down and only perform as per spec sheet above 10Mhz or so. There are a lot of unknowns which make this a very challenging endeavor.

Thankfully, the signal of interest is LF, which means it's within the usable BW of most high-Z instruments like a scope + HV probe, so if you had the free choice then scope and HV differential probe would be the best way to measure on a live line.

Since you said it's for business, I'd seriously consider just renting a suitable scope + probe for a day.
 
Last edited:

Thread Starter

Ashraf_Almumbarak

Joined Sep 25, 2017
9
It potentially can, yes. The problem is that your test instrument has an low-Z input which expects a matched line, and if you can't provide a matched line then pretty much all bets are off. For conducted EMI measurements a LISN helps because it provides a specific impedance, and because it acts as a low-pass it filters out the crap that can be found on a typical mains line, which is what you need for EMI measurements because you only want to see the interference caused by your UUT (and not the crap that sits on a real mains line). And because a LISN provides a specific impedance for the mains line it's now pretty easy to provide some impedance matching for a low-Z instrument (most LISNs have an 50ohms output for a measurement device so you just connect your SA or EMI receiver and you're good to go).

However, if I understand your original posting right then you don't want to do conductive EMI testing, you want to measure a specific frequency on a real (unmatched) mains line, which is something completely different. The lack of proper impedance matching will lead to side effects which might very well impact the 122kHz signal which you're interested in. Your "probing" (transformer) itself will impact the signal, because of its resistance and high inductance. Without matching you'll also have a hard time getting realistic amplitude measurements (and comparing an unmatched line with an unknown Z with a generator signal with a different Z won't necessarily give you realistic results). Also, while the frequency range of most general purpose SAs goes down to 9kHz, they often perform pretty badly so low down and only perform as per spec sheet above 10Mhz or so. There are a lot of unknowns which make this a very challenging endeavor.

Thankfully, the signal of interest is LF, which means it's within the usable BW of most high-Z instruments like a scope + HV probe, so if you had the free choice then scope and HV differential probe would be the best way to measure on a live line.

Since you said it's for business, I'd seriously consider just renting a suitable scope + probe for a day.
Yes it is for business but unfortunately I can't hire a suitable scope + the probe this is not available in my country. I would rather buy it but it will take awhile. So since the SA is the only thing available I will start with it and then when the scope comes I will do it again. Thank you for your effort.
 
Top