5142 real data

[COMSIS]

Hello

I have a 5600 and 5142 using RFSA or Ni-scope I just want to have real data (0 to positive axis frequency)

My situation is I want to stream real Bandwidth say 4 MHz (0 MHz-4 MHz) and in this bandwidth I have 3xFM signals at 1Mhz,2Mhz,3Mhz,4Mhz now using Ni modulation toolkit I want to downconvert 1/2/3Mhz Signal to base band only one at a time and then demodulate it using FM demodulator VI.

1- What I have done is I have decimated the interleaved array for complex acquisition(16 bit) then make a Waveform out of it and then downconvert and demodulate it using FM demodulator (Using RFSA)

2- I have only used 5142 using niscope I set the data type to real and with frequency translation Enable and DDC enable.i am only getting real data out of which I make a waveform and then Downconvert  it and demodulate it using same FM demodulator (using niscope)

Now in both the cases I see spectrum perfectly ok  centered at 98 MHz (0 to 4MHz are mapped as 0 MHz=FM98, 1 MHz=FM99 2 MHz=FM100 3 MHz=FM101).I downconvert  1 MHz signal and demodulate it but what happened is I hear FM 97Mhz instead of FM99.Does it mean that real signal is still carrying the negative frequency signals since I am acquiring from 98 MHz onwards 

Is it that frequencies in the negative frequency axis are overlapped on positive axis frequencies . when I demodulate I hear Signal with more strength as in my case FM97 is more strong then FM99(as seen on RFSA demo Panel). I have tested it on different FM signals results are same the signal seems to be overlapped with negative frequency  and which ever is stronger signal it is played (demodulated).

My question is what is the actual use of real acquisition data using in-scope if real data still has negative part. Or what I am doing wrong to get only real data.

Regards

[Dan_K]

Hello COMSIS,

If you are looking at real data at an intermediate frequency and then downconvert it you will get a combination of signals from both sides of the center frequency. So, if you have a signal at 97 MHz and a signal at 99 MHz and downcovert with a center frequency of 98 MHz you will get a combination of the signal at 97 MHz and 99 MHz at 1 MHz due to the reflection of the 97 MHz signal across the origin.

There are a couple of easy ways to avoid this behavior:

• Don't acquire the 97 MHz signal. Use an analog bandpass filter before injecting the signal into your instrumentation.
• Bring the signal down to IF and filter it out in software. Use a center frequency that results in your signals of interest being further away from the center, say at 10, 11, 12 and 13 MHz. Then use a bandpass filter in software to get rid of all of the data except your signal of interest and downconvert to baseband.

Either of these will work, let me know if you have any questions.

[COMSIS]

Hello

Tnx for the reply

I am not getting it clearly why is it so. The reason u gave might seams to work in the case where I am acquiring Complex interleaved data from RFSA,then the decimate to get only real data since it is containing samples from +ve and -ve frequency axis so it is possible to have FM 97 reflection on FM 99.

Why in the case where I am using digitized only with niscope has same result.

As u can see from OSP block diagram there is a separate path for real data passing through
filtering and decimation block and then real data is returned. What this filtering and decimation doing if it is not rejecting negative frequency component in my acquisition. What kind of real data is returned over here

[chris_amadasun]

Hello COMSIS,

Yes when you use the real processing mode of the OSP chip than only real data (I values only) are returned.

What Dan was saying is that the 97MHz and 99MHz frequencies are probably aliasing.

If we only had the 97MHz channel and we multiplied it by 98MHz we would have a signal at 1 MHz. This is the same happens if you multiply a 99MHz signal with 98MHz.

What Dan is suggesting is that when you want to demodulate the 99MHz channel you will need to filter out the 97MHz channel before you downconvert

Does this help?

[Andy_Hinde]

Hello,

Perhaps I can help clarify a few things:

The goal is to acquire a chunk of spectrum, say 95-99 MHz, which contains multiple FM signals. In software, you wish to pick off one of these FM stations and ignore the others.

What you should due is set the center frequency of the 5661 (5600+5142) to the middle of the band you wish to capture. In this example it would be 97 MHz. Then, you would need to set the IQ sample rate such that you capture a minimum of 4 MHz of spectrum. The relationship between BW and IQ sample rate is:

IQ Datal BW = 0.8 * I Sample Rate

So, your IQ rate would need to be >= 5 MS/sec. However, your FM signals at the bottom and top of the spectrum have some bandwidth to them. Broadcast FM uses 200 kHz channels, so we'll use that here. I would set the IQ Rate to 6 MSPS in this case to give me an acquisition BW of 4.8 MHz, to make sure I capture all of the BW of the FM signals on the bottom and top ends here.

So, at this point, you will have IQ data in software, centered at 0 Hz, and with a BW spanning 4.8 MHz. You have complex IQ data now, but you can still in software move the cventer frequency around so that 0 Hz corresponds to frequencies other than 97 MHz. You can use the MT Upconversion and Downconversion VIs to recenter the IQ data at your choice of FM carriers - the routines are polymorphic so select the Complex version of the subVIs. One of the inputs to the Upconversion and Downconversion routines is passband bandwidth, so makes sure to filter the signal to remove the unwanted carriers,

At this point you should have IQ data centered at the desired FM carrier with the remaining FM signals removed. This signal would then be ready for FM Demodulation via the Modulation Toolkit.

To clarify some previous points made:

"If you are looking at real data at an intermediate frequency and then downconvert it you will get a combination of signals from both sides of the center frequency. So, if you have a signal at 97 MHz and a signal at 99 MHz and downcovert with a center frequency of 98 MHz you will get a combination of the signal at 97 MHz and 99 MHz at 1 MHz due to the reflection of the 97 MHz signal across the origin."

This is either incorrect, or not phrased properly. Either way this is confusing. If you have signals at 97 and 99 MHz and use the 5661, the 5661 analog downconverts the signal (via the 5600) to an IF center frequency of 15 MHz, then the 5142 samples the 5600 output (which now has signas at 14 and 16 MHz) and the OSP produces complex IQ data with the IQ data centered at 0 Hz and the original signal at -1 MHz and 1 MHz.

"Yes when you use the real processing mode of the OSP chip than only real data (I values only) are returned."

Real data is not the same as I data. you must have complex IQ data to have I data. I data is the real portion of complex IQ data, not real data. IF data is real data.

"the 97MHz and 99MHz frequencies are probably aliasing."

There is no aliasing going on here.

Regards,

Andy Hinde

RF Systems Engineer

National Instruments

[COMSIS]

Hi

Really good and explanatory answer. Just a little pinch for confirmation since lots of Probably and Possibly been

there in the reply posts. As I think so 5600 gives 20 Mhz real bandwidth to 5142 which in turns make it complex at OSP NCO multiplication stage So weather I am going to use RFSA to acquire complex data or I am going to use Niscope for complex mode data the results will be same(in terms of getting I out of complex).

 What I under stood from last post is when I set My OSP mode to real data setting my Sampling rate to10Mhz I am getting 4Mhz bandwidth (since for real case BW=IQ(rate)x0.4) which is having no aliased or reflected signal and when doing Freq translation and DDC enable then this 4 Mhz bandwidth comes to 0 Mhz - 4 Mhz depending on my DDC CF (e.g. my IF is 10Mhz and I set DDC C.F to 10Mhz). The data is perfectly fine and can be demodulated perfectly without any doubt of aliasing and reflection just wana be sure about data integrity that's all.

Regards

[Andy_Hinde]

Hello,

Re: 5142 real data

07-20-2010 02:52 AM

Hi

"As I think so 5600 gives 20 Mhz real bandwidth to 5142 which in turns make it complex at OSP NCO multiplication stage"

This is correct. The 5600 outputs a 20 MHz wide IF signal, centered at 15 MHz, to the 5142 input. The 5142 OSP performs digital downconversion, filtering, and decimation to provide you complexx IQ data, centered at 0 Hz, with the requested final IQ sample rate. The bandwidth of the IQ data is 80% of the requested IQ sample rate, and is filtered down from 20 MHz to the end bandwidth in the OSP.

"So weather I am going to use RFSA to acquire complex data or I am going to use Niscope for complex mode data the results will be same(in terms of getting I out of complex)."

Do not use NI-Scope for programming this. You are using the PXI-5661, which is a combination of PXI-5600 and PXI-5142, and is programmed via NI-RFSA. NI-RFSA handles all the details of perating the two devices together, such as applying calibration correction for the PXI-5600 IF passband and the 5142 frequency reponse. It also makes sure the 5142 OSP is always tuned to the exact IF output center frequency of the 5600. Using NI-Scope to do this is theoretically possible, as this is what is done within the NI-RFSA driver (under the hood), but it will only complicate your situation without providing you any additional benefits. Everything you want to do is possible via NI-RFSA.

" What I under stood from last post is when I set My OSP mode to real data setting my Sampling rate to10Mhz I am getting 4Mhz bandwidth (since for real case BW=IQ(rate)x0.4) which is having no aliased or reflected signal and when doing Freq translation and DDC enable then this 4 Mhz bandwidth comes to 0 Mhz - 4 Mhz depending on my DDC CF (e.g. my IF is 10Mhz and I set DDC C.F to 10Mhz). The data is perfectly fine and can be demodulated perfectly without any doubt of aliasing and reflection just wana be sure about data integrity that's all."

It appears you are still approaching this is a dificult manner. I am attaching an NI-RFSA shipping example (LV 2009 format) that is all you really need. You can set your center frequency to 97 MHz (or whatever you need, basically the center of the spectrum containing your multiple FM signals) and set your IQ rate to 125% of the desired bandwidth. If you want to acquire IQ data with 5 MHz of BW, set your IQ sample rate to 6.25 MSPS. The NI-RFSA drier will return you corrected (for calibration) IQ data with 0 Hz representing the configured RF center frequency of 97 MHz.

What you would need to do next is simply add Modulation Toolkit code to operate on the complex IQ data output from 'niRFSA Read IQ' in the VI. There may be an FM carrier right at 97 MHz, there may not be, I'm not making assumptions about your situations. But all you need to do is use MT Downconvert Passband and MT Upconvert Passband to shift the 5 MHz (or whatever) wide IQ data in frequency so that it is centered on the desired FM carrier. If you have a carrier at -2 MHz (meaning it was at 95 MHz originally) you would use MT Upconvert Pasband to shift this 2 MHz higher in frequency to be centered at 0 Hz. You would use the Passband Bandwidth input to filter out all other FM carriers, leaving you with IQ data centered on an FM carrier (at 0 Hz) with all other FM carriers filtered out.

Another option you have is to simply set your RF center frequency and IQ sample rate controls for NI-RFSA to acquire a single FM carrier from the start, which would allow you to demodulate it directly. I am attaching MT niRFSA Demodulate FM to show how to do this. The difference between the two approaches is in one case you're grabbing all the FM carriers at once and processing them individually in SW, versus the other case where you both acquire and process the FM carriers one at a time.

Regards,

Andy Hinde

National Instruments

[COMSIS]

Tnx Andy

For a detailed post. Why I am insisting on Ni_scope or 5142 real acquisition. What I actually want to do is I want to

demodulate multiple FM signals in My 4MHZ bandwidth using a bandpass filter for every signal. Now what happen here in RFSA case is i have to do upsampling then Upconversion  then downconversion and then band pass filtering  So basically 4 steps are involved which actually slows down the Vi and I think so there is a possibility that Backlog starts to increase and digitizer memory over flows (in RFSA case)So using NI-scope I was thinking that I will get the 4MHZ downconverted between 0-4MHZ inside my 5142 and all I need to do is bandpass filtering for demodulating my signals My demodulation process will avoid 4 steps and only 1 will be required any possibility what should be DDC

 frequency and What should be enabled or disabled. I have attached a  rough depiction of what I am aiming to do

[Andy_Hinde]

Hi COMSIS,

You can do exactly the same thing in NI-RFSA that you describe. Just center your original RF center frequency below the lowest frequency carrier so all your carriers are centered above 0 Hz at baseband.

Example - Assume you have FM carriers at 95, 97, and 99 MHz.

Set your RF Center frequency to 94 MHz and capture 12 MHz of BW (to cover 88 to 100 MHz). To get 12 MHz of BW, you need to set your IQ Rate to 15 MSPS.

This puts the RF carriers at +1, +3, and +5 MHz in your complex IQ data returned from NI-RFSA Read/Fetch.

Then take the IQ Data out of the Fetch/Read routine and brach off four paths to four (Downconvert Passband VIs + Demodulate FM) from Modulation Toolkit. Set the Center frequency parameters on the Downconvert Passband VIs to shift each carrier down to be centered at 0 Hz.

I want to impress upon you there is no beneift to using NI-Scope in the use case you are describing. Hope this helps.

Regards,

Andy Hinde

RF Systems Engineering

National Instruments

[COMSIS]

Hello

THX for the reply I have been testing what u have suggested  using RFSA . But the problem is now that vi crashes

with digitizer memory overwrite error. My Vi only includes one downconversion vi with no FM demodulation vi.

How would I incorporated 3 or 4 downvconversion if vi returns an error within one minute of acquisition. I have tested it

with producer consumer loop which makes it more obvious that in consumer loop where I downconvert the signal runs more slower than acquisition loop. Due to which elements in queue starts to increase as a result of which memory backlog starts to increase and ultimate result is " ERROR" have a look at my VI if u have any suggestions this regards plz suggest. i am acquiring 4Mhz bandwidth with center at 100MHZ. Vi attached is amended from RF Stream and Playback proj.

Regards