The Rohde & Schwarz RTM3004 which I was fortunate enough to RoadTest has been a very capable and versatile instrument. Aside from doing serious work with it now that I’ve been working from home, it even featured in a number of “fun” posts such as my “making music with an oscilloscope” submission to the Project14 Acoustics challenge. Despite this, the review had never been entirely complete, as we were promised the RTM-K18 Spectrum Analysis option but due to a technicality, it could not be provided to units sourced in the USA.

 

Since my last update, firmware revisions V1.501 and 1.550 introduced the ability to perform new “track” functions, ability to disable date/time, configuration of colours for math waveforms, new pattern-generator waveforms. The latest V1.600 firmware adds the ability to configure action on trigger, adds support for modular probes, adds delay to trigger measurement, the ability to disable the R&S logo in screenshots and the new RTM-K37 Spectrum Analysis Option.

 

Thanks to a tip-off from a fellow RoadTester (Fred27) , I was alerted about this and after quickly reaching out to our contact at Rohde & Schwarz, I was soon on my way to evaluating the new RTM-K37 Option.

 

The Story Behind the RTM-K18/RTM-K37 Spectrum Analysis Option

At the commencement of the RoadTest in June 2018, I was expecting that the RTM-K18 Spectrum Analysis option would be included, as the RTM3K-COM4 specification includes the application bundle. Unfortunately, it turned out that this was not the case for US and Canadian units – the true cause for this was not clearly explained, but there was an initial hope that there would be a different version of this option for those markets which would be coming soon.

 

That promise was left unfulfilled for a while, with no new information forthcoming until just recently when thanks to a tip-off and careful reading of the firmware update notes, it was noticed that a new RTM-K37 Spectrum Analysis option was added. Checking the manual shows that both RTM-K18 and RTM-K37 options appeared to be identical in function and display.

 

Faced with this confusion, I reached out to our contact at Rohde & Schwarz and requested some clarification. From his explanation, it was concluded that:

  • RTM-K18 has been replaced by RTM-K37. RTM-K18 is no longer for sale, with RTM-K37 being functionally identical to the former RTM-K18. Those with RTM-K18 might see their option automatically changed to RTM-K37.
  • US and Canadian RTM3K-COM4 units did not have RTM-K18 included with their application bundles, but moving forward, will have RTM-K37 included worldwide. Those who have purchased an RTM3K-COM4 or RTA4K-COM4 who do not have access to RTM-K37 can request a license free of charge from their Rohde & Schwarz representative.

 

Installing RTM-K37

Installing the option begins with updating the firmware to 1.600. This is achieved by copying the correct .FWU file to a USB flash drive, inserting it into the oscilloscope and navigating to the Firmware Update page and executing the update. This process takes a few minutes, after which the oscilloscope reboots automatically.

After this, assuming you have a key for the option, you need to head to the Options page and Input option key manually. Assuming all goes well, the key will be accepted and the option will be added onto the list, unlocking its capabilities.

 

RTM-K37 Features and Interface

To access the Spectrum Analysis features, you need to open up FFT mode. There is no separate application for this mode, instead, it is enabled in FFT mode by the Spectrogram toggle. This also brings a number of other features including Automatic RBW, Peak List and Display configurations.

The display menu allows you to change the spectrum colour, toggle magnitude mode, automatically determine the signal thresholds for colouration (or manually define them below). The marker options allow for you to change the levels and marker types which are useful when using the peak list function.

Different types of reference marker are available for selection, and other advanced peak list search options are also available. This is detailed in their updated User Manual, now up to V.08.

The various colour palettes are displayed above. Each pane can be moved using the drag handles, which means that you can have just the FFT, just the waterfall, just the time-domain representation or any combination of the above. The bars allow for access to the parameters of the FFT/spectrum, with new values keyed in using the on-screen keypad. The value ranges permitted are limited due to the sample rate, memory depth and processing power of the oscilloscope – a key limitation is the resolution bandwidth (RBW) which seems to be sample-rate dependent (and hence, centre-frequency dependent) and may reach only as low as 144kHz at the upper frequencies.

Magnitude mode allows for the FFT trace to be coloured based on the magnitude of the signal. A curious result occurs if you turn off magnitude mode – you can still select the colour for the trace but it is locked at the highest value (i.e. red, white or yellow). By default, when Magnitude Mode is on, you can adjust the upper and lower limits for the colouration using drag handles on the FFT trace which is quite convenient.

 

The Peak List function is also quite useful, allowing for automatic searching for peaks which meet a certain amplitude criteria and automatically listing them in the table underneath. Unfortunately, it doesn’t seem to respect the viewed frequency range and seems to always start from zero. I’ve reported this to Rohde & Schwarz, so hopefully an improvement may be in the works.

One limitation seems to be that the scroll rate for the spectrum doesn’t seem to be configurable. Depending on the RBW selected and the frequency, scroll rates for the screen could range from around 37 seconds to 2 minutes 34 seconds and in extreme cases, up to almost 4 minutes. This, along with the limited RBW selections, results in limited viewing resolution for transient signals with no other analysis tools being offered for signal characterisation.

There are time cursors available for use with the waterfall display, however, these are only accessible when the acquisition is stopped. This was not entirely obvious to me at first glance.

While testing, I was not able to get anything useable from a directly attached antenna, so I went to connect the oscilloscope to an amplified antenna source. While perusing the spectrum, I came across this incident where changing the central frequency by 1MHz resulted in the appearance and disappearance of a phantom aliased signal (an LTE carrier). After consulting with Rohde & Schwarz, I learned that the unit uses the regular oscilloscope input chain, thus when changing the centre frequency, the sampling rate may also change which may cause aliasing depending on the input. However, because the oscilloscope probe input goes through the input bandwidth filter before being sampled, it is sometimes possible to avoid aliasing by engaging the probe bandwidth filter at an appropriate setting. Otherwise, it is something that perhaps users need to be aware of and take steps to avoid (e.g. by changing the centre frequency and watching the behaviour of aliased signals).

 

During testing, some visual graphical glitches were encountered and reported to Rohde & Schwarz. They have since acknowledged at least one of the causes and a fix is also in the works.

 

Analysing Some Spectrum!

Perhaps this will be a good opportunity to take a “drive” around my RF neighbourhood using the amplified antenna input to see how various types of signals look on the RTM3004’s RTM-K37 option.

My local UHF DVB-T fill-in station broadcasts in the UHF band. Here, a number of 7MHz wide COFDM carriers can be seen side-by-side separated by guard band and an empty channel.

Various LTE and UMTS carriers below 1GHz can be seen – for example, in Band 28 (700MHz), Band 5 (850MHz) and Band 8 (900MHz).

The 400MHz band belongs to various land mobile radio systems, mostly trunked digital radio with carriers of about 8-16kHz width. Unfortunately, the limited RBW means that these channels are hard to visualise and measure accurately.

Finally, the broadcast FM band with wide FM carriers of about 200kHz width. At lower frequencies which results in lower sampling rates, a narrower RBW is possible, providing the ability to visualise the FM modulation directly. However, the scroll rate does not seem to be configurable and narrower RBWs typically result in slower scroll rates.

 

Comparison with a Dedicated Spectrum Analyser

As a former RoadTester for the Tektronix RSA306 Real-Time Spectrum Analyser, I felt it would be a good exercise to compare the RTM-K37 option to what a dedicated RF spectrum analyser would be capable of.

Given that I had just recently done a search for NB-IoT carriers on the air with the RSA306, it was probably good to try the same thing with the RTM-K37 to see how it would fare. Focusing on the Telstra NB-IoT carrier, we can see it to be the small “shoulder” on the right side of the LTE carrier. Not very clearly identified, in part, due to the limitations of the hardware – the minimum RBW was selected.

Contrast this to the RSA306’s performance which allows us to see the NB-IoT carrier clearly off the side of the LTE carrier, along with the Optus LTE carrier below. It was even possible to get “closer” as an RBW of 1kHz could easily be achieved in DPX view. As a result, it seems that the limitations of the RTM-K37 option makes it more difficult to analyse narrow-band signals due to the limited RBW capability. This can be seen in some of the screenshots in the previous section where land-mobile radios with 8-16kHz bandwidth emissions are hard to visualise clearly.

 

Another key metric that is used to compare spectrum analysers is the Displayed Average Noise Level (DANL) which represents the noise floor of the spectrum analyser. This is usually measured with the input terminated in a 50-ohm resistance at the narrowest RBW setting at the widest bandwidth with the lowest reference level. As a result, I used a 50-ohm BNC terminator on the input of Channel 1, set to 50-ohm termination, and the most sensitive setting of 500uV/div.

The average noise level is about -108dBm at an RBW of 145kHz. If my calculations are correct, the difference between 145kHz and 1Hz is a difference of 5.16dB, so the DANL is around -113dBm/Hz. This figure is probably not bad considering that it is from an oscilloscope input channel, but compared to dedicated spectrum analysers which often claim -150dBm/Hz to -160dBm/Hz, this explains why receiving off-the-air signals from an unamplified antenna was a bit of a fruitless exercise.

 

Conclusion

The RTM-K37 Spectrum Analysis option adds the capability to perform spectrum analysis tasks, supercharging the FFT mode with a waterfall display and adding features such as a peak search and magnitude mode colour display. This makes identifying time-varying phenomena in the frequency domain a lot easier than trying to mentally work it out from a wiggly FFT trace. For the most part, it functions as expected with some room for improvement, although suffers from some limitations such as limited minimum RBW/span settings and inability to configure scroll rates. It analyses a portion of the memory buffer to display the spectrum, but is not truly time-frequency correlated. In some rare cases, the screen did show some unusual behaviour which was resolved after a reboot and aliasing was observed at certain frequencies due to automatic changes in sampling rate depending on centre frequency selection which could be reduced somewhat with a change in probe input filter bandwidth settings.

 

However, it is important to remember that this option is a software feature, thus the capabilities are ultimately limited by the characteristics of the oscilloscope’s input signal chain. To that effect, the input is not optimised for low-level RF signals, thus observing signals from an unamplified antenna can be quite difficult due to the high displayed average noise level (DANL) of -113dBm/Hz as compared to a proper RF spectrum analyser which may have as low as -160dBm/Hz. As a result, perhaps it is best used where the sensitive inputs of an RF spectrum analyser may be overwhelmed – e.g. for analysing the EMC potential of switching waveforms or the output of sub-GHz transceiver ICs. Likewise, the bandwidth available for spectrum analysis is also limited by the bandwidth of the oscilloscope, limiting it to applications below 1.2GHz.

 

Looking at their competitors, it seems the Tektronix 3-series MDO have gone a different route, with their spectrum analysis feature “shoehorning” a proper spectrum analyser into the same box. While this provides the full RF spectrum analyser experience including a wider bandwidth capability, it does also mean a separate input channel, requiring separate RF probes that connect to the N-connector and the care in ensuring signal levels are not too high. It shares the disadvantage of not being time-frequency correlated, with this feature being reserved for the higher-end models.

 

As a result, the RTM-K37 option is best seen as an enhancement to the FFT mode by providing some spectrum analysis capabilities. There are still many good reasons to keep a conventional RF spectrum analyser on your bench. As for which approach is best – that really depends on your needs. After all, I heard Rohde & Schwarz are famous for their spectrum analysers …