RoadTest: Rohde & Schwarz Oscilloscope Kit RTM3K-COM4
Author: Fred27
Creation date:
Evaluation Type: Test Equipment
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: The competitor to this that springs to mind is Keysight's 3000T X-series. As part of my review I'm comparing it to lower budget 'scopes to illustrate what improvements you can expect as price (and hopefully quality) goes up.
What were the biggest problems encountered?: Only the huge range of features and applications to get my head around - if this can be considered a problem.
Detailed Review:
I'm sure most Element14 community members are very interested in this oscilloscope but in all honesty is probably not something many of us can justify purchasing for ourselves. If that sounds like you and you just want to get an idea of how impressive this piece of test equipment is, then my Overview Video is for you. It's everything I can summarise about the RTM3004 all wrapped up in one 30 minute video.
If you want to know all the details or you're at work and can't watch videos without disturbing your colleagues, then don't worry. I'll cover everything in a written review. You can skip the overview video and you won't miss anything. Head straight for the written review. I'll describe everything I can with just text, photos and screenshots. You may want to check out the video where I'm covering usability though. That's hard to get across without.
I’m also going to try to save your time throughout this road test by not telling you about things that would apply to any ‘scope. Where appropriate I’m going to contrast it with those for a more modest budget – a Rigol DS1074Z (the entry level 'scope everyone's familiar with) and Keysight DSOX1102G Product LinkProduct Link (a more mid range choice). Obviously, these aren't direct competition to the RTM3004 but I'd like to show what improvements you see as quality and price go up. I'm sure many Element 14 community members will have equipment of this calibre so I’m hoping that contrasting with these should be more useful that an isolated review. As the RTM3000 series is the next step up from the RTB2000 series Product LinkProduct Link I’ll also mention where I see significant changes from its younger sibling. It's most obvious direct competitor is the Keysight MSOX3014T Product LinkProduct Link but unfortunately I can't give you a direct comparison with that.
I’m also going to spare you an unboxing. The RTM3004 came in a cardboard box much like any other ‘scope. It was packed well enough to survive the abuse that UPS gave it, with some nice dense foam and none of that expanded polystyrene that really fills up your bin and can’t be good for the planet. If you really have a burning need to know any more about the box, please ask!
I’m obviously very grateful to Rohde & Schwarz and to Element14 for giving me the opportunity to test this amazing ‘scope. It’s a step above the sort of equipment I’m used to. However, I’d be doing you all a disservice if I let this colour my review. If there’s a reason not to give it 60/60 then I’ll find it. I’ll try my best to be an ungrateful git and review it like I spent my own hard-earned cash on it. Talking of which, if you do spend your own cash on it (and do so before the end of June 2018) you’ll also get a 3GHz FPC1000 Spectrum Analyser Product LinkProduct Link bundled in with it. This particular spectrum analyser has been really well road tested here on Element14 already. 3GHz Spectrum Analyzer - R&S® FPC1000. Hey... if I’m reviewing it like I bought it, then where’s my free Spectrum Analyser? (Only joking.)
If you're getting bored or there's a particular bit that interests you, here's some bits you can skip forward to:
00:00 | What you get. A look around the RTM3004 and accessories. |
06:15 | Switching it on. First impressions and general use. |
12:09 | Exploring the 10-bit ADC and memory depth. |
16:15 | Digital channels and protocol decoding. |
17:44 | Power analysis features. |
20:55 | FFT features. |
24:57 | Connectivity and remote control. |
28:15 | Summary. |
The 'scope itself is fairly unassuming. The large screen is the only thing that hints that it’s something different from a more modest piece of test equipment. The fact that there’s only one set of vertical controls (more on that later) and the lack of buttons around the sides of the screen give the front a nice simple clean look. This is also helped by the fact that the logic analyser ports are now round the side and use a tidier HDMI style connector. Its little brother the RTB2000 series had 0.1” headers on the front panel and whilst these are fine they are likely to be unused a lot of the time and make the RTB2004 look a little unfinished.
Round the back you’ll find a USB 2.0 type B jack to connect to a PC, gigabit ethernet for LAN connection and an AUX out BNC connector for pass/fail and trigger output. It’s nice to also found a hard power switch here in addition to the soft switch round the front. I’m not keen on devices with just a soft switch as I like to ensure things are properly powered off when not in use. My Keysight E36312A power supply is a good example of how not to do this. Power down using soft switch and it will still quietly tick away consuming 10W. There’s no way to actually switch it off except at the mains plug. Getting back to the back of the RTB3004 there's an interesting panel cutout that looks like it would hide some interesting expansion. Unfortunately not. There's nothing under there and it doesn't seem likely there would ever be. The RTB2004 shares the same case and the same unused cutout. Perhaps other equipment also shares the same case and actually makes use of that cutout.
This 4-channel version comes with 4 nice 500MHz passive probes. Obviously, there’s just 2 for the RTM3002. They’re fixed 10:1 so no more accidentally changing to 1:1. There’s a pin on the outside edge of the BNC connector that connects to the outer ring on the ‘scope to auto set the 10:1 range and switch the inputs 1Mohm if you've previously been using them as 50ohm. This pin is spring loaded and doesn't interfere with or scratch anything if you decide to use them on another 'scope. They come with all the usual probe accessories that I won’t bore you by listing. Underneath the usual sprung hook you’ll find a nice sharp gold-plated pogo-pin type tip. Some people don’t like sprung tips so there is a solid tip supplied too. The only thing worth noting here is that the channels on the ‘scope are yellow, green, orange and blue. The plastic rings are yellow, green, pink and blue. What happened to orange? Not a big deal really, but I said I’d be picky!
There are some further probe accessories that are available from R&S but not included. Things like caps for probing various pitch ICs, spare coloured rings (still no orange). I thought these might be useful so I went online to buy a pack. Over £200? Can that be right? For not much more than some plastic and springs? A plastic cover for the front Product LinkProduct Link is £68 and a soft carry case Product LinkProduct Link is £173, so maybe that is right. I con't find anyone who has them in stock anyway.
If you’ve gone for the RT-ZL04 Logic Analyser option you’ll have two 8-channel logic pods. These connect to the side of the ‘scope with HDMI B connector. Each one has 8 smaller connectors with a ground and signal connector and these can be connected to your device under test with either a grabber (8 supplied per pod) or directly to a pin header. They don't seem to be available as just one 8 channel pod even though this would in theory be possible. The logic analyser channels are 400Mhz (rather than 250MHz for the RTB2000). Perhaps this is why the ribbon cable on now coaxial.
You get a small R&S branded “pencil case” to store your accessories in. Apparently, it can be used for probes but I feel it’s a little small for that. The lead would be strained a bit too much for my liking. The little bits and pieces like trimmers and ground springs – yes. Logic analyser pods – yes. Probes – I don’t think so. They have their own sturdy Ziploc bag anyway. You'd be better off using that. I'm sure everyone has their own preferred probe-management procedures anyway. Stick with that.
It boots up fast. It’s reasonably quiet. Wow - that screen is impressive! The only minor niggle that is initially apparent is that the screen is shiny. You can clearly see me and the camera in the first photo below. I couldn’t use a flash or extra lighting as it reflected back badly. I think I’m OK with a shiny screen but suspect this is going to annoy some people. Other R&S devices have matte screens so it's not a company-wide approach. I'm thinking of the FPC1000/1500 that have been road tested here on E14, but they aren't touchscreen devices. The screen is a 10.1” diagonal so I think it should be reasonably easy to get a tablet screen protector to reduce glare and fingerprints if you wanted.
So, now it's powered up, let's see what it can do. I decided to tackle this test by going with the age old male tradition of not reading the manual unless I had to. If something isn't obvious or it doesn't work straight away then I'll complain that it's broken and then perhaps read the manual. I'm hoping that it will be intuitive enough for that not to be necessary and that the RTM in RTM3004 doesn't actually stand for Read The Manual. (If it did then I'm sure they'd have called it the RTFM3004.)
{gallery} First impressions |
---|
Lovely screen: Perhaps a bit shiny for some tastes though |
Trace display: Showing a sample signal from the generator |
Probes: Very nice, but the accessories are expensive! |
Logic probes: One half of the set (connected to a development board) |
Round the back: Nothing too surprising. Unfortunately that interesting panel cutout goes nowhere. |
As far as a first impression goes there are a few things that you can directly compare – how fast it boots up to a usable state and once it has, what you can see and hear. The results for all are pretty much in the order you’d expect. For noise measurements I’ve only really got a relative figure. I don’t have a professional setup but I used a sound meter on my phone from 6” away. There was very little background noise (the odd bird tweeting in the garden) but this read 10dB. Take all these dB figures as a relative indication rather than an absolute.
Model | Boot-up time | Noise level | Screen size | Used for waveforms | Price range (excl. VAT) |
---|---|---|---|---|---|
Rohde & Schwarz RTM3004 | 12s | 11dB | 10.1" | 9.2" | £2,912 - £17,439 |
Keysight DSOX1102G | 21s | 16dB | 7" | 5.6" | £363 - £1.115 |
Rigol DS1074Z | 37s | 18dB | 7" | 5.3" | £308 - £617 |
The RTM3004 comes out on top by a fair margin. It takes 12 seconds to boot up to a screen that is displaying traces - with settings remembered from when it was shut down. This applies whether using the soft button or from full power off. There is initial burst of noise from the large (3.5” diameter) fan for a second or so as it boots up – reminding me of how PCs tend to do that before the thermal management kicks in. Then it drops to a very quiet 11dB (from the background of 10dB). Obviously, the fact that the sound level reduces means that it is possible it may increase again under load. The fact that I don't remember the fan kicking in throughout this road test indicates that either it doesn't or that it remains fairly quiet. At its nosiest moment it was still quieter than the competition. That lovely screen has a 10.1” diagonal with a full 9.2” used for displaying traces rather than menus, etc.
This is obviously the most expensive of the three too. It wouldn’t be fair to compare without an idea of what you’re paying for this. The base 100MHz 2 channel RTM3002 comes in at £2912. (I’m going to go with UK prices excluding VAT throughout.) This particular configuration 1GHz 4+16 channel with all software options retails at a significantly increased £17,439.
By comparison the DSOX1102G takes considerably longer at 21 seconds to fill its 7” screen with 5.6” of wiggly lines. It’s significantly louder at 16dB and this noise level does not seem to depend on usage. The fan is smaller at 2.25” which I’m sure is responsible for this. Smaller fans have to spin faster to move the same amount of air. Whilst comparing the obvious, the Keysight is sold as a 2 channel scope with external trigger. In fact, it wouldn’t be fair to call it anything other than a 2+1 as the external trigger can be used as a digital channel in exactly the same way the RTM3004 16 digital channels can. The "external trigger" input can be displayed and it can be used to decode as a serial bus. It really is a full digital channel once you buy a third probe. I found a cheap one from eBay was enough for this digital-only functionality.
The lower end of Keysight’s 1000-X series comes in at £363 for the 50MHz EDUX1102A but can increase to £1,115 for the 100MHz DSOX1102G with all options.
Continuing the “you get what you pay for” theme we have a 37 second boot up for the DS1074Z to use 5.3” of its 7” screen. Again, it’s louder at 18dB. People do complain about this 2” fan and often replace it.
The lower end of this series is the DS1054Z which retails in the UK for £308. Let’s be fair to the Rigol – the top end DS1104Z with all decoder options rarely costs anyone more than this. You know what I mean. If you want the 4+16 channel MSO option you will have to pay at least £617 for a MSO1074Z.
The RTM3000 is available from 100MHz all the way up to 1Ghz. Whilst bandwidth is obviously really important it’s not something I’m going to dwell on. This is for a number of reasons.
• The specification pretty much covers it. If you need a certain bandwidth then you know you need it and you’ll pay for it.
• I can only compare against my 100Mhz ‘scopes, and there's no way to temporarily uninstall the preinstalled 1GHz option.
• The included probes are 500MHz and this will be the limiting factor. You’ll really need an active probe to go above 500MHz anyway and this is something you’ll certainly have to factor into the cost of your setup.
Now this is something I can take a close look at and I feel is a valid comparison with the other two 8-bit ‘scopes. This test would be valid on any RTM3000 model - from the base 100MHz 2-channel version up. In order to demonstrate what a difference it makes, I’m going to return to a road test I did a while ago for a DC/DC convertor. At the time I wanted to measure the ripple on start-up and looking back on this I don’t feel I did a great job. I'll repeat this measurement again and compare the results when using each of the 3 'scopes. As my initial road test used the DS1074Z that's where I'll start
Here’s what I showed using my DS1074Z. It's not great, is it? Now I could have switched to AC coupling, cranked up the sensitivity and shown the steady state ripple (as I did in the third screenshot). For the sake of comparison, lets stick with trying to measure the start-up ripple and overshoot. Here you can see that if I have the vertical sensitivity low enough to capture the whole 0 to 3.3V+ range then the detail of the ripple is lost. Switching to AC coupling in the final screenshot shows that the ripple is there and the 'scope can measure it, just not alongside the whole start-up waveform.
{gallery} DS1074Z 8-bit ADC |
---|
DS1074Z Capture: Capturing the output voltage (yellow) and power good (blue) as we switch on. |
DS1074Z expanded: Expanded timebase, but showing low vertical resolution |
DS1074Z AC Coupled: AC coupling shows steady stage ripple, but that's not what we need. |
The Keysight is another 8-bit ‘scope. You can see things are better, but there’s not a lot of detail in there at 500us/div. Switching to 50ns/div reminds me of that old 80s video game Defender. The peaks and troughs are there an you could probably even measure the size to some degree but the detail is missing. Once again, AC coupling to inspect steady state ripple is much better.
{gallery} DSOX1102G 8-bit ADC |
---|
DSOX1102G Capture: Capturing the output voltage (yellow) and power good (green) as we switch on. |
DSOX1102G expanded: Expanded timebase, but showing low vertical resolution |
DOSX1102G: Defender! |
DSXO1102G AC Coupled: AC coupling shows steady stage ripple, but that's not what we need. |
I used the same setting for the initial waveform - 1V/div vertical sensitivity and 500us/div timebase. There’s a nicer higher resolution screen and a couple more divisions showing (12x10 matching the Rigol but better than the Keysight's 10x8) but I’m going to cut the other scopes some slack and say that it’s a similar result.
But then I tap the little Zoom icon (on the dedicated front panel button), drag with my finger to select the area of interest and everything changes. The bottom half of the screen shows incredible detail. It can be panned and scaled independently of the top. Zooming right down to an equivalent 40ns/div shows how much more detail we captured at the same 1V/div vertical sensitivity. Not just that, but we can still see the overall waveform. This is ideal if you’re looking at overshoot or other noise as part of a larger picture. All of this is done whilst still having a clear view of the overall trace, and no AC coupling at all! I’ll dig further into the Zoom functionality later when addressing usability, but this simple example already gives an idea of how useful and intuitive this is.
{gallery} RTM3004 10-bit ADC |
---|
RTM3004 Capture: Capturing the output voltage (yellow) and power good (green) as we switch on. |
RTM3004 Zoom: All you need to do is select the area of interest |
RTM3004 Zoom More: No need to change the timebase on the main trace - just on the zoomed one |
RTM3004 Zoom again: Keep going - more detail |
RTM3004 Still zooming: No need for AC coupling at all! |
Memory – you can never have enough. Everyone always wants more detail and over a longer time. A recent project of mine was a custom NFC reader for reading the NFC implant in my hand and then logging me into my PC if it was detected. If you're interested in this project then you'll find more details over on my blog. The more unusual part of it is a custom wire wound inductor based antenna. Other than that it’s a fairly typical microcontroller project involving a MSP430 microcontroller and a TRF7970A NFC reader. The communication between the two happens over SPI and needs fairly long pauses between commands in order to let the device power up oscillators, etc. To be honest, at the time I decided the digital side of things was better handled with a PC based logic analyser. However, I’m going to recreate this debugging scenario using an oscilloscope alone. One of the key promotional points of the RTM3004 is its 40Msample of memory - or 80Msample if only using 2 channels. The K15 History and Segmented memory option bumps this up to 460Msample although this is captured in chunks of the same 40/80Msample maximum.
OK, time to see if all this memory make the process any easier. I’m going to level the playing field a bit by using only the 4 analog channels and ignoring the digital channels for now. The goal is to see one full cycle: making some SPI calls to fire up the NFC reader, scanning (using a 13.56MHz signal) for the presence of an NFC tag, giving up when a tag isn't detected and using SPI to switch the reader off again. The whole process takes about 200ms so I set the horizontal scale to 2ms/div and the sample size to the full 40Msamples. (Dropping down to two channels would increase this to 80Msamples. The segmented memory allows this to be repeated - up to the limit of 460Msamples.)
The difference was amazing. I could see the whole capture and then zoom in on any part I needed. All I had to do was Zoom in to see the protocol decoding working perfectly and the reader output clearly visible with the frequency measured as 13.56MHz. Once again the zoom feature makes inspecting the detail of what you're measuring a doddle - whether it's vertically or horizontally. I've hardly scratched the surface on this road test but I'm tempted to say already that Zoom is the standout feature of this 'scope in general use. It's that good.
{gallery} RTM3004 memory depth |
---|
RTM3004 showing a view of the full 200ms sample: You can just see the markers at the bottom for the snippets of protocol decoding |
RTM3004 zooming in: The SPI commands that kick off the NFC reader's output - and the result (yellow). |
RTM3004 NFC signal: A clear 13.56MHz according to the frequency measurement bottom left |
RTM3004 SPI stop: Here we can see the SPI command that turns the NFC reader off. |
This ‘scope comes with 1Msample of segmented memory. Whilst sold as a 2 channel ‘scope with an external trigger, this trigger can actually be used as a 3rd digital channel. I started off using the 3 channels for SPI alone, but I also wanted to see the 13.76Mhz NFC reader output – otherwise a cheap logic analyser is arguably a better tool for the job. The traces shown are SPI clock (blue), MOSI (green) and NFC (yellow). Unfortunately whilst this is a nice device overall, the 1MSample memory depth crippled it in this test. I set it up at 20ms/div as with the others and this meant the sample rate dropped to 2.5MSa/s which wasn't enough to capture the detail. Also, having only 10 divisions horizontally rather 12, it didn't quite capture the lot. Zooming in shows that the Keysight also failed to capture the SPI signal adequately. The 13.56MHz NFC output is also unusable. Whilst it's a fail on this particular memory depth test, I added a faster sample in the final screenshot. When capturing a smaller piece of the process the Keysight work well. This after all, is exactly what segmented memory is used for and it does have segmented memory.
{gallery} DSOX102G memory depth |
---|
DSOX1102G full trace: Well, almost a full trace. The 10 divisions don't quite get it all. |
DSOX1102G SPI fail: 2.5MSa/s isn't enough to capture a 4MHz SPI signal |
DSOX1102G NFC fail: Same again. You can't capture 13.56MHz at 2.5MSa/s |
DSOX1102G comfort zone: Use segmented memory and all is good again. |
I set the Rigol to capture the whole trace. It has 12MSamples of memory but this is split across all traces. As I was using all 4 of them it has 3MSamples for each and it could sample as 12.5MSamples/s. It turns out that this isn't quite enough. As you can see from the screenshots, once I zoom in the detail isn't there. (And I miss that split zoom screen already!) The SPI signal has just about been captured but there's a real risk of missing data. The decoded output is wrong, but that's another issue I'll get to later. Also, I always thought NFC was 13.56MHz? Oh yeah - it is. The NFC output has been undersampled and is showing a false 1.78MHz signal. I suppose at least with the Keysight it was obvious that this wasn't correct. Once again, this can all be fixed by capturing smaller parts of the output, but that's not what this memory depth test was all about.
{gallery} DS1074Z memory depth |
---|
DS1074Z full trace: Looks OK from this height |
DS1074Z SPI start: Not great resolution, let's zoom in more. The decode is wrong too! |
DS1074Z NFC: I always thought NFC was 13.56MHz? |
I've never used a mixed signal oscilloscope before - instead relying on an oscilloscope alongside a logic analyser. This has always got me by, although the digital and analog realms stayed fairly separate. I might find that I checked a serial bus on a logic analyser first, only swapping over to a 'scope if I suspected the quality rather than the content of the signal was off. Once I was happy that the 1s looked like 1s and the 0s looked like 0s, then it was back to the logic analyser. I was looking forward to checking this out on the RTM3004 and found myself pleasantly surprised by how easy it was. I connected up one logic pod and attached some of the mini grabbers exactly where the probes were. I didn't even disconnect the probes for this test, although it's nice to know I'd just freed up 3 channels if I needed them.
Next I pressed "Logic" and 8 digital channels appeared. I used the "Label" menu and labelled them - Clock, CS, MOSI and MISO seemed more useful that D0, D1, D2 and D3. I used the slightly strangely named "Threshold and deskew" menu to trim things down to the 4 channels I needed. I have since noticed that the channels can also be deselected in a more intuitive way from the logic probe area marked D7-D0 at the bottom of the screen! A lot of things on the RTM3004 can be done in a many ways. As I'm sticking to my approach of ignoring the manual I sometimes find them in an unusual order.
In some ways the LA option is described simply as "there are also 16 digital channels" but there's a bit more to it than that. The fact that they're in sync with the analog channels is easy to forget but important. Obviously the processing differs between the analog and digital channels and this difference is small but potentially significant - hence the ability to deskew the digital channels. This allows you to adjust the timebase offset between a digital channel and the analog ones. The range for this is plus or minus 200ns in 400ps steps. Each of the 16 digital channels can be deskewed separately which I assume must be to allow for any skew in the device under test as I can't imagine the RTM3004's processing differs between them. I found that if I connected one analog and one digital channels to the pattern generator output then I needed to deskew the digital channel by +10.8ns. There is also the essential threshold to determine exactly what is considered to be a 1 or a 0. There are presets for this covering TTL (1.4V), ECL (-1.3V) and CMOS (2.5V) and it can also be set manually anywhere between -8V and +8V. I'd have thought that should cover most use cases. Hysteresis can be set to Small, Medium or Large without any specifics to say exactly what these are.
{gallery} Mixed Signal operation |
---|
Pressing Logic: This adds in the digital channels |
Labelling the channels: "Clock" is more descriptive than "D0" |
Disabling unused channels: I'm only using D0-D3, so I'll hide D4-D7 |
Deskewing: An offset of 10.8ns seemed about right. |
After I had the digital channels working, it was then a matter of switching the protocol decode over to use the digital channels. These channels are treated just like any other. From the protocol configuration menu it's just as easy to select D0-D15 as it is to select C1-C4. You can even select and decode on a mix of digital and analog channels if you wish. Protocol decoding isn't strictly related to the logic analyzer option at all, but they do fit very nicely together.
Another thing worth mentioning about the protocol decoding is that the trace for the decoded channel doesn't have to be displayed on the screen. There is a tiny "bits" trace displayed above the decoded output, which can of course be turned off if you don't like it. Personally I think it's better to use this and instead to hide the channel it's derived from. I was only decoding from 4 channels, but if you start approaching 16 then even this large screen will begin to get crowded. On top of that you, could in theory add all 4 analog channels, 4 decoded protocols and maybe some maths channels.
From a protocol decoding point of view I found it all very easy to use. Every signal and bus can be individually labelled. These labels appear over on the right hand side, with the fixed B1-4 and signal name labels over on the left. It's possible to decode 4 buses at once, including identical protocols at different bit rates. All the settings I could think of were there. The only setting I found necessary to use was the falling edge clock option for the TRF7970A's unusual SPI setup. This option was missing on the Rigol but there on the Keysight. That rather than the sample rate was why the SPI data in the memory depth test was wrong on the Rigol.
The options to trigger on the decoded signal are fairly extensive. In my memory depth test I was trying to capture some long running SPI communication and to do so I used a single trigger when I knew the cycle was about to start. If I hadn't done this then I might have triggered in the middle of the process or on the SPI commands that were being sent to turn off the NFC reader. I know that this whole process is started off by a 2-byte command of 0x00 and 0x20. It's easy enough to set a pattern of up to 32 bits and to specify whether each of them has to be 1, 0 or if it is to be ignored. This made it even easier to trigger on the specific SPI command that i was interested in. For SPI there is also an option to trigger on the start, end or N bits through a frame. All of the serial buses had similar triggering options and many included triggering on error conditions too. Strangely, I couldn't see any options to trigger on a parallel bus patterns.
The decode protocols are separated into a number of options - K1 includes both SPI and I2C, K3 include CAN and LIN, but all the others (UART, I2S, MIL-STD-1553, ARINC 429) are available separately. Whether you think this is a good idea (only buy what you need) or bad value (having to buy each one) does I suppose depend on how many you need.
In summary, I found the decode side very good - much better that the other 'scopes. I found the mixed signal functionality a real time saver over having to swap over between two different tools. In some respects though, if you're only looking at digital signals and have a PC nearby then a cheap logic analyser can sometimes still hold its own against an expensive oscilloscope.
{gallery} Protocol decoding |
---|
Swapping the SPI clock: Here it's using analog channel C2 |
Swapping the SPI clock: Click on "C2" and you chose the source from a choice of 20! |
Swapping the SPI clock: Now we're using D0. Next I'll swap MOSI and MISO... |
Protocol triggering: Setting a 16-bit trigger pattern on MOSI |
Protocol decoding: Note how the trigger is at the end of the 0020h 16-bit pattern |
The K31 Power Analysis option provides loads of functionality if you're working on power supplies. I found very little information on this and what I did concentrated on Switched Mode Power Supplies. DC/DC converters are mentioned almost as an aside so I decide to see what it could do with the same TPS62135 DC/DC converter I used earlier. First though, I think it's worth documenting what all the available power analysis applications are. Below are screenshots from the application menu that describe the applications just before you select them.There are 14 in all, grouped by input, output, switching ad power path. There are also tools to help with deskewing and zeroing any current or differential probes.
{gallery} Power analysis options |
---|
The input side of things seems only to apply to SMPS work, so I'll concentrate on the output measurements for my DC/DC convertor. The first is obviously ripple, which I was trying to characterise on startup for my earlier 10-bit ADC analysis. Unfortunately the ripple application only works with steady state ripple. I tried to zoom in on the area of startup ripple that I was interested in, but the analysis was done over the whole of the captured trace. To be honest this is really what most people would be trying to measure so this seems OK. (Transient response is covered in a separate application.)
I fired up the ripple application, which showed on the description screen - a voltage probe and a current probe connected to the output. I pressed again and found it odd that three channels are enabled and that channel 3 was selected for analysis. I'm not sure exactly why. Maybe it's is expected that channels 1 and 2 are still measuring input voltage and current. I had connected channel 1 to the input and channel 2 to the 3.3V output. Anyway, after switching to channel 2 I was presented with a screen showing exactly what you'd need to know. It was easy enough to switch over to channel 1 to characterize the ripple on the 16V input to the DC/DC convertor coming from my Keysight E36312A supply. In addition to screenshots I could also save this data as a "report". The RTM3004 dumped all the relevant data out as XML and also as a screenshot of just the waveform area. It also includes everything else you might need for a report - details of the equipment used, time, date, and any user entered data like the description of the device under test.
This is all very useful data, but the real bonus comes when you download the "R&S Report" generator application and fire it up on your PC. This takes all the data and formats it all as a customizable PDF. The default report configuration has R&S branding and minimal styling but would be easy enough to apply any corporate branding required. Note that on Windows you must run the installer for R&S Report as administrator or it will error.
{gallery} Ripple measurement |
---|
Ripple analysis: The start screen |
Ripple analysis: Input voltage |
Ripple analysis: Output voltage |
Ripple analysis: Report PDF |
Spectrum analysis of the output was similarly easy. Selecting the option splits the screen 3 ways, the top section being the traces, the second being the FFT output and the third being tabular output for the power analysis. Not much to say here other than it was easy and my DC/DC converter didn't really have anything nasty up its sleeve. As with ripple, it's possible to easily generate a professional looking report. This clearly makes use of the FFT application that I'll cover later. That middle third of the screen looks very familiar.
{gallery} Power spectrum analysis |
---|
Power Spectrum Analysis: The start screen |
Power Spectrum Analysis: Input voltage |
Power Spectrum Analysis: Switching to the output voltage |
Power Spectrum Analysis: Output voltage |
I tried following the instructions to measure the transient response of my DC/DC converter. However, I only managed to get myself into a situation where things didn't seem to match the instructions in the user manual. Menu items that were listed in the instructions didn't quite seem to match what was on screen. Trying to adjust the Top Level and Base Level voltages causes the value setting dialog to pop up and then immediately disappear. Channel 3 kept being enabled whether I intended to use it or not. The instructions in the manual were minimal and I spent a while assuming that I hadn't understood them properly.
In the end I decided that one small item was the cause of my issues, and that was "Required probes: Differential voltage probe". Now I was only trying to measure the output of as isolated DC/DC converter so I assumed that one of the regular probes would do the job just fine. I can't really see why it wouldn't. My best guess it that there is an assumption that a differential probe is being used and that certain items (like the Top and Base Level voltage settings) are just refusing to work unless they find a Rohde & Schwartz differential probe connected to the R&S probe interface. To be honest this is the first point where I can say that the usability here is sub par. If a differential probe is required (and I don't see why it should be) and it must be used on channel 3, then it would be much better if the UI let me know that, rather than just behaving strangely.
The RTB3004 comes with a Fast Fourier Transform application as standard which works up to a nominal 1.2GHz. I would have liked to check whether the FFT frequency range is limited by the purchased bandwidth, but unfortunately (for this road test at least) this model came with the 1GHz option pre-installed. As the FFT is a mathematical analysis of the stored time domain waveform, I'm going to jump to the conclusion that you can only get a reasonable FFT output for the bandwidth range of your 'scope. If you can't see it in oscilloscope mode then you won't be able to see it in FFT mode.
My current project is porting the same custom NFC reader over to TI's sub-1Ghz CC1310 ARM microcontroller and this seemed like an ideal test device for a 1Ghz FFT / Spectrum Analyzer. I decided that I needed to bypass the 500MHz probes and look at the full range. The first thing I tried was to attach a simple 868MHz whip antenna directly to the 'scope and switch Channel 1 from 1M ohm to 50 ohm input impedance. The most flexible way I found to output a known and controlled test signal was to use TI's SmartRF Studio 7. With a few clicks I could set the device to sweep from 800MHz to 900MHz in 10MHz steps. Ideal for testing the FFT functionality.
The FFT initially claims a maximum range from 0 to 1.2GHz. However, when you start changing this range things get a little odd. On my initial playing around with the 'scope using firmware 1.200 I managed to get the device to span negative frequencies. Nothing was displayed until 0Hz, but it was possible to select this odd range. Firmware version 1.300 came out shortly afterwards with a fix for this, but I'm not sure the range selection is completely sorted. Depending on the lower and center frequency, sometimes I managed to get an upper limit of 1.2GHz, sometimes 1.8Ghz. I even got 2.3Ghz offered as an upper limit a few times, but this never stuck - it reverted to 1.8Ghz when selected. You can see this quirk in the first screenshot. Is 600MHz - 1.8Ghz a valid setting? Unfortunately I don't have a reliable RF source over 1GHz to test if this works. I couldn't actually find anything in the manual to specify the maximum frequency range. However if I send the SCPI command SPECtrum:FREQuency:FULLspan then the range is reset to 0 - 1.2GHz. This makes me suspect that being able to select 1.8GHz is a bug. Being offered 2.3GHz definitely is.
The display defaults to a Flat Top window which definitely provides a nice fast display. The other options available are Hamming, Hanning (shown), Blackman and Rectangle. In the screenshots I've also turned on the peak hold functionality so you can see each of the swept peaks.
{gallery} FFT analysis |
---|
FFT range error?:Allowing selection of 2.3GHz |
FFT Flat Top: Including the max hold (blue) |
FFT Hanning: Clearer peaks shown |
DSOX1102G: Spectrum analysis from 0 - 20kHz showing 10kHz signal generator output |
The FFT performance is definitely a step above that on the DSOX1102G although with a bandwidth limit of 100Mhz on the DSOX1102G and a minimum range of 0 to 340MHz on the RTM3004 it was difficult to make a meaningful comparison. I've added a screenshot of the DSOX1102G's FFT display for comparison but, well... there's just no comparision! I couldn't find any signal that would be high enough to register on the RTM3004 (other than at the left hand edge) but not too high for the DSOX1102G. In fact, if you need FFT on a signal below about 50MHz then you may need the Keysight!
It was useful being able to attach a whip antenna directly, but obviously this isn't ideal if you're trying to measure a signal amongst ambient RF background noise. I know my Tado thermostat also uses sub-1Ghz radio and it could throw off any measurements I make if it decided to chat at just the wrong time. I decided the easiest way to cut this out and another test of the RTM3004's RF capability was to couple the transmitter output directly to the 'scope. The CC1310 Launchpad uses what I thought to be a uFL connector but it turns out it's a JSC connector and getting cables / adapters for this is awkward and expensive. I decided to buy one of TI's CC1312 Launchapds which has a beefier ARM M4 processor but more importantly a nice standard SMA output. At least it does once I'd moved an 0402 capacitor to swap from the on board PCB antenna to the SMA output. This capacitor really didn't want to move. Maybe it was glued down before soldering? Anyway hand-soldering 0402 is tricky enough at the best of times without it needing a reasonable amount of force before it..... Damn! Where did that go? Yes - I had to play "hunt the 0402". The things I do to produce a road test for you all!
Anyway, once it was located and resoldered I was off. A quick check of the same FFT shows a very similar output. It looks a bit cleaner and you can see that each of the peaks in the blue max hold trace are now at same height. This would indicate that the aerial was indeed picking up a small amount of noise and that it has a better response at the lower end of the range. The improved accuracy and noise immunity would help a lot once I was looking at the more detailed Spectrm Analysis features.
Rather than being a real swept spectrum analyser, the K18 option is more of an enhancement to the FFT functionality, and as such it appears as extra menu items under the FFT application rather than as something completely separate. Even the manual states that "the option R&S RTM-K18 offers additional functionalities to the basic FFT calculation: Spectrum mode, Spectragram, Peak list and markers".
To be honest in comparison to the Keysight's FFT, the RTM3004's FFT is starting to look like a real spectrum analyser anyway. With the K18 option installed it should get closer still. From the manual again "The spectrum mode increases the dynamic range and the update rate. In spectrum mode, no time-based waveform is acquired for the channel of the spectrum source, and this time-based waveform cannot be analyzed." Whilst technically it may be enhanced FFT, I don't feel that describing it as a spectrum analyser option is misleading.
Whilst the FFT functionality is standard, full Spectrum Analysis requires the K18 option which at the time of writing is not yet available in North America yet. I'm in the UK, but unfortunately it also wasn't present on this road test model as it had been shipped via Rohde & Schwarz in the US. I know that rich@rohdescopesusa tried his best battling corporate bureaucracy to get the option enabled in time for this road test but unfortunately it looks like I'll have to skip reviewing the Spectrum Analyser for now. When the K18 is available to the US, then I'll certainly come back and add to this.
Well, it may have been almost 2 years in the making, but it seems that a new K37 option has replaced the K18 that was missing from the US units, and R&S have kindly supplied me with this new option. The RTM3000 user Manual has separate sections for the K18 and K37 options but they appear to be the same. When I check an earlier version and the only difference that I could see was that K18 used to have "Spectrum Mode" but that has now disappeared. This mode apparently "increased the dynamic range and update rate at the expense of removing the time domain display" (i.e. the normal oscilloscope trace). Never having had K18 I can't say what difference this made. I suspect that it's permanently in "spectrum mode" but the time domain now remains.
The option extends the standard FFT functionality rather than replacing it. It's still the case that all analysis is done on the stored trace. This can indeed be quite useful as it's possible to catch a transient signal and analyse the frequency domain. The standard FFT menu has options for Source, FFT Window, Automatic Resolution Bandwidth, Waveform and Vertical Scale (as you can see above). Installing K37 adds options for Spectrogram, Peak List and Display.
This is the primary functionality of this option. It lets you see how the frequency domain analysis of the signal varies over time. You can chose a colour scheme (using the new Display menu item) independently for the frequency domain and the spectrogram sections of the display. When running, the spectrogram display scrolls upwards showing you the most recent data at the bottom.Here's it's easy to spot how the amplitude (colour) and frequency (horizontal position) vary over time. I couldn't see any way to vary the speed at which it scrolls. Rather than trying to explain, it's probably easiest to show you a few different sorts of signals and how they are displayed. Unfortunately I couldn't show Frequency Modulation as the RTM3000 signal generator's upper limit for FM is lower than the spectrum analyzer's lower limit. Unfortunately I don't have a suitable RF signal generator. That would however be a nice red sine wave, which R&S demonstrate nicely in their K37 demo video. https://www.youtube.com/watch?v=glmBMksk4Ps
{gallery} Spectrogram |
---|
Amplitude modulation: This signal starts steady (solid line) but then AM begins. |
ASK: This shows Amplitude Shift Keying at a fixed frequency |
FSK: Frequency Shift Keying can be shown as the signal hops between two frequencies |
Frequency Sweep: The signal is swept from one frequency to another |
Sub-1GHz frequency hopping: This is a TI CC1312 hopping between frequencies in the Sub-1GHz band |
Obviously it's hard to convey a display that changes over time with just a still image. Here is the same CC1312 hopping between frequencies in real time.
This option turns on the automatic sensing and display of the peak frequencies of your signal. With peak list on, the Marker option appears to allow you to configure this. You can set thresholds, whether it uses the current spectrum signal or uses the peak hold, average, etc. You can also set whether the markers represent a single frequency or whether this is taken as a larger span covering potential sidebands.
Display lets you independently set the colour scheme for both the spectrogram and frequency domain displays. It also lets you see and set the minimum and maximum signal levels used to set these colour thresholds.
All in all, the K37 option is a great extension of the FFT functionality as opposed to a completely new feature. It won't compete with a dedicated Spectrum Analyzer, but if definitely the absolute top end of what you could get with a software (FFT) based device. It's a great addition to the oscilloscope.
As you'd expect, it's easy to connect remotely to the RTM3004. You have two options - USB or Ethernet. Both work well so it may come down to personal preference or exactly what you're trying to do.
Simply connect your PC via a USB A cable (not supplied) and the 'scope announces itself as a media transfer (MTP) device and appears in Windows 10 at least as Portable Device "Rohde&Schwarz RTM3004" without any problems or driver installation. There is the option to change this to a USB Serial port if you want to communicate using SCPI commands or to a Test & Measurement Class (TMC) device if you want to use the R&S VISA application to control it. However, I found the default MTP mode to be most useful so I switched back to that.
The first thing you'll see is that the RTM3004 exposes three folder - Internal Storage, Live Data and Upload. Internal storage allows you to retrieve any data saved on the device such as screenshots, waveform data, bus tables - basically anything you might save to the device's internal storage. As I started this road test using 3 different 'scopes I had got into the habit of sticking a USB flash drive into the front of each one to take screenshots. I realised it might be easier using the screenshots folder here and it was - to a point. Unfortunately the contents of the "screen" folder don't refresh unless you disconnect and reconnect the USB cable. F5 does nothing. There was apparently a bug in version 1.200 of the firmware that stopped MTP file transfer working in Windows 10, but I must admit this didn't seem to affect me. However the refresh problem exists in the current 1.300 version of the firmware.
The Live Data folder does mitigate this problem to some degree. It contains just one file of each type the device produces - a BMP and PNG for screenshots, waveforms for each analog and digital channel, etc. However, these are dummy files and they are only created on demand. Drag screenshot.png to your PC and a screenshot is taken. Drag the same file again and it will be a different screenshot. This is far more useful and became my go to way of taking screenshots for this road test.
Firmware updates can also be done over USB (or flash drive). I found it very annoying that my Keysight E36312A power supply needed an application installed on my PC just to do a simple firmware update. This is how all devices should be updated!
USB works well, but Ethernet is even better. Once again setup is simple and most importantly free of any applications you have to install. The only thing worth doing is going into your Ethernet settings on the RTM3004 and picking a suitable hostname. I was boring and just chose "RTM3004" as I'm very unlikely to have more that one of these on my network. After that you just need to go to http://rtm3004 and you're away! Here is where you realise why the ethernet port on the back of the device is gigabit. There is basic device information page of course, and a page which allows you to view and save a screenshot of the device, along with 5 rather out of place buttons to control triggering, autoset and preset. There's a page for entering SCPI commands and another for loading and saving the same sort of data that is available over USB. There is also simple pages for network configuration and setting the password. All of these are very useful, but the real gems are the next two.
Livescreen is a real time view of the oscilloscope screen. i can imagine this is perfect for a classroom or presentation situation. As all this requires is a web browser, adding an external HDMI output to your RTM3000 would need nothing more than a RaspberryPi and a network cable. I fired up Chrome's dev tools and it seems that this is done with some simple but very chatty javascript that repeatedly requests a screen image from the 'scope. Basic and robust, but perhaps a little heavy on network traffic. I simulated a slow 3G connection and unsurprisingly things got a little sluggish. To be honest, who would try this over a slow connection anyway.
Finally there is Remote Front Panel. This extends livescreen to add a fairly faithful representation of the front panel controls to the right of the screen. These are all clickable / dragable (to rotate). I've taken a screenshot here with the browser fairly small, but obviously at more reasonable screen resolution the trace displays as just well as on the device. The only thing worth noting is that if you are annotating via the remote front panel, the lines that you draw with a mouse appear dotted as the updates are passed to the 'scope. That's about all I can complain about, so you can see how picky I'm getting here! It's hard to do this justice without video, so perhaps you should check out my overview for this.
I have to say that usability and the general feel of a quality instrument is where Rohde & Schwarz really shine. This is the stuff that's hard to get onto a datasheet or even a brochure but really make all the difference. I'll try my best to describe it, but this is the sort of thing that you'll really have to see for yourself - either in a video or better yet in person.
When writing this road test I often had one or more oscilloscopes powered up as I typed. It was here that I really noticed how significant the difference in volume was. If just the RTM3004 was on, then I didn't really notice it. If either of the other 'scopes was on then it irritated me just enough that unless there was some setting or data I needed I had to go switch it off. They weren't really loud and the RTM3004 wasn't completely silent but they just fell either side of that line. The quality of screen itself and the other front panel controls is excellent. The shininess that I first noticed when taking a photograph of the powered down device didn't turn out to be an issue. Maybe for some people with unfortunately placed lighting it would be, but I just didn't notice it after that first few minutes.
I liked the fact that the logic probes connect around the side. I can imagine there's a lot of personal preference here and it may depend on where you have your test equipment. They don't seem to take too much room as the cable is quite flexible. Most people probably won't use them that often and it looks so much cleaner than the RTB2004.
The touchscreen is amazing. Touchscreens are everywhere these days but I've never used one on an oscilloscope before. Sometimes and for some things I still went for the usual controls, but I found more and more that for much of the moving and zooming that is inherent in its usage that the touchscreen was my go-to control. Even when precisely entering a value (where capacative touch doesn't shine) the onscreen keyboard made its presence felt and appreciated. Luckily this has not completely replaced the usual rotary encoders and buttons. I did find one particular use case where I looked for a touch enabled action but it wasn't there - going back to normal scope mode - usually when closing the Zoom function. This I found was really only practical using the physical Zoom button on the front panel. I'm sure it's possible using the screen alone but I didn't see a quick and easy way. When touching the screen to select an item such as a trace to move, sometimes it's not possible to be clear which trace you're selecting. In this case a nice disambiguation menu pops up asking which control you intended to select.
The shininess of the screen and its tendency to attract fingerprints seemed obvious when I first got the device out of the box. They stare you right in the face when the 'scope is off. You can probably see me or the camera in much of the video I've taken. However, I have to say that it never seemed an issue when I was actually using the 'scope. Not once. Your opinion may differ, but I'm going to make a call and say it's fine.
The front panel controls are all nice to use. The colour coding of the single set of vertical controls as they move between channels is very neat. The only thing lacking here is that there's no obvious colour coding of the BNC inputs themselves. It's there on the channel markers at the bottom of the screen and I made sure to put the coloured rings on both ends of the probe cables, so it's a minor issue. The Rigol DS1074Z also has single vertical controls so the concept didn't feel awkward. The Keysight doesn't and I found that to be fine to. Both approaches seem to work . Considering the RTM3004 with LA option gives you potentially 26 signals if you add up the analog channels, digital channels, maths channels and buses then I can't see any other way to have done it.
The blue for channel 4 is perhaps a little similar to the logic channels. And why is CH3 orange onscreen but pink for the probes? Another thing I'd say about the front panel is that the smaller rotary controls could probably do without being detented. I know that some people replace the rotary selector on the DS1074Z with a detented one, but that's just so it doesn't change when it's being pushed to select a value. The rotary controls also have a linear response. When I spin them fast, it's probably because I want them to move fast. Maybe I've changed the horizontal scale and want to scroll quickly. The linear response makes them seem a little slow. The touchscreen comes to the rescue again as swiping does the job much faster.
Another perhaps unexpected control is the fact that you can plug a USB mouse or keyboard in. I didn't feel a great need for a keyboard, but a mouse did an excellent job of selection for zooming - much better than fat fingers. I must admit that I thought this was brilliant when I discovered it, but haven't plugged one in very often since then. Time will tell whether I come back to this or not. I love that it's yet another option though.
There are lots of little things that remind you this is a quality 'scope. Small stuff that might not seem important but all add up to a feel of quality and of engineers who care about the product they produce. I get the feeling they perhaps have input into the UI rather than just implementing what they're asked to do.
I don't like to see road tests with "In progress. I'll complete this later". Often it means that however well intentioned the reviewer is, it's never really going to get finished. That's why I haven't posted this road test until I feel it's complete. However, there may be things that you wish I'd covered or you would like covered in more depth. I can't guarantee to get to them all, but feel free to ask in the comments. If I can, then I'll try to test it out for you. So, the only "to do" I'll leave this with is things you'd like me to do. And of course the Spectrum Analyser option.
Whilst researching the RTM3004 I also found other good information on it online. Other reviewers have covered different thing to me and in different ways. There's obviously two other road tests here on E14. If you prefer videos, then I can recommend Mike Harrison's ones on the RTM3004 https://www.youtube.com/watch?v=-siuk4p4dO0 and also his reviews of the RTB2004 with which it shares many features. If you like a more technical analysis then nctnico has done a nice review over on EEVBlog: Rohde & Schwarz RTM3000 review - Page 1
Well, it's no surprise to find my conclusion is that this is an incredible oscilloscope. I said I wouldn't let myself be dazzled by it and blindly give it 60/60 and I haven't. I was close, but I haven't. The features of the RTM3004, the usability and the general quality feel to everything are absolutely spot on. The performance exceeds what I could ask for and I'm sure that this can be said of most Element14 community members. I'm not saying there aren't some of you who need a 6GHz 'scope or a full spectrum analyser with RF decoding. However, these are rather specialist needs and I'm sure you'll be well aware of exactly what will and what won't fit the bill. As a general purpose but high end oscilloscope you really couldn't ask for anything more.
The only things I've really marked it down on is the price / performance ratio. This sort of test equipment is expensive. A comparable 'scope from another manufacturer would also be expensive. However, this premium pricing does seem to carry over into everything about it. Simple accessories are also expensive. Each protocol decoder that you might need has to be purchased separately. The probe accessories, bag, cover - they're all very expensive. The logic analyser pods are great if you want 16 channels, but there seems to be no option to buy just one set and have 8 channels.
Other than the price I have been thoroughly impressed by all aspects of it. I found the standout feature was the Zoom. It might not seem like a major thing, as all 'scopes will let you expand the timebase on vertical resolution on a captured waveform. It's just the ease with which this frequent task is now done. Being able to see the detail in the context of the whole waveform was so useful. One of my goals was to see if the logic analyser has a real benefit over a nice PC-based dedicated one like a Saleae. The logic analyser option is something I feel I'll genuinely use - even if I'm looking only at digital signals. If I'm using the PC anyway - maybe for debugging microcontroller code - then it might be easier to grab the USB one. The usability of the logic channels on the RTM3004 is right up there, and obviously if you (or think you might) have to delve into something more than perfect 1s and 0s then you only have one real choice. The screen is great, and the way you can use or ignore the touchscreen as you see fit is good. All in all, its a joy to use.
Well, that's my road test wrapped up - until it's time to test the Spectrum Analyser. I hope I covered the bits that interest you. If not, please let me know.
Well, it's been a while and the RTM3004 has settled into regular use rather than "road test" mode. As you can expect, it's performing just fine but daily use can be different and a little less predictable that carefully evaluating a device.
I had wondered before starting the road test how the logic analyser features would fare compared to a dedicated PC-based logic analyser. Last week I had some fairly simple SPI debugging to do. What would I reach for? I tried not to just let the shiny new toy win, but I genuinely went for the RTM3004 over the PC-based option. I found that:
I'm genuinely finding that it's a worthwhile and useful option. If you are going for a RTM3k (or a RTB2k) then I'd suggest the LA option is well worth considering.
I've now updated theroad test with the newly-arrived K37 Spectrum analyzer and Spectrogram option that's now available on devices destined for the US market. Previously K18 was available, but not to the US.
Top Comments
Hi David - thanks for the review and your thoughts on the RTM3000! It's a very thorough review and we greatly appreciate your time spent on it. I'm still reading through it all, but one quick comment…
Beautifully done road test.
John
The long-awaited K18 option for the US models has arrived! It's now called K37 but it's essentially the same. Expect an update to my road test to cover this in the next week or so.