Is there a good midrange ($3k to $10k) scope?

[nctnico]

I'm not looking to buy anything right now but I do like to keep taps on what is available and for some reason that list is awfully short. My basic requirements for a daily use scope would be 300 to 500 MHz, 4 channels,10Mpts per channel minimum in all usage scenarios, peak-detect (also in roll mode), big screen and protocol decoding/triggering. These requirements are based on the kind of usage scenarios I came across. For obvious reasons I rather rule out the Chinese B-brands (bugs and checkbox features).

I came up with the following list:
Keysight's 2000/3000/4000/6000 series already fails on the memory requirement (1Mpts or less in real usage scenarios).
Lecroy's Wavesurfer 3000 series seems nice and affordable but it doesn't have peak detect
Hameg HMO3000 also falls short on memory and a 640x480 display was nice in 1990
Tektronix' MDO3000 series does tick all the boxes but appearantly the UI is rather slow
Rhode & Schwarz's RTM2000 series also ticks all the boxes although I have no idea how well these perform.

So all in all the R&S RTM2000 may seem to be the only scope which may fit my requirements. IMHO that is not much choice.

[mikeselectricstuff]

Do you really need that much memory?
Versatile triggering options and segmented memory can often reduce the need for huge memory.
beware of missing out on other useful features by eliminating possible options too quickly.

[Someone]

Don't worry too much about the speed of the UI and updating, it can be overblown by some people. The Maths and Measurements are slow on the Tektronix mid range 3000/4000 platforms but most other operations are smooth and easy to use, they also have Peak Detect and Envelope as distinct capture modes while maintaining the full 10Mpts although its not clear exactly how they interact with the intensity and persistence controls the graded display is very good. As an everyday scope for looking at signals the Tek 4000 series are more than adequate and worth looking at.

[jjoonathan]

Is there some sort of strange amnesia in this forum about mixed-domain or does it have pitfalls that make it much less useful than it looks? I keep seeing posts about Tek being a joke in the midrange, but if Tek is a joke, then what do they think of Keysight missing the IoT bandwagon? If I were designing RF trinkets, time-correlated spectrum analysis sounds pretty damn handy...

[G0HZU]

Yes, I wouldn't mind having a play with an MDO 4000.

At work we have loads of the Tek 4000 series scopes as general purpose 'advanced' scopes and they are quite popular. But I've not seen the MDO 4000 model at work yet. Possibly because we already have quite a few Tek RTSAs and they are always in demand for waveform analysis.

[EEVblog]

If I were designing RF trinkets, time-correlated spectrum analysis sounds pretty damn handy...

It can be, sure. But apart from that the MDO scopes are a pain to use, so horribly slow when you try and do more than one thing at once.
I would not want to use one as my everyday use scope.

[MarkL]

Don't worry too much about the speed of the UI and updating, it can be overblown by some people. The Maths and Measurements are slow on the Tektronix mid range 3000/4000 platforms but most other operations are smooth and easy to use, they also have Peak Detect and Envelope as distinct capture modes while maintaining the full 10Mpts although its not clear exactly how they interact with the intensity and persistence controls the graded display is very good. As an everyday scope for looking at signals the Tek 4000 series are more than adequate and worth looking at.

I had an MDO4104B in the lab for a test drive a couple of years ago.

The slow GUI drove me nuts.  The most insidious thing was that the A/B knobs had no detent AND the GUI would get randomly behind in updating the screen, sometimes by several seconds.

The lack of a detent forces a user to rely on visual feedback.  I was constantly watching the cursor whiz by what I thought I had selected and select the wrong thing or whip the waveform off the screen in one direction or the other.

And when 10Mpts or 20Mpts is on is when the GUI is at it's worst.

In my testing it took 50 (fifty) seconds to decode a 20M I2C capture and update the screen.  And during that time it was essentially locked up.  Perhaps not so overblown.

Plus about 20 bugs found in the course of testing.

Totally unusable in my opinion.  I'm a big Tek fan and this was the most disappointing thing I'd ever seen from them.

[linux-works]

midrange scopes really don't work so well on their own.

see, there's the cost of the crossover, then you have to find matching tweeter scopes and maybe even subwoofer scopes.

[Someone]

Don't worry too much about the speed of the UI and updating, it can be overblown by some people. The Maths and Measurements are slow on the Tektronix mid range 3000/4000 platforms but most other operations are smooth and easy to use, they also have Peak Detect and Envelope as distinct capture modes while maintaining the full 10Mpts although its not clear exactly how they interact with the intensity and persistence controls the graded display is very good. As an everyday scope for looking at signals the Tek 4000 series are more than adequate and worth looking at.

I had an MDO4104B in the lab for a test drive a couple of years ago. The slow GUI drove me nuts.  The most insidious thing was that the A/B knobs had no detent AND the GUI would get randomly behind in updating the screen, sometimes by several seconds.
The lack of a detent forces a user to rely on visual feedback.  I was constantly watching the cursor whiz by what I thought I had selected and select the wrong thing or whip the waveform off the screen in one direction or the other. And when 10Mpts or 20Mpts is on is when the GUI is at it's worst.
In my testing it took 50 (fifty) seconds to decode a 20M I2C capture and update the screen.  And during that time it was essentially locked up.  Perhaps not so overblown. Plus about 20 bugs found in the course of testing. Totally unusable in my opinion.  I'm a big Tek fan and this was the most disappointing thing I'd ever seen from them.

This is the difference between hardware and software driven analysis, once you're doing the analysis offline the processing time scales with the memory depth. You can see the Lecroy WaveSurfer 3000 dropping its capture rate as serial decode is enabled here:

And thats with a 50,000 point capture, so you could expect it to be 200 times slower at 10M points which is right around the same figures you are mentioning. The scopes aren't slow all the time, only when you've enabled CPU intensive operations.

[MarkL]

...
The scopes aren't slow all the time, only when you've enabled CPU intensive operations.

The MDO4k I tested was sluggish and had UI (human interaction) problems even without decoding turned on, just to be clear about that.

nctnico didn't say why he wanted 10Mpts/ch minimum, but it's typically for serial data capture.  The MDO4k performed horribly for that application.  And honestly, I would expect some delay on a long record; maybe 5 seconds or 10 seconds tops (and give me a progress bar, please).  But not enough to go get a cup of coffee.

I don't have any experience with LeCroy, but I don't think it's fair to extrapolate two packets on the screen to how it can handle a 10Mpt decode.  The MDO had major issues with moving around in the capture and zooming.  And you had to know what NOT to move or zoom into, otherwise it would decide it was time to decode the whole thing again for another 50 seconds.  LeCroy and other manufacturers may have spent a lot more effort on usability.

I don't see the utility of large capture buffers if the scope can't deal with it reasonably.  I think it's a waste of money.

[Wuerstchenhund]

Tektronix' MDO3000 series does tick all the boxes but appearantly the UI is rather slow

I have (again, as my colleague brought it back    )  a MDO3054, and I can honestly say it's the worst big brand scope I've ever had. The UI is horrible (instead of touch as on modern scopes, you have to make your way through the various functions and settings with soft keys and not one but two multi-purpose knobs with no detents ). Like all Tek scopes, it only reaches high waveform rates in a special acquisition mode (now called 'FastAcq') in which you can't do measurements. And like all Tek scopes, it's UI freezes once it does something that is more taxing than sitting in idle, i.e. FFT, or even just deep memory acquisitions.

Serial decode is slow, I mean really slow. Most scopes drop the waveform rate in decode mode but the Tek is woeful.

The built-in "spectrum analyzer" is a sad joke, all Tek did was adding another input with N connector and feed it to the scope's 8bit ADC system, where it employs 1Mpts FFT to get the spectrum. The dynamic range is woeful (well, it's 8bit), as is the RF performance in general, and not only worse than bottom-of-the-barrel SAs like Rigols DSA815 but even than FFT on other 8bit scopes  . The only advantage over FFT on another scope is BW (the Tek 'SA' port goes to 3Ghz if the MOD3SA option is installed).

The upside is that it's triggering system is really good, it has some basic memory search functions (although slow), the screen is nice and bright, and the scope is pretty silent. And it comes with a numeric keypad which is nice. But that doesn't even start to compensate for the fact that pretty much everyone who used this scope hated it. It really is that bad.

Rhode & Schwarz's RTM2000 series also ticks all the boxes although I have no idea how well these perform.

Well, it's a very low update rate scope (11k wfms/s?), which for a modern scope is really low (hell, even the old LeCroy WaevRunner2 LT264M I had years ago had higher update rates), but other than that they are fine scopes. The UI is mostly simple (although some settings are not where you'd expect them) and responsive, the screen (XGA) is nice and bright, and there are many areas where it shows that whoever designed it paid attention to details. You also get an Z mode (brightness modulation) for X/Y, something most other DSOs don't have.

The RTM is based on an older SoC (Blackfin) which shows in some areas, i.e. FFT is only 128kpts max for the RTM2000.

Depending on your budget the R&S RTE could be an alternative as well.

So all in all the R&S RTM2000 may seem to be the only scope which may fit my requirements. IMHO that is not much choice.

Yes, it's not. But if your want Peak Detect then the RTM or the RTE is probably your best choice. I agree that the outlook isn't exactly stellar. It shows that the market could well use another strong competitor, unfortunately it seems Tek isn't up to it and there's no-one else left really.

[snoopy]

Tektronix' MDO3000 series does tick all the boxes but appearantly the UI is rather slow

I have (again, as my colleague brought it back    )  a MDO3054, and I can honestly say it's the worst big brand scope I've ever had. The UI is horrible (instead of touch as on modern scopes, you have to make your way through the various functions and settings with soft keys and not one but two multi-purpose knobs with no detents ). Like all Tek scopes, it only reaches high waveform rates in a special acquisition mode (now called 'FastAcq') in which you can't do measurements. And like all Tek scopes, it's UI freezes once it does something that is more taxing than sitting in idle, i.e. FFT, or even just deep memory acquisitions.

Serial decode is slow, I mean really slow. Most scopes drop the waveform rate in decode mode but the Tek is woeful.

The built-in "spectrum analyzer" is a sad joke, all Tek did was adding another input with N connector and feed it to the scope's 8bit ADC system, where it employs 1Mpts FFT to get the spectrum. The dynamic range is woeful (well, it's 8bit), as is the RF performance in general, and not only worse than bottom-of-the-barrel SAs like Rigols DSA815 but even than FFT on other 8bit scopes  . The only advantage over FFT on another scope is BW (the Tek 'SA' port goes to 3Ghz if the MOD3SA option is installed).

That's not true. When Dave did an in-depth review of this scope he explained why the scope performs better with the spectrum analyser input compared to the ordinary scope inputs. Check it out. There are other reviews by the "The Signal Path" who explain why it is better as well

[Wuerstchenhund]

The built-in "spectrum analyzer" is a sad joke, all Tek did was adding another input with N connector and feed it to the scope's 8bit ADC system, where it employs 1Mpts FFT to get the spectrum. The dynamic range is woeful (well, it's 8bit), as is the RF performance in general, and not only worse than bottom-of-the-barrel SAs like Rigols DSA815 but even than FFT on other 8bit scopes  . The only advantage over FFT on another scope is BW (the Tek 'SA' port goes to 3Ghz if the MOD3SA option is installed).

That's not true. When Dave did an in-depth review of this scope he explained why the scope performs better with the spectrum analyser input compared to the ordinary scope inputs. Check it out. There are other reviews by the "The Signal Path" who explain why it is better as well

Well, I haven't seen the reviews (and frankly, I can't be arsed to spend time watching them), so I rather rely on the unit I have here at home plus what Tek tells me about how it works. Which is that the 'SA' input goes to the same 8bit acquisition system (4x 2.5GSa/s) which in SA mode operates in 10GSa/s interleaved mode to acquire 1Mpts of signal with max 3Ghz BW. That's also the reason the MDO3k can't run in TD and FD mode at the same time.

The reason the SA input works "better" is that in SA mode the scope uses 10GSa/s to satisfy Nyquist-Shannon for a 3Ghz signal, which it can't do through the BNC inputs as they are limited to 100MHz, 200MHz, 500MHz, 1Ghz or whatever the analog bandwidth of the scope is, plus each input can only sample at 2.5GSa/s (5GSa/s on the 1Ghz model), which is insufficient for a 3Ghz signal.

[Wuerstchenhund]

Is there some sort of strange amnesia in this forum about mixed-domain or does it have pitfalls that make it much less useful than it looks?

There's no amnesia. MDO just isn't as big as a deal because you pay through the nose for a mediocre scope with a very poor SA.

Tek's marketing is pretty misleading as none of the MDOs offer anything like a "real" SA. All they do is offering an additional high BW input that feeds to the scope's 8bit acquisition system and then use FFT to capture the spectrum. The MDO3K can't do TD and FD at the same time because it's ADC is too slow, the MDO4K can but you still end up with an 8bit FFT which isn't any better than FFT on any other scope,  with very poor dynamic range. The limitation in terms of sample size (1Mpts on the MDO3k) also means there are limitations which combinations of span and RBW are possible. On top of it, it's very slow, too.

Don't forget that none of the MDO scopes are particularly cheap but all pretty much located at the upper end of the price spectrum for a scope of that class.

I keep seeing posts about Tek being a joke in the midrange, but if Tek is a joke, then what do they think of Keysight missing the IoT bandwagon? If I were designing RF trinkets, time-correlated spectrum analysis sounds pretty damn handy...

I'm not sure that Keysight is "missing the IoT bandwagon" (which at the moment is pretty much overhyped anyways with lots of solutions in search of a problem).

If you're into IoT and designing RF trinkets then you'd be unlikely to redesign the RF part from scratch but rather use approved/certified RF modules, and for that you don't really need an SA.

However, if you were designing RF components yourself, then you'd certainly want something a lot better than a poor 8bit FFT SA like the one in the MDOs. If you want brand new, get one of the bottom-of-the-barrel SAs from the Chinese B-brands Rigol or Siglent, and you get a way better RF performance for around the same money Tek wants to unlock the 3GHz BW on the MDO3k.

And if you can live with 2nd hand, there are lots of options like an old HP SA, or one of the Anritsu or Advantek SAs that can often be found for low prices. Heck, even the SA in old cell phone testers like the R&S CMU200 or Agilent E7495A/B perform better than the one in the MDO scopes (they all come with 10+bit samplers), and can often be found for little money.

[nctnico]

I think it is easy to get a Scope Month prize MSOX3104T on eBay for less than $8k on best offer. It has 4Mpts memory at single channel.

That would be a huge step back from the 8Mpts (reduced to 2Mpts or less in real usage scenarios with 4 channels on) my Agilent/Keysight DSO7104A has. Ofcourse you can always 'make do' but capturing one trace and then examine the parts of interest using zoom is much easier than having to make several seperate measurements (which likely need different trigger settings/delays). Also the data from multiple measurements isn't related (assuming it is possible to make the measurement again which in some cases can't be done).

Rhode & Schwarz's RTM2000 series also ticks all the boxes although I have no idea how well these perform.

Well, it's a very low update rate scope (11k wfms/s?), which for a modern scope is really low (hell, even the old LeCroy WaevRunner2 LT264M I had years ago had higher update rates), but other than that they are fine scopes.

Sure the 12000 waveforms/s of the RTO2000 doesn't look like much but as long as the trace gets updated 10 times per second a scope is good to work with (assuming the update rate isn't limited by the time/div setting). IMHO one of the most overrated (hyped) specs is waveforms/s because at some point it just doesn't serve any practical purpose.

[HighVoltage]

Well, I really like my 6000X Scope and it has become my every day scope of choice.
So far, I have not run in to any limitations.

If you are ever in northern Germany, come by and play with it, if you like.

[Wuerstchenhund]

Sure the 12000 waveforms/s of the RTO2000 doesn't look like much but as long as the trace gets updated 10 times per second a scope is good to work with (assuming the update rate isn't limited by the time/div setting). IMHO one of the most overrated (hyped) specs is waveforms/s because at some point it just doesn't serve any practical purpose.

In general I'd fully agree that the update rate is vastly overhyped and that excessive update rates are pretty useless in reality.

However, for a scope that at the end of the day lacks any advanced analysis tools (and which therefore relies much more on conventional methods for finding glitches and issues), 12k wfms/s is really very low (as I said, my old WR2 LT264M measured >23k wfms/s, and that was a scope from 2004 which came with plenty of advanced toolsets). For a new scope with standard tools I'd want at least 100k wfms/s as otherwise the probability to catch a rare glitch drops significantly.

I guess the other question is if you plan to invest in active probes, which means you'd bind even more money to a specific platform (i.e. if you were to change to another brand later you'd have to re-buy probes as well). Something worth keeping in mind.

The RTM1054 was a fantastic compact 500Mhz scope - but then I only paid $1900 or so for it. My verdict would be different if I were asked to pay $6k for it (plus the options).

Overall it's not exactly great. Unfortunately there's no scope that can do everything. If you want PD then this pretty much takes out everything from LeCroy aside from the WaveJet Series (which is an Iwatsu rebadge and if I remember right offers PD). I think Dave reviewed the WaveJet T (T=Touch) not too long ago. Keysight has the InfiniiVision X which are good scopes but come with tiny memory. Tek has the MDO which has decent memory but otherwise is a real pain to use. R&S has the HMO3000 and RTM2000, both pretty old platforms (the RTM was R&S' first own scope and came out in 2009).

It doesn't seem as there's a lot that fits your requirements Maybe you already have the best option in form of your DSO7104A?

[Someone]

It doesn't seem as there's a lot that fits your requirements Maybe you already have the best option in form of your DSO7104A?

The answer is probably more scopes! No point trying to have a universal do everything tool when two or more radically different tools can do their jobs better without compromise, just a shame that its so hard to share active probes between brands.

[nctnico]

That is my current solution: two scopes
I'll probably always want two scopes just in case one breaks down and one of them has to be cheap enough to bring along to customers.

I think the suggestion about keeping probes generic is a good one. I have a couple of passive divider HF (3GHz) probes from Tektronix but since they have BNC plugs they work on any scope with a BNC input (and 50 Ohm input impedance) or a spectrum analyser.

[mikeselectricstuff]

That would be a huge step back from the 8Mpts (reduced to 2Mpts or less in real usage scenarios with 4 channels on) my Agilent/Keysight DSO7104A has.

Memory isn't everything... every scope will have its share of good and bad points.
The MSO3000 can do two simultaneous serial decodes (same or different  types) -  Not sure I've seen any other scope that can do this) and has a ton more trigger and measurement options.

As regards slow UI - you are ALWAYS using the UI so slowness or bad design will rapidly become very annoying. Other features like deep memory etc. are very useful in a relatively small number of situations,  and the deeper the memory, the fewer times you'll really need it.
In most cases can be worked around.

[nctnico]

The RTM2000 can display 4 bus decode traces (for example UART RX/TX and SPI MISO/MOSI together).

At some point you have to see if 'making do' is like using a Pozidrive screwdriver on a Philips screw or using a hammer to drive a screw into wood. In my experience good tools help to do work more quickly & better (less errors). Sometimes I buy equipment for a specific project just to make sure I have to right tool.

Anyway having lots of memory makes life much easier. I have a relatively high-end Tektronix logic analyser (which wasn't expensive) and it has a 32Mpt memory per channel. In many cases I don't care about setting a trigger condition; I just let it run until the memory is full (which can be long when using transitional recording) which often gives me all the data there is to collect from a system. From there I just go through the data to analyse it and maybe save it for later so I can load the data again and check some other things if necessary. On a scope the same approach also works. Capture one trace and analyse it and/or save it for future reference (which is easy to manage because all the data for one measurement is in one file). Needing to do more measurements just takes more time and introduces more chances for errors.

Still.. do I use the logic analyser daily? No not by a long shot (2 times per year is more likely) but still it has paid for itself several times over.

[MarkL]

...
I guess the other question is if you plan to invest in active probes, which means you'd bind even more money to a specific platform (i.e. if you were to change to another brand later you'd have to re-buy probes as well). Something worth keeping in mind.

One way to approach the active probe problem is to buy Tek probes that have the TekProbeII (TP2) interface.  Tektronix is still selling probes with this interface, and used ones are abundant.

Both LeCroy and Agilent sell small adapters so you can plug TP2 probes into their proprietary connectors.  Tek also sells an adapter for TP2 to their new proprietary connector.  I don't think Rohde & Schwarz has one.

If you want to use a TP2 probe on a spectrum analyzer or other piece of gear without an active probe interface, you can get a Tek 1103 which is a TP2 standalone power unit.

The adapters aren't cheap, but I have taken this approach and it saved me a bundle not having to re-buy all the probe types when it was time to refresh a scope.  Probes obsolesce at a much slower rate.

[ivan747]

midrange scopes really don't work so well on their own.

see, there's the cost of the crossover, then you have to find matching tweeter scopes and maybe even subwoofer scopes.

Made my day

[rfbroadband]

I am happy with the RTM 1000. I haven't found any serious limitations for this scope in this mid level scope class. I would probably look for a 1000 series scope on the used market first before I purchase a new RTM 2000 if price is an issue.