MSO vs MDO - What's going on?

[EEVblog]

No zone, just mask. Should usually be useful in same ways since it'll save fails in history. Or I don't know what people use zone trigger for, I've never used it even in scopes that had it.

Zone trigger lets you just draw a box around the area on the waveform you want to trigger on and it just does it. very handy.

[Howardlong]

Interestingly the previous Megazoom III used in the older generation 5000, 6000 and 7000 scopes supports XGA and 8Mpts, with a ~1000x640 pixel waveform area, almost the entire width of the display is used unlike the Megazoom IV which looks like a waveform area of about 640x384 pixels.
Megazoom III also supported 8Mpts compared to 4Mpts of Megazoom IV.

IIRC the Megazoom III used external memory. The IV 4Msamples is on the die, that's why it's so quick.

Sounds reasonable.

I assume by "quick" you're referring to the waveform update rate, and not the scope UX.

If that is the case, then just like memory depth, I'd suggest that we're back to diminishing returns: once you have, say, 100kWfm/s (Megazoom III speed), practically speaking how often are you going to benefit from 1MWfm/s?

[Mr. Scram]

Sounds reasonable.

I assume by "quick" you're referring to the waveform update rate, and not the scope UX.

If that is the case, then just like memory depth, I'd suggest that we're back to diminishing returns: once you have, say, 100kWfm/s (Megazoom III speed), practically speaking how often are you going to benefit from 1MWfm/s?

Every day to never, depending on your workload. A testing station in a production environment would benefit from doing a large amounts of masking tests within a few seconds.

[nctnico]

Interestingly the previous Megazoom III used in the older generation 5000, 6000 and 7000 scopes supports XGA and 8Mpts, with a ~1000x640 pixel waveform area, almost the entire width of the display is used unlike the Megazoom IV which looks like a waveform area of about 640x384 pixels.
Megazoom III also supported 8Mpts compared to 4Mpts of Megazoom IV.

IIRC the Megazoom III used external memory. The IV 4Msamples is on the die, that's why it's so quick.

That is not it. This scope uses some kind of parallel acquisition system which helps to increase waveform update speed (note to others: this high rate only works at a few time/div settings). Math is done on sub-sampled data so that increases performance too (probably combined with a DSP inside the ASIC). A modern FPGA can do better (both in memory bandwidth and calculations). It is just that you get in an area with diminishing returns versus investment in FPGA development.

[EEVblog]

Interestingly the previous Megazoom III used in the older generation 5000, 6000 and 7000 scopes supports XGA and 8Mpts, with a ~1000x640 pixel waveform area, almost the entire width of the display is used unlike the Megazoom IV which looks like a waveform area of about 640x384 pixels.
Megazoom III also supported 8Mpts compared to 4Mpts of Megazoom IV.

IIRC the Megazoom III used external memory. The IV 4Msamples is on the die, that's why it's so quick.

Sounds reasonable.

I assume by "quick" you're referring to the waveform update rate, and not the scope UX.

If that is the case, then just like memory depth, I'd suggest that we're back to diminishing returns: once you have, say, 100kWfm/s (Megazoom III speed), practically speaking how often are you going to benefit from 1MWfm/s?

It's not just that. It can do everything faster like the multiple get trigger point mode, masking, serial decode info and a host of other stuff, all without slowing down the UI.
I'm not sure if the Megazoom III did all the memory->screen mapping in the chip like the IV does?

[nctnico]

The UI responsiveness is just a matter of how the processes in the CPU are prioritised. Every DSO uses an ASIC or FPGA to map the waveforms onto the screen so that is also not new. Using the CPU to map waveforms onto the screen is going to be too slow. But there are many drawbacks to the Megazoom ASIC. Try to move / scale reference traces for example. On the Agilent MSO7104A I had this wasn't possible (probably the same for using cursors on the reference traces). Also the UI on the MSO7104A was horrible to use compared to the other oscilloscopes I have now. Fast UI but still slow to operate.

[SWR]

Tek is the only one that makes an "MDO" scope, it's pretty much a niche product in the mid range scope category.

Weeeell... It's not to be pedantic, but other MDO options are available in the market.

GW Instek makes a MDO-2000 series of scopes that addresses some of the issues discussed in this thread:
- High resolution 1Mpoint FFT with quick update rate.
- 40Msample memory.
- SA type UI where you don't have to adjust resolution etc. in the time domain.
- Nice broad selection of math functions on Ch1-4, Ref1-4: (the Teks limited math ability was mentioned earlier in this thread)
      Measurement Pk-Pk,Max,Min,Amp,High,Low,Mean,CycleMean,RMS,CycleRMS,Area,CycleArea,
                          ROvShoot,FOvShoot,Freq,Period,Rise,Fall,PosWidth,NegWidth,Dutycycle,
                          FRR,FRF,FFR,FFF,LRR,LRF,LFR,LFF,Phase,RPRFShoot,FPREShoot,
                          +Pulses,-Pulses,+Edges,-Edges
      Operators      +,-,*,/,<,>,<=,>=,==,!=,||,&&,
      Functions       Intg,Diff,Log,Ln,Exp,Sqrt,Abs,Rad,Deg,Sin,Cos,Tan,Asin,Acos,Atan

The UI is pretty snappy to work with on my MDO-2004EX. I assume it's equally fast on the "high end" scopes, but I have no experience (I'm always on a budget). We've got Tek and Keysight at work and they are not faster to work with in the daily work. IMO they found a nice low cost way to make the scope fast by integrating both the processor and FPGA on a Xilinx single chip solution. It's both cheaper and faster to communicate on the same chip than going via the PCB.

It might be worth considering if your funds are limited and you have to maximize your value for money?

Best regards
Soren

[2N3055]

Tek is the only one that makes an "MDO" scope, it's pretty much a niche product in the mid range scope category.

Weeeell... It's not to be pedantic, but other MDO options are available in the market.

GW Instek makes a MDO-2000 series of scopes that addresses some of the issues discussed in this thread:
- High resolution 1Mpoint FFT with quick update rate.
- 40Msample memory.
- SA type UI where you don't have to adjust resolution etc. in the time domain.
- Nice broad selection of math functions on Ch1-4, Ref1-4: (the Teks limited math ability was mentioned earlier in this thread)
      Measurement Pk-Pk,Max,Min,Amp,High,Low,Mean,CycleMean,RMS,CycleRMS,Area,CycleArea,
                          ROvShoot,FOvShoot,Freq,Period,Rise,Fall,PosWidth,NegWidth,Dutycycle,
                          FRR,FRF,FFR,FFF,LRR,LRF,LFR,LFF,Phase,RPRFShoot,FPREShoot,
                          +Pulses,-Pulses,+Edges,-Edges
      Operators      +,-,*,/,<,>,<=,>=,==,!=,||,&&,
      Functions       Intg,Diff,Log,Ln,Exp,Sqrt,Abs,Rad,Deg,Sin,Cos,Tan,Asin,Acos,Atan

The UI is pretty snappy to work with on my MDO-2004EX. I assume it's equally fast on the "high end" scopes, but I have no experience (I'm always on a budget). We've got Tek and Keysight at work and they are not faster to work with in the daily work. IMO they found a nice low cost way to make the scope fast by integrating both the processor and FPGA on a Xilinx single chip solution. It's both cheaper and faster to communicate on the same chip than going via the PCB.

It might be worth considering if your funds are limited and you have to maximize your value for money?

Best regards
Soren

GW Instek MDO was mentioned already.  I agree with you as it being quality product and very good, but it is not really same as Tek MDO. There is no time domain correlation between time domain channels and FFT.
That is the distinctive feature, of Tek MDO, and Keysight and LeCroy gated FFT are similar to it. Still, Tek  can trigger on RF event, that others can't ...

[EEVblog]

Tek is the only one that makes an "MDO" scope, it's pretty much a niche product in the mid range scope category.

Weeeell... It's not to be pedantic, but other MDO options are available in the market.

GW Instek makes a MDO-2000 series of scopes that addresses some of the issues discussed in this thread:
- High resolution 1Mpoint FFT with quick update rate.
- 40Msample memory.
- SA type UI where you don't have to adjust resolution etc. in the time domain.

It uses a standard scope front end, so it's just a software solution, not the same as the Tek MDO that has a real RF front end.

is it any different to Keysight gated FFT etc really?

[2N3055]

Tek is the only one that makes an "MDO" scope, it's pretty much a niche product in the mid range scope category.

Weeeell... It's not to be pedantic, but other MDO options are available in the market.

GW Instek makes a MDO-2000 series of scopes that addresses some of the issues discussed in this thread:
- High resolution 1Mpoint FFT with quick update rate.
- 40Msample memory.
- SA type UI where you don't have to adjust resolution etc. in the time domain.

It uses a standard scope front end, so it's just a software solution, not the same as the Tek MDO that has a real RF front end.

is it any different to Keysight gated FFT etc really?

SA mode is made same as on Tek MDO3000, it's an exclusive screen mode and no correlation with other data..

[ogden]

It uses a standard scope front end, so it's just a software solution, not the same as the Tek MDO that has a real RF front end.

Gains from RF frontend of Tek MDO3000 are - better sensitivity, SNR and freq response. That's it. Main value is signal processing. I would not include MDO4000 in comparison because it is another class of instrument.

is it any different to Keysight gated FFT etc really?

They are different things. AFAIK gated FFT of Keysight is about frequency analysis of recorded samples in the gated time window.

I see that Keysight (like 3000T X-Series) have same FFT features - Start Frequency, Stop Frequency, Center Frequency and Span controls. Seeing very good FFT update rate of FFT GW Instek MDO-2000 I can tell that they use digital downconversion (DDC) before FFT. Dunno about Keysight. It is much more efficient to downconvert frequency range of interest into baseband, calculate FFT-1024 and show every of 1024 freq bins on the screen rather than calculate 1M FFT points, waste 99.9% compute resources to show just 1024 points. Performance difference is gigantic between two. With DDC+FFT1024 one can achieve realtime spectrum analysis, using 1M FFT - never.

SA mode is made same as on Tek MDO3000, it's an exclusive screen mode and no correlation with other data..

IMHO it's cheating to call it Mixed Domain Oscilloscope without correlation between time and frequency domain displays.

[2N3055]

IMHO it's cheating to call it Mixed Domain Oscilloscope without correlation between time and frequency domain displays.

I agree, both Tek MDO3000 and GW Instek are not really MDO in my opinion. Only Tek MDO4000.

[Howardlong]

I had a quick look at a couple of videos on the MDO2000E, one concentrating on the SA, and the close in phase noise looked pretty bad, I assume it's from their reference clock and/or PLL. Perhaps it was just the example they gave.

[Howardlong]

IMHO it's cheating to call it Mixed Domain Oscilloscope without correlation between time and frequency domain displays.

I tend to agree, and I fell into that trap when I bought an MDO3000: I assumed that there was some operational integration between the scope and the SA, like the MDO4000 has, but all it is is an SA and scope in the same box, and they cannot run concurrently, and as such it certainly doesn't offer any time/frequency domain correlation. That missing feature is noticeable only by its absence in the marketing blurb.

[Mr. Scram]

The UI responsiveness is just a matter of how the processes in the CPU are prioritised. Every DSO uses an ASIC or FPGA to map the waveforms onto the screen so that is also not new. Using the CPU to map waveforms onto the screen is going to be too slow. But there are many drawbacks to the Megazoom ASIC. Try to move / scale reference traces for example. On the Agilent MSO7104A I had this wasn't possible (probably the same for using cursors on the reference traces). Also the UI on the MSO7104A was horrible to use compared to the other oscilloscopes I have now. Fast UI but still slow to operate.

The current generation is the Megazoom IV ASIC which allows scaling, cursors and measurements on reference waveforms in the same expedient manner as any waveform.

[2N3055]

The UI responsiveness is just a matter of how the processes in the CPU are prioritised. Every DSO uses an ASIC or FPGA to map the waveforms onto the screen so that is also not new. Using the CPU to map waveforms onto the screen is going to be too slow. But there are many drawbacks to the Megazoom ASIC. Try to move / scale reference traces for example. On the Agilent MSO7104A I had this wasn't possible (probably the same for using cursors on the reference traces). Also the UI on the MSO7104A was horrible to use compared to the other oscilloscopes I have now. Fast UI but still slow to operate.

The current generation is the Megazoom IV ASIC which allows scaling, cursors and measurements on reference waveforms in the same expedient manner as any waveform.

Correct. In my opinion, 3000T is close to being perfect except FFT being only 64kpoints (not a big deal for me anyway), and not having larger acq memory. If they had 10Mpoints per ch (like GW Instek) it would be pretty much fantastic, current memory is OKish, but sometimes it needs more... Also, one thing that is not intuitive is that if you capture something, you see that it is something resembling some protocol, you cannot just enable protocol decode after the fact and see what it says. You need to recapture. Also if you are not sure about parameters, you will have to keep changing parameters and recapturing until you get it right, and then start your work. Not a problem when working on something that spits out 100 packets per second, but when working at something that occasionally sends a packet it is not that nice..
Those are the moments when I switch to Pico that has 500MPoints, and decodes in software, so I can capture and analyse offline..

[Howardlong]

That's pretty much my modus operandi also. Use the go-to scope for 90% to 95% of things, then for the remaining cases leverage those key differentiating features.

The downside of this approach is that as a result you tend not to be as experienced in using those other instruments, and, in the case of complex instruments, you can easily forget how to do things, or not be fully aware of their capabilities and quirks.

[2N3055]

That's pretty much my modus operandi also. Use the go-to scope for 90% to 95% of things, then for the remaining cases leverage those key differentiating features.

The downside of this approach is that as a result you tend not to be as experienced in using those other instruments, and, in the case of complex instruments, you can easily forget how to do things, or not be fully aware of their capabilities and quirks.

When I got 3000T, it was quite a work to absorb manuals and materials. It was very useful that scope has demo/teaching mode and corresponding two educators manuals that are excellent to quickly get an idea of features. Also they have applications notes for pretty much every protocol and feature that also show how to do it and even some little tricks that are not obvious for first time users

But you will only know from top of your head things that you do often.
One thing that I find useful, though, is that most of the time for questions like: "does it have this source for this trigger? can I apply this measurement on this etc.." that quickest and most concise source is not the user manual, but the SCPI commands programmers manual.