EEVblog #587 - Tektronix MDO3000 Mixed Domain Oscilloscope Teardown

[digital]

https://www.eevblog.com/forum/Smileys/default/icon_smile_thumbsup.gif One of your best videos yet also informative commentary I learned a lot from this video.Thanks Dave

[kudzo]

Hi Dave, did you use the tagarno for the close-ups?

[tom66]

That you can't use scope and spectrum analyzer together as a true mixed domain tool?

I was more referring to "why doesn't ever spectrum analyser do this?" most of them have RF front ends about half the instrument wide, and then a cheap(ish) ~100MSa/s 12-bit ADC.

It seems they've got a really low cost RF front end but a high ADC cost - but that's not a problem if you already need that for your scope functionality.

It would be nice to have 2ch ADC and 2ch spec-analyser, with the analyser having half bandwidth, that's a reasonable compromise IMO.

[open loop]

I wonder if the scope is measuring he temperature of the AWG chip as well and as it was disconnected the firmware would have quickly issued the high temperature error?

Another thing could be that the firmware tried to set the fan speed but failed and then issued a temp warning.?

[Hydrawerk]

Dave, could you please measure the waveform update rate of MDO3000? I wonder how it changes when you turn on all analog channels and decoding or digital channels...

[Hydrawerk]

Here is a competitive fact sheet made by Tektronix guys. MDO3000 vs. DSOX3000. Yes, the MDO3000 is more feature rich, but it cannot reach 1000000 waveforms per second... Well, somebody does not need that update rate and prefers longer memory and that famous waveform inspector feature.

[Hydrawerk]

Does MDO3000, MSO3000 or MSO2000 have a segmented memory or something similar like that Rigol waveform recorder? It seems that not at all. :-(

[TMM]

I was more referring to "why doesn't ever spectrum analyser do this?" most of them have RF front ends about half the instrument wide, and then a cheap(ish) ~100MSa/s 12-bit ADC.

Because there's only so much noise floor you can get out of an 8bit ADC by dithering and 10GSa/s @ 12bits doesn't come cheap.

[EEVblog]

For all those who guessed what that "touch sensor" on the PCB was, it's nothing of the sort. The PCB designer got bored and wanted to put a symbol of a sine wave on an old school round scope screen. The vias connect to the internal ground plane.
FAIL IMO

[EEVblog]

Here is a competitive fact sheet made by Tektronix guys. MDO3000 vs. DSOX3000. Yes, the MDO3000 is more feature rich, but it cannot reach 1000000 waveforms per second... Well, somebody does not need that update rate and prefers longer memory and that famous waveform inspector feature.

Typical marketing comparisons, they simpyl won't show the competitive product winning on anything, if it does they simply leave it out.

[Dave]

Your teardown experience is really showing, with the datasheet add-ins and all of that good stuff. This episode was a real joy to watch.

[tom66]

Here is a competitive fact sheet made by Tektronix guys. MDO3000 vs. DSOX3000. Yes, the MDO3000 is more feature rich, but it cannot reach 1000000 waveforms per second... Well, somebody does not need that update rate and prefers longer memory and that famous waveform inspector feature.

I have to admit, I was weighing up scopes recently and was considering the 2000X but I'm disappointed that a £850+ scope only has a megapoint of sample memory.  It was pretty much a deal breaker for me as I frequently analyse serial comms over long time periods. Anything less than about 10 megapoints (for a high end scope) is just not worth it, IMO.

[Dave Turner]

Dave,

As a point of interest how many planes are there on the boards?

[Hydrawerk]

I have to admit, I was weighing up scopes recently and was considering the 2000X but I'm disappointed that a £850+ scope only has a megapoint of sample memory. 

When you turn on serial decode, the memory drops to 500kpoints per channel at my DSOX2000. Still not bad.

[TVman]

If it must blow in, would a filter have added so much to the cost...?

That was probably something they forgot?

[Hypernova]

Here is a competitive fact sheet made by Tektronix guys. MDO3000 vs. DSOX3000. Yes, the MDO3000 is more feature rich, but it cannot reach 1000000 waveforms per second... Well, somebody does not need that update rate and prefers longer memory and that famous waveform inspector feature.

Agilent's Apr 1st promotion makes a lot of sense now. In raw performance Agilent may no longer win but if you price them out fully optioned the MDO3000 is no competition. Outside of RF heavy uses protocol decoding is much more useful.

Does make you wonder what Agilent is going do with their next gen Magazoom ASIC.

[EEVblog]

Does make you wonder what Agilent is going do with their next gen Magazoom ASIC.

It will be interesting to see. They are surely working on it, they wouldn't have just been sitting back for the last 3 years gloating about the success of the X series.
They will almost certainly have to go to an external memory architecture, as they won't be able to fit enough memory on the ASIC die to compete. 4MB was enough 3-4 years ago, but the market has changed a lot since then.

[hikariuk]

For all those who guessed what that "touch sensor" on the PCB was, it's nothing of the sort. The PCB designer got bored and wanted to put a symbol of a sine wave on an old school round scope screen. The vias connect to the internal ground plane.
FAIL IMO

10/10 for style; 1/10 for implementation?

[tom66]

Does make you wonder what Agilent is going do with their next gen Magazoom ASIC.

It will be interesting to see. They are surely working on it, they wouldn't have just been sitting back for the last 3 years gloating about the success of the X series.
They will almost certainly have to go to an external memory architecture, as they won't be able to fit enough memory on the ASIC die to compete. 4MB was enough 3-4 years ago, but the market has changed a lot since then.

I think Rigol use commodity DDR2/3. It's got a 64 bit or 128 bit bus. If you have enough pins on the FPGA you can clock in up to 16 bytes per cycle, so you can use cheap 400MHz DDR and be able to capture up to 6.4GSa/s.

Probably want a small capture buffer on the ASIC for the intensity grading, but doesn't need to be any more than (points on screen) x (num waveforms per display refresh).

Wonder what it would take to do that million wfms/s on an FPGA like Rigol do (getting about 50,000 on the DS2000 series.)

[deweff]

Why is everyone complaining about the thermal design.  Tek didn't design it to be run with the case open.  Why is anyone surprised?  It overheats when you take the fan away from the heatsinks.  Duh.  Don't do that.

[Rigby]

Duh.  Don't do that.

Yeah, I agree.  I don't think folks are complaining about the airflow design alone, it's how dust will accumulate inside.  The owner's manual doesn't have a procedure for cleaning dust from inside, so it would appear that they don't see it as an issue.

[echen1024]

For all those who guessed what that "touch sensor" on the PCB was, it's nothing of the sort. The PCB designer got bored and wanted to put a symbol of a sine wave on an old school round scope screen. The vias connect to the internal ground plane.
FAIL IMO

How did you find this out? Tek surely didn't release a document stating that "PCB Engineer gets bored designing latest product, adds vintage touch"

[Rigby]

There are Tek folks on this board, one springs to mind immediately.  And we know that Dave has gone to that person for Tek information before.

We also know that he has enough inroads into Tek that he got a fully specc'd MDO3000 for free.

[deweff]

Yeah, I agree.  I don't think folks are complaining about the airflow design alone, it's how dust will accumulate inside.  The owner's manual doesn't have a procedure for cleaning dust from inside, so it would appear that they don't see it as an issue.

Well that makes more sense.  But still, I don't think any of the main scope makers give directions on how to open up their scopes.  In general they frown upon users opening up scopes.  And it's not like they could have done anything other than air cooling (even "water cooling" has a radiator through which you blow air).  I don't think this is different for Tek than it is for any other scope maker.  If you're really worried about dust, you could rig it up with a car air filter on the intake 

We also know that he has enough inroads into Tek that he got a fully specc'd MDO3000 for free.

Tek probably sent him a scope for marketing purposes.

[aramosfet]

Hi,

I have a basic question. It was mentioned in the video that this scope does ~200,000 waveform updates per second. But the LCD display refresh rate is likely to be far less. No more than 60-100 hz. How does the synchronisation work? Or am I missing something here?

Thanks,