EEVblog 1545 - World's Fastest Oscilloscope, MXO4 TEARDOWN

[EEVblog]

The world's fastest oscilloscope, the Rohde & Schwarz MXO4 teardown!
Wow, this is an impressive beast in more ways than one.

https://www.rohde-schwarz.com/us/products/test-and-measurement/oscilloscopes/rs-mxo-4-oscilloscope_63493-1164992.html

ADC: https://www.ti.com/lit/ds/slvsdr3c/slvsdr3c.pdf
FPGA: https://www.digikey.com/en/products/detail/amd-xilinx/XCZU7EG-1FBVB900I/7034695

[YurkshireLad]

When I first read the title, I thought it said "world's fastest teardown". Doh!

[SiliconWizard]

Great engineering!

[pdenisowski]

I've had an MXO4 longer than almost anyone and I learned a lot from that video   Great stuff!

And I also completely understand the joy in cutting through a "Calibration Void if Broken" sticker ...

[nctnico]

Nice! A bit surprising that the MXO4 uses off-the-shelve ADCs. Didn't expect that but if they do the job for the right price then why not.

[TheSteve]

It certainly was pretty inside. The only negative I really noticed was if you wanted to clean the dust off the fan - quite the disassembly required. Maybe R&S wants to trade one for my modified 1 GHz MSOX3024t. There aren't many things I'd trade my special scope for, but this is one of them.

[Fgrir]

Minor nitpick, but at 16:51 and 22:20 you circle 1K resistors and call them 1M.

But yeah, great teardown.  I need to find a project to justify this scope now.

[globoy]

This might be a naive question (I'm no metrologist) but does the 200 MHz version have entirely different electronics?  I ask because the price for the ADC and ZINQ chips from distribution are very pricey (at least in low qty) and more than the list price of that version of the scope.  Otherwise R&S must be getting a helluva price on those chips.

[Kleinstein]

Given that they still give the same 5 GS/s sampling rate the ADC is likely the same, maybe a slightly more noisy bin / grade.
For the ZINQ chip kind of has to be the same or they would need separate firmware.
The digikey price for such special parts with an expected downward trend in the price are rather inflated.

So R&S will earn way less money from the 200 MHz version - though it is still not a cheap scope.

[nctnico]

This might be a naive question (I'm no metrologist) but does the 200 MHz version have entirely different electronics?  I ask because the price for the ADC and ZINQ chips from distribution are very pricey (at least in low qty) and more than the list price of that version of the scope.  Otherwise R&S must be getting a helluva price on those chips.

You can get hefty discounts from Xilinx if you register the project and Xilinx deems you worthy.

[globoy]

Thanks.  If it is all the same electronics then one does wonder about hacking the 200 MHz version :-)

[SiliconWizard]

This might be a naive question (I'm no metrologist) but does the 200 MHz version have entirely different electronics?  I ask because the price for the ADC and ZINQ chips from distribution are very pricey (at least in low qty) and more than the list price of that version of the scope.  Otherwise R&S must be getting a helluva price on those chips.

You can get hefty discounts from Xilinx if you register the project and Xilinx deems you worthy.

Yes, and it's also obvious that they don't pay the ADCs 3000 bucks each.

[Martin72]

After the teardown Dave needed a cigarette I guess..
Beautiful inside...This is building quality at it´s best, fron this point of view, the scope´s entry-price is cheap.

This might be a naive question (I'm no metrologist) but does the 200 MHz version have entirely different electronics?

You can upgrade the 200Mhz up to 1.5Ghz just via license key, so the hardware is the same.
Buy the 200Mhz "cheapo" and hack it to 1.5Ghz..

[boB]

Isn't this scope faster than the MX04 ?

https://hackaday.com/2018/09/24/tearing-into-a-1-3-million-oscilloscope/

https://www.keysight.com/us/en/product/UXR1004A/infiniium-uxr-series-oscilloscope-100-ghz-4-channels.html

I thought that was the one you were talking about when I saw the title.


boB

[EEVblog]

You can get hefty discounts from Xilinx if you register the project and Xilinx deems you worthy.

Yes, and it's also obvious that they don't pay the ADCs 3000 bucks each.

Of course, but it's still funny.

[SiliconWizard]

You can get hefty discounts from Xilinx if you register the project and Xilinx deems you worthy.

Yes, and it's also obvious that they don't pay the ADCs 3000 bucks each.

Of course, but it's still funny.

Yes, you gotta wonder who is ever going to buy these at this price, so why bother displaying them in the catalog?
(If you're going to make prototypes you're probably still not going to buy them through this channel.)

[pdenisowski]

Isn't this scope faster than the MX04 ?

https://www.keysight.com/us/en/product/UXR1004A/infiniium-uxr-series-oscilloscope-100-ghz-4-channels.html

I thought that was the one you were talking about when I saw the title.

"Fastest" means "highest waveform update rate" (number of waveforms per second that can be acquired and displayed).  By that measure, the MXO4 is an order of magnitude faster than the UXR.

A lot of people think that sampling rate = speed but don't realize that most scopes actually discard a very large percentage of the samples they acquire and thus are "blind" most of the time. 

I did an entire video (which has also been turned into a whitepaper) on this topic:



A loose analogy:  a "fast" or "high speed" camera is one that acquires and stores a large number of frames per second, not one that has a very high number of pixels per frame.

[pdenisowski]

From Keysight's own whitepaper on the topic

"Waveform update rate can be extremely important when evaluating oscilloscopes for purchase. Although this specification is often overlooked, it can have a direct impact on your ability to capture a random and infrequent event which occurs just once in a million occurrences of your signal. There are three reasons why fast update rates are important for today’s oscilloscopes"

https://www.keysight.com/us/en/assets/7018-01745/white-papers/5989-7885.pdf

[Mechatrommer]

Isn't this scope faster than the MX04 ?
https://www.keysight.com/us/en/product/UXR1004A/infiniium-uxr-series-oscilloscope-100-ghz-4-channels.html

"Fastest" means "highest waveform update rate" (number of waveforms per second that can be acquired and displayed).  By that measure, the MXO4 is an order of magnitude faster than the UXR.

isnt fastest means "the highest fart count in a day"? it was in the dictionary of the future right?

[Fungus]

How much would it cost to manufacture a PCB of that size/complexity with that many components on it?

(I mean just pure manufacturing costs, ignoring the price of the components)

[boB]

From Keysight's own whitepaper on the topic

"Waveform update rate can be extremely important when evaluating oscilloscopes for purchase. Although this specification is often overlooked, it can have a direct impact on your ability to capture a random and infrequent event which occurs just once in a million occurrences of your signal. There are three reasons why fast update rates are important for today’s oscilloscopes"

https://www.keysight.com/us/en/assets/7018-01745/white-papers/5989-7885.pdf

I would still like to see the insides of a one $million dollar scope.   

Compared to $8000, there must be something interesting in there.

boB

[Detlev]

I would still like to see the insides of a one $million dollar scope.   

Compared to $8000, there must be something interesting in there.

boB

https://youtu.be/DXYje2B04xE

😎😉

[nctnico]

How much would it cost to manufacture a PCB of that size/complexity with that many components on it?

In volume: probably around $100 to $200 for placement costs. You'll only be paying for time needed by a P&P machine and it is not like the board is filled to the brim with components. The board is mostly green.

[julian1]

The bga footprints have exposed bottom-layer pads for probing. And in places, there are two traces routed between each pad.

[boB]

From Keysight's own whitepaper on the topic

"Waveform update rate can be extremely important when evaluating oscilloscopes for purchase. Although this specification is often overlooked, it can have a direct impact on your ability to capture a random and infrequent event which occurs just once in a million occurrences of your signal. There are three reasons why fast update rates are important for today’s oscilloscopes"

https://www.keysight.com/us/en/assets/7018-01745/white-papers/5989-7885.pdf

I guess it depends on your interpretation of "fast" then.   Evidently Keysight themselves said this....

"Keysight has introduced what is claims is the world’s fastest oscilloscope, operating at 256Gsample/s on signals up to 110GHz."

So, I guess you could take that either way.

I would think that rise and fall time might also be a factor that determines what you call faster or slower

boB

[Fungus]

Keysight said...

Who cares what some Keysight marketer said?

I would think that rise and fall time might also be a factor that determines what you call faster or slower

No, that's called "bandwidth".

[EEVblog]

"Fastest" means "highest waveform update rate" (number of waveforms per second that can be acquired and displayed).  By that measure, the MXO4 is an order of magnitude faster than the UXR.
A lot of people think that sampling rate = speed but don't realize that most scopes actually discard a very large percentage of the samples they acquire and thus are "blind" most of the time. 

AFAIK the MXO4 is the only scope to actually display the blind time on screen.
If I was them I'd be bragging too.

13 years now since the megazoom IV ASIC... 

[EEVblog]

Yes, you gotta wonder who is ever going to buy these at this price, so why bother displaying them in the catalog?
(If you're going to make prototypes you're probably still not going to buy them through this channel.)

Military or government. They need to fix a bit of kit and need a spare chip from stock and an already approved vendor.

[TheSteve]

"Fastest" means "highest waveform update rate" (number of waveforms per second that can be acquired and displayed).  By that measure, the MXO4 is an order of magnitude faster than the UXR.
A lot of people think that sampling rate = speed but don't realize that most scopes actually discard a very large percentage of the samples they acquire and thus are "blind" most of the time. 

AFAIK the MXO4 is the only scope to actually display the blind time on screen.
If I was them I'd be bragging too.

13 years now since the megazoom IV ASIC... 

Pretty impressive it took 13 years for someone to beat the Megazoom IV ASIC.

[nctnico]

Mainly because high waveform update rates are more of a marketing gimmick than actually usefull. Features like reverse brightness are much more handy to spot deviations in a signal. But other than that, using triggers is the way to go. Together with segmented recording so you can capture each individual event with several other signals if you like and study cause & effect in detail afterwards.

[pdenisowski]

Pretty impressive it took 13 years for someone to beat the Megazoom IV ASIC.

The ASIC is one of the most time-consuming parts, but there are other components / systems that also take time.  We started planning the MXO4 in 2016

(That's not a secret, we have a nice Linkedin post about it )

https://www.linkedin.com/feed/update/urn:li:activity:6993907988830822400

[thm_w]

It certainly was pretty inside. The only negative I really noticed was if you wanted to clean the dust off the fan - quite the disassembly required. Maybe R&S wants to trade one for my modified 1 GHz MSOX3024t. There aren't many things I'd trade my special scope for, but this is one of them.

Compressed air.

[boB]

"Fastest" means "highest waveform update rate" (number of waveforms per second that can be acquired and displayed).  By that measure, the MXO4 is an order of magnitude faster than the UXR.
A lot of people think that sampling rate = speed but don't realize that most scopes actually discard a very large percentage of the samples they acquire and thus are "blind" most of the time. 

AFAIK the MXO4 is the only scope to actually display the blind time on screen.
If I was them I'd be bragging too.

13 years now since the megazoom IV ASIC... 

Blind Time (didn't know it was called that)  must be pretty long for our 8 channel Pico-Scope.

[EEVblog]

Mainly because high waveform update rates are more of a marketing gimmick than actually usefull. Features like reverse brightness are much more handy to spot deviations in a signal. But other than that, using triggers is the way to go. Together with segmented recording so you can capture each individual event with several other signals if you like and study cause & effect in detail afterwards.

It's hard to trigger on something you never saw in the first place.

[David Hess]

Mainly because high waveform update rates are more of a marketing gimmick than actually usefull. Features like reverse brightness are much more handy to spot deviations in a signal. But other than that, using triggers is the way to go. Together with segmented recording so you can capture each individual event with several other signals if you like and study cause & effect in detail afterwards.

It's hard to trigger on something you never saw in the first place.

Back when Tektronix came out with their Digital Phosphor Oscilloscopes which capture a histogram directly to the frame buffer, the idea was to use DPO mode to find the glitch, and then advanced triggering and segmented memory to capture it.

Mask testing should be able to do it all however at least back then, mask testing was much slower than advanced triggering, and I suspect that is still the case.

[EEVblog]

Mainly because high waveform update rates are more of a marketing gimmick than actually usefull. Features like reverse brightness are much more handy to spot deviations in a signal. But other than that, using triggers is the way to go. Together with segmented recording so you can capture each individual event with several other signals if you like and study cause & effect in detail afterwards.

It's hard to trigger on something you never saw in the first place.

Back when Tektronix came out with their Digital Phosphor Oscilloscopes which capture a histogram directly to the frame buffer, the idea was to use DPO mode to find the glitch, and then advanced triggering and segmented memory to capture it.
Mask testing should be able to do it all however at least back then, mask testing was much slower than advanced triggering, and I suspect that is still the case.

Yes, you can also use infinite persistence mode on any scope.
The problem of blind time still remains through on DPO or infinite persistence mode. The longer the blind time the longer your have to theoretically wait for your glitch to become visible.
And that's only IF you suspect an infrequent glitch in the first place and enable those modes.

The whole push towards fast updating started by Keysight Megazoom and now bested by R&S MXO-EP ASIC is that you stand a better chance of seeing something infrequent when you are just regularly using the scope.

[Mechatrommer]

Mainly because high waveform update rates are more of a marketing gimmick than actually usefull. Features like reverse brightness are much more handy to spot deviations in a signal. But other than that, using triggers is the way to go. Together with segmented recording so you can capture each individual event with several other signals if you like and study cause & effect in detail afterwards.

It's hard to trigger on something you never saw in the first place.

if its still not obvious to them... the other term is... dead time... in which time triggering circuit is shutoff and acquisition memory is not enough to capture the rest... https://www.ni.com/en-my/shop/electronic-test-instrumentation/oscilloscopes/what-are-oscilloscopes/every-measurement-starts-with-a-trigger.html but instead of marketing confusion as "fastest" scope, its more practical and "educational" to say "the least" blind time scope, or 10% or 5% blind, or even better if it can achieve 0% blind/dead time... 0% blind time is not a gimmick, its a 100% guarantee you'll capture glitch on first observation occurrence (if infinite persistence feature is turned on, a feature that is common in even cheap dsos today).

[pdenisowski]

if its still not obvious to them... the other term is... dead time

When I made my video and whitepaper explaining these concepts (in anticipation of the MXO4 launch), I struggled with what terminology to use: "blind time" vs. "dead time" and then all of the possible permutations of the words "acquisition," "update," "capture," "waveform," and "rate", e.g.

- acquisition rate
- waveform acquisition rate
- update rate
- waveform update rate
- capture rate
- etc. etc.

So much easier making content on spec ans and VNAs where everyone uses (more or less) the same terminology for everything ...

[David Hess]

The whole push towards fast updating started by Keysight Megazoom and now bested by R&S MXO-EP ASIC is that you stand a better chance of seeing something infrequent when you are just regularly using the scope.

When did HP release their first Megazoom DSO?  Tektronix released their first DPO DSO in 1995 (TDS500B and TDS700A InstaVu) but they did not call it that then.  There is a gap in HP catalogs from 1994 to 1996 available online.  The HP 1997 catalog discusses HP Megazoom but not the 1993 catalog.

[jonpaul]

Bonjour a tous,  I  use my 1970s Tektronix 7104/7A29/7B10/7B15 analog mainframe.

1 GHz analog with MCP plate CRT.

https://w140.com/tekwiki/wiki/7104

Jon

[nctnico]

Mainly because high waveform update rates are more of a marketing gimmick than actually usefull. Features like reverse brightness are much more handy to spot deviations in a signal. But other than that, using triggers is the way to go. Together with segmented recording so you can capture each individual event with several other signals if you like and study cause & effect in detail afterwards.

It's hard to trigger on something you never saw in the first place.

I can't keep my eyes open and stare at a screen for that long. A glitch is gone in a blink of an eye. If I really want to know whether a signal is not going out of bounds, I set a scope to infinite persistence and let it run overnight as a last resort. Otherwise I set a trigger to a limit (pulse width, height, runt, etc). The more time / periods you can fit on a screen, the better the ratio blind time versus capture time becomes. So a lower waveform update rate can actually be better in terms of blind time to captured time ratio! I demonstrated that in the RTM3004 review I did a couple of years ago (combined with the rather unique reverse brightness feature which I suspect is also present in the MXO4).

[ballsystemlord]

What is the thing near the center of the picture that has blue near the terminals, a black body, and three red dots on it? I haven't the faintest. Offset in video 8:44.

Thanks!

[KE5FX]

What is the thing near the center of the picture that has blue near the terminals, a black body, and three red dots on it? I haven't the faintest. Offset in video 8:44.

Thanks!

Those are SMT inductors, probably from Coilcraft and probably in the 10 nH-10 uH range.

They are sort of a pain in production because they are wound with wire that makes the proverbial RCH look like LMR600 coax.  The windings are covered with blue plastic on top but are exposed on the bottom.  They are easily damaged during assembly and the welds at the contact ends aren't very stout either.  As if that weren't enough, they are difficult to inspect because the solder contact area may not be visible from overhead.

[Fungus]

I can't keep my eyes open and stare at a screen for that long. A glitch is gone in a blink of an eye. If I really want to know whether a signal is not going out of bounds, I set a scope to infinite persistence and let it run overnight as a last resort.

If it's only one or two fast glitches and you've got dead time in your 'scope then it might not be enough. 

[EEVblog]

Mainly because high waveform update rates are more of a marketing gimmick than actually usefull. Features like reverse brightness are much more handy to spot deviations in a signal. But other than that, using triggers is the way to go. Together with segmented recording so you can capture each individual event with several other signals if you like and study cause & effect in detail afterwards.

It's hard to trigger on something you never saw in the first place.

I can't keep my eyes open and stare at a screen for that long. A glitch is gone in a blink of an eye. If I really want to know whether a signal is not going out of bounds, I set a scope to infinite persistence and let it run overnight as a last resort. Otherwise I set a trigger to a limit (pulse width, height, runt, etc). The more time / periods you can fit on a screen, the better the ratio blind time versus capture time becomes. So a lower waveform update rate can actually be better in terms of blind time to captured time ratio! I demonstrated that in the RTM3004 review I did a couple of years ago (combined with the rather unique reverse brightness feature which I suspect is also present in the MXO4).

Again, the point is that a faster updating scope has a much greater chance of showing you something you are not expecting, and it has a much greater chance of doing that during normal operation of the scope.
I don't know why this simple fact is questionable. Sure, you can argue about the need, and bang-per-buck and what have you, but all things being equal a faster updating scope is better than a slower one.
This is why scopes manufacturers have been focussing on and improving this over many decades now.

[David Hess]

I know how early Tektronix DPOs work, however are there any detailed descriptions of how HP Megazoom and now whatever R&S is doing work?  Google results were not very informative.  I suspect HP released a detailed article but it is too old to be considered relevant in search results.

[switchabl]

The Feb. 1992 issue of the HP journal was all but dedicated to pre-Megazoom 54600 series DSOs.
http://hparchive.com/Journals/HPJ-1992-02.pdf

Apr. 1997 issue has several articles on the first generation Megazoom 54645A/D but doesn't go quite as deep.
http://hparchive.com/Journals/HPJ-1997-04.pdf

[Mechatrommer]

When I made my video and whitepaper explaining these concepts (in anticipation of the MXO4 launch), I struggled with what terminology to use: "blind time" vs. "dead time" and then all of the possible permutations of the words "acquisition," "update," "capture," "waveform," and "rate", e.g.

- acquisition rate
- waveform acquisition rate
- update rate
- waveform update rate
- capture rate
- etc. etc.

So much easier making content on spec ans and VNAs where everyone uses (more or less) the same terminology for everything ...

memory depth / acquisition rate = time frame of signal captured (T1), time to rearm trigger (T2). 100 x (T2 - T1) / T2 is blind time percentage. i'm sure you are a lot better than me, or maybe you did provide in datasheet i havent look, i cant afford it anyway. cheers.

[David Hess]

memory depth / acquisition rate = time frame of signal captured (T1), time to rearm trigger (T2). 100 x (T2 - T1) / T2 is blind time percentage. i'm sure you are a lot better than me, or maybe you did provide in datasheet i havent look, i cant afford it anyway. cheers.

When I was looking at this in detail a couple years ago with the Tektronix DPO design in mind, I concluded that double buffering with two separate acquisition records was required to minimize blind time, which means halving record length.

I think the old Tektronix DPO design was limited by the rearm time of its analog trigger, which could have been made much faster.

[nctnico]

memory depth / acquisition rate = time frame of signal captured (T1), time to rearm trigger (T2). 100 x (T2 - T1) / T2 is blind time percentage. i'm sure you are a lot better than me, or maybe you did provide in datasheet i havent look, i cant afford it anyway. cheers.

When I was looking at this in detail a couple years ago with the Tektronix DPO design in mind, I concluded that double buffering with two separate acquisition records was required to minimize blind time, which means halving record length.

Dual buffering is the generic approach to do acquisitions in every DSO. If you look close to the specs of a DSO you can see some have more memory in single shot and/or sequence mode.

[David Hess]

Dual buffering is the generic approach to do acquisitions in every DSO. If you look close to the specs of a DSO you can see some have more memory in single shot and/or sequence mode.

None of my DSOs double buffer simply because it would have required two completely separate memory banks with duplicate hardware.  A single memory bank could not sustain access from the digitizer and the processor simultaneously.  On older DSOs double buffering might not yield any improvement if the analog trigger took too long to read and rearm.

[nctnico]

Then your DSOs must be really old  Even the Keysight Megazoom Asic does double buffering.

[Mechatrommer]

Then your DSOs must be really old  Even the Keysight Megazoom Asic does double buffering.

how much percentage of blind time do you know?

[egonotto]

Hello,

the MXO 4 has at least 4 GB for temporary information storage for operating system and instrument firmware and >= 7 GB for waveform data and measurement data.
Such information can be found in the "R&S®MXO 4 Series Oscilloscope Instrument Security Procedures"

Best regards
egonotto

[David Hess]

Then your DSOs must be really old  Even the Keysight Megazoom Asic does double buffering.

With an ASIC it makes a lot of sense because the extra multiplexing hardware can be integrated.  I am curious about their fast RAM implementation.

With older DSOs it was a struggle just to make the acquisition RAM fast enough without wider word widths, and this shows up with multiple channels not only halving the record length but the sampling rate also, so double buffering would have come at considerable cost.

All of the designs I came up with a couple years ago were either limited by the RAM which could fit on the acquisition FPGA, or the processor's cache size.  The cost of external memory was too high for what I was considering.

[nctnico]

Depends a bit on what you want to do where it comes to samplerate. Over 15 years ago I already managed to get to 1.6GB/s with DDR2 memory on an FPGA using a soft DDR2 controller. With a hard IP DDR controller on a more modern FPGA, you can easely get way more bandwidth. And typically these hard IP blocks support simulating dual/multi port access to make life easy. Pretty neat stuff! See the many oscilloscopes that are build around the Xilinx Zync FPGA for example.

[David Hess]

Over 15 years ago I already managed to get to 1.6GB/s with DDR2 memory on an FPGA using a soft DDR2 controller. With a hard IP DDR controller on a more modern FPGA, you can easely get way more bandwidth. And typically these hard IP blocks support simulating dual/multi port access to make life easy. Pretty neat stuff! See the many oscilloscopes that are build around the Xilinx Zync FPGA for example.

I was looking at DPO style designs so the data is not streamed to memory but instead used to update a 2D histogram which requires a read, modify, and write cycle, so standard DRAM would run at only a fraction of its rated speed.  That is why the alternative of using processor cache was so interesting.  An FPGA can do it because the operation can be split up at the cost of multiplying the required RAM, and then combining the acquisition histograms later.

The segmented memory method has the advantage of preserving the original acquisition record, but at the cost of extra complexity.

[Wolfgang]

Nice! A bit surprising that the MXO4 uses off-the-shelve ADCs. Didn't expect that but if they do the job for the right price then why not.

I made some tests about how many bits of the 12Bit Texas Instruments ADC are really usable. The answer is a bit above 10 at best, at 10MHz signal frequency and in high resolution mode.
Details here:

https://electronicprojectsforfun.wordpress.com/to-enob-or-not-to-enob/

[switchabl]

First of all, thank you for posting this, this looks like a formidable effort!

I will have to look at it again in more detail later but you seem to be using something like SFDR in your ENOB definition except you are counting noise bins as "peaks", is this correct? I see two problems with this:
1) The power of a noise bin is dependent on the (arbitrarily chosen) FFT length. In case the largest "peak" is a noise bin, the ratio is essentially meaningless. A normal SFDR measurement would use a sufficiently long signal to ensure that the largest spur/harmonic is clearly above the noise floor and the ratio is always taken between two tones.
2) This would actually be a strictly less stringent measure than "conventional" SINAD. You are comparing the signal power to that of the largest non-signal bin. SINAD (the way I would define it) involves the sum of all the other bins.

[Wolfgang]

First of all, thank you for posting this, this looks like a formidable effort!

I will have to look at it again in more detail later but you seem to be using something like SFDR in your ENOB definition except you are counting noise bins as "peaks", is this correct? I see two problems with this:
1) The power of a noise bin is dependent on the (arbitrarily chosen) FFT length. In case the largest "peak" is a noise bin, the ratio is essentially meaningless. A normal SFDR measurement would use a sufficiently long signal to ensure that the largest spur/harmonic is clearly above the noise floor and the ratio is always taken between two tones.
2) This would actually be a strictly less stringent measure than "conventional" SINAD. You are comparing the signal power to that of the largest non-signal bin. SINAD (the way I would define it) involves the sum of all the other bins.

Comment is correct, but that is why I used standard values for frequency, memory depth and sample rate. The point is when you see a "spur" or glitch or whatever, its normally not possible to determine what exactly caused it. In case of strong spurs, its clear, if noise is the main reason its getting fuzzy. You can also see this from the standard deviation values on my plots. Strictly speaking the total power of ALL spurs (plus noise) would go into the SINAD figure, but the practice I used is common for ENOB measurements AFAIK. The hard limit, at least as larger amplitudes, is always the 3rd or 5th harmonic of the signal frequency. In fact, the "largest peak" method overestimates the ENOB value by a small amount. For the TI ADC in the MXO4, the values found are very plausible. If what you want to measure are spectra, the SFDR approach is more practical than integrating over all non-signal components.

[ballsystemlord]

@EEVblog / Dave,
I think you were very pressed for time when you did this video, there's so much stuff in this scope that's not in other scopes, so I'd like to ask if you could do a follow up video on this scope so that you could go more in depth about the various components and stuff. What is believed to be Coilcraft inductors are just the tip of the iceberg in cool stuff this oscilloscope has inside of it.

Thanks!