New 2GHz touchscreen scope from Tek, June 6th

[fcb]

Congratulations to Tek on the new 'scope series.

Presumably they plan on launching a 54 (4 channel) at some point? That will be very interesting to me.  My last Tek scope was a Telequipment D64 (when I was 13!), I've been an HP/Agilent/Keysight chap 'scope user since then. Perhaps my next big scope will be a Tek.

[BrianH_Tektronix]

Hi

BrianH_Tektronix wrote:
"
Its also is fastest scope on the market with vertical resolution higher than an 8-bit ADC
"
No Keysight Technologies
Infiniium S-Series has 10 Bit and 10GS/s per channel

But your new scope has with 6.25GS/s only 8 Bit and not 12 Bit.
12 Bits only with 3.125 GS/s

Best regards
egonotto

My mistake for not being clear enough. Fastest was in reference to Waveform Capture Rate. Which increase your probability of finding rare events. Keysight's S-Series and the HDO-A series won't mention the fact their scopes are under a max of 1000 wfms/sec only when in specific settings. While the 5 Series MSO can capture over 500,000 wfms/sec and is fully optimized at the press of a button.

The 5 Series MSO does have a true 12-bit ADC on each channel. The fact that 6.25GS/s on 8 channels is a lot of data to push through the system and we don't store memory on the chip does require us to push data in 8-bits or 1 byte transfers at max sample rate. We auto switch when you go into the new High Res mode or manually set the sample rate at 3.125GS/s or lower.

I'm not sure if you're aware, but the S-Series does only transfer 10-bits at 20GS/s and 10GS/s. At 5GS/s and below its running in 8-bit transfer mode, and their peak-detect and HighRes modes are based off 8-bit data. They then have to box car average the 8-bit data to get up to 12-bit in HighRes. I guess you could say they have more 8-bit modes than 10-bit modes.... 

[technogeeky]

edit: Mistakes were made in terminology. See below.

[P_Doped]

The Y-axis is cost, not cost/channel.  The cost appears to go up slower than linear:
Ex:
Bandwidth
6-channel 500MHz ~ $19k
6-channel 2GHz ~ $30k (~1.5x cost for 4x the BW)

Channels
4-channel 350MHz ~ $12.5k
8-channel 2GHz ~ $21k  (~1.7x cost for 2x the channel count)

Assuming I'm interpreting your charts correctly, logarithmically is the correct term.

[Hydron]

Just a quick query about these low-C probes - are they really 10x attenuation? Elsewhere I've seen that they use 250k termination (plus 9.75M probe resistance) when plugged in, and and are actually 40x (which makes the 4pF number possible).

I believe the scope input itself might be 250k (there is a "termination" test in the UI), but the net effect is a 10x probe with low capacitance.  There is a special ASIC which enables it.

There is some confusion because other vendors (Tek included) ALSO offer a Zo probe which is low capacitance but also high attenuation (or 10x attenuation but low DC resistance).  LeCroy PP066, Tektronix P6150, Keysight N2874/6A - all are "transmission line" probes and typically have only about 1k of input impedance, but low capacitance and high bandwidth.

The TPP1000 and other probes are a new architecture in passive probes and require a scope with the ASIC.  There is a white paper here on the technology:
http://www.tek.com/document/application-note/low-capacitance-probes-minimize-impact-circuit-operation-1

Two other benefits of the low capacitance probes.  Yes they are 1GHz with a short ground lead, but lower input capacitance means you can even get 500MHz with a ground lead, which is not the case for most 10M passive probes.

And, there is the TPP0502 which gets you 500MHz of BW but only 2x attenuation - great for low noise applications!

http://www.effectivebits.net/2011/09/probe-attenuation-overlooked.html

I agree that low-C, high-Z passive probes are great, and quite a different beast from "transmission-line" probes. My question is that if the scope input is set to 250k and the probe appears as 10Meg, surely this is a 40x probe, not a 10x? If the scope does some magic to eliminate the noise and sensitivity disadvantage of the extra attenuation (low noise pre-amp or similar?) then sure, this is fantastic, but if not, isn't it really just a 40x probe with a lower-than-1M scope input impedance?

[Omicron]

The 5 series is truly totally new.  Not just a new box or a new UI, but a totally new ASIC chipset running the scope.

I hope this is a promise of things to come and this new ASIC and large touchscreen UI will find its way into the successors of the MDO4000 and the MDO3000 series?

[nfmax]

If I understand this correctly, the ASIC-per-channel design means the there is no tradeoff in either sample rate or memory depth as additional channels are enabled. But on the other hand, turning off unused channels doesn't increase either of these parameters either.

If memory bandwidth is now the limiting factor (it seems to be the root cause of the bit depth limitation at the 6.5 Gsps rate), can disabling channels increase bit depth at the fastest sample rate?

[tautech]

If I understand this correctly, the ASIC-per-channel design means the there is no tradeoff in either sample rate or memory depth as additional channels are enabled. But on the other hand, turning off unused channels doesn't increase either of these parameters either.

If memory bandwidth is now the limiting factor (it seems to be the root cause of the bit depth limitation at the 6.5 Gsps rate), can disabling channels increase bit depth at the fastest sample rate?

Watch Sharihar's vid.
https://www.eevblog.com/forum/testgear/overview-of-the-tektronix-mso58-8-channel-6-25gss-2ghz-mixed-signal-scope/msg1226890/#msg1226890

[rf-design]

The 5 Series MSO does have a true 12-bit ADC on each channel. The fact that 6.25GS/s on 8 channels is a lot of data to push through the system and we don't store memory on the chip does require us to push data in 8-bits or 1 byte transfers at max sample rate. We auto switch when you go into the new High Res mode or manually set the sample rate at 3.125GS/s or lower.

Thank you for your technical response. It will help us to understand the merits of the ASIC design for the scope.

If you could operate the 12bit ADC with 450fs jitter it will achieve an ENOB of 7.2bit at 2GHz. That is a good number for the jitter of the ADC and the clock source together. The spec values are 7.0bit@1GHz for the 2GHz model and 7.6dB@1GHz for the <2GHz models. That seems very conflicting because the 1GHz-BW noise at 5mV/DIV is 202uVe for the 2GHz model and 262uVe for the <2GHz models. Natural I would assume the lower noise of the 2GHz AFE indicate a better ENOB. But it is reverse. Why?

If you are able to design and produce a 12bit 6.25GS/s ADC why do you not simply extend the memory interface to the bandwidth of the ADC? I could not believe that the 12bit are true at 6.25GS/s and wasted!

The noise values for the 2GHz model at 1mV/DIV and 2mV/DIV are only stated for 175MHz and 350MHz bandwidth. Does this mean that I could not measure 1mV/DIV and 2mV/DIV at the full bandwidth?

Is the FlexChannel based on an ASIC which combine 12/8bit-ADC and 8bit-Digital signals in one digital trigger engine? That could be some kind of cost optimization for chip design. But what is the benefit for the scope designer to have one channel ASIC?

[fonograph]

lets compare RMS noise at 50ohm input            Keysight Infiniium S 1GHz                  Tektronix MSO 5 1GHz                  LeCroy HDO 4000 1GHz
                                                               
                                                     1 mV/div                               90 uV                                            254 uV                                       145 uV

                                                     1 V/div                                6.8 mV                                           13.0 mV                                      4.90 mV

[Muxr]

After seeing Shariar's video, that scope looks great. Tek looks to be back!.. hopefully some of this tech trickles down to some more affordable scopes over time.

[w2aew]

One of the things that I find very compelling about the Flexchannels is when you consider their use in non-digital applications.  Each bit of the 8-bit input has logic threshold than can be varied between +/-40V.  I have worked with many customers where this is a very valuable feature.  Consider avionics applications where there are dozens of electro-mechanical actuators, sensors, etc. that needed monitoring, or RADAR applications where there are many discrete signals (gating signals, power for various modules, RF switches, etc.) present in a system.  These flex channels give the engineer the ability to get a system level view of their analog signals, logic command and control, as well as multiple supply, sensor, and other system level signals.  Features like this are a huge development investment, and certainly are not undertaken in a vacuum - they're result of years of customer application focus and feedback.

One of the nicest features of this scope, that unfortunately doesn't show up well in video, is the *high-resolution* display.  Not only is the display big, it's 1920x1080 resolution is better than my laptop!  This gives you really nice, non-intrusive vertical and horizontal axis labels, crisp and informative dialogs and menus, and of course, high-res waveform quality.

Like Joel and Brian have stated, it's hard not to be a bit biased since I work for Tektronix.  However, I spent 25 years as an engineer, as a test equipment customer/user prior to joining Tek - and this is certainly one of the nicest scope developments that I can remember in my long career.

[Eric_S]

Nice new scope that looks like it could be used in some nice ways for electric motor analysis.

So it's eight channels and it's got a 12bit ADC (@3.125 GS/s, 8bit @ 6.25). Fine enough, the relevant Lecroy does 2.5GS/s http://teledynelecroy.com/oscilloscope/oscilloscopeseries.aspx?mseries=547 (There's allso the Yokogawa 8-chan, and the Yoko scopemeter ... and some scopemeter'ish Keysight offering iirc that are skulking about in this market)

It's got up to 64 digital channels at the sacrifice of analogs. Bad design choice for end users IMO. No matter if it makes it neatly modular when designing a new product around the platform. But but.

The mem is bigger than the Lecroy's standard, but shorter than the Lecroy's max.

Screen looks very nice.


Now what to use this for?

Well one application may be that you are looking at an electrical motor, and you want to look at a three phase input (that's six channels right off the bat if you don't want to interpolate), and you may also be interested in looking at torque and angle (+1 analog and three digitals, x2 if you got some internal sensors that you want to tap into at the same time as the reference ones). And all of the suddenly those flex channels are all used up.

Another thing I wonder about is the ability to do write your own analysis channels in python/matlab and get it displayed straight off in the scope?

I'm assuming that things like an electrical motor analysis package etc will come in time, to offer the nice people from Tesla, Toyota, MAN, ... and so on.


A bit rambly, but it's my first post, so be nice.

[TheSteve]

1920x1024? I would think it is actually 1920x1080 (and you need a new laptop if it doesn't have a FHD LCD )

[tautech]

Excellent overview video by Shahriar. 

It really is a drool-worthy scope.  Large high def touchscreen with what appears to be a fast and responsive, well thought out interface. 12 bits, deep memory and full 6.5 Gs/S on every channel - impressive.  But the price even at the low end is in the "if you have to ask..." territory.  And the Flex Channel thing - while perhaps offering some advantages - has several disadvantages as Dave points out - and for that price?  Clear corporate profit motive behind it IMO.

Anyways - a scope of this class is way beyond me and any hobbyist, yet I think its release along with the recent R & S release points to larger, higher def touch screens and higher bit ADCs are the future for scopes and eventually will likely trickle down into the modest spec, affordable for home use scopes.

And yeah - those anti-collapse feet are the bomb!  I don't know why other manufacturers haven't done this.  That feature cannot be high cost and should be on every scope in any price range (and probably soon will be).

I thought so too.
Such a simple and elegant solution that has been totally overlooked by manufacturers for too bloody long. 
Well done Tek. 

[w2aew]

The new measurement system is remarkable and I THINK it can edge count, but perhaps W2AEW can check it for me since he has a scope available and I don't.

Yes, a couple of ways.  Since the measurement system makes measurements of the entire acquired record, it counts the number of occurances of each measurement (to compute stats), so you've got the count of edges or whatever measurable feature you want.

[w2aew]

It would have been really nice to have a dedicated channel button above each input connector.

It does!

They don't vanish.  If you're not a fan of swiping the channel badge icon to the trash, there's always the regular, color coded hard button per channel on the side of the scope that you can use to add/select/remove any of the channels without using the touchscreen.

[w2aew]

1920x1024? I would think it is actually 1920x1080 (and you need a new laptop if it doesn't have a FHD LCD )

Yeah - I probably need a new laptop!  ;-)

[egonotto]

Hi,

the scope is great.

But I can not understand why they gives the random noise with 3 digits.

In table "MSO5X < 2 GHz models, High Res mode" page 17 it seems to be a print error 200MHz (at 1MOhm) should perhaps 20MHz?

Best regards
egonotto

[technogeeky]

One of the things that I find very compelling about the Flexchannels is when you consider their use in non-digital applications.  Each bit of the 8-bit input has logic threshold than can be varied between +/-40V.  I have worked with many customers where this is a very valuable feature.  Consider avionics applications where there are dozens of electro-mechanical actuators, sensors, etc. that needed monitoring, or RADAR applications where there are many discrete signals (gating signals, power for various modules, RF switches, etc.) present in a system.  These flex channels give the engineer the ability to get a system level view of their analog signals, logic command and control, as well as multiple supply, sensor, and other system level signals.  Features like this are a huge development investment, and certainly are not undertaken in a vacuum - they're result of years of customer application focus and feedback.

One of the nicest features of this scope, that unfortunately doesn't show up well in video, is the *high-resolution* display.  Not only is the display big, it's 1920x1024 resolution is better than my laptop!  This gives you really nice, non-intrusive vertical and horizontal axis labels, crisp and informative dialogs and menus, and of course, high-res waveform quality.

Like Joel and Brian have stated, it's hard not to be a bit biased since I work for Tektronix.  However, I spent 25 years as an engineer, as a test equipment customer/user prior to joining Tek - and this is certainly one of the nicest scope developments that I can remember in my long career.

This is something I didn't understand until I saw Shahriar's video. +/- 40V is more like the specification for a normal oscilloscope input rather than a logic analyzer input. You could conceivably watch the switching pattern of dozens of power FETs at once. The only case I can think of is using the logic channels through resistive dividers to watch all 20 or 40 FETs inside of a big inverter/charger (like 3000-4000W) and look for timing mismatches that need to be trimmed out.

I do wonder what kind of safety mechanisms those logic probe packs have built in, and if they offer complete protection to the scope frontend.

[Fungus]

Despite the criticism, it looks like there are a lot of things to really like about this scope

Yep. It will come down to usage and UI design but I haven't seen anything that puts me off*. Most of the 'complaints' here are just people with a lack of imagination thinking it has to work a certain way when it doesn't have to at all, it can work in all sorts of ways they simply didn't think of.

If Tex has sat down and done the UI right then it could easily be a game-changing device and what we're looking at here is the future of oscilloscopes.

If I have to criticize it, it would be that "vertical" isn't a good orientation for a touch screen that you use all day long.


(*) Apart from the price, obviously. I don't think I'll ever own one of these (unless I won the prize draw).

[technogeeky]

The new measurement system is remarkable and I THINK it can edge count, but perhaps W2AEW can check it for me since he has a scope available and I don't.

Yes, a couple of ways.  Since the measurement system makes measurements of the entire acquired record, it counts the number of occurances of each measurement (to compute stats), so you've got the count of edges or whatever measurable feature you want.

As a test of your honesty (and we all know you, and that you're honest)... one of those annoying interview questions:

What bugs/annoyances have you found in using the unit that you've reported, and have been fixed?

What bugs/annoyances have you found in using the unit that haven't yet been fixed?

[BrianH_Tektronix]

The 5 Series MSO does have a true 12-bit ADC on each channel. The fact that 6.25GS/s on 8 channels is a lot of data to push through the system and we don't store memory on the chip does require us to push data in 8-bits or 1 byte transfers at max sample rate. We auto switch when you go into the new High Res mode or manually set the sample rate at 3.125GS/s or lower.

Thank you for your technical response. It will help us to understand the merits of the ASIC design for the scope.

If you could operate the 12bit ADC with 450fs jitter it will achieve an ENOB of 7.2bit at 2GHz. That is a good number for the jitter of the ADC and the clock source together. The spec values are 7.0bit@1GHz for the 2GHz model and 7.6dB@1GHz for the <2GHz models. That seems very conflicting because the 1GHz-BW noise at 5mV/DIV is 202uVe for the 2GHz model and 262uVe for the <2GHz models. Natural I would assume the lower noise of the 2GHz AFE indicate a better ENOB. But it is reverse. Why?

If you are able to design and produce a 12bit 6.25GS/s ADC why do you not simply extend the memory interface to the bandwidth of the ADC? I could not believe that the 12bit are true at 6.25GS/s and wasted!

The noise values for the 2GHz model at 1mV/DIV and 2mV/DIV are only stated for 175MHz and 350MHz bandwidth. Does this mean that I could not measure 1mV/DIV and 2mV/DIV at the full bandwidth?

Is the FlexChannel based on an ASIC which combine 12/8bit-ADC and 8bit-Digital signals in one digital trigger engine? That could be some kind of cost optimization for chip design. But what is the benefit for the scope designer to have one channel ASIC?

There is a difference in acqboards. The 1GHz scopes have a single front-end custom ASIC up to 1GHz which allows full bandwidth at high gain settings. On the 2GHz models there is an added front-end ASIC that brings the bandwidth up. Because the 2GHz have an extra front-end ASIC the system noise is a little higher. Regarding the BW at high gain settings the 1GHz and lower BW models have full bandwidth at 1mV/div.

The FlexChannel inputs are based on a completely new designed Tektronix ASIC, there is a 12-bit ADC behind each channel and no 12/8-bit combination as you mentioned. This allows us to use both analog and digital inputs per channel through the same chip. Allowing tighter integration (uses less board space), better performance, higher accuracy and lower system noise. Each FlexChannel input regardless of analog or digital shares has the same specs of up to 125Mpts at 6.25GS/s which is very impressive, all other scopes on the market have different record memory/sample rates on the MSO channels and will have worse accuracy when correlating analog, digital and trigger circuit correlation.  To answer your other question, the ADC ASIC serves more than 1 channel and was designed with the fastest memory controllers at the time of design.

Anyways, much is new, much is still coming. Good questions everyone. 

[w2aew]

The new measurement system is remarkable and I THINK it can edge count, but perhaps W2AEW can check it for me since he has a scope available and I don't.

Yes, a couple of ways.  Since the measurement system makes measurements of the entire acquired record, it counts the number of occurances of each measurement (to compute stats), so you've got the count of edges or whatever measurable feature you want.

As a test of your honesty (and we all know you, and that you're honest)... one of those annoying interview questions:

What bugs/annoyances have you found in using the unit that you've reported, and have been fixed?

What bugs/annoyances have you found in using the unit that haven't yet been fixed?

I've only found a few minor annoyances and bugs, but this is based on pre-released firmware.   There's one touchscreen target area (when grabbing the edge of a window to resize it) that is a little small and difficult to grab sometimes.  Another is that the collapsible measurement badge area will sometimes collapse itself after I delete one of measurement badges.  No big deal, easy to bring back with a quick swipe.  There are a few improvements I'd like to see - for example, I'd like to be able swipe a channel or measurement badge off the edge of the screen as an alternative way to delete it.  Other than these minor nits, the new interface is a joy to use.

[rf-design]

The 5 Series MSO does have a true 12-bit ADC on each channel. The fact that 6.25GS/s on 8 channels is a lot of data to push through the system and we don't store memory on the chip does require us to push data in 8-bits or 1 byte transfers at max sample rate. We auto switch when you go into the new High Res mode or manually set the sample rate at 3.125GS/s or lower.

Thank you for your technical response. It will help us to understand the merits of the ASIC design for the scope.

If you could operate the 12bit ADC with 450fs jitter it will achieve an ENOB of 7.2bit at 2GHz. That is a good number for the jitter of the ADC and the clock source together. The spec values are 7.0bit@1GHz for the 2GHz model and 7.6dB@1GHz for the <2GHz models. That seems very conflicting because the 1GHz-BW noise at 5mV/DIV is 202uVe for the 2GHz model and 262uVe for the <2GHz models. Natural I would assume the lower noise of the 2GHz AFE indicate a better ENOB. But it is reverse. Why?

If you are able to design and produce a 12bit 6.25GS/s ADC why do you not simply extend the memory interface to the bandwidth of the ADC? I could not believe that the 12bit are true at 6.25GS/s and wasted!

The noise values for the 2GHz model at 1mV/DIV and 2mV/DIV are only stated for 175MHz and 350MHz bandwidth. Does this mean that I could not measure 1mV/DIV and 2mV/DIV at the full bandwidth?

Is the FlexChannel based on an ASIC which combine 12/8bit-ADC and 8bit-Digital signals in one digital trigger engine? That could be some kind of cost optimization for chip design. But what is the benefit for the scope designer to have one channel ASIC?

There is a difference in acqboards. The 1GHz scopes have a single front-end custom ASIC up to 1GHz which allows full bandwidth at high gain settings. On the 2GHz models there is an added front-end ASIC that brings the bandwidth up. Because the 2GHz have an extra front-end ASIC the system noise is a little higher. Regarding the BW at high gain settings the 1GHz and lower BW models have full bandwidth at 1mV/div.

The FlexChannel inputs are based on a completely new designed Tektronix ASIC, there is a 12-bit ADC behind each channel and no 12/8-bit combination as you mentioned. This allows us to use both analog and digital inputs per channel through the same chip. Allowing tighter integration (uses less board space), better performance, higher accuracy and lower system noise. Each FlexChannel input regardless of analog or digital shares has the same specs of up to 125Mpts at 6.25GS/s which is very impressive, all other scopes on the market have different memory/sample rates on the MSO channels and will have worse accuracy when correlating analog, digital and trigger circuit correlation.  To answer your other question, the ADC ASIC serves more than 1 channel and was designed with the fastest memory controllers at the time of design.

Anyways, much is new, much is still coming. Good questions everyone. 

If you design a new 1GHz AFE ASIC it will possible incorporate Tek famous overdrive recovery patent:

https://www.google.com/patents/US5075636

M377 IC used in some 11000 plug-ins


I am a two decade user and owner of the 11000 system with some 11A33 plug-ins. Up to now there is no other AFE which allow an offset up 16000 DIV and fast oberdrive recovery. I found that special on-state measurement on SMPS are only accuracte with this old scope system. Is there a chance that the new AFE allow similar recovery specs. This will help to send the outdated scope into the bin instead of reserving them for the special measurements.

Thanks in advance.
Reiner