WTF Keysight labeling 750 MHz scopes as 1 GHz !!!

[rhb]

Bill and Dave must be spinning in their graves.

My MSOX3104T arrived today.  I put one of Leo Bodnar's 40 pS pulsers on it and measured a rise time of 436 pS.  This just had a fresh cal before it shipped from the factory in Malaysia.  So I pulled up the datasheet.  For the 3104 the rise time is given as <450 pS.   All the other models in the line are labeled as BW = 0.35/RT.  But they labeled the 750 MHz model as 1 GHz.  After an hour or so warm up, the rise time dropped to 430 pS, 814 MHz.

May those responsible be consigned to the lowest level of hell.

[nctnico]

I can't imagine the fault is at Keysight. Usually their scopes have a slightly higher bandwidth than specified.
What is your setup like? What kind of coax cable? Switched the input to 50 Ohms? Zoomed in far enough for the math to work OK?

[TheSteve]

Tek suggests that for modern DSO's a value of 0.45/rt is suggested.

https://www.tek.com/support/faqs/how-bandwidth-related-rise-time-oscilloscopes

[DaJMasta]

Do you have an RF gen capable of producing a 1GHz tone?  It would be easy to find the -3dB point of the input if you could sweep from under to over and make a note of where the rolloff starts.

[Rich@RohdeScopesUSA]

Tek suggests that for modern DSO's a value of 0.45/rt is suggested.

https://www.tek.com/support/faqs/how-bandwidth-related-rise-time-oscilloscopes

This.  I'm guessing the 3000X has a brickwall filter on the 1GHz model (so it can achieve 1GHz on 4ch @ 2.5GS/s), so 0.45/risetime is the correct formula.

-Rich

[egonotto]

Hello,

it seems that the bandwidth (-3dB) of your scope is well over 1GHz.
It is a warranted specification.
And they say it should be 1GHz with a rise rime  <=450ps.

As your rise time is <450ps why have you doubt over the bandwidth?

Don't worry, have fun with your new scope

Best regards
egonotto

PS:
displace the poor innocent out of the hell

[HighVoltage]

I am totally happy with my 1GHz MSOX3104A
It shows a shortest rise time of 400 ps
So 1 GHz BW  * 400ps = 0.40 seems to be right

[rhb]

I'm very happy with my MSOX3104T.  It's marvelous.  The only issue I have had is with the FFT function.   The limitation to 64K points limits the RBW.  But I've got an 8560A, so it's not all that important.  I wish it had more memory, but I'm guessing that large memory and rapid waveform update rates don't go together.  And 7% overshoot is a bit above the 3% I like to see.  I really wish they would let you set it to a Gaussian edge step response such as R&S use.

I bought this because it was a very good deal on a very high spec instrument (at least for anything I'm going to do).

What I am *not* happy with is marketing.  For the calculated rise time for the lower end models they use 0.35/BW but for the top end they use 0.45/BW.  Changing the rules part way through the spec is not something I find acceptable.  I'd greatly prefer they gave measured rise times.

According to my out of cal 8648C  set at -20 dBm output level it is about 2 dB down from 900 to 1000 MHz and  usable to about 1.4 GHz.  That's very good.  I'll know more once I check my 8648C against my freshly cal'd 438A, 8481D and 8482A.

I've not yet read the manual.  I deliberately just plugged it in and started fiddling.  My chief complaint about the Instek MSO2204EA is the UI.  Keysight did a really good job on this.

[rf-loop]

What I am *not* happy with is marketing.  For the calculated rise time for the lower end models they use 0.35/BW but for the top end they use 0.45/BW.  Changing the rules part way through the spec is not something I find acceptable.  I'd greatly prefer they gave measured rise times.

This is NOT marketing "trick". It is just basic fundamentals.

Have you ever think where from this old 0.35/BW is coming.

Diffrent scopes have different BW shape. 0.35 is for gaussian type BW shape. It is math.
These scopes what you now talk have "brick wall" type BW shape,  quite flat top until nominal freq 1G and after then fast decaying "brick wall" shape. THis also good for level measurement if compare with gaussian type what may start decay after quite low freq. (also brick wall filter is better for keep ADC input free from over Nyquist frequencies for minimize aliasing.)

It ia also very well explained in Hewlett-Packard/Agilent/Keysight Application notes and other papers decades of years.
http://literature.cdn.keysight.com/litweb/pdf/5989-5733EN.pdf

It can also quite easy find more deep fundamentals and math if need.
Different shapes are optimal for different things including user side and manufacturing side.

Different front end frequency response "filter" shapes is one reason why risetime measurement with enough fast risetime pulse source is not good for measure oscilloscope real bandwidth.  It really is like "poor hobby grade method" for measure scope BW independent of who try tell it is "nice and good method". It is not. Only what it is - it is easy and fast what may give some rough imagine. But it is of course good for characterize risetime (wih some notes).

Best method for characterize BW is sinewave RF generator and better if it have level head what keep real level constant there in oscilloscope input connector. (other method is calibration grade splitter connected to scope input port and signal generator and power meter for control true level in this point.

[MarkL]

No need for separate app notes.  It's all in the 3000T User Guide.  Rise time and BW are described starting with the section "Oscilloscope Bandwidth and Sample Rate" (pg 218):

  http://literature.cdn.keysight.com/litweb/pdf/75037-97049.pdf

A few pages later in "Realtime Sampling and Oscilloscope Bandwidth" (pg 228), it goes on to say:

When Realtime sampling is on, the bandwidth of the oscilloscope is limited because the bandwidth of the reconstruction filter is set to fs/4. For example, a MSOX410xA oscilloscope with channels 1 and 2 on has a bandwidth of 625MHz when Realtime sampling on and 1GHz when Realtime sampling is off.

Yes, that's a 4000A example in the 3000T manual, but maybe the 3000T behaves the same way if Realtime sampling is on?

[nctnico]

So it is not a 4 channel 1GHz scope then.

[KE5FX]

If you need high bandwidth and have a huge budget, you should buy the nicest top-tier DSO you can find.  Party on.  Call up Keysight or R&S or Tek or LeCroy and make the salesperson's day.  Flash your cash, knock yourself out.

For everyone else who doesn't have a massive budget to play with, here's an unsolicited suggestion: buy a new middle-of-the-road MSO or DSO model for use 90+% of the time when you don't need GHz+ bandwidth.  Select it on the basis of its feature set, memory size, update rate, vendor reputation, UI quality... anything but bandwidth.   

Then, pair it with a used GHz+ scope.  Something that was once a coveted high-end instrument, but is now considered obsolete because it doesn't have all the latest triggering modes, protocol decoders, or software support.  You can save a LOT of money that way while still getting what you want. 

Most people should not be buying these pseudo-"1 GHz" scopes at the top end of a mid-grade family.  $1K for a TDS 694C gets you an honest 3 GHz at 10 GS/s on all four channels.  Sure, it's 20 years old, and shows its age in a lot of ways.  Big deal.  If it breaks, go back to eBay and buy 3 more of them.  You will still come out WAY ahead in terms of both money and performance. 

Don't like the TDS models?  That's fine -- HP and LeCroy have similar models in their back catalog, just waiting for a new home.  Whatever floats your boat.  Just don't shop for bandwidth at factory-new prices.  Let someone else take the depreciation.

[chris_leyson]

My HP54610B says 500MHz BW on the front panel but the rise time is 470ps when checked with a Tek S52 pulse gen, 25ps rise time. If I assume a Gaussian passband response then that gives me 750MHz BW. Haven't got a sig gen available to double check but the measured BW exceeds the spec.