Show us your square wave

[TunerSandwich]

Square wave is series of sinewaves

Indeed.......with a fundamental, and upper and lower order harmonics....here are two plots with a higher interpolated sample rate.....one showing the dynamic fundamentals of the center frequencies and the other showing the total roll-off of those fundamentals....

I am a bit puzzled by the topic of "show us your square wave".....too many variables between the oscillator/gen and the scope reading.....I don't think it says a hell of a lot about the quality of a sig gen, without analyzing the entire signal path

I did the plots posted using coax (properly terminated), but the results will be completely different with a more stable probing solution....and different i/o & measurement settings......to have any kind of real meaning in reality, there has to be a standard/definition of measurement parameters....

maybe I am missing the point of the question?

[EV]

The 4162 is pretty respectable - congrats and enjoy!

Thanks! Here is picture of 40MHz sync out square wave from Tek R7103 scope with 7A29 plugin (BW = 1GHz).

[Wuerstchenhund]

I am a bit puzzled by the topic of "show us your square wave".....too many variables between the oscillator/gen and the scope reading.....I don't think it says a hell of a lot about the quality of a sig gen, without analyzing the entire signal path

I agree re. the variability of external parameters like cabling and termination, and that without a proper test setup the screenshots are worthless.

Cables and termination play a huge role which is too often underestimated. Just for fun I repeated some of my measurements with a cheap China crap RG-59U BNC cable I found in a box I was given, and thanks to the mismatch and the overall poor quality of the cable the results were pretty poor. I then put the RG-59U where it belongs (trash can) and re-tested with a Pasternack PE333 SMA cable (rated to 18GHz), and got the same results as with my original measurements (which was done with proper cabling and termination).

It just shows that the best scope (or any other mesurement device) is useless without proper consideration for the measurement path.

maybe I am missing the point of the question?

See it as a bit of fun, i.e. who can produce the worst square wave ever 

[rx8pilot]

I am a bit puzzled by the topic of "show us your square wave".....too many variables between the oscillator/gen and the scope reading.....I don't think it says a hell of a lot about the quality of a sig gen, without analyzing the entire signal path

It does show the variables which is specifically what I am interested in learning. Looking at the anomalies and trying to best understand the origin. I agree the screen shots isolated are of limited value, but if the whole setup is described it can be interesting information about what/how signals can be altered by taking a measurement.

[c4757p]

Here is SDS1072CML frequency response.

There's not even a section of that that even remotely resembles "flat".

I'll have to remember that - Siglent wouldn't know a proper analog frontend if it bit them in the ass, apparently.

[TunerSandwich]

See it as a bit of fun, i.e. who can produce the worst square wave ever 

I think I might have some 0.1ohm per foot flat nichrome laying around here....and some BNC to binding post adapters.....maybe I could fashion a split end, moist, wooden stick as the dielectric..... O0

[alterbaron]

There's not even a section of that that even remotely resembles "flat".

I think it looks bad because he's doing a linear frequency sweep.

Here's two plots from LTSpice showing the same 72MHz first-order RC low-pass filter.
The first image looks like your usual flat response up to the corner frequency, followed by the usual 20dB/dec roll-off.
The second image portrays the same data using a linear frequency scale, and it definitely looks less "flat".

[TunerSandwich]

I am a bit puzzled by the topic of "show us your square wave".....too many variables between the oscillator/gen and the scope reading.....I don't think it says a hell of a lot about the quality of a sig gen, without analyzing the entire signal path

It does show the variables which is specifically what I am interested in learning. Looking at the anomalies and trying to best understand the origin. I agree the screen shots isolated are of limited value, but if the whole setup is described it can be interesting information about what/how signals can be altered by taking a measurement.

The largest uncertainty will be introduced by the reflections in the cable......unless the gen and scope are utter crap, or we are looking at 100Hz square waves, with massive vertical scales...

Anything useful (by todays standards) is going to suffer the greatest degradation in the cable.....you can see in my plots where the cable capacitance was causing the charge/discharge effect....and subsequent phase shift.

To really resolve such signals, rise time is everything....so we need lots of bandwidth, and active, differential measurements....

My plots were of a 15MHz function @ 10mV....zero offset.  Scope input BW is 600MHz, with a samplerate of 200GS interpolated sinx/x....

As you can see the cable decimated the measurement.....

At a certain point quantum tunneling through silicon will start to become the issue.....it just depends on how deep down the rabbit hole you want to go.....generally we use square waves as a clocking source....so as long as the jitter in the clock chain can still resolve a binary step response it's "all good"......you would be surprised how badly you can mangle a square wave and still not need buffering to resolve a bit

[TSL]

@TunerSandwich that's a mighty fine machine you have there, not often you see an upper limit of 1THz !!

What scope is that ?

regards

Tim

[TunerSandwich]

@TunerSandwich that's a mighty fine machine you have there, not often you see an upper limit of 1THz !!

What scope is that ?

regards

Tim

Hi Tim,

It's a LeCroy WR 64MXi.  Keep in mind that the 200GS is RiS (random interleaved sampling) and the measurement is sin x/x....so there is a great deal of interpolation happening there.  You can see the input noise roll off @ 100GHz (and disappear by 200GHz).  However you can see another ELF fundamental beginning to form.....if we took the timebase out further and were able to maintain that 200GS, we would see the complete ELF fundamental.  Thankfully I know there isn't enough energy present to worry about interference from that tone....but it's amazing that even with the ERES turned off, we can resolve that dynamic range through the noise....I'm quite happy with the scope.

[Wuerstchenhund]

I think I might have some 0.1ohm per foot flat nichrome laying around here....and some BNC to binding post adapters.....maybe I could fashion a split end, moist, wooden stick as the dielectric..... O0

Considering some the things we've seen here I'd say give it a go! 

[rx8pilot]

Silliness aside....

If anyone is willing and technically capable. It would be interesting and informative to show the range of errors between a good scope and SG. Various BNC cables, probes with long and short grounding, good/bad terminations, impedance mismatches, etc.

[TunerSandwich]

Silliness aside....

If anyone is willing and technically capable. It would be interesting and informative to show the range of errors between a good scope and SG. Various BNC cables, probes with long and short grounding, good/bad terminations, impedance mismatches, etc.

The issue we will run into on the SG/FG will be range.....for these modern scopes an old dinosaur just isn't going to give us 10Hz to 1GHz....and a modern Gen that will do that, and still keep quantization noise and step response phase would cost more than most of the scopes we can afford. 

Again seeing perfect square waves isn't that important, in the bigger scheme of things.  They are pulses, and really only useful for clocking.  We can mangle them pretty bad and still resolve a low jitter clock source.....and if that fails we can buffer and oversample to correct potential dynamic range errors. 

I just can't imagine why an utterly pure square wave would be that important.  However, where it might become critical, is in low voltage control signal/clock....but again that can be more easily overcome (than measuring the perfect pulse) by simply oversampling and buffering. 

I will throw a couple more measurements up, that are somewhat practical.  If I have to break out any differential amplifiers and calibrate DC offsets, then it will have exceeded ay kind of real world clock signaling.....a bad PCB layout will utterly destroy rise time, and even a good layout is going to have planar impedance/frequency issues. 

Most real world devices have clock signals, that look even worse than the scope plots in this thread....and still manage to work damn well.....even with shit 5mHZ $0.10 crystals we can get very stable 100MHz clocks......this seems to be one place where software and on chip counters has truly trumped the need for pristine analog square waves. 

What would interest me is seeing just how low of a signal level we could get to, and still derive a clock.....I think even at 1mV pk-pk, with shit interconnects and ludicrous bandwidth we could still pull a clock from (practically) a sine wave.....

Did you see just how bad the ringing was in the plots I posted earlier?  I counted no less than 35 fundamental tones in 1Hz to 100GHz, and that is not even factoring in upper and lower order harmonics.  An interesting experiment would be to feed that garbage back out into a clock ref input of a microprocessor or a/d and see how much jitter it really equals.....and then how much we can clean it up with proper buffering/oversampling.....

[tautech]

Silliness aside....

If anyone is willing and technically capable. It would be interesting and informative to show the range of errors between a good scope and SG. Various BNC cables, probes with long and short grounding, good/bad terminations, impedance mismatches, etc.

I tried to display some in post:
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg588472/#msg588472

Sure it is with only 10 MHz, but there is 4 differing connections:

RG-58C/U 50 Ohm source & DSO internal termination
RG-58C/U 50 Ohm source & Tek 50 Ohm feed-through
Hi Z source 10x probe
Hi Z source 1x probe
Both probes into DSO 1M of course. 

Any further requests? 

[Jay_Diddy_B]

Hi group,

We can look to simulation to get an idea of the waveforms we should expect from various combinations of pulse generators and scopes.

I have created a LTspice model that has 200ps and 5ns risetime pulse generators. I have added some 50 Ohm (RG58 type transmission lines) and 75 Ohm transmission lines (RG59 type) The 10ns length represents about 2m of cable.

I have also generated 20 and 50 MHz Bessel and Butterworth filters. The Bessel filters would be the best choice for square waves, the Butterworth filters would give better sinewave performance.

I have also added a (near) Gaussian filter with 100MHz to simulate a traditional oscilloscope response. A modern oscilloscope might not have a Gaussian response.

Here is the model:



And the results:












I have attached the LTspice model in a zipfile.

Regards,

Jay_Diddy_B

*** This is my 666 th. message ***

[tautech]

*** This is my 666 th. message ***

[T3sl4co1l]

*** This is my 666 th. message ***

And that one (or at least you're showing it at this instant) was your 1669th... say what you will?  

[Rupunzell]

There is a divide between the importance of pulse fidelity between analog folks and digital folks.

For digital folks, accurate, reliable, consistent transmission and reception of ones or zeros is what matters most. As clock rates and data rates have gone well into the Ghz range common practice and techniques from the RF-microwave world have been incorporate into much of modern digital design. Varying degrees of pulse distortion is often acceptable as information is transmitted as basic on or off over time.

Pulse, transient fidelity can be quite different for all that analog stuff as impulse response can tell a LOT about circuit and system behavior when test like this are applied properly and interpreted properly. Doing this really, really well require a great deal of theoretical and real world experience.

 If one were to dig into the entire field of pulse fidelity, transmission lines, impedance matching and all related there were a number of Tektronix & Hewlett Packard application notes published back in the 1960 and 1970's that does a good job at explaining this along with  text books materials. Little if any of that physics has changed to this day. The foundations of this technology remains much the same.

What could be more relevant in this discussion of pulse response and all related has been covered in a Linear Technology app note# 47 written by Jim Williams in 1991.
http://cds.linear.com/docs/en/application-note/an47fa.pdf

The moment a probe or interconnecting cable has been attached to a DUT  to examine and gather information about it's behavior with a time or frequency domain instrument, the actual DUT behavior can be easily skewed and distorted if all involved is not properly understood and applied.

As for old dinosaur pulse generators not being able to go from 10 Hz to 1Ghz.. Not really true, there are quite a number of snap diode, tunnel diode pulse generators that can easily cover 10 Hz to over 1 Ghz with low noise from the 1960's. It is so easy to forget and discount what has been done and replace it with what is the latest and greatest, part of the not invented here syndone. All instrumentation is relative to need and expectations limited by humanity's current understanding of physics.



Bernice

[TunerSandwich]

As for old dinosaur pulse generators not being able to go from 10 Hz to 1Ghz.. Not really true, there are quite a number of snap diode, tunnel diode pulse generators that can easily cover 10 Hz to over 1 Ghz with low noise from the 1960's. It is so easy to forget and discount what has been done and replace it with what is the latest and greatest, part of the not invented here syndone. All instrumentation is relative to need and expectations limited by humanity's current understanding of physics.

Bernice

The problem there is assuming they are still in spec....and then having an order of magnitude higher resolution signal chain, to even answer that question. 

Believe me I am not saying there is NO merit to hi-fidelity analog pulse measurements.....but in a DSO world, and a world filled with oversampling and "cheap" clocks....the quest for fidelity becomes somewhat of an academic endeavor. 

It's all good sport, but I highly doubt a bean counter overlooking a project is going to justify the cost of going down that road, when there are so many "band-aids" post A/D to "clean up" those potential "problems". 

It's one of those things where we've just gotten too deep into the band-aid approach and all the signals are good for is proving or disproving some theoretical ideal.  I would think the more relevant test today is an artifact based PWM signal, which has coincidental parameters to the way in which a DSO acquires, and then interpolates/represents a signal. 

This is one of those few occasions where I would discount the overall fidelity of an analog pulse, and it's usefulness as a reference.  Maybe that's not a good thing, but one does have to admit it's at least a valid reality?   

P.S. I have never bought into the digital vs analog statement.....at the end of the day it's all still analog....the 0s and 1s have to ride on something right?    There just happens to be a lot more room for "error" in that application

[c4757p]

*** This is my 666 th. message ***

And that one (or at least you're showing it at this instant) was your 1669th... say what you will?  

2669?

[AlfBaz]

Rise time is measured at just over 5ns, which is pretty good. (Bandwidth can be approximated as 0.35 / (rise time), so we can't expect any faster than a 5ns rise on a 70MHz scope.)

I would have guessed that rule of thumb was flushed down the toilet long ago.

Is this true?

I recall looking at this with simulations, some time ago.

Given the unity gain bandwidth specifications of LT opamps and their available models in ltspice, I firstly checked their bandwidth with ac analysis, used the -3db point to determine the rise time using that approximation then fed different pulses above and below the calculated figure to observe differences in output responses to the input pulse and got good correlation

[T3sl4co1l]

*** This is my 666 th. message ***

And that one (or at least you're showing it at this instant) was your 1669th... say what you will?  

2669?

Damn, the only thing missing is you needing 6996 posts.  (You broke it!)  Remind me of this thread in a few weeks / months, would you?

Tim

[Jay_Diddy_B]

Hi group,

I want to extend my previous post with some measurements.

I apologize for the blurry pictures.

I want to show just how bad the cable can be:

Here is a reference waveform made with a 50 Ohm cable. The scope bandwidth is 1 GHz and the HP8112A pulse generator has 4.5ns rise and fall times.





I then replace the cable with a really bad cable made from 75 Ohm RG6 coax with binding post adapters for the connections:

Here is the waveform:



It is similar to the waveform from the LTspice model. The notch on the leading edge is twice the time it takes for the signal to travel from one end of the cable to the other. The velocity is 66% of the speed of light.



Here is a photograph of the (terrible) construction of the cable.



If I increase the rise and falltimes to 10ns (greater than or equal to the electrical length of the cable) most of the distortion caused by the impedance miss match are eliminated.



So generally the impedance matching is important if the risetime is less than the propagation time.

Regards,

Jay_Diddy_B


 

[BravoV]

So generally the impedance matching is important if the risetime is less than the propagation time.

Thanks, its written in my personal notebook now ! 

[rx8pilot]

I tried to display some in post:
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg588472/#msg588472
..............
Any further requests? 

Thank you for that.

I then replace the cable with a really bad cable made from 75 Ohm RG6 coax with binding post adapters for the connections:
Here is the waveform:

I was expecting worse from the binding post contraption.