Show us your square wave

[HighVoltage]

I just came across this old Tektronix video from 1961, explaining really nicely about the square wave.
This might be a real good video for the young players...
The Square Wave 1961

[Tomorokoshi]

That's not a square wave...

That's a square wave.

[gf]

A 10MHz square wave from the built-in DDS of my Hantek 6074BD USB oscilloscope, displayed on the same oscilloscope. Directly connected with a coax cable, with external 50Ohm terminator at the scope. The DDS is specified DC~25MHz, using a DAC clock of 200 MHz. The rise time approximately reflects the 25MHz bandwidth limit of the function generator. All in all not spectacular.

The 2nd image shows the same same signal, but at a low amplitude of 10mV. The apparent ripple can hardly be overlooked.  A FFT from this signal shows a decent peak at 200MHz, so I'm pretty confident that we see the DDS clock, leaking through to the signal output In reality, the ripple amplitude is likely even higher, since the scope's analog bandwidth is only 70MHz.

[tggzzz]

OK, here's the edges from a very simple pulse generator:

• the source is 5V with no load, but obviously it is used to drive 2.5V into the scopes' 50ohm input
• 4GHz LeCroy HDO4904, connected directly to the scope input with a BNCfemale-BNCfemale adaptor
• 1GHz Agilent MSOS104A, connected via a 1m piece of coax of untested quality

It looks like the 10%-90% risetime is 256ps with a 6.3% overshoot, and the falltime is 453ps with a 3.8% overshoot.

Considering the simplicity of the circuit, that is remarkably fast. It is a simple demonstration that modern jellybean logic (74LVC1G*) generates significant power into the microwave waveband - and hence RF practices are appropriate.

In this circuit a major contributor to the performance is the decoupling capactors, especially the 0V/5V planes and short wide wires, and not forgetting that MLCCs have a very significantly reduced capacitance when there's a DC bias voltage.

My apologies for the quality of the photos; they had to be taken relatively quickly and in non-ideal conditions.

[MaxFrister]

Measurements Corporation Model 71 Square Wave Generator displayed on a Hewlett Packard model 122AR.

[T3sl4co1l]

Weird, the sweep linearity on that scope seems to be wrong?

Tim

[Microcheap]

Only fine equipment in this thread so I've decided to bring some "toys" to compare. The first is an unlocked Hantek HDG2002B, the output is a 15MHz 2Vpp signal with a 50ohm pass through terminator on scope and then a 1MHz signal directly connected on the scope using a BNC cable.
The second function generator is a MHS-5200A, both signal showed are connected to the scope with a BNC cable only.

[dzseki]

Source for all tests below: EMG 12563 (an old hungarian made) ECL pulse generator the rise time was measured to be 425ps (including the scope's self rise time) with a LeCroy WavePro 725 2.5GHz 40GS/s scope. The interconnect cable was (the same for all testing) a vintage Amphenol RG223U coax 1m long.

Yokogawa DL1740 4CH 1Gs/s 500MHz:

HP 54503A 4CH 20Ms/s 500MHz, the measured rise time varies somewhat between the channels, but for the good...

[0culus]

Here's a frequency domain 10 MHz square wave from my 3325B, showing the harmonics tapering off in power out to around 330-340 MHz.

[dzseki]

Looking at the manual, are they saying <1.5ns for the edge rate? 
http://www.emg.hu/gepkonyvek/EMG_12563.pdf

That is for the trigger output, for the main output they say <1ns.
Here is a fresh shot, although my HP 1720A does not feel very well today...
The first rise is the trigger output of the EMG 12563, the second rise is the main output, the time base was set to 2ns/div. Unfortunately even the main output is measured beyond 1.6ns this time.
Anyway the trigger rise time is not that far behind, and considering that it also has a selectable amplitude between 1V (in reality 1.5V terminated) or 100mV and a fixed 50% duty cycle it is very much usable even alone.

[MarkL]

Here's something a little different.

Here's a picture of the eye scan tool on an Agilent 16702B logic analyzer with a 16756A analysis card and E5382 probe.  It's looking at the output of an Analog Devices SiGe ADCMP580 comparator eval board.

The ADCMP580 has a rise/fall of 37ps (typ.) for 20% to 80%.  Applying a simple extrapolation puts it around 50ps for 10% to 90%.

The eye scan has a slope tool which I've placed at 10% and 90% in the center of the transition area.  The time difference between the two markers is 130ps.

So, that would give the analyzer an estimated rise time of sqrt(130ps**2 - 50ps**2) = 120ps, and a BW somewhere around 2.9GHz.  The analyzer datasheet says 2.33GHz, so I guess it's not too far off.

It works but it makes a really lousy sampling oscilloscope.  It took several minutes to gather all these data points (180M samples, if I'm reading the stats right).  The clock rate was 100MHz.

[David Hess]

That reminds me of the testing I did adding a pretrigger to my PG506.  The fast rise output shown below was taken with a Tektronix 7T11A in sequential sampling mode and a 14GHz S-4 sampling head through 1 or 2 nanoseconds, I forget, of RG400 cable on a analog storage 7834 mainframe.  The photograph was processed to produce inverse gray scale to make it suitable for printing.

The edge itself is almost perfect with a transition time of about 550 picoseconds.  This particular S-4 sampling head suffers from excessive blow-by this is not visible at this time scale.  The tilt may be due to dribble up in the RG400 cable; I did not notice it at the time or I would have verified if that was the case.

The massive amount of pattern dependent jitter is caused by supply voltage variation caused by the TTL counter chain inside the PG506 getting into the TTL based 75 nanosecond pretrigger delay circuit.  This was unnoticeable on a typical oscilloscope of up to 300MHz bandwidth; a 500MHz DSO might just see it.  This circuit would need to be corrected to be usable for its intended application but it serves as a lesson as to why single ended logic including CMOS is not suitable for low jitter applications.

The second photograph was taken much later and is the same output being used to test my 100MHz 2232.  The displayed aberrations are typical and produced completely within the oscilloscope and within the specifications although I think the performance could be improved slightly.

[Keysight DanielBogdanoff]

So, not infinite bandwidth but...

how about 3.5 ps edge @ 113 GHz realtime bandwidth? 

[TheSteve]

So, not infinite bandwidth but...

how about 3.5 ps edge @ 113 GHz realtime bandwidth? 

Show off! 

[_Wim_]

So, not infinite bandwidth but...

how about 3.5 ps edge @ 113 GHz realtime bandwidth? 

Is that a "hacked" DSOX1102, or do we need to buy something better for that?   

[HighVoltage]

So, not infinite bandwidth but...

how about 3.5 ps edge @ 113 GHz realtime bandwidth? 

WOW!
That would be a scope for Wave 2019!

What pulse generator are you using for this?

[JPortici]

Can't see the edge. I dare you to expand the timebase to 10ps/div

[Keysight DanielBogdanoff]

WOW!
That would be a scope for Wave 2019!

What pulse generator are you using for this?

It's our calibration pulse, so essentially the fastest edge we can muster. It's a laser - it turns out optical is much easier at these frequencies.

[bson]

It's our calibration pulse, so essentially the fastest edge we can muster. It's a laser - it turns out optical is much easier at these frequencies.

That's pretty incredible! 

Are there complete optical front ends for the UXR series?

[tggzzz]

WOW!
That would be a scope for Wave 2019!

What pulse generator are you using for this?

It's our calibration pulse, so essentially the fastest edge we can muster. It's a laser - it turns out optical is much easier at these frequencies.

Well, 110GHz is within a factor of 3 of being out of RF bands and into the IR bands!

[David Hess]

NIST used what was effectively an optically driven sampling oscilloscope as their reference for calibrating pulse generators and I assume they still have it.  I always assumed HP/Agilent/Keysite had something similar to use in-house.  I think Tektronix sent their pulse reference to NIST for calibration.

At lower frequencies, low being up to 10s of GHz, it is possible to use 3 sampling heads to calibrate each other without any outside reference by measuring their kick-out pulses which are otherwise just annoying during normal use.

[Electro Fan]

So, not infinite bandwidth but...

how about 3.5 ps edge @ 113 GHz realtime bandwidth? 

Wow - very cool and Thanks/Congrats to Daniel and Keysight for sharing!

I think that beats the previous best posted around here by over 10 ps.  We are now within less than 4 ps of seeing our first sub-picosecond rise time.  Maybe Keysight or someone has something laying around that can do that.....

Actually, we might only be about 2.5 ps away as under 1 ps would maybe qualify as sub PS?

[Electro Fan]

Probably should go for the extra ~3.5 ps and get firmly into the realm of 3 digit femto seconds.

[Wolfgang]

So, not infinite bandwidth but...

how about 3.5 ps edge @ 113 GHz realtime bandwidth? 

Ahem. Can we see this with a usable horizontal scale ?

[TheSteve]

I can't beat Daniel but am still pleased.