Yet another fast edge pulse generator

[Leo Bodnar]

If you are here you probably like fast scopes too.

IMSAI Guy gets himself a 50GHz DCA-X 86100D scope:


and runs LBE-1322 pulser through its paces:


Leo

[analogcurious]

Here's a snapshot using the latest addition to my lab. My lab is def growning slowly 

[David Hess]

If your lab can afford that, then you are a better person than I am.

[KE5FX]

If your lab can afford that, then you are a better person than I am.

Huh?  I'm sure Jeffrey Epstein could've bought an 86100C or three.

[TERRA Operative]

I got mine today. 

My TDS794D 2GHz scope (as fast as I can go currently) is indeed 2GHz.

I also have the older BNC pulse gen (and the older 2-channel GPSDO which I love), but this new pulse gen being adjustable will be great for adjusting oscilloscope front ends. (I was waiting for this adjustability for a long time!)

[analogcurious]

Looks pretty neat noiceeeeeeeeeeee

[jonpaul]

Dear TerraOp: Very fine , but Why the 16% overshoot on leading edge on the TEK 2 GHz scope, perhaps a digital filter artifact? Try different sample Windows, pulse response options?

Our old Yokogawa 500 MHz DL7440 does better.

We see way less pulse aberration on our  1990s vintage TEK  1 GHz analog scope

using the great Leo Bodnar 40 pS BNC (old version) pulser>>Mini Ckt 20 db 50 Ohm attn HAT-20+>>TEK 1 GHz vert  7A29 >>TEK 7104 1 GHz mainframe 


The 20 db attenuator reduces possible reflections due to scope VSWR.

I confess that I  calibrated and tweaked the  TEK 7A29s VERT and TEK 7104 scope vert amps for optimized transient response. Many tiny caps and coils, iterative, takes experience but very satisfying!
See the excellent TEK circuits concepts book Oscilloscope Vertical Amplifier Circuits.

https://davmar.org/TE/TekConcepts/TE/TekConcepts/TekVertAmpCircuits.pdf

HAVE AN ABSOLUTELY FANTASTIC DAY!

Jon in Paris

PS: also See old (long) thread "show us your squarewave"
https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg5988861/#msg5988861

[TERRA Operative]

No idea, I didn't do any setup on the scope beyond 'make a trace on the screen'.
I'll have to drag the scope out again later once I have space on the bench to play with it more.

[David Hess]

Dear TerraOp: Very fine , but Why the 16% overshoot on leading edge on the TEK 2 GHz scope, perhaps a digital filter artifact? Try different sample Windows, pulse response options?

I would trust the sampling oscilloscope examples which do not show that, but it could be variation between different Leo pulsers.  A digital filter artifact would likely show preshoot, and I do not think the TDS794D has adjustments for that except during calibration.

[rhb]

Dear TerraOp: Very fine , but Why the 16% overshoot on leading edge on the TEK 2 GHz scope, perhaps a digital filter artifact? Try different sample Windows, pulse response options?

I would trust the sampling oscilloscope examples which do not show that, but it could be variation between different Leo pulsers.  A digital filter artifact would likely show preshoot, and I do not think the TDS794D has adjustments for that except during calibration.

I don't understand.  i thought that:

https://www.eevblog.com/forum/projects/yet-another-fast-edge-pulse-generator/msg5808045/#msg5808045

shows that the actual rise time, using an SD-32 head is 19 ps for the LBE-1322 vs 15 for the Tek pulsers.  I know of nothing affordable that comes close to that accuracy and resolution.

At these speeds, connector life is a serious issue as is proper torque and honest BW specifications.  The Keysight data sheet for the MSO product line gave the rise time as  0.35/BW *except* that at 1 GHz it's  a larger numerator.

As the owner of 6 of Leo's pulsers.  4 BNC square wave units, one Gaussian pulse unit and an 1322 , I don't think there is much likelihood that it is anything but the scope.   Leo is 100% test before ship. All my readings show as good as or better than Leo's probably because of setting  difference.

The 0.35/BW risetime of my SD-32 is 8.25 fs which makes my 15 vs 19 ps measurement errors insignificant.

So, what you see from one of Leo's pulsers is your scope.  I just wish he'd offer a Gaussian impulse version of the 1322.

A variable rate minimum phase Gaussian impulse would be a hugely useful diagnostic tool by creating very broad BW  system impulse responses.

Have Fun!
Reg

[jonpaul]

RHB: agree , Leo has excellent testing and consistency.

The poor overshoot on the TerraOP is an artifact of his scope or test fixtures/term/setup.

My TEK 7104 /7A29 has excellemnt transient resp. on any of several LB pulsers and TEK PG506 CAL fizt fast rise

Jon

[David Hess]

Dear TerraOp: Very fine , but Why the 16% overshoot on leading edge on the TEK 2 GHz scope, perhaps a digital filter artifact? Try different sample Windows, pulse response options?

I would trust the sampling oscilloscope examples which do not show that, but it could be variation between different Leo pulsers.  A digital filter artifact would likely show preshoot, and I do not think the TDS794D has adjustments for that except during calibration.

I don't understand.

I mean there is likely something out of calibration with the TDS794D, because I trust the sampling oscilloscope examples from the Leo pulse generators as far as flatness.

At these speeds, connector life is a serious issue as is proper torque and honest BW specifications.  The Keysight data sheet for the MSO product line gave the rise time as  0.35/BW *except* that at 1 GHz it's  a larger numerator.

As bandwidth increases above 500 MHz, it is less likely that the oscilloscope has a Gaussian or first order response where the 0.35/BW rule applies.  I think the 7104 manages it, but some of the later Tektronix mainframes starting with the 11K do not with certain plug-ins.  This by itself might explain the TDS794D's response.

The 0.35/BW risetime of my SD-32 is 8.25 fs which makes my 15 vs 19 ps measurement errors insignificant.

Sampling inputs have a non-linear frequency response so the 0.35/BW rule does not apply to them, and the rule that does apply to them is somewhat equivocal.

My TEK 7104 /7A29 has excellemnt transient resp. on any of several LB pulsers and TEK PG506 CAL fizt fast rise

I really like the output from my PG506 because it is so flat with no aberration.  I keep dreaming of designing a faster version, like the 1GHz 7000 standardizer,  but PNP RF transistors are no longer available.

[jonpaul]

David: agree 100%

Our work since 1980s for high res digital audio involved digital signal pulse jitter and aberration effects on the reconstructed analog audio audio.

(see my AES and SMPTE papers and conference presentation 1994...2017)

Low abberation fast rise transient resp  pulse gens and scopes were basic tools.

Indeed the TEK 7104 + 7A29 are  excellent on transient resp, IF you have trimmed the VERT amps and preamps for it.

On my two  PG506/A, fast rise is very good. You  MUST use a precision  TEK 50 Ohm 50 cm cable   for true  trans resp.

The standardizers are hard to maintain, the delicate elastomeric switches and cams  get intermittent, special cleaning procedure.
Was  very lucky to acquire and get repaired and calibrated, several of the   standardizers for TEK 7000, 67-0587-02

https://w140.com/tekwiki/wiki/067-0587-02

Have an absolutely fantastic weekend!


Jon

[David Hess]

On my two  PG506/A, fast rise is very good. You  MUST use a precision  TEK 50 Ohm 50 cm cable   for true  trans resp.

I have been using 2 nanosecond RG-400 cables for it which seem to work out well.  I grade my homemade cables by delay, usually 1, 2, and 5 nanoseconds.

The only photograph that I have handy is from when I was messing with a pretrigger for my PG506, shown below where I used my sampling plug-ins and an S-4 sampling head, so about 14 GHz.  Note what looks like what might be a 600 picosecond "dribble-up" feature, possibly from my 2 nanosecond RG-400 cable?

The standardizers are hard to maintain, the delicate elastomeric switches and cams  get intermittent, special cleaning procedure.
Was  very lucky to acquire and get repaired and calibrated, several of the   standardizers for TEK 7000, 67-0587-02

I only have the -00 250 MHz standardizer.

[jonpaul]

Dear David: RE cables:

With connexions friends at Belden cable, we compared   18 types for a SMPTE paper.
 best 75 Ohm was Belden 1394, 12 GHz, but needs special connectors and crimp tool/die.

Epay sometimes has TEK cable for  SG503, SG504,  Rich C of AP got us a bunch long ago....
012-0482-00 Tektronix Precision Coax   50 Ohm 
https://w140.com/tekwiki/wiki/012-0482-00

We avoid them altogether and place the gen /LeoB/TEK PG506 near the scope and use combos of BNC elbows and adapters.

Of course the best mainframe 7000 test is the Cal plugns.

Jon

[David Hess]

Epay sometimes has TEK cable for  SG503, SG504,  Rich C of AP got us a bunch long ago....
012-0482-00 Tektronix Precision Coax   50 Ohm 
https://w140.com/tekwiki/wiki/012-0482-00

We avoid them altogether and place the gen /LeoB/TEK PG506 near the scope and use combos of BNC elbows and adapters.

Of course the best mainframe 7000 test is the Cal plugns.

I checked the performance of my SG503 using my 2 nanosecond RG-400 cables with my S-4 sampling heads and flatness was well within the specifications, like 4 times better.

I love sampling oscilloscopes; they are so useful for calibrating other oscilloscopes.

[Bernd_2]

Hello,

here are a few pictures of the Leo Bodnar fast pulser LBE-1322 on MXO5 at 2 GHZ Bandwith and 50 Ohms Termination. The LBE-1322 is connected using the SMA-BNC adapter supplied.

The images show the acquisition modes Sample, Envelope, and Average with 8 counts in succession.

The measured rise time is approximately 203 ps. Using the formula bandwidth = 0.45/rise time, this results in a bandwidth of approximately 2.2 GHz.

Best Regards,
Bernd

[zike]

I ran across a similar 30 ps pulse generator offered by a Denver company called "Voltative" https://www.voltative.com/.
I had not heard of them before; if anyone knows of (or owns) the company, please feel free to share your story. I bought their PG-2 model to play with. Looks and behaves very much like the Bodnar LBE-1320A. 

Quick test with my antique Tek S-6/7S11/7T11/7904 sampling rig displays 40ps rise and fall. After RSS regressing the known 30ps of the S-6, this indicates about 27ps net for the pulse edge. Indeed this is about what's printed on the factory test certificate (evidently calibrated against the maker's Tek 11k rig). Note however that the unit sends 1Vpp into 50 ohms, so I had to insert an SMA attenuator to keep my sampling head happy. This may have degraded measured rise time. My MCL VAT-20 pad is nominally only rated for 6 GHz (60 ps). However these attenuators are typically better than this, as evidently seen here.

One quirk is the PG-2 displays a fair bit of overshoot, about 13% in the first 50ps. This is within spec for the unit and my measurement matches the test report; whether it matters depends on your application. It shouldn't make any difference if you're characterizing a GHz-class or slower scope, for example. FWIW both my Bodnar units overshoot around 5% (as seen on a 12 GHz-equivalent sampling rig), with my S-52's and homebuilt tunnel diode pulsers in that ballpark.   

Here are a couple snapshots of the rising edge. Please excuse the antique e-beam painting.

[Tantratron]

Few years ago before Brexit, I've purchased Leo Bodnar fast rise pulser which is really great product.

Lately I'm comparing four TDS700/C/D oscilloscope in my lab which were reconditioned and upgraded (TDS784C TDS754D TDS784D and TDS794D).

Attached are pictures showing same Leo Bodnar pulser test done on all four TDS with same extended sampling 100 GSa/s (see pictures name for reference of who is who).

My first question is about my TDS754D test about what is correct bandwith formulae because these are analog then DSP enhanced sampling. Should we keep the classic formulae BW x Rise = 0.35 with the specific HW/SW of these TDS500/700 oscilloscope ?

The value is roughly 450 ps so do you consider it is close from 1 GHz bandwith or good 500 MHz ?

My second question is about TDS784C and TDS784D where I'm bit surprised of some top overshoot ringing at the end of the rise time. Do you know if all 1 GHz from this serie have the same form factor versus the TDS754D (500 MHz) being softer with less overshoot, same question about my TDS794 (2 GHz)  ?

[tautech]

Do the numbers on this:

Using the formula bandwidth = 0.45/rise time

Which many think is more appropriate these days.

[2N3055]

Tektronix specified this.
In Performance verification documents  for TDS 700C/D it is stated that:

Calculated risetime in ns is 400/BW (in MHz)

[TERRA Operative]

...same question about my TDS794 (2 GHz)  ?

Further back up the thread I posted pics from my TDS794D which look very similar, which means both of our instruments are either screwed up in the same way, or they are both working as intended (or both options may be true... ).

[2N3055]

...same question about my TDS794 (2 GHz)  ?

Further back up the thread I posted pics from my TDS794D which look very similar, which means both of our instruments are either screwed up in the same way, or they are both working as intended (or both options may be true... ).

That pulse response is normal for that model. Overshoot is result of brickwall AA filters used on inputs of higher BW scopes. Please see attached Keysight technical note.

[srihari]

iam learning and want to design the hv pulser with subnano rise time,but when i started simualtion in ltspice two stage with fmmt417 the output is not understanding

[Tantratron]

That pulse response is normal for that model. Overshoot is result of brickwall AA filters used on inputs of higher BW scopes. Please see attached Keysight technical note.

Regarding the AA filter (Anti-Aliasing), not sure if these are the same found say on TDS754C/TDS754D versus the TDS784C/TDS84D. On the forum, there were discussions and instructions about 4 small SMD caps on the acquisition board near the hybrids. It is said that desoldering the 4 caps does extend the global bandwidth from 500 MHz to 1 GHz but might require re-calibration.

I've never done this removal but I remember once purchasing on eBay one TDS784C from Belgium (see this thread). When I open to clean and recondition, I notice the caps were there so it was maybe a lie from the vendor because many cheat that way. But applying the Leo Bodnar pulser did show much higher bandwith, still wondering if tektronix label TDS784D and left the 4 caps.

What is bit strange with my untouched TDS754D, see the picture few post before in this thread, there is overshoot which is why I wonder if other members with their TDS784C/D do have overshoot or not.