Probe into probes. What's up?

[mawyatt]

Tektronix took advantage of their CCD based sampling technology to make a small and low cost sampling front end for that series of DSOs. 

Tek also used their custom CCD semiconductor fab process to produce some of the better CCD imaging chips back then.

Best,

[tautech]

Found some time to play.....

All following square wave/step response checks are from one of Leo Bodnar's fine 10 MHz pulsers and in this one's case, a 30ps pulser.
First, the pulser alone into internal 50 Ohm input attenuation in SDS6204A:
(used for later reference trace)


Next is with Siglents SAP1000 active probe on Ch1 with ground spring and tip connected to the back side of the BNC pads on Leo's pulser where we can plainly see even its 1.2pF input capacitance can impact on the fidelity of the pulse.


So with some simple trickery with a reference trace captured as a fidelity reference (leftmost) and pulser on ch4 and active probe on ch1 the propagation delays from the different input paths have all 3 traces nicely staggered for us to easily compare them all.
Edges: Reference trace then Ch4 50 Ohm input to Bodnar pulser and last, ch1 Siglent SAP1000 active probe.


Where to from here ?
Well only a 1 GHz active probe has been used thus far and I have 3 passive probe's step responses yet to check, Siglent 500 MHz SP5050A that comes with SDS6204A, Pintek 500 MHz CP-3501R/pro and Pintek 600 MHz CP-3601R.

This is where things get interesting as we must compare each one by one using the Bodnar pulser terminated into 50 Ohms with the probe tip to BNC adapter for the connection and the probe step response is just what it is....providing LF and HF compensation has been set correctly.

Here we have a teaser with a couple of NIB probes, freshly LF compensated and compared to the reference trace which indicates HF compensation needs also be checked.



To be continued......

[jonpaul]

Rebonjour à tous!  Bravo pour Tautron, Very fine work on the Leo B 40 ps.

Years ago we posted some similar results on Tektronix 7104/7A29, 7B10, 7B15, and 2467B, and digital Yokogawa DL7440. ("show us your square wave"?) We tested sans probes, our trusty Leo B 40 ps direct to scope 50 Ohms, with a few setup improvements:

1/Pulser at 10 MHz rate,  Fine transient response shown at faster sweep rates, to the fastest on each scope.
The Tektronix 7104 1 Ghz fastest analog scope is amazing.....

2/ trim of scope ( and plug-in) vertical response  can be optimized for best BW or cleanest transients response. BUT  Not both! Our work in digital audio transmission and wideband transformers required best transients response. the modern  Chinese scopes preffered to optimize first BW specs, not transient abberation.

We have spent many happy hours tweaking fast scope verticale amps for best transients..many interacting trims, some are fragile tiny ceramic disc trimmers, or a few turns nano Hy  coils, not for the faint of heart...

3/Observéd Transient abberation is affected by the pulser and  scope VSWR, due to reflections.
Our old trick is to insert a wideband 20 db passive attenuator between the pulser and the scopes input.
We reccomend Mini-Circuits HAT-20 BNC in-line.  Many Ghz band. The results always show some improvement.   

Textmatic.....
LeoB 40 pS>>>HAT-20 attn>>Scope 50 Ohm input.

https://www.minicircuits.com/WebStore/dashboard.html?model=HAT-20%2B

So, have an Absolutely fantastic day!

Bon Soirée

Jon

[bdunham7]

This is where things get interesting as we must compare each one by one using the Bodnar pulser terminated into 50 Ohms with the probe tip to BNC adapter for the connection and the probe step response is just what it is....providing LF and HF compensation has been set correctly.

To be continued......

Very nice!  Since you have the equipment to show off, perhaps you could demonstrate the effect of probe loading on the actual signal by connecting a passive probe to the pulser as you've shown and then probing the backside with your active probe.  The 1.2pF of the active probe did have a slight effect on the rise time (~30ps or so) of the direct signal, but that was pretty subtle.

[tautech]

Very fine work on the Leo B 40 ps.

This one is actually 28ps according to the screenshot provided with it from Leo.

3/Observéd Transient abberation is affected by the pulser and  scope VSWR, due to reflections.
Our old trick is to insert a wideband 20 db passive attenuator between the pulser and the scopes input.
We recommend Mini-Circuits HAT-20 BNC in-line.  Many Ghz band. The results always show some improvement.   

Yes I know these tricks too but......to not get too complex and confuse the readers we have not used one but that is coming. 
For now when we need external termination of the Bodnar pulser a Tek 1x 50 Ohm BNC feedthrough has been used but you've only seen its effects in the last screenshot and it's loaded with 2 passive probes. 
IIRC Tek rated these feedthrough's to 1 GHz and I also have a 10x version that we will later use as a pad but higher input sensitivities will be needed of course but we save this for a later installment only using the 10x pad so to reduce the chance of reader confusion.

For interest I got a new Pintek 1x 50 Ohm BNC feedthrough a couple years back and the VNA indicates it's better than these great old Tek terminations so we might also be comparing how it and the Tek affect the step response waveform.
http://www.pintek.com.tw/customer/pintek/upload/PL-50pro-DM.pdf

This is where things get interesting as we must compare each one by one using the Bodnar pulser terminated into 50 Ohms with the probe tip to BNC adapter for the connection and the probe step response is just what it is....providing LF and HF compensation has been set correctly.

To be continued......

Very nice!  Since you have the equipment to show off, perhaps you could demonstrate the effect of probe loading on the actual signal by connecting a passive probe to the pulser as you've shown and then probing the backside with your active probe.  The 1.2pF of the active probe did have a slight effect on the rise time (~30ps or so) of the direct signal, but that was pretty subtle.

Yes but worthy of demonstrating as the old adage; every measurement changes the measurement, is very valid when we examine results closely.

As time permits we certainly will look at many more things in detail and maybe even differential probes of which I only carry 25 MHz models to display although their rise times are slow they are quite adequate for many tasks......then there's current probes too which no power engineer shouldn't be without for which a Tek P6021 with passive termination is probably still the best bang for buck if you can find a good one after how many decades in production ? 
Like Jon I have the 6021 and 6022 < both highly valued pieces of kit.

[Njk]

It has been done at various times.  The Tektronix A6902 is a good example which is fully documented, and there are modern faster probes.

Thanks for the link, it was interesting to see how they did it back in eighties in a pure analog way. Indeed, the isolation not something new but to-day it can be done in a number of different ways, because digital semiconductor technology is much more advanced now. For instance, google for isolated ADC or something like that. And it does not look like the price for T&M equipment is closely related to the cost of its internal components (RAM, etc). Some related discussions on this forum as well. What is remarkable, however, is that a battery, i.e., a thing that uses kilograms of precious raw materials, is still much more economical solution.

[jonpaul]

NJK, Tautech, Bdunham7:

Found my 2019 Tektronix 7104/&A29/LeoB 40 pS shots with Mini Ckts 20 db 50 Ohm atten.

After selection of 7A29 and transient response adjustment both  7A29 and 7104, perhaps the best I can do.

Your thoughts appreciated!


Lower pix   BNC inline atten and 50 Ohm  term collection.

Vive l'ANALOG! Vive TEKTRONIX!

Bon week-end!

Jon

[jonpaul]

Rebonjour,
 
New scope shots of Leo B 40 pS with/sans Mini Ckts -20 db to the reduce reflected energy from scope t pulser due to VSWR or Zo mismatch.

 Yokogawa DL7440 500 MHz BW, FS = equv 100 G/sec, sweep  at 1 nS/div

upper LeoB40>>Scope 50 Ohm

lower:  LeoB40>>Mini Ckts HAT-20 atten>scope 50 Ohm input

 Your thoughts?


Jon

[David Hess]

As time permits we certainly will look at many more things in detail and maybe even differential probes of which I only carry 25 MHz models to display although their rise times are slow they are quite adequate for many tasks

When using faster ones the limiting factor tends to be common mode rejection, so there may be no benefit to a higher bandwidth differential probe.

[tautech]

As time permits we certainly will look at many more things in detail and maybe even differential probes of which I only carry 25 MHz models to display although their rise times are slow they are quite adequate for many tasks

When using faster ones the limiting factor tends to be common mode rejection, so there may be no benefit to a higher bandwidth differential probe.

Of course unless you need that resolution.
I was heading towards eventually demonstrating propagation delays with various probing techniques where I may need to revert to 100 MHz rated probes to get somewhere nearer the good ol' DP-25 differential probes we carry.

Thanks for the poke and hopefully can provide some more pretty pics soon. 

[jonpaul]

Differential probes are by definition active, and with rare exceptions like P6046 TEK, have limited BW.

They have their place esp the HV 4kV CATIII but are not useful in high BW work.

Jon

[BillyO]

Your thoughts?


Jon

There seems to be a reasonable reduction in ringing.

[David Hess]

Differential probes are by definition active, and with rare exceptions like P6046 TEK, have limited BW.

They have their place esp the HV 4kV CATIII but are not useful in high BW work.

The Tektronix P6046 was particularly useful beyond its bandwidth rating because it had useful common mode rejection beyond its bandwidth.

[jonpaul]

Rebonjour, David, BillO: Mille mercis pour tes idées et feedbacks.

1/ We used P6046  in 1971..1975  with the Sequerra Tuner at 10.7 and 100 MHz. Entire tuner RF/IF was full differential.
 https://hometheaterreview.com/sequerra-model-1-fm-tuner-reviewed/

2/ We found P6046 was delicate mechanically, electrically and has a thick cable. We made special PCB and test fixtures to mate to the probes inputs.
https://w140.com/tekwiki/wiki/P6046

3/ The max voltage range +/- 25V  is critical as even a sub uSec transient or static can blow the sensitive  probe inputs. Very hard to repair!

"Use the shorting plugs on the tips when not in use:  Ground the probe! Seriously. Sticker the probe "Ground Me!" ."
We always protect the tips with the 10X attn head. That head increases the range to +/- 250V.

4/ As most of our work high speed afterwards was single ended, thus CMRR was not an issue.

5/ All questions re the 120/230V PSU connections, testing, iuse of ground lead are in the manual:
https://w140.com/Tek_P6046_Differential_Probe-Operation_Manual.pdf

6/ By analyzing the measurement requirement and system block, we seldom find a need for the differential probes,  high speed or HV, even  though we have both.
99%of times we use single ended V and current probes.

Being retired, my work is reduced to SMPS/Electronic Ballast (use mains isolation transformer)  digital audio transmission and magnetics,  eg AES/EBU/SP/DIF, baud rate to 24.576 MHz >> BW ! 125 MHz.




Hope this is of interest,

Have an ABSOLUTELY FANTASTIC day!

Jon

PS:   for WB fast work, we use just  a coax to BNC to make a Zo probe or TEK Zo probe  P6136.

10X: Of all the TEK 10X probes accumulated since 1968 (!) the most useful, robust and practical (tips, ground return lead, BW, aberration) we use is P6137, way better than the smaller or larger TEK  series.



 

[tautech]

.................

This is where things get interesting as we must compare each one by one using the Bodnar pulser terminated into 50 Ohms with the probe tip to BNC adapter for the connection and the probe step response is just what it is....providing LF and HF compensation has been set correctly.

Here we have a teaser with a couple of NIB probes, freshly LF compensated and compared to the reference trace which indicates HF compensation needs also be checked.



To be continued......

Continuing
By trial and error the best feed through was found from 3 for the lowest ringing to terminate the Bodnar pulser into for testing HF probe step response. The Tek 1x was the winner with the least overshoot to initially do the HF compensation to a direct BNC connection from the pulser.

Pulser, feedthrough's, Siglent PS3050A 10x and Pintek CP-3501R 10x probe


SAP1000 and Pintek CP-3601R 10x probe


3 passive probes were checked against the Ref waveform captured from the Bodnar pulser into Ch4 input.
Each of these required HF compensation (see above screenshot) for which I used the Bodnar 10 MHz pulser whereas all probe pamphlet instructions state 1 MHz should be used and peaks (overshoot) should be within or no higher than max amplitude.
However I chose on this occasion to best replicate the Ref waveform with all HF adjustments and the results are as follows.
Then a check with the SAP1000 1 GHz active probe on the rear of the pulser BNC with it connected to the termination to check signal fidelity was near to our previously saved Reference waveform. Near enough is the verdict and gives us an indicative risetime of some 340ps to compare passive probe performance against roughly 80ps slower than the direct BNC connection.
SAP1000



Siglent SP3050A 500 MHz 10x sense probe proved the most difficult to perform the HF compensation upon for the following screenshot.



Pintek CP-3601R 600 MHz 10x probe



Pintek CP-3501R 500 MHz 10x probe.




The surprise is the Siglent SP3050A while not providing the best step response did provide what appears to be the slightly better rise times.

Next up are some other probe checks, PP215, a SP2030A we have lying about and a couple more Pintek 10x sense probes, CP3351R and CP3101R and while at it we should also test 100x CP3308R 2kV and 1000x HVP-08 8kV probes.

[jonpaul]

Hello, very fine but.....

1/ Your shots  Time/div much too slow  to see fine structure of transient response and aberration.

Use a 1 ns or faster /div. on DL7440 Yokogawa, or TEK 7904/7104/7B10/7B15.

Unsure what the Sigeilnt scopes are capable of.

2/ There are Many poorly made or old   50 Ohm terms/feedthrough  are about. Check the BW and Zo of your terms on a wideband network analyzer or impedance analyzer  at least 1 Ghz.

We found the Mini Circuits and old EGG are the best, most robust and widest BW.


3/ No need for any feedthrough in scope testing if scope of plugin has a50 Ohm inut.

That is better than  an external feedthrough term.

Enjoy,

Jon

[tautech]

Hello, very fine but.....

1/ Your shots  Time/div much too slow  to see fine structure of transient response and aberration.

Use a 1 ns or faster /div. on DL7440 Yokogawa, or TEK 7904/7104/7B10/7B15.

Unsure what the Sigeilnt scopes are capable of.

2/ There are Many poorly made or old   50 Ohm terms/feedthrough  are about. Check the BW and Zo of your terms on a wideband network analyzer or impedance analyzer  at least 1 Ghz.

We found the Mini Circuits and old EGG are the best, most robust and widest BW.


3/ No need for any feedthrough in scope testing if scope of plugin has a50 Ohm inut.

That is better than  an external feedthrough term.

Enjoy,

Jon

Please see and understand first 2 screenshots in this earlier post:
https://www.eevblog.com/forum/testgear/probe-into-probes-whats-up/msg4678768/#msg4678768

We require the Bodnar pulser to be externally terminated into 50 Ohms to provide a 25 Ohm source for probe HF calibration, risetime and step response tests against a previously saved reference trace that used the scopes internal 50 Ohm input.

We are using a 2 GHz rated 4ch DSO and for you the EU link:
https://www.siglenteu.com/digital-oscilloscopes/sds6000a/
Fastest timebase is 100ps/div

[jonpaul]

Tautech:

Many thanks, good link for the scope.

still do not see any shots at fast sweep, only the 10 nS/div, can you show shots like those references with fine detail at 1 nS..200 pS/div?

All my work was at 50 Ohms, no probes jut the pulser>>scope at 50 Ohms internal.

Thus a bit different than your very interesting probe tests.

Be good to do a check of feedthru 50 Ohm BNC/SMA terms, of various brands and ages:  LeoB>>50 Ohm Term>>1 M scope to see the cheges in term VSWR at fast edge.

Enjoy,

Jon

[Bud]

1/ We used P6046  in 1971..1975  with the Sequerra Tuner at 10.7 and 100 MHz. Entire tuner RF/IF was full differential.
 

How important the ground lead length on P6046 is for signal integrity? Same as for a single ended probe or is it more for electrical safety of the input FETs?

[jonpaul]

bud, as P6046 is differential, indeed ground lead legnth is not as,critical as in single ended probes.

However the HF CMRR rolloff limit the effectiveness thus the ground clip is,useful.

Since the FET front end has a CM and DM voltage limit, the CM input 1 r input 2 to ground voltage can be easier exceeded if no ground or a,long ground,lead with attendant high inductance is present.

We always used the standard ground lead

So not as much for safety (only low,V use of these probes) but to protect the ET inputs.

Jon

[Bud]

Thanks Jon. The standard ground lead you mean the one you snap to the ground clip on the probe casing? (Not on input 1 or 2 tip ground)

[tautech]

Bud, this is P6046:

[Bud]

I know what P6046 is, I have one 
There is a short ground post on the other side of the probe head, to which you snap a ground lead of a required length. This is what i was asking about.

[tautech]

I know what P6046 is, I have one 
There is a short ground post on the other side of the probe head, to which you snap a ground lead of a required length. This is what i was asking about.

Gotcha. 

Yep our P6021 and 6022 have those too.

[Bud]

There it is