开云体育

Has anyone managed to calibrate a 2465B without using the Tek specific test gear (PG506, TG501)?
I have an HP8112A pulse generator and an HP33120 function generator, can they be used instead?

- Keith

I’d be *very* surprised: You need a relatively low frequency (1MHz or so) square wave with an edge speed << 1ns for risetime/BW check and adjustment and low-and high-frequency pulses (up to 1GHz) for horizontal timing with an accuracy much better than the 2465’s spec, for starters.
The instruments you mention can’t do that, unfortunately.

Raymond

Thanks Raymond, yes the fast edge rate is a problem, the 8112 is about 5nS min.

However... I do have the Tek fast pulse PCB that is used in a PG506, I believe that has < 1nS edges and could be driven by the 8112 to address the edge rate?

Also have access to a siglent SDG6052X which is a 500MHz function generator.

Hey Keith,

I calibrated the CRT, timing, vertical and the triggers on my 2467 using a
TG501 and a PG506. I didn't touch the transient response calibration, but I
don't think the PG506 is fast enough for that. You need a much faster edge,
and IIRC Tek specifies a tunnel diode pulser.

I think if you have a decent signal generator or ARB, it should be able to
produce all the signals the calibration calls for^W^Wneeds. The horizontal
(timing) calibration is all done by eye, using the technician as a visual
comparator. TG501 is very convenient, as it gives you fairly narrow,
fast-rising pulses for timing marks and it's quick and easy to toggle to
the requested settings.
A square wave of the required frequency would work, though, even a sine at
a pinch. There'd be an awful lot of faffing about with controls, though :).

The vertical calibration is similar to the horizontal, e.g. the technician
is used to adjust levels until traces line up. IIRC you need a max of a 5V
peak to peak (ground to peak) signal for the vertical calibration, which is
easily done with a digital signal generator or an ARB that has 12bit
resolution or better. At the lower end it might be getting hairier, I don't
remember offhand what the signal levels are there, but to get good
resolution and low noise you might need to use attenuators, depending on
how your siggen/ARB works.
The trigger calibration is however automated, and is somewhat picky on the
signal provided, but again nothing you wouldn't be able to provide with an
ARB or a digital siggen.

Siggi

On Fri, Jan 17, 2025 at 7:00?AM Keith via groups.io <keith=
peardrop.co.uk@groups.io> wrote:

Thanks Raymond, yes the fast edge rate is a problem, the 8112 is about 5nS
min.

However... I do have the Tek fast pulse PCB that is used in a PG506, I
believe that has < 1nS edges and could be driven by the 8112 to address the
edge rate?

Also have access to a siglent SDG6052X which is a 500MHz function
generator.

On Fri, Jan 17, 2025 at 12:59 PM, Keith wrote:

However... I do have the Tek fast pulse PCB that is used in a PG506, I believe
that has < 1nS edges and could be driven by the 8112 to address the edge rate?

Also have access to a siglent SDG6052X which is a 500MHz function generator.

Keith,
The fast output stage of the PG506 has been / is to be adjusted while *in* a PG506. Because of the speeds (high frequency signal components) involved, its wave shape will be different if outside that PG506. So, you'll need to verify/calibrate it (!) before using it to adjust a 2465B. If you adjust without knowing its exact wave shape (which should be as square and flat as possible), you're adjusting the *combination* of the PG506's output stage as-is with the 2465B, which of course is of very little use. I don't believe you'll get even close and you might end up with significant adjusted-in over- or undershoot in your 2465B. And I left out the fact that a PG506 isn't really fast enough; you need something like the tunnel diode pulser as proposed in the 2465B's service manual. A well-tuned PG506 with known exact wave shape may do for "reduced requirements" though for rise time adjustment.
In case you'd consider a "Jim Williams pulser" or the like, which some people propose or use and which is easy to build, remember that the edge you need is a *step* i.e. a signal that contains relatively low frequencies (the flat parts, both *immediately* before and *immediately* after the step), to high frequencies (the steep range in between and its beginning and end). In effect, with a bare JWP you see (and adjust) a *slew rate*, not a *step time*, so *not* the correct frequency behaviour of your 2465B. "Rise time measurements" and adjustments using a JWP in most combinations of 'scope and pulser make no sense, even if some serious Youtube publishers try and make you believe otherwise.
You may consider purchasing a Leo Bodnar pulser for adjusting the step behaviour.

For the faster timing adjustments, the Siglent SDG6052X isn't fast enough.

For vertical sensitivity adjustment, you need a low frequency (1 kHz square wave or so) source with good amplitude accuracy (at the very least 2%). I don't think any of the instruments you mentioned have that, as far as specified, although there are (rather complicated) ways to correct for it.

Adjusting a 2465B is rather complicated and involved, especially first time. Add to that that the instrument is rather picky with regard to following the prescribed procedure (completely) and you'll realise that not using convenient instrumentation may be quite an uphill battle.

I trust you have replaced the NVRAM or you're sure it has enough life left? Your calibration results are stored there. I assume this whole exercise has become necessary as a result of NVRAM replacement.

You may be lucky in finding a TekScopes member close enough to you who may be willing to lend out his/her PG506/TG501/SG504 set...

Raymond

Hi Keith,
I saw Siggi's reply just after I sent mine.
He has a lot of experience with 2465 'scopes and always makes a lot of sense in what he says.

As an example, he may well be right re. transient response adjustment: After NVRAM replacement you don't need to do it, only after repair, exchanging parts or bending leads like those to the CRT. I don't know why you want to calibrate your 'scope, just assumed "full calibration".

Although horizontal (timing) employs verification by eye, you do need enough time resolution to be able to do that. A highest-resolution 2ns is used, right at the end of the Siglent's range of *sine wave* signals. The Siglent's square wave/pulse ranges end at 120/150MHz. Note 2465B SM 070-6863-01, Table 5-3, starting at point 28. Some adjustments only apply to some S/N ranges, see SM.

Vertical: I didn't check if your synthesised generators provide enough accuracy. The Siglent does provide enough resolution but I couldn't immediately find anything about its accuracy, except for DC.

The HP 8112A is analog, except for its controls. It is useless for anything that needs accuracy.

Raymond

This is all necessary as the original DS1225 had parity errors (it was datecode '96). I don't know of anyone locally who has the right Tek test gear so it's either try with what I have or leave the scope uncalibrated.

I could buy one of the Leo Bodnar fast pulse generators but they cost more than I paid for this scope!

- Keith

Each Cal group is self contained, as long as it completes with no errors, the CAL for that procedure is saved to NV Ram.

Transient and CTT are most critical CAL, but PS,Hor and Vert are easily done and more important.

We use the original Tek cal, and the LeoBodnar for transient trim.

You can CAL without the Tek CAL, but it is long and doubtful of complete success.

Note that NV Ram defaults are available

Suggest you repost on tekscopes2, as our 2465B maven Chuck H was banned here.

Chuck may have more precise information.

Amicalement

Jon

I wonder if a Leo Bodnar pulse generator would work. IIRC, it can produce
pulses with 350 ps rise times.

DaveD
KC0WJN


On Fri, Jan 17, 2025 at 08:42 Sigur?ur ?sgeirsson via groups.io <siggi=
sort.is@groups.io> wrote:

Hey Keith,

I calibrated the CRT, timing, vertical and the triggers on my 2467 using a
TG501 and a PG506. I didn't touch the transient response calibration, but I
don't think the PG506 is fast enough for that. You need a much faster edge,
and IIRC Tek specifies a tunnel diode pulser.

I think if you have a decent signal generator or ARB, it should be able to
produce all the signals the calibration calls for^W^Wneeds. The horizontal
(timing) calibration is all done by eye, using the technician as a visual
comparator. TG501 is very convenient, as it gives you fairly narrow,
fast-rising pulses for timing marks and it's quick and easy to toggle to
the requested settings.
A square wave of the required frequency would work, though, even a sine at
a pinch. There'd be an awful lot of faffing about with controls, though :).

The vertical calibration is similar to the horizontal, e.g. the technician
is used to adjust levels until traces line up. IIRC you need a max of a 5V
peak to peak (ground to peak) signal for the vertical calibration, which is
easily done with a digital signal generator or an ARB that has 12bit
resolution or better. At the lower end it might be getting hairier, I don't
remember offhand what the signal levels are there, but to get good
resolution and low noise you might need to use attenuators, depending on
how your siggen/ARB works.
The trigger calibration is however automated, and is somewhat picky on the
signal provided, but again nothing you wouldn't be able to provide with an
ARB or a digital siggen.

Siggi

On Fri, Jan 17, 2025 at 7:00?AM Keith via groups.io <keith=
peardrop.co.uk@groups.io> wrote:

Thanks Raymond, yes the fast edge rate is a problem, the 8112 is about

5nS

min.

However... I do have the Tek fast pulse PCB that is used in a PG506, I
believe that has < 1nS edges and could be driven by the 8112 to address

the

edge rate?

Also have access to a siglent SDG6052X which is a 500MHz function
generator.

If you have CTT option the cal requires a precise 50% DC precise frequency and level.

If no CTT option ignore

Jon

On Fri, Jan 17, 2025 at 9:44?AM Jean-Paul via groups.io <jonpaul=
ix.netcom.com@groups.io> wrote:

Transient and CTT are most critical CAL, but PS,Hor and Vert are easily
done and more important.

I really should look at doing the transient calibration on my 2467 now as
my GAS (Gear Acquisition Syndrome) has left me in possession of both a
home-made tunnel diode generator and e.g. an S52.

You can CAL without the Tek CAL, but it is long and doubtful of complete
success.

I think first-time calibration from the instructions in the manual is
always going to be a long and doubtful process :). I know I had to run
through the process at least twice, and I probably repressed a couple of
run-throughs. I remember the B time base requires a second scope, and it
wasn't until years later and I'd scoured the schematics that I understood
the purpose of the step (for 15 minutes, don't ask me now).
I think for many here the journey is the reward, and at least OP is certain
to learn some things.

I should also note that the calibrator in the 2465-series is intended to
allow checking most of the time base calibration with nothing but the scope
and a 10X probe, that's why its output frequency follows the time base (to
a point).
The output of the calibrator is also specified to be 0.4V peak-to-peak to
sufficient accuracy to allow checking at least a couple of vertical ranges.

On Fri, Jan 17, 2025 at 06:41 PM, Sigur?ur ?sgeirsson wrote:

I really should look at doing the transient calibration on my 2467 now as
my GAS (Gear Acquisition Syndrome) has left me in possession of both a
home-made tunnel diode generator and e.g. an S52.

(Home made) tunnel diode generators obviously need very careful physical layout with minimal dimensions and preferably, checking and adjusting with a fast (sampling) 'scope.
In my experience, S-52's obviously have a fast (spec. is 25ps max.) rising edge but not such very nice flat behavior immediately after it: Spec. is -7%,+7% with max 10%pp within the first 1.8ns after the step. For its purpose (TDR), where only the edge speed counts, this was irrelevant. Leo Bodnar's pulser, which as I just found out is no longer being produced BTW, shows about -3%, +3% under/overshoot.
When putting my first TD pulser together, I was lucky enough to have S-2, S-4 and S-6 sampling heads in a Series 7000 Tek mainframe at my disposal. They mostly agreed and after quite some tweaking, the signal ended up being pretty fast (< 200ps) and flat. I'm talking about mm-size tweaks to get there.

I think for many here the journey is the reward.

It certainly has been for me, which is a wonderful thing when acquiring dozens of 'scopes.

Raymond
Raymond

Jim WIlliams wrote at least one application note where he presents a pulse generator with ~350 ps rise times. I took a quick look through my collection of Linear Technology application notes and found that Williams has a circuit for such a device in Application Note 45, page 18.



I would also look at Application Note 47.



If I get some more time later I'll look more closely.

DaveD

--
This email has been checked for viruses by Avast antivirus software.
www.avast.com

On Fri, Jan 17, 2025 at 07:56 PM, Dave Daniel wrote:

Jim WIlliams wrote at least one application note where he presents a pulse
generator with ~350 ps rise times. I took a quick look through my collection
of Linear Technology application notes and found that Williams has a circuit
for such a device in Application Note 45, page 18.

Jim Williams used the phrase "verifying rise time capability" in relation to his pulser. In my opinion, often the meaning of this is misinterpreted. I'd like to present the most important (or maybe only the most obvious) aspect of this. This is on the one hand to verify the validity of my argument, on the other hand to be told where I'm wrong, if so.

Suppose a Jim Williams pulser with a "rise" and "fall" time of 350ps each, as in AN45. One could think that such a pulser may be perfectly used for *adjusting* most 'scopes, like a relatively slow 100MHz 465.
Realizing that the pulse *duration* is much less than 1ns, it is obvious that the pulse has come and gone before the 465, with its 3.5ns rise time, is able to show any of it. This is the main reason for incorrect complaints about the 1ns pulse on a TG 501 being "broken": Even a 2465B (400MHz BW, 875ps rise time) cannot show that pulse to any significant amplitude. The bit of straight slope, if at all is visible, just shows the slew rate of the 'scope, not the rise time.
On a correctly adjusted 'scope with a significantly faster (>=5 x or so) slew rate than the pulse's, the slew and amplitude of the pulse are shown "correctly". On slower 'scopes, at least the peak value shown is as less than the real peak value and the slope (slew) results from the 'scope only.

Regarding the correct interpretation of what is generally called "rise time", its definition is something like the duration of the response to an infinitely fast step between two distinct voltage levels, measured at the 10% and 90% level for analog 'scopes. To be able to see the two distinct voltage levels and to allow for the low-frequency content to be seen, the transient must be clearly shorter in time than the end levels. That's why a relatively low signal frequency is needed, compared to the rise time and why a nice, high frequency signal with a nice, short transient is useless for rise time measurement, *unless* the 'scope is much faster than the square wave, in which case you're measuring the square wave, not the 'scope...

Any comments welcomed.

Raymond

Hello again Raymond,

A few notes on the CAL and fast rise pulses.

1/ Tek service manual CAL procedure is vague in a few places, see the notes from Chuck Harris on tekscopes2.

2/ since 1980s, we designed and researched low jitter digital transmission and transformers.

The transient rise time is important but especially the aberrations, eg percent overshoot, ringing, etc.

The PG506 fast rise is OK, and older Tek fast pulse adapter (TD, Avalanche) and Jim Williams Lin Tech can be used.

My standard is the Leo Bodnar 40 pS pulser, fortunately have two of the BNC model, USB powered
10 MHz. Highly recommended!

To have best results, use no cables, buffer pulser output with a Mini-Circuits HAT-20 20 db 50 ohm BNC attenuator (to minimize reflections from VSWR of the scope input).

That setup has worked for 2465/7/B and also 1 ghz scopes and plugins like 7104/7A29, to optimize the vertical transient response.

Note that transient response tweaking is in steps of interacting adjustments, affecting different time scales. The procedure should be repeated a few time for optimal results.

Can post some photos if any interest.

Amicalement


Jon

Hi Jon,
Please find my responses interleaved.

Raymond no replys seen in your last msg.

Forgot that latest Leo Bodnar site update no longer lists the BNC pulser version,

Their new pulsers are SMA connectors, with SMA to BNC adapter

Jon

Hi all,
As I am incompetent in this field you will take this with indulgence please but as Raymond above asks for "Any comments welcomed.
" I add my 2€ cents...
To generate a square signal at any frequency with a fast rise time I was thinking of using an Analog Devices HMC363S8G/GE SMT GaAs HBT MMIC DIVIDE-BY-8, DC - 12 GHz.
Output transition time claimed is 100 pS. GE is around 25€ with Mouser.
Output level is -6 to -9 dBm so with a good attenuator following to ascertain the 50 ohm the level would not be large but would that be silly?
Best
Renaud

Ray, Bonjour

No idea about the divider, the output must drive a 50 ohm load, at leas 1 V

Easier with discrete parts. No freq divider needed.

Usually a 1-10 MHz rate.

many fast pulser designs on EEVBLOG, see " show me your fast square wave ", as well as Jim Williams as mentioned.

See old Tek pulser adapters, with TD and avalanche transistors.

Amicalement

Jon

On Sat, 18 Jan 2025 at 10:33, Jean-Paul via groups.io <jonpaul=
ix.netcom.com@groups.io> wrote:

Ray, Bonjour

No idea about the divider, the output must drive a 50 ohm load, at leas 1
V

Easier with discrete parts. No freq divider needed.

Usually a 1-10 MHz rate.

many fast pulser designs on EEVBLOG, see " show me your fast square wave
", as well as Jim Williams as mentioned.

As I'm sure you know, for scope risetime you need a step, not an avalanche
generator pulse. If a pulse is shorter than the scope's risetime then a
100ps 1V pulse will look exactly the same as a 1ns 100mV pulse.

If you need 1V into 50ohms, then consider homebrewing 3*(74lvc1g14+143ohms)
in parallel. That will drive 2.5V into 50ohms with <300ps risetime. You
need to be very careful of layout and decoupling.

I ought to respin mine; I think I could reduce the ~6% overshoot.

For lower voltages, either reduce the Vcc and accept the increased
risetime, or use a "GHz" RF attenuator.

Be aware that even short lengths of coax will attenuate higher frequencies;
if possible "mount" the step generator on the scope's input.

If using a coax cable for a conventional scope input (1Mohm//15pF), be
aware that the 15pF will cause the standard problems with mis-terminated
transmission lines. The problems can be reduced by having an inline pad
mounted on the scope's (BNC) input. Even 3dB is useful, but 6dB is better
(and a 2.5V step would still be >1V).