Tek 2465 calibration question

[analogRF]

I need to calibrate my 2465A and newly restored 2465B and I was wondering if I can do the whole calibration except the bandwidth calibration using an Arbitrary Waveform generator (connected to GPSDO) like Siglent SDG2042 or Tek AFG3021?

Also since I have a chance to pick up a BK precision 1400 (same as Heathkit IG-4244) would that be enough to do the calibration on these scopes? The manual is very complicated and seems only those Tek equipment can do the job but I am sure that should not be the case. The BK Precision only generated fixed values of amplitude/period in 1-2-5 steps. not continuously variable.

Any clarification is highly appreciated

[med6753]

A few questions first....

Are you sure you lost all your cal data on both scopes?

Are you getting power up fails indicating calibration problem?

Depending on what you need to cal you should be able to get reasonably close or even dead nuts without all the expensive equipment Tek says is required. But some of the calibration constants can be real picky as to what they will accept.

For example: I was able to properly calibrate the Counter/Trigger option on a 2465 DMS with a Heath IG-4244 Oscilloscope Calibrator and that was one particular calibration constant that needed to be right on or it wouldn't accept it.

[analogRF]

no cal data is lost and no fail 04 errors but the scopes are just far from calibration. It is easy to spot specially for time base and when you look more carefully the vertical gain is also off.

so I can go through the cal process with a good arb generator like siglent (which I have) or I need the IG-4244? I still prefer not to but the latter.

my 2465B has CTT option as well (I think option 06 or 09)

[med6753]

I've never had to calibrate the 2465 mainframe itself but I see no reason why the Siglent couldn't accurately calibrate both the timebase and vertical gain.

Most likely your CTT option is OK. If it's out of cal it will flag an error on power up.

[analogRF]

apparently the CAL process needs time markers at 5ns and 2ns as well as longer periods. If I understand correctly it means having pulse trains (or rather impulse trains) with 200MHz and 500MHz repetition frequency!?

How can I generate that? neither Siglent SDG2042 nor the BK precision 1400 can come even close.

[med6753]

Hmmm...lemme go look at the service manual myself and take a look.

Gee, I guess I'll have to eat my words with respect to the horizontal time base. Apparently some sort of time mark generator is a requirement or it's a no go. Perhaps someone can chime in with a possible alternative. 

[tautech]

apparently the CAL process needs time markers at 5ns and 2ns as well as longer periods. If I understand correctly it means having pulse trains (or rather impulse trains) with 200MHz and 500MHz repetition frequency!?

How can I generate that? neither Siglent SDG2042 nor the BK precision 1400 can come even close.

Hmmm, yes not even an 'improved' SDG2042X mightn't be good enough. (Pulse 25 MHz max)
An 'improved' SDG6022X should do it with a 3.3ns pulse width out to its 150 MHz repetition rate.

While I've never had a 2465 I've followed the thread on them for years and noted the comments rf-loop made about calibrating/adjusting them particularly in reply #14
https://www.eevblog.com/forum/testgear/tektronix-2465b-oscilloscope-teardown/msg73972/#msg73972

[analogRF]

150MHz repetition rate is not nearly enough. The manual asks for time markers all the way down to 2ns and 5nsec.
So if I understand correctly, it is asking for narrow impulses at every 2nsec and 5nsec which is 500MHz and 200MHz rate.

something doesn't sound right here to me because I really don't believe those old Tek equipment like PG506 (or whatever it is according to the manual) can really generate pulses at 500MHz rate.

I was thinking maybe I can just use the zero crossings of a 500MHz sinusoidal from GPSDO locked RF generator....

EDIT: if we are gonna need time markers every 2nsec, then what would be the width of those pulses? like 200ps or something?! this is crazy. I am sure I am missing something here...

[tautech]

I've always checked the spec sheet of any 'required' Cal equipment to properly determine what is called for then used equipment that has met those specs.
I'm sure with a little deeper digging into the specs and the Cal processes/steps then exactly what Tek requires will become clearer.

Maybe there's clues in the previously linked thread.

[JS]

150MHz ARB repetition rate could be enough, if you set an arb with a few markers on it's own.

Then there's the question, is the jitter in such AWG good enough to calibrate your smelloscope?

JS

[tautech]

150MHz ARB repetition rate could be enough, if you set an arb with a few markers on it's own.

Then there's the question, is the jitter in such AWG good enough to calibrate your smelloscope?

JS

100 ps
Check for yourself:
https://www.siglenteu.com/wp-content/uploads/dlm_uploads/2018/04/SDG6000X_DataSheet_DS0206X-E02A.pdf

[JS]

150MHz ARB repetition rate could be enough, if you set an arb with a few markers on it's own.

Then there's the question, is the jitter in such AWG good enough to calibrate your smelloscope?

JS

100 ps
Check for yourself:
https://www.siglenteu.com/wp-content/uploads/dlm_uploads/2018/04/SDG6000X_DataSheet_DS0206X-E02A.pdf

The 3.3ns minimum with kills it, I don't know if in a user arb it could be set lower, with few samples making each peak, few times per cycle. Also, specified is inter cycle jitter, not inside the cycle, I guess it should hold or be better to that number, but just a guess.

JS

[Johnny10]

The Tektronix TG501 will do the job.

Goes to 1GHz

[analogRF]

i did a little bit more digging at 2465B manual and the manual for TG501 time marker generator and also CG5011 calibration generator.
It turns out that for 5ns,2ns,1ns,0.5ns time markers they are actually creating SINE waves by using frequency doublers followed by high-Q bandpass filters at 200MHz,500MHz,1GHz,2GHz. So basically a GPSDO locked RF gen should do the same job
However, for 10nsec and slower markers they do actually create sharp pulses (rise times are not that fast though ~1ns I think. i have to check again) using high speed discrete transistor ECL logic circuits. Still 10ns is hardly doable, if at all, with Siglent SDG2042 or AFG3021 etc...those guys have several nsec rise time and a minimum pulse width that makes is impossible but I guess even for 10ns perhaps an RF gen can still be used

However, another issue is that when calibrating for vertical gain, pulses with amplitudes up to 50V (or was it 100V ?) are required and those CG5011 and PG506 are capable of doing that (up to 200V in case of CG5011) these are not high frequency or very fast edge pulses, just with accurate amplitudes.

EDIT: for vertical gain I guess even sine waves are ok as some calibrators like heathkit IG-4244 (not PG506) produce sine waves for gain calibration However the amplitudes required for calibrating 2465 are from 10mv to 100vpp with better than 0.25% accuracy and frequency up to 100KHz

[analogRF]

@Johnny10

yes, my problem was and remains that I dont have those TG501 and PG506 and just for one time calibration of two scopes it is not worth spending. So I thought maybe it can be done with other equipment but now that I read a bit more it seems that those equipment are indeed required as per manual (or equivalently a CG5011/5010 which I think it was not available when 2465 manual was written)

could you post pictures of th5ns, 10ns and 20ns time markers as well. thanks

EDIT: oops! you already showed 10ns, sorry.

[Johnny10]

OK

[BravoV]

Alternative Tek 2901 time mark generator, 2ns to 50ns are in sine waves, rest longer time bases are in sharp pulses.



Examples ..





Did a teardown -》 HERE

[analogRF]

@BravoV and @Johnny10  thanks for the info

I read the cal manual for 2465 and had a look at manuals of calibration generators (CG5011, TG501, PG506) and I am now 100% confident that calibration can be easily accomplished with an accurate function generator and an RF generator plus a fast pulser for BW adjustment.

For vertical CAL and performance checks only sine waves with pp amplitudes of 10mv to 20v (HiZ and up to 10v 50ohm) are required with low frequencies like 1KHz or so. That's easy to do with any modern arb gen. Only the amplitude must be accurate (better than 0.25%).

for trigger performance checks also sine waves up to 500MHz are required which any RF gen can produce.

For horizontal CAL and performance checks, time markers of 2nsec to 0.5sec are needed and 2nsec, 5nsec are certainly sinusoidal (500MHz and 200MHz) and even 10nsec looks very sinuosoidal as well. So, a combination of a good function gen and an RF gen (up to 500MHz) can easily do the job. The timing needs to be accurate which is easily accomplished by locking to a GPSDO or OCXO etc...

[MarkL]

...
EDIT: for vertical gain I guess even sine waves are ok as some calibrators like heathkit IG-4244 (not PG506) produce sine waves for gain calibration However the amplitudes required for calibrating 2465 are from 10mv to 100vpp with better than 0.25% accuracy and frequency up to 100KHz

Just a minor note here that the IG-4244 does produce square waves for gain calibration.  It can do 1mV to 100V pk-pk, and has a duty cycle around 50% with a fixed frequency of approximately 1kHz.

The voltage output accuracy is only 1%, so it's not really a good match for the calibration requirements.  The 1% comes from their selection of 1% resistors in the divider.  It could probably be easily improved by selecting more precise resistors.

The SINE setting on it produces a low distortion sine output with a fixed ~1v pk-pk at ~1kHz, which is useful in adjusting the trigger circuits on older scopes.

Also, on the calibration itself, if you find that the cursor and other measurements are always accurate and it's just the graticule calibration that's off, you might only need to adjust the CRT XY deflection gains and offsets.

I'm interested in your success using various RF and arb generators; please post back with your results!

[analogRF]

thanks @MarkL. My mistake, yes, it produces pulse for vertical gain but still sine wave is perfectly fine for vertical gain CAL as far as I understand from 2465 manual. Maybe pulses are better and more accurate to match to horizontal graticule lines so it is better suited but sine wave works.

Sure I will post everything here when I do the CAL, but I have 2-3 repairs in queue which are bothering me too much right now and I have to get rid of them first. specially this weird thing with this E3631A driving me crazy right now...

[Proto]

A quick thing to do is check the +10V (reference) supply and adjust it if necessary.  Be sure to move the cal jumper with the power off before any calibration steps are made.  The timebase is a DAC controlled current source charging specific capacitors based on the time/div setting.  You need to spot check all the other supplies and perform the calibration steps for the DAC before looking at the timebase error and linearity.

Though it is nice to have a fast time mark generator, you can learn a lot in checking a 10nS/div setting by  using a 10MHz square wave from you function/arb generator.  A full cycle at 10MHz should line up with the first, middle and last graticule line on screen.  Using 20MHz which most function generators can manage will give you more transitions that line up.  Your generator square waves will morph into sine waves as you increase the frequency and it is common for sine wave output to go the farthest. 

The point is that 10nS ticks from a time mark generator (when checking 10nS/div, for example) allow you to see the linearity of the sawtooth waveform that drives the horizontal deflection plates of the CRT.  However, you can glean most of the linearity tracking or lack of it from fast square waves at frequencies that that cause the edge transitions to line up with a least a few graticule lines.  Then you can switch to sine wave for the highest frequencies you generator can manage.  At slower sweep rates, you func/arb generator can can gradually emulate more of what a time mark generator can do.