Topics

Calibrators vs Signal/Function Generators


David Berlind
 

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David


Stephen
 

On Tue, Jul 14, 2020 at 03:13 AM, David Berlind wrote:


Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point
as a part of their resurrection. I see that the primary function of some Tek
calibration fixtures is to generate square and sine waves. How is this
significantly different from what a well-calibrated signal/function generator
can do?

Thank you.

David
I’m nowhere close to a metrologist expert,
But I believe that if your generator has a very fast rise time and a very steady and true output signal, that’ll do the job just fine. As a matter of fact, I think that modern gear, when well calibrated to start with, are probably at least as good, if not better, than the older gear from the 60’s or 70’s, which themselves may need to be recalibrated. Provided they can do the same things required in the procedure, that is.


Dave Daniel
 

Dave,

DescriptionDescriptionQuis custodiet ipsos custodes?

On what basis is a “well calibrated” instrument calibrated? How accurate and with what precision was it calibrated? Against what standards?

The problem is that the calibration instruments, or the instruments used to calibrate the instruments, etc., etc., need to be pretty much dead-on wrt to the actual values being measured. Metrology labs and other people who offer calibration services calibrate their calibration instruments such that they are traceable back to a root source, such as NIST.

I try and keep one indepentably calibrated ‘scope (my 2465B) around for measurement comparison. But then there is my 8566B, my 3456A, and my 8660D which have are uncalibrated.

DaveD

On Jul 14, 2020, at 10:13, David Berlind <david@...> wrote:

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David



Harvey White
 

Typically:

Frequency is crystal controlled.

Risetimes are held to a particular spec (see PG506, 1 ns or so rise and fall times on certain waveforms).

Amplitudes are in the same sequence as the scope gain control, making it easier to check calibration.

Amplitudes do not have to be readjusted per run, they're fixed by the generator calibration while the function generator (some of them) are variable.  Also note that the PG506 can put out high amplitude square waves while many function generators do not.

Timebases?

Ditto on crystal controlled.

ditto on frequencies.

Now for frequency standards, a CPLD or FPGA would work fine with a known frequency for the timebase generator.  You could make that and possibly use a pre-built demo board.  Amplitude?  you're on your own.

Oh, and some of the cal generators, say the 5000 series, can be remote controlled and therefore lend themselves to automated testing and calibration.

Harvey

On 7/14/2020 10:13 AM, David Berlind wrote:
Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David



David Berlind
 

DaveD,

But wouldn't you have the "standard" problem in both cases? (calibrator vs Signal generator)

On July 14, 2020 10:35:04 AM "Dave Daniel" <kc0wjn@...> wrote:

Dave,

DescriptionDescriptionQuis custodiet ipsos custodes?

On what basis is a “well calibrated” instrument calibrated? How accurate and with what precision was it calibrated? Against what standards?

The problem is that the calibration instruments, or the instruments used to calibrate the instruments, etc., etc., need to be pretty much dead-on wrt to the actual values being measured. Metrology labs and other people who offer calibration services calibrate their calibration instruments such that they are traceable back to a root source, such as NIST.

I try and keep one indepentably calibrated ‘scope (my 2465B) around for measurement comparison. But then there is my 8566B, my 3456A, and my 8660D which have are uncalibrated.

DaveD

On Jul 14, 2020, at 10:13, David Berlind <david@...> wrote:

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David



Dave Daniel
 

Yes. That’s my point.

Suppose one uses a Tektronix SG-504 to calibrate a ‘scope. But how does one know that the SG-504 is properly calibrated?

DabeD

On Jul 14, 2020, at 10:51, David Berlind <david@...> wrote:

DaveD,

But wouldn't you have the "standard" problem in both cases? (calibrator vs Signal generator)

On July 14, 2020 10:35:04 AM "Dave Daniel" <kc0wjn@...> wrote:

Dave,

DescriptionDescriptionQuis custodiet ipsos custodes?

On what basis is a “well calibrated” instrument calibrated? How accurate and with what precision was it calibrated? Against what standards?

The problem is that the calibration instruments, or the instruments used to calibrate the instruments, etc., etc., need to be pretty much dead-on wrt to the actual values being measured. Metrology labs and other people who offer calibration services calibrate their calibration instruments such that they are traceable back to a root source, such as NIST.

I try and keep one indepentably calibrated ‘scope (my 2465B) around for measurement comparison. But then there is my 8566B, my 3456A, and my 8660D which have are uncalibrated.

DaveD

On Jul 14, 2020, at 10:13, David Berlind <david@...> wrote:

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David





Chuck Harris
 

If you use a function generator, at the very minimum, you will have
to use it in conjunction with an accurate frequency counter, and an
accurate level meter of some sort... say an AC True RMS voltmeter.

Tektronix made their calibration equipment easy to calibrate to
national standards. For instance, their voltage calibrator PG506,
uses very, very low impedance transistor switches to switch a zener
derived DC reference voltage on and off to make a square wave.

Because of this topology, the PG506 can be calibrated by setting
its low impedance switch to ON, and measuring the DC voltage output
of each output voltage position.

The switch's resistance is a tiny, tiny fraction of the 1M input
impedance of the scope, so as to be a part per million, or so, which
is insignificant.

Your function generator is built in an entirely different direction,
and cannot be calibrated to this sort of accuracy... so it will need
to be measured. Most AC meters aren't anywhere near this accurate,
either... or, if they are, it is only at one frequency... likely 60Hz.

You can do what you want, but it will be slow, tedious, and error prone,
and when you are done, it won't be as accurate as your AC voltmeter..

For timing measurements, amplitude isn't important, but a very fast
rise time pulse is, so small errors in where you line up the graticule
on the pulse will be insignificant. The square wave output of most
function generators is not even close to fast enough... And can in
no way make high enough frequency signals to align the higher time base
speeds. An RF synthesizer can, but it makes a sine wave, which is
hard to line up on the graticule, and to make adjustments for linearity
of the sweep...

Your time ought to be worth something. If you have a "pile of scopes"
that you will be looking to calibrate, then you ought to either hire it
out, or get the proper equipment to do the job effectively.

-Chuck Harris

David Berlind wrote:

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David




Stephen
 

The time base is IMHO, the most important thing that has to be accurate in a scope. You don’t use it to accurately measure amplitude, you have a DMM for that. So I personally use a TEK 184 to calibrate the time bases of all my scopes. And using the Delayed Time Base method, I can calculate a random frequency from my Function Generator down to about 0.1%, by eye, and sometimes less. That’s good enough for me. When I tried to calibrate a scope with a regular function gen, I was Way off, and by a lot more.
FWIW


Tom Gardner
 

One prime use for scopes is to assess signal integrity, i.e. whether a receiver will correctly interpret an analogue voltage as a digital signal.

Prime prerequisites for that are solid triggering and the flatness of the frequency response. The correctness of the phase response is usually assured by the scope's design in conjunction with the correct frequency response.

A scope's time domain accuracy is trivial to measure nowadays. The amplitude accuracy is less important than the flatness.

On 14/07/20 16:58, Stephen wrote:
The time base is IMHO, the most important thing that has to be accurate in a scope. You don’t use it to accurately measure amplitude, you have a DMM for that. So I personally use a TEK 184 to calibrate the time bases of all my scopes. And using the Delayed Time Base method, I can calculate a random frequency from my Function Generator down to about 0.1%, by eye, and sometimes less. That’s good enough for me. When I tried to calibrate a scope with a regular function gen, I was Way off, and by a lot more.


Siggi
 

On Tue, Jul 14, 2020 at 10:59 AM Dave Daniel <kc0wjn@...> wrote:

Yes. That’s my point.

Suppose one uses a Tektronix SG-504 to calibrate a ‘scope. But how does
one know that the SG-504 is properly calibrated?
SG504 AFAIK is used only to validate the flatness of a scope's frequency
response and its bandwidth.
It isn't used for amplitude or timing calibration.
As such, the only requirement on the SG-504 is to be amplitude STABLE
across frequency (for relatively short durations of time). This in turn is
ensured mostly by design and construction, where it has a temperature
compensated amplitude control loop that extends practically all the way out
to the output connector. This eliminates the major problems with cable
losses and reflections.

So, taking the amplitude out of the equation, all you need to calibrate is
the frequency scale, and accurate-enough timing is easy enough to come by.


 

Weighing in here - the most common issue that you'll hit is that the flatness of the levelled sine wave generators is specified to a much higher standard than the typical levelled sig-gen.

For example the SG504 with levelling head accuracy specification is well down in the sub +/- 0.1dB area (within 3%) whereas a typical RF sig-gen is +/- 1dB

David

-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of David Berlind
Sent: 14 July 2020 15:14
To: TekScopes@groups.io
Subject: [TekScopes] Calibrators vs Signal/Function Generators

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David


Roy Thistle
 

On Tue, Jul 14, 2020 at 08:58 AM, Stephen wrote:


you have a DMM for that
at 219.5 MHz?


Stephen
 

On Tue, Jul 14, 2020 at 06:54 AM, Roy Thistle wrote:


On Tue, Jul 14, 2020 at 08:58 AM, Stephen wrote:


you have a DMM for that
at 219.5 MHz?
Up to 219.4 MHz only. Sorry, I should’ve been more specific.


 

The differences between calibrators and generators are profound.

1. the calibrator is defined in terms of steps that are *exactly the needed intervals for calibration.*
2. values are *peak to peak* on calibrators, *rms* in generators, a huge issue.
3. most generators cannot easily be set to the required intervals over the huge dynamic range needed (mV to V), or to the precise time intervals (uS to S).
4. calibrators do ONLY the required task, nothing else.

low end generators do not even have calibrated output voltage (sometimes not even any markings, just a knob), and may have only dial frequency calibration. if you are lucky enough to have a higher end digital generator you may have a better grip on the time and frequency values, but considering the  1dB or worse amplitude cal of even high end hp generators, and the need for proper termination, it's quite easy to make some pretty big errors in scope calibration. also, generators are general instruments, and are not really optimized for any task, and have divergent performance in many settings, but adequate for their use. calibrators, do only a specific task, and often have design implicit accuracy that can be set instrument wide with only one or two adjustments.

the argument about ultimate calibration is a bit circular, literally *NOBODY* has truly accurate items at their disposal, only *adequate* ones. I am lucky enough to have some very end cal equipment acquired over many years, AND I send it to a good cal lab, which *in turn* sends their gear out annually to a national accredited lab for cal. but as any metrology guy will tell you, there always remains some uncertainty in all measurements, even with known good standards. I can be confident my work is good enough for the task at hand, but only because I have tools with performance far beyond the required need for the work, *AND* I regularly cross-check what I am doing between different standards

scope cal is a modest undertaking, and 1% is adequate literally for any function, in the world of calibration, that is NOTHING in terms of precision, where the real discussion is in terms of ppm (parts per million) or better.

to be clear, if you have a source of known good time and voltage, and you know the impact of the terminating impedance, you can cal a scope, but it may be excruciatingly painful with a generator, and quick and easy with a dedicated calibrator. on something like a 2465, you really can't fool around, you need exactly the right tools, or you are just damaging the scope to attempt a "cal". do not go there.

anyway, just my $0.02 worth, you can get my industry articles on calibration from our site on this page:
https://www.sphere.bc.ca/test/data.html
look in the yellow bar at the left, under the AEA logo.

all the best,
walter

--
Walter Shawlee 2
Sphere Research Corp. 3394 Sunnyside Rd.
West Kelowna, BC, V1Z 2V4 CANADA
Phone: +1 (250-769-1834 -:- http://www.sphere.bc.ca
+We're all in one boat, no matter how it looks to you. (WS2)
+All you need is love. (John Lennon)
+But, that doesn't mean other things don't come in handy. (WS2)
+Nature is trying very hard to make us succeed, but nature does not depend on us.
We are not the only experiment. (R. Buckminster Fuller)


 

Oops 3% isn't quite sub 0.1dB, but definitely small at 0.2567dB

David

-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of David C. Partridge
Sent: 14 July 2020 18:17
To: TekScopes@groups.io
Subject: Re: [TekScopes] Calibrators vs Signal/Function Generators

Weighing in here - the most common issue that you'll hit is that the flatness of the levelled sine wave generators is specified to a much higher standard than the typical levelled sig-gen.

For example the SG504 with levelling head accuracy specification is well down in the sub +/- 0.1dB area (within 3%) whereas a typical RF sig-gen is +/- 1dB

David
-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of David Berlind
Sent: 14 July 2020 15:14
To: TekScopes@groups.io
Subject: [TekScopes] Calibrators vs Signal/Function Generators

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David


Chuck Harris
 

When you are tweaking the 2465's vertical amplifier networks
to perfect the transient response, you can get into an area
where the pulse looks very nice and flat, with few ripples,
but, if you do a quick check on frequency response, you
will see that it is below specification by quite a bit...
Instead of 350MHz, it is more like almost 300MHz.

You can tell this is going to be true by other observations,
but the SG504 gives you a nice qualitative picture with a
couple of cranks of the dial.

You don't need to look at it all that hard, but if you set the
SG504 to 6 divisions p-p at 6MHz, and still get between 4.4
and 6 divisions in the range from 350MHz to 420MHz, you have
achieved the desired transient response... and not a false
response.

[You can also tell the false response because the leading edge
is thicker for the first division below the corner... indicating
that the rise time is slower there than it has to be. You want
that thick edge not to roll below about 0.1-0.2 divisions.]

The ultimate accuracy of the SG isn't all that important,
and indeed, I used a S-D synthesizer before I got my SG504, and
racing the two instruments shows no indication that the
synthesizer was doing a substandard job. Just that the SG makes
the test easier.

-Chuck Harris

David C. Partridge wrote:

Oops 3% isn't quite sub 0.1dB, but definitely small at 0.2567dB

David
-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of David C. Partridge
Sent: 14 July 2020 18:17
To: TekScopes@groups.io
Subject: Re: [TekScopes] Calibrators vs Signal/Function Generators

Weighing in here - the most common issue that you'll hit is that the flatness of the levelled sine wave generators is specified to a much higher standard than the typical levelled sig-gen.

For example the SG504 with levelling head accuracy specification is well down in the sub +/- 0.1dB area (within 3%) whereas a typical RF sig-gen is +/- 1dB

David
-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of David Berlind
Sent: 14 July 2020 15:14
To: TekScopes@groups.io
Subject: [TekScopes] Calibrators vs Signal/Function Generators

Question for the hive:

I have a pile of scopes here that I'll be looking to calibrate at some point as a part of their resurrection. I see that the primary function of some Tek calibration fixtures is to generate square and sine waves. How is this significantly different from what a well-calibrated signal/function generator can do?

Thank you.

David










scm@...
 

In bringing an SG503 back to life, I observed a significant difference in the positive going amplitudes and the negative going amplitudes at many frequencies. The reason for this is a significant amount of second (and, possibly other even) harmonic content in the output. This difference is not present in my 067-0532-01 (65-500 MHz, based on a GR Unit Oscillator, which generates a much purer sine wave). Since the SG503 leveling circuit regulates the peak-to-peak amplitude of the output, this is of no consequence for calibrating 'scopes since we're interested in the amplitude and not the RMS value.


Dave Daniel
 

So, did you want to hijack this thread or start a new one?

On Jul 14, 2020, at 18:11, scm@... wrote:

In bringing an SG503 back to life, I observed a significant difference in the positive going amplitudes and the negative going amplitudes at many frequencies. The reason for this is a significant amount of second (and, possibly other even) harmonic content in the output. This difference is not present in my 067-0532-01 (65-500 MHz, based on a GR Unit Oscillator, which generates a much purer sine wave). Since the SG503 leveling circuit regulates the peak-to-peak amplitude of the output, this is of no consequence for calibrating 'scopes since we're interested in the amplitude and not the RMS value.