Author Topic: Frequency measurement question?  (Read 1372 times)

0 Members and 1 Guest are viewing this topic.

Offline Zenwizard

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Frequency measurement question?
« on: July 07, 2019, 02:53:06 pm »
I am repairing a Tek type 184 and a TG 501. I need to measure and internal clocks to make sure they are in spec before I Cal the time bases on a few 7000 series scopes. My issue is I do not have a known good frequency counter in the lab yet. I have a Rigol scope and its internal counter says they are in spec and bang on. However this does not have enough zeros in the counter for my liking. My issue now is since I don’t have a frequency standard in the lab either to check in a counter I think I have 2 options one is acquire a new frequency counter with a current fresh Cal and trust it as known good. Or get a counter that is working and hook it up to an external reference like a GPS disciplined oscillator at 10 Mhz Is this correct? I am new to high precision frequency measurements but trying to really check in the bench and cal the scopes as they go back on. I would send them off to Tek but Tek wants $1000.00 to Cal a 7000 series mainframe OUCH
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 6232
  • Country: de
Re: Frequency measurement question?
« Reply #1 on: July 07, 2019, 03:23:46 pm »
One could use the series 7000 scope to measure a known frequency. I don't know the scope, how good it is to measure a frequency. However if it needs a very accurate frequency, it should be somehow capable do a good frequency measurement. So the check of the internal clock would be indirect. There could still be some jitter problem - but this is something that a normal counter would not notice either.
 

Online ArthurDent

  • Frequent Contributor
  • **
  • Posts: 875
  • Country: us
Re: Frequency measurement question?
« Reply #2 on: July 07, 2019, 03:29:24 pm »
Most GPSDOs will have a 10Mhz output far more stable and accurate than you would need and they are cheap. Triggering a scope from the GPS frequency then watching output from the device under test connected to the vertical input channel will display drift which is the difference in frequency between GPS and the device. After it has warmed up and stabilized you can adjust the device timebase for the lowest drift possible.  Knowing that the drift of one complete cycle at 10Mhz is equal to 100ns, you can time how long it takes the adjusted timebase to complete one cycle of drift and calculate the timebase exact frequency.

This counter is using GPS for the external timebase and checking the frequency of an Rb oscillator. Once you know you have an adjusted counter you'll feel much better about the readings you're taking. Of course you'll probably want a second and third counter to check the other ones, it is addictive. ^-^
 

Online Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1126
  • Country: us
    • The Messy Basement
Re: Frequency measurement question?
« Reply #3 on: July 07, 2019, 03:55:04 pm »
Don't know how good you need to be, but I cal my counters just by zero beating the oscillators against WWV with a shortwave. Put a 4" antenna on the BNC oscillator output, or maybe even shorter, then you can pick up both signals at the same time.
 

Offline thinkfat

  • Supporter
  • ****
  • Posts: 279
  • Country: de
    • Matthias' Hackerstübchen
Re: Frequency measurement question?
« Reply #4 on: July 07, 2019, 07:19:29 pm »


Most GPSDOs will have a 10Mhz output far more stable and accurate than you would need and they are cheap. Triggering a scope from the GPS frequency then watching output from the device under test connected to the vertical input channel will display drift which is the difference in frequency between GPS and the device.


Neat trick I learned just recently: If your counter can output its reference oscillator, connect it and the gpsdo output to different channels of an oscilloscope and switch to xy mode. Any drift will become visible immediately.

Gesendet von meinem Nokia 6.1 mit Tapatalk

 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 9902
  • Country: us
  • DavidH
Re: Frequency measurement question?
« Reply #5 on: July 08, 2019, 02:14:59 am »
Analog oscilloscope timebases are not accurate enough to require that accurate of a frequency counter for calibration.  I would be completely satisfied with 0.1% accuracy which is almost an order of magnitude better than an analog timebase and any frequency counter can meet that requirement unless it is broken.  The timebase in your Rigol is completely adequate for this.
 
The following users thanked this post: capt bullshot

Online ArthurDent

  • Frequent Contributor
  • **
  • Posts: 875
  • Country: us
Re: Frequency measurement question?
« Reply #6 on: July 08, 2019, 04:36:31 am »
Analog oscilloscope timebases are not accurate enough to require that accurate of a frequency counter for calibration.  I would be completely satisfied with 0.1% accuracy which is almost an order of magnitude better than an analog timebase and any frequency counter can meet that requirement unless it is broken.  The timebase in your Rigol is completely adequate for this.

If you're using any scope in X-Y mode you're not using the scope time base at all. Triggering a scope externally with a standard and watching drift of a oscillator to be calibrated on a vertical channel is comparing one external time base against another and any scope, analog or digital will work just fine. Using this method I can easily detect drift of a few nanoseconds over any length of time and the scope time base is immaterial. Therefor calibrating any scope time base is almost a waste of time. Just check it occasionally against your time mark generator every so often.

There are other ways of using a scope as a high accuracy indicator as well without worrying about the scopes time base.    http://www.to-way.com/tf/osccalib.pdf
« Last Edit: July 08, 2019, 04:45:32 am by ArthurDent »
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 6232
  • Country: de
Re: Frequency measurement question?
« Reply #7 on: July 08, 2019, 08:28:06 am »
....
 Therefor calibrating any scope time base is almost a waste of time. Just check it occasionally against your time mark generator every so often.

For an analog scope, calibration of the time base is no more than an occasional check against a mark generator.
It does make sense to check, as the time base is usually just set by RC pairs and X-gain, so it can very well drift over time.

For a digital scope with functions like frequency counting, it makes sense to have higher requirements. If resolution is sufficient one could use the scope itself and thus only needs a stable reference signal - this could be a strong local radio station. In the old days of analog TV the horizontal frequency was also good in some places.
 

Offline Zenwizard

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: Frequency measurement question?
« Reply #8 on: July 08, 2019, 10:56:12 pm »
The spec of the 7000 scopes direct off of cal is only 3% as they are 100% analog I do have a 100 Mhz bk counter but the cal spec for the 184 wants 2 more decades they the counter has. The reference clock in the 184 is 10 Mhz with +/- 100 Hz and a drift of 3 PPM. The TG501 is even tighter. But it's reference is only 1 Mhz. The counter says it is within 3 Hz so that one I think is good and after everything stabilizes the 184 is measuring about 30Hz low but still in spec I am tempted to leave sleeping pots be and not mess with them. Checking the pulses on the Rigol scope MSO 1104Z the pulses seem right where they should be on the screen at each major division. The 184 has a 2 hour warm up time due to the ovenised crystal.
 

Offline Zenwizard

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: Frequency measurement question?
« Reply #9 on: July 08, 2019, 10:57:41 pm »
Also to be clear I am trying to cal the time mark generators on the way to scope time bases.
 

Online Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1126
  • Country: us
    • The Messy Basement
Re: Frequency measurement question?
« Reply #10 on: July 09, 2019, 12:19:09 am »
A shortwave with an S meter will get you within maybe 1 Hz or better.
 

Offline tkamiya

  • Super Contributor
  • ***
  • Posts: 1007
  • Country: us
Re: Frequency measurement question?
« Reply #11 on: July 09, 2019, 01:10:29 am »
One (very) nice thing about GPSDO is it is self-calibrating.  It takes a bit to set it up but once done, it's a very trusty source of a standard signal.
 

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 9898
  • Country: gb
    • Having fun doing more, with less
Re: Frequency measurement question?
« Reply #12 on: July 09, 2019, 07:29:29 am »
I am repairing a Tek type 184 and a TG 501. I need to measure and internal clocks to make sure they are in spec before I Cal the time bases on a few 7000 series scopes. My issue is I do not have a known good frequency counter in the lab yet. I have a Rigol scope and its internal counter says they are in spec and bang on. However this does not have enough zeros in the counter for my liking. My issue now is since I don’t have a frequency standard in the lab either to check in a counter I think I have 2 options one is acquire a new frequency counter with a current fresh Cal and trust it as known good. Or get a counter that is working and hook it up to an external reference like a GPS disciplined oscillator at 10 Mhz Is this correct? I am new to high precision frequency measurements but trying to really check in the bench and cal the scopes as they go back on. I would send them off to Tek but Tek wants $1000.00 to Cal a 7000 series mainframe OUCH

You need to understand the accuracy you need for your job, using numbers not adjectives.

What's the 7000's timebase spec?
How far can the 184's crystal oscillator be pulled by the capacitor?
Compare the two and decide what is sufficient. I'll bet that if it works the 184 will be more than adequate.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 9902
  • Country: us
  • DavidH
Re: Frequency measurement question?
« Reply #13 on: July 10, 2019, 04:33:58 am »
You need to understand the accuracy you need for your job, using numbers not adjectives.

What's the 7000's timebase spec?
How far can the 184's crystal oscillator be pulled by the capacitor?
Compare the two and decide what is sufficient. I'll bet that if it works the 184 will be more than adequate.

With a 7B80/7B85, the worst case best is about 1.5%.  Calibration is assumed to be against a 0.25% source which not coincidentally is the accuracy specification of the built in calibrator; this leaves a lot of leeway for a modern calibration source.

Slideback measurements in delta delayed mode can be made to about 0.5% which is unusually accurate for an analog timebase; the improvement comes from removing all errors contributed by the CRT display and explains why this feature even exists.
« Last Edit: July 10, 2019, 04:39:00 am by David Hess »
 

Offline Zenwizard

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: Frequency measurement question?
« Reply #14 on: July 11, 2019, 02:23:35 am »
The 184 and the tg501 both "work" now for the 501 this means that it is at least passing the internal checks. The cal documents spec either time mark generator for the time bases calibration so either will work and from testing I am reasonable sure the 501 is at least in the ball park.

What I am ACTUALLY trying to measure is the internal clocks for the 184 and the 501. I need to make a 10Mhz measurement for the 184 this need to be 10 Mhz +/- 100 Hz with less then 3 PPM drift over time after a 2 hour warm up for the crystal. This measurement is taken with the unit in case warmed up at the correct time/div setting. The same is true for the TG 501. This ones needs a 1Mhz Measurement with an even tighter tolerance then the 184. However the 501 has self correcting circuits to get the drift down. As I am dealing with gear that has a birth date before mine I am just trying to check it and verify BEFORE I use them to cal some scopes.

My PROBLEM - I only have a BK precision 1803C that I cant get to last through the warm up time of the 184 as it has been converted to run on batteries. This frequency counter only goes to 100Mhz and the cal documents for the time mark generators want 2 to 3 decades more digits on the counter to check their reference frequencies.

A possible solution. I don't mind adding some gear to the lab and a good frequency counter would not be a bad thing. But if I hit the used market how do I trust the cal on the counter is correct. And can I get around this issue with a GPSDO either as a sanity check or as an external reference for the counter so I can trust what it is telling me when I measure the clocks of the 184 an 501?
 

Online Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1126
  • Country: us
    • The Messy Basement
Re: Frequency measurement question?
« Reply #15 on: July 11, 2019, 03:15:06 am »
IMO, you can't trust any frequency counter you buy, until you calibrate it. So that gives you the same problem you've already got. GPS is great, but it's not fast and easy unless you spend some money or want a new DIY project. You appear to be in the USA; why don't you want to use WWV at 10 MHz? It's way better than you need. WWVB at 60 kHz is probably better, but you have to convert it to a useful frequency.
 
The following users thanked this post: kj7e

Offline tggzzz

  • Super Contributor
  • ***
  • Posts: 9898
  • Country: gb
    • Having fun doing more, with less
Re: Frequency measurement question?
« Reply #16 on: July 11, 2019, 07:17:00 am »
What I am ACTUALLY trying to measure is the internal clocks for the 184 and the 501. I need to make a 10Mhz measurement for the 184 this need to be 10 Mhz +/- 100 Hz with less then 3 PPM drift over time after a 2 hour warm up for the crystal.

We thought you were trying to calibrate your 7000. Has that changed?

Why are you trying to "trying to measure is the internal clocks for the 184"? All the clocks are derived from the crystal, so you only have to measure the frequency of one of them. Choose one frequency that is within the range of your existing counter, or compare it with a GPS or WWV etc.

But if you are only trying to calibrate a 7000, why do you need that precision in the first place?

My suspicion is that if the 184 is working (and its internal analogue frequency dividers can be pernickity!) then it will be sufficient to calibrate the 7000.

Yes, it can be fun to make sure the extra digits are correct, but it is easy to fall into the timenut rathole without good cause.
There are lies, damned lies, statistics - and ADC/DAC specs.
Glider pilot's aphorism: "there is no substitute for span". Retort: "There is a substitute: skill+imagination. But you can buy span".
Having fun doing more, with less
 

Offline Zenwizard

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: Frequency measurement question?
« Reply #17 on: July 11, 2019, 03:02:07 pm »
We thought you were trying to calibrate your 7000. Has that changed?
This is still the end goal yes I still have 3 scopes and 5 time bases that need checked and adjusted.

Why are you trying to "trying to measure is the internal clocks for the 184"?
Working through the service manual performance check section to make sure it is still in spec to be useful for calibrating the scopes.

But if you are only trying to calibrate a 7000, why do you need that precision in the first place?
I have to "check in" the 184 and the 501 both time mark generators just different generations the 501 is a TM500 plugin.

My suspicion is that if the 184 is working (and its internal analogue frequency dividers can be pernickity!) then it will be sufficient to calibrate the 7000.
This is exactly what I want to check it. It is out putting pulses sure but I have no clue if they are in time.
A 184 or 501 is part of the required equipment for the cal procedure of the 7000 scopes as noted in the service manual for the mainframes and the plugins. I am trying to make sure it IS working correctly.

Yes, it can be fun to make sure the extra digits are correct, but it is easy to fall into the timenut rathole without good cause.
Extra digits are speced and required by the service manual for the 184 and the 501. The different clock measurement is because the 184 has a 10Mhz reference clock and the 501 has a 1Mhz reference clock.
 

Offline Zenwizard

  • Regular Contributor
  • *
  • Posts: 54
  • Country: us
Re: Frequency measurement question?
« Reply #18 on: September 02, 2019, 01:52:02 pm »
Thanks for the help guys. What I ended up doing is getting an Agilent 53131A that I was surprised to find had the HS oven in it though it did not have the 3 Ghz add on board. I then fed that with the Bodner GPSDO at 10 Mhz as an external reference to the counter. Just using the internal counter it was within 2 Hz at 10 Mhz so the counter is even pretty close. The 501 was really close on its reference clock. However the 187 was a pain in the butt. The spec on that one was +- 100 Hz on a 10 Mhz base clock. Ok not to bad mine was off by about 300 Hz so needed adjustment. But you have to adjust it to an even swing around the 10 Mhz clock on mine that was +-15Hz as the oven was cycling. Super touchy and figgly adjustment. Also it uses a fast double pulse to make the time mark peaks that needs to be even relative to its self. But all good to go and did what I needed. So working well.

Thanks again.
 

Online SilverSolder

  • Frequent Contributor
  • **
  • Posts: 691
  • Country: 00
Re: Frequency measurement question?
« Reply #19 on: September 02, 2019, 09:16:20 pm »
IMO, you can't trust any frequency counter you buy, until you calibrate it. So that gives you the same problem you've already got. GPS is great, but it's not fast and easy unless you spend some money or want a new DIY project. You appear to be in the USA; why don't you want to use WWV at 10 MHz? It's way better than you need. WWVB at 60 kHz is probably better, but you have to convert it to a useful frequency.

Most younger people don't own a shortwave radio, would be my guess -  so it isn't as cheap and easy as all that, even if it does sound like a pretty accurate method!
 

Online Conrad Hoffman

  • Super Contributor
  • ***
  • Posts: 1126
  • Country: us
    • The Messy Basement
Re: Frequency measurement question?
« Reply #20 on: September 03, 2019, 12:10:42 am »
Quite true. You can find a very inexpensive radio that will pick up 10 MHz, but the price of a ready to go GPSDO isn't that much more.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf