Lincoln said that it was "Better to remain silent and be thought a fool than to open one's mouth and remove all doubt." OK, here we go.. In my quest to understand and make good use of the above mentioned counter, I've taken what I believe to be a residual measurement.  That is, taking the ref out and measuring it with itself.  I get different answers depending on the sample interval.  Is that what I should expect?  If I were a mathamagitian, I suppose it would be obvious from the ADEV equation, but I can't see it.  Am I setting up the counter incorrectly? The data sheet lists ADEV at 1s, which is 0.01ppb ( I have option 10 - high stability OCXO).  Not sure how to equate that to the ADEV plot. (1E-11?) And why do they only give that one point? Maybe I should have been a "volt nut".  The math would be easier! -- Jim

Hi This actually splits into two rabbit holes very quickly: The ADEV side is probably the easy one. If you look at the counter as a device with X.X ps of error on each reading. The longer the reading, the less that jitter is relative to the measure. If you have some sort of averaging on, the more samples you average, the more noise you throw out. The other rabbit hole involves the 5323x counters in general. Their ref out ports have issues. One could be a bit more descriptive, but this is a public list. Simple answer is not to use the ref out for anything important. As long as you do that, they work pretty well. Don’t do that and you will be dealing with nutty issues all over the place. On a different topic (we never stay on topic for long), the 5323x devices very much like a fast slew rate at the input. If you drive them with a fast edge, they perform quite a bit better than the spec sheet suggests. Keysight has a video up demonstrating this on YouTube. If anything, the video understates the level of improvement possible. https://www.youtube.com/watch?v=4DfLkxVe7Lk Keysight 53230A Single Shot Resolution Demo youtube.com So, it’s not a horrible counter. You just want to avoid that ref out. Bob > On Jan 19, 2023, at 9:51 AM, Jim Muehlberg via time-nuts <time-nuts@lists.febo.com> wrote: > > Lincoln said that it was "Better to remain silent and be thought a fool than to open one's mouth and remove all doubt." > > OK, here we go.. > > In my quest to understand and make good use of the above mentioned counter, I've taken what I believe to be a residual measurement. That is, taking the ref out and measuring it with itself. I get different answers depending on the sample interval. Is that what I should expect? If I were a mathamagitian, I suppose it would be obvious from the ADEV equation, but I can't see it. Am I setting up the counter incorrectly? > > The data sheet lists ADEV at 1s, which is 0.01ppb ( I have option 10 - high stability OCXO). Not sure how to equate that to the ADEV plot. (1E-11?) And why do they only give that one point? > > Maybe I should have been a "volt nut". The math would be easier! > > -- > Jim<53230A residual.png>_______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com

On some counters (I know of at least one) the ADEV "noise floor" differs if you are exactly "at" the internal reference versus being on a slightly different frequency. So a better test is to use an external reference, a fairly stable oscillator, and send it to both the ref input and the counter input. If you can, you should offset the external reference to the internal reference at least 0.1 Hz (easy to test by temporarily not connecting to the ref input). This will cause a phase jump every at most 10 seconds if there is any leakage from the internal reference into the PLL generating the internal high frequency reference when using an external reference. One extra problem is that the linear regression used in some counters may produce overly optimistic stability if the slope of the regression is almost zero Therefore it makes sense  to also do this test using  a signal generator also running from the same external reference at a frequency that is guaranteed not to have a simple mul/div relation with the external reference Erik. On 19-1-2023 17:10, Bob Camp via time-nuts wrote: > Hi > > This actually splits into two rabbit holes very quickly: > > The ADEV side is probably the easy one. If you look at the counter as a device with X.X ps of > error on each reading. The longer the reading, the less that jitter is relative to the measure. If > you have some sort of averaging on, the more samples you average, the more noise you throw > out. > > The other rabbit hole involves the 5323x counters in general. Their ref out ports have issues. > One could be a bit more descriptive, but this is a public list. Simple answer is not to use the > ref out for anything important. As long as you do that, they work pretty well. Don’t do that and > you will be dealing with nutty issues all over the place. > > On a different topic (we never stay on topic for long), the 5323x devices very much like a fast > slew rate at the input. If you drive them with a fast edge, they perform quite a bit better than > the spec sheet suggests. Keysight has a video up demonstrating this on YouTube. If anything, > the video understates the level of improvement possible. > > https://www.youtube.com/watch?v=4DfLkxVe7Lk > Keysight 53230A Single Shot Resolution Demo > youtube.com > > So, it’s not a horrible counter. You just want to avoid that ref out. > > Bob > >> On Jan 19, 2023, at 9:51 AM, Jim Muehlberg via time-nuts <time-nuts@lists.febo.com> wrote: >> >> Lincoln said that it was "Better to remain silent and be thought a fool than to open one's mouth and remove all doubt." >> >> OK, here we go.. >> >> In my quest to understand and make good use of the above mentioned counter, I've taken what I believe to be a residual measurement. That is, taking the ref out and measuring it with itself. I get different answers depending on the sample interval. Is that what I should expect? If I were a mathamagitian, I suppose it would be obvious from the ADEV equation, but I can't see it. Am I setting up the counter incorrectly? >> >> The data sheet lists ADEV at 1s, which is 0.01ppb ( I have option 10 - high stability OCXO). Not sure how to equate that to the ADEV plot. (1E-11?) And why do they only give that one point? >> >> Maybe I should have been a "volt nut". The math would be easier! >> >> -- >> Jim<53230A residual.png>_______________________________________________ >> time-nuts mailing list -- time-nuts@lists.febo.com >> To unsubscribe send an email to time-nuts-leave@lists.febo.com > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com

Hi The 531xx and 532xx counters in frequency average (=enhanced resolution) mode both suffer from the “right on frequency” effect. Probably the worst example is a SR620 counter. They do very odd things error wise when hooked to their internal reference. The issue with the Ref Out on the 532xx is a whole different thing. You can feed the signal into an analyzer (of whatever sort) and see that it’s been messed up. Normally the Ref Out and Ref In on this sort of device stays somewhere near same / same. Not in this case ….. Bob > On Jan 20, 2023, at 9:29 AM, Erik Kaashoek via time-nuts <time-nuts@lists.febo.com> wrote: > > On some counters (I know of at least one) the ADEV "noise floor" differs if you are exactly "at" the internal reference versus being on a slightly different frequency. > So a better test is to use an external reference, a fairly stable oscillator, and send it to both the ref input and the counter input. > If you can, you should offset the external reference to the internal reference at least 0.1 Hz (easy to test by temporarily not connecting to the ref input). This will cause a phase jump every at most 10 seconds if there is any leakage from the internal reference into the PLL generating the internal high frequency reference when using an external reference. > One extra problem is that the linear regression used in some counters may produce overly optimistic stability if the slope of the regression is almost zero > Therefore it makes sense to also do this test using a signal generator also running from the same external reference at a frequency that is guaranteed not to have a simple mul/div relation with the external reference > Erik. > > On 19-1-2023 17:10, Bob Camp via time-nuts wrote: >> Hi >> >> This actually splits into two rabbit holes very quickly: >> >> The ADEV side is probably the easy one. If you look at the counter as a device with X.X ps of >> error on each reading. The longer the reading, the less that jitter is relative to the measure. If >> you have some sort of averaging on, the more samples you average, the more noise you throw >> out. >> >> The other rabbit hole involves the 5323x counters in general. Their ref out ports have issues. >> One could be a bit more descriptive, but this is a public list. Simple answer is not to use the >> ref out for anything important. As long as you do that, they work pretty well. Don’t do that and >> you will be dealing with nutty issues all over the place. >> >> On a different topic (we never stay on topic for long), the 5323x devices very much like a fast >> slew rate at the input. If you drive them with a fast edge, they perform quite a bit better than >> the spec sheet suggests. Keysight has a video up demonstrating this on YouTube. If anything, >> the video understates the level of improvement possible. >> >> https://www.youtube.com/watch?v=4DfLkxVe7Lk >> Keysight 53230A Single Shot Resolution Demo >> youtube.com >> >> So, it’s not a horrible counter. You just want to avoid that ref out. >> >> Bob >> >>> On Jan 19, 2023, at 9:51 AM, Jim Muehlberg via time-nuts <time-nuts@lists.febo.com> wrote: >>> >>> Lincoln said that it was "Better to remain silent and be thought a fool than to open one's mouth and remove all doubt." >>> >>> OK, here we go.. >>> >>> In my quest to understand and make good use of the above mentioned counter, I've taken what I believe to be a residual measurement. That is, taking the ref out and measuring it with itself. I get different answers depending on the sample interval. Is that what I should expect? If I were a mathamagitian, I suppose it would be obvious from the ADEV equation, but I can't see it. Am I setting up the counter incorrectly? >>> >>> The data sheet lists ADEV at 1s, which is 0.01ppb ( I have option 10 - high stability OCXO). Not sure how to equate that to the ADEV plot. (1E-11?) And why do they only give that one point? >>> >>> Maybe I should have been a "volt nut". The math would be easier! >>> >>> -- >>> Jim<53230A residual.png>_______________________________________________ >>> time-nuts mailing list -- time-nuts@lists.febo.com >>> To unsubscribe send an email to time-nuts-leave@lists.febo.com >> _______________________________________________ >> time-nuts mailing list -- time-nuts@lists.febo.com >> To unsubscribe send an email to time-nuts-leave@lists.febo.com > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com