Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
On 28 August 2015 at 21:48, Dr. David Kirkby (Kirkby Microwave Ltd) < drkir...@kirkbymicrowave.co.uk> wrote: > My LCR meter came back from Keysight UK last week, where it was > calibrated. This instrument works at various frequencies from 20 Hz to 1 > MHz, so obviously has some sort of oscillator in it. But I don't think the > absolute accuracy on frequency is important on this, as it does not even > have the ability to set to an arbitrary frequency. There are only 8000 or > so steps, and at the high end, some of those steps are more than 100 kHz > apart!!! So clearly frequency accuracy on this instrument is not that > important. > > Anyway, the cal certificate, a copy of which I put here > > > http://www.kirkbymicrowave.co.uk/cal_certificates/Keysight-standard-calibration-with-uncertainties-for-4284A-precison-LCR-meter-18-08-2015.pdf > > shows on page 5 that it was checked at 1, 8, 20, 80, 400 kHz, and 1 MHz. > But the uncertainty reported (7.6 Hz) seems extremely high, given they used > a 53132A counter as a working standard, and a 5071A primary frequency > standard. Why should the uncertainty be so high? Am I missing something? > I don't know if this thread is closed, but I did get a telephone call from someone at Keysight today, in response to my query on this matter. Essentially 1) At the particular calibration centre in the UK, the frequency counter used would have been locked to an external reference, as are all the counters they use at that particular calibration centre in the UK. The reference is an old HP 5071 cesium, which is regularly compared with GPS. Their best uncertainty is 5 x 10^-13. 2) The LCR meter tolerance on the 1 MHz is +/- 100 Hz. This means it does not need a particularly good counter. 3) To enable Keysight to use one calibration procedure world-wide, that calibration procedure is written to allow the uncertainty to be as high as the worst case with a counter running on its own internal oven. So in essence, the meter when calibrated in the UK would have been measured to far better than the +/- 7.6 Hz indicated on the calibration certificate. but by specifying +/- 7.6 Hz, it allows the same procedure to be used anywhere in the world. Dave ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
Look at the service manual, the calibration setup listed is usually pretty close to what Keysight is using in their local calibration facility. For an LCR meter I doubt they would be using a time nuts grade counter Content by Scott Typos by Siri On Aug 29, 2015, at 10:02 PM, Bob Camp kb...@n1k.org wrote: Hi Well, since we all have made our totally uninformed guesses, the only thing to do now is to give Keysight a call and see what the real answer is. Bob On Aug 29, 2015, at 1:46 PM, Dr. David Kirkby (Kirkby Microwave Ltd) drkir...@kirkbymicrowave.co.uk wrote: On 29 August 2015 at 12:59, Javier Herrero jherr...@hvsistemas.es wrote: Hello, The calibration certificate does not indicate that the measurements were done with the frequency counters referenced to the 5071A at the time of calbiration (if so, it would be listed under the Calibration Equipment Used table). It says that the 53132A were calibrated against the 5071A. If for your calibration they have used 53132A witout the oven oscillator option it is very probable that its uncertainity is 7.6ppm as indicated in the certificate. Since the maximum error tolerable for the LCR meter is 100ppm (+/-100Hz @ 1MHz), it makes sense to perform the measurement with an instrument with an uncertainity of 7.6ppm, and not to use the better counter in the lab for that purpose. Regards, Javier I would find it a bit hard to believe they would use a counter without an oven in their lab, as it would seriously restrict what they can do with it, making it more difficult to replace one counter with another. I would have thought that within reason it best to have the lab have reasonably high spec kit, so more than one instrument could be done on the same line. They did for example use a pair of 3458As, despite I'm sure the voltage accuracy requirements could be met with a multimeter with far greater uncertainty than an expensive 3458A. It makes more sense (within reason) to have 3458As in the cal lab, as it allows a wider range of instruments to be calibrated. Also, if you consider the spec on the 53132A without an oven, it is 3 x 10^-7 per month. So after 12 months that could be 12 * 3 10^-7 or 3.6 10^-6, so if it did drift the maximum amount each month for a year, the uncertainty would higher than it actually is. I intended to contact Keysight about the calibration for a couple of other reasons 1) I would like to know if it was adjusted or not. That is not clear from the cal certificate, since the * As received condition - Not applicable, as this calibration certificate applies to the initial calibration of a new, refurbished or upgraded equipment. * Action taken - The equipment was upgraded. I doubt it has seen a cal lab in ages. The upgrade was just a software one, to enable cable lengths of 2 m and 4 m (option 006) to be used to connect the DUT, which they kindly provided free of charge, on the condition I paid for the calibration. 2) They never put any stickers over the screws that prevent the covers being removed, which struck me as a bit odd. Since I was going to ask about those two issues, I will ask about the uncertainty on frequency too. It will be interesting what response I get. I'm just interested - I realize that this instrument does not demand much of the counter used to calibrate it. The demanding calibration devices would be the resistance and capacitance standards. Dave ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
On 28 Aug 2015 23:05, Oz-in-DFW li...@ozindfw.net wrote: The uncertainly listed seems to be 7.6 mHz (milliHertz, or .0076 Hz. A bit better that you mention.. No, please look again. The first line does show an uncertainty of 7.6 mHz, but that is when the LCR meter was set to 1 kHz. The last line shows an uncertainty of 7.6 Hz (1000x higher) when measuring 1 MHz, which is obviously 1000x higher in frequency than 1 kHz. So the original uncertainty I quoted was correct. The uncertainty rises proportionately with frequency. Dave. Dave. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
Hi, On 08/29/2015 10:01 AM, Dr. David Kirkby (Kirkby Microwave Ltd) wrote: On 28 Aug 2015 23:05, Oz-in-DFW li...@ozindfw.net wrote: The uncertainly listed seems to be 7.6 mHz (milliHertz, or .0076 Hz. A bit better that you mention.. No, please look again. The first line does show an uncertainty of 7.6 mHz, but that is when the LCR meter was set to 1 kHz. The last line shows an uncertainty of 7.6 Hz (1000x higher) when measuring 1 MHz, which is obviously 1000x higher in frequency than 1 kHz. So the original uncertainty I quoted was correct. The uncertainty rises proportionately with frequency. I agree. That's how I read it too, and it is very obvious as you look at the table. The calibration is in line with Chapter 10 of the manual: http://wiki.epfl.ch/carplat/documents/hp4284a_lcr_manual.pdf It might be that the 7.6 ppm number was a practical uncertainty measure found in deeper analysis along the lines of GUM and was added to the calibration chart afterwards. For the peak-error of 100 ppm, this provide a good margin anyway. So, it's the instrument and the way it behaves, nothing in their test-setup which is way better than needed for this particular instrument. In short, don't worry about it, it's OK. Cheers, Magnus ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
Hello, The calibration certificate does not indicate that the measurements were done with the frequency counters referenced to the 5071A at the time of calbiration (if so, it would be listed under the Calibration Equipment Used table). It says that the 53132A were calibrated against the 5071A. If for your calibration they have used 53132A witout the oven oscillator option it is very probable that its uncertainity is 7.6ppm as indicated in the certificate. Since the maximum error tolerable for the LCR meter is 100ppm (+/-100Hz @ 1MHz), it makes sense to perform the measurement with an instrument with an uncertainity of 7.6ppm, and not to use the better counter in the lab for that purpose. Regards, Javier On 28/08/2015 22:48, Dr. David Kirkby (Kirkby Microwave Ltd) wrote: My LCR meter came back from Keysight UK last week, where it was calibrated. This instrument works at various frequencies from 20 Hz to 1 MHz, so obviously has some sort of oscillator in it. But I don't think the absolute accuracy on frequency is important on this, as it does not even have the ability to set to an arbitrary frequency. There are only 8000 or so steps, and at the high end, some of those steps are more than 100 kHz apart!!! So clearly frequency accuracy on this instrument is not that important. Anyway, the cal certificate, a copy of which I put here http://www.kirkbymicrowave.co.uk/cal_certificates/Keysight-standard-calibration-with-uncertainties-for-4284A-precison-LCR-meter-18-08-2015.pdf shows on page 5 that it was checked at 1, 8, 20, 80, 400 kHz, and 1 MHz. But the uncertainty reported (7.6 Hz) seems extremely high, given they used a 53132A counter as a working standard, and a 5071A primary frequency standard. Why should the uncertainty be so high? Am I missing something? When they done my VNA last year http://www.kirkbymicrowave.co.uk/cal_certificates/Keysight-standard-calibration-with-uncertainties-for-8720D-vector-network-analyzer-16-09-2014.pdf the uncertainty on frequency was about 5 orders of magnitude better than that. The 10 MHz timebase was measured with an uncertainty of 0.0010 Hz. I checked the Keysight UK accreditation (by UKAS) for frequency http://www.keysight.com/upload/cmc_upload/All/UKAS_S_2015-08-14_Eng.pdf and see over the range 0.1 Hz to 500 MHz, which covers the LCR meter, their accreditation is 6.0 in 10^11 + 0.020 nHz. I can't believe they are unable to measure better than 7.6 ppm on frequency, so are wondering why the uncertainty is so high, even though I am sure such an uncertainly is very acceptable for this application. It is either an error on the cal certificate, or I am missing something. I expect it is the latter, and hoping someone here can fill me in. Dave ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
Hi On Aug 29, 2015, at 7:09 AM, Magnus Danielson mag...@rubidium.dyndns.org wrote: Hi, On 08/29/2015 10:01 AM, Dr. David Kirkby (Kirkby Microwave Ltd) wrote: On 28 Aug 2015 23:05, Oz-in-DFW li...@ozindfw.net wrote: The uncertainly listed seems to be 7.6 mHz (milliHertz, or .0076 Hz. A bit better that you mention.. No, please look again. The first line does show an uncertainty of 7.6 mHz, but that is when the LCR meter was set to 1 kHz. The last line shows an uncertainty of 7.6 Hz (1000x higher) when measuring 1 MHz, which is obviously 1000x higher in frequency than 1 kHz. So the original uncertainty I quoted was correct. The uncertainty rises proportionately with frequency. I agree. That's how I read it too, and it is very obvious as you look at the table. The calibration is in line with Chapter 10 of the manual: http://wiki.epfl.ch/carplat/documents/hp4284a_lcr_manual.pdf It might be that the 7.6 ppm number was a practical uncertainty measure found in deeper analysis along the lines of GUM and was added to the calibration chart afterwards. For the peak-error of 100 ppm, this provide a good margin anyway. If it’s a +/- 100 ppm error and that’s all due to temperature, you get a linear “best guess” of 2 ppm / C. That would equate to needing to know the temperature to about 3.5 C. That’s doing pretty well in an open room. If you are guessing a temperature inside a piece of gear, that’s doing pretty well running inside a chamber. Bob So, it's the instrument and the way it behaves, nothing in their test-setup which is way better than needed for this particular instrument. In short, don't worry about it, it's OK. Cheers, Magnus ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
On 29 August 2015 at 12:59, Javier Herrero jherr...@hvsistemas.es wrote: Hello, The calibration certificate does not indicate that the measurements were done with the frequency counters referenced to the 5071A at the time of calbiration (if so, it would be listed under the Calibration Equipment Used table). It says that the 53132A were calibrated against the 5071A. If for your calibration they have used 53132A witout the oven oscillator option it is very probable that its uncertainity is 7.6ppm as indicated in the certificate. Since the maximum error tolerable for the LCR meter is 100ppm (+/-100Hz @ 1MHz), it makes sense to perform the measurement with an instrument with an uncertainity of 7.6ppm, and not to use the better counter in the lab for that purpose. Regards, Javier I would find it a bit hard to believe they would use a counter without an oven in their lab, as it would seriously restrict what they can do with it, making it more difficult to replace one counter with another. I would have thought that within reason it best to have the lab have reasonably high spec kit, so more than one instrument could be done on the same line. They did for example use a pair of 3458As, despite I'm sure the voltage accuracy requirements could be met with a multimeter with far greater uncertainty than an expensive 3458A. It makes more sense (within reason) to have 3458As in the cal lab, as it allows a wider range of instruments to be calibrated. Also, if you consider the spec on the 53132A without an oven, it is 3 x 10^-7 per month. So after 12 months that could be 12 * 3 10^-7 or 3.6 10^-6, so if it did drift the maximum amount each month for a year, the uncertainty would higher than it actually is. I intended to contact Keysight about the calibration for a couple of other reasons 1) I would like to know if it was adjusted or not. That is not clear from the cal certificate, since the * As received condition - Not applicable, as this calibration certificate applies to the initial calibration of a new, refurbished or upgraded equipment. * Action taken - The equipment was upgraded. I doubt it has seen a cal lab in ages. The upgrade was just a software one, to enable cable lengths of 2 m and 4 m (option 006) to be used to connect the DUT, which they kindly provided free of charge, on the condition I paid for the calibration. 2) They never put any stickers over the screws that prevent the covers being removed, which struck me as a bit odd. Since I was going to ask about those two issues, I will ask about the uncertainty on frequency too. It will be interesting what response I get. I'm just interested - I realize that this instrument does not demand much of the counter used to calibrate it. The demanding calibration devices would be the resistance and capacitance standards. Dave ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
Hi Well, since we all have made our totally uninformed guesses, the only thing to do now is to give Keysight a call and see what the real answer is. Bob On Aug 29, 2015, at 1:46 PM, Dr. David Kirkby (Kirkby Microwave Ltd) drkir...@kirkbymicrowave.co.uk wrote: On 29 August 2015 at 12:59, Javier Herrero jherr...@hvsistemas.es wrote: Hello, The calibration certificate does not indicate that the measurements were done with the frequency counters referenced to the 5071A at the time of calbiration (if so, it would be listed under the Calibration Equipment Used table). It says that the 53132A were calibrated against the 5071A. If for your calibration they have used 53132A witout the oven oscillator option it is very probable that its uncertainity is 7.6ppm as indicated in the certificate. Since the maximum error tolerable for the LCR meter is 100ppm (+/-100Hz @ 1MHz), it makes sense to perform the measurement with an instrument with an uncertainity of 7.6ppm, and not to use the better counter in the lab for that purpose. Regards, Javier I would find it a bit hard to believe they would use a counter without an oven in their lab, as it would seriously restrict what they can do with it, making it more difficult to replace one counter with another. I would have thought that within reason it best to have the lab have reasonably high spec kit, so more than one instrument could be done on the same line. They did for example use a pair of 3458As, despite I'm sure the voltage accuracy requirements could be met with a multimeter with far greater uncertainty than an expensive 3458A. It makes more sense (within reason) to have 3458As in the cal lab, as it allows a wider range of instruments to be calibrated. Also, if you consider the spec on the 53132A without an oven, it is 3 x 10^-7 per month. So after 12 months that could be 12 * 3 10^-7 or 3.6 10^-6, so if it did drift the maximum amount each month for a year, the uncertainty would higher than it actually is. I intended to contact Keysight about the calibration for a couple of other reasons 1) I would like to know if it was adjusted or not. That is not clear from the cal certificate, since the * As received condition - Not applicable, as this calibration certificate applies to the initial calibration of a new, refurbished or upgraded equipment. * Action taken - The equipment was upgraded. I doubt it has seen a cal lab in ages. The upgrade was just a software one, to enable cable lengths of 2 m and 4 m (option 006) to be used to connect the DUT, which they kindly provided free of charge, on the condition I paid for the calibration. 2) They never put any stickers over the screws that prevent the covers being removed, which struck me as a bit odd. Since I was going to ask about those two issues, I will ask about the uncertainty on frequency too. It will be interesting what response I get. I'm just interested - I realize that this instrument does not demand much of the counter used to calibrate it. The demanding calibration devices would be the resistance and capacitance standards. Dave ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
The uncertainly listed seems to be 7.6 mHz (milliHertz, or .0076 Hz. A bit better that you mention.. On 8/28/2015 3:48 PM, Dr. David Kirkby (Kirkby Microwave Ltd) wrote: My LCR meter came back from Keysight UK last week, where it was calibrated. This instrument works at various frequencies from 20 Hz to 1 MHz, so obviously has some sort of oscillator in it. But I don't think the absolute accuracy on frequency is important on this, as it does not even have the ability to set to an arbitrary frequency. There are only 8000 or so steps, and at the high end, some of those steps are more than 100 kHz apart!!! So clearly frequency accuracy on this instrument is not that important. Anyway, the cal certificate, a copy of which I put here http://www.kirkbymicrowave.co.uk/cal_certificates/Keysight-standard-calibration-with-uncertainties-for-4284A-precison-LCR-meter-18-08-2015.pdf shows on page 5 that it was checked at 1, 8, 20, 80, 400 kHz, and 1 MHz. But the uncertainty reported (7.6 Hz) seems extremely high, given they used a 53132A counter as a working standard, and a 5071A primary frequency standard. Why should the uncertainty be so high? Am I missing something? When they done my VNA last year http://www.kirkbymicrowave.co.uk/cal_certificates/Keysight-standard-calibration-with-uncertainties-for-8720D-vector-network-analyzer-16-09-2014.pdf the uncertainty on frequency was about 5 orders of magnitude better than that. The 10 MHz timebase was measured with an uncertainty of 0.0010 Hz. I checked the Keysight UK accreditation (by UKAS) for frequency http://www.keysight.com/upload/cmc_upload/All/UKAS_S_2015-08-14_Eng.pdf and see over the range 0.1 Hz to 500 MHz, which covers the LCR meter, their accreditation is 6.0 in 10^11 + 0.020 nHz. I can't believe they are unable to measure better than 7.6 ppm on frequency, so are wondering why the uncertainty is so high, even though I am sure such an uncertainly is very acceptable for this application. It is either an error on the cal certificate, or I am missing something. I expect it is the latter, and hoping someone here can fill me in. Dave ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. -- mailto:o...@ozindfw.net Oz POB 93167 Southlake, TX 76092 (Near DFW Airport) ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Why would Keysight UK uncertainty measuring 1 MHz be as high as 7.6 Hz?
Hi If it is an un-compensated crystal oscillator (I’d bet it is) there will be a maximum temperature slope to the part. In order to have absolute calibration, you would need to know both the frequency read *and* the temperature of the inside of the device. I’d bet somebody did the math, looked at the “rated” temperature control in the cal lab area and came up with a proper number for the ambiguity of the calibration. If the slope is 2 ppm / C then the number you have implies a 3.5C temperature control in the area. That seems pretty tight. A 1 ppm / C number gets you to 7.6 C. That seems about right for the room plus the ambiguity in the internal temperature rise inside the box. Bob On Aug 28, 2015, at 4:48 PM, Dr. David Kirkby (Kirkby Microwave Ltd) drkir...@kirkbymicrowave.co.uk wrote: My LCR meter came back from Keysight UK last week, where it was calibrated. This instrument works at various frequencies from 20 Hz to 1 MHz, so obviously has some sort of oscillator in it. But I don't think the absolute accuracy on frequency is important on this, as it does not even have the ability to set to an arbitrary frequency. There are only 8000 or so steps, and at the high end, some of those steps are more than 100 kHz apart!!! So clearly frequency accuracy on this instrument is not that important. Anyway, the cal certificate, a copy of which I put here http://www.kirkbymicrowave.co.uk/cal_certificates/Keysight-standard-calibration-with-uncertainties-for-4284A-precison-LCR-meter-18-08-2015.pdf shows on page 5 that it was checked at 1, 8, 20, 80, 400 kHz, and 1 MHz. But the uncertainty reported (7.6 Hz) seems extremely high, given they used a 53132A counter as a working standard, and a 5071A primary frequency standard. Why should the uncertainty be so high? Am I missing something? When they done my VNA last year http://www.kirkbymicrowave.co.uk/cal_certificates/Keysight-standard-calibration-with-uncertainties-for-8720D-vector-network-analyzer-16-09-2014.pdf the uncertainty on frequency was about 5 orders of magnitude better than that. The 10 MHz timebase was measured with an uncertainty of 0.0010 Hz. I checked the Keysight UK accreditation (by UKAS) for frequency http://www.keysight.com/upload/cmc_upload/All/UKAS_S_2015-08-14_Eng.pdf and see over the range 0.1 Hz to 500 MHz, which covers the LCR meter, their accreditation is 6.0 in 10^11 + 0.020 nHz. I can't believe they are unable to measure better than 7.6 ppm on frequency, so are wondering why the uncertainty is so high, even though I am sure such an uncertainly is very acceptable for this application. It is either an error on the cal certificate, or I am missing something. I expect it is the latter, and hoping someone here can fill me in. Dave ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.