Скачать презентацию 16 th July 2013 NCSLI 2013 Nashville 1 Скачать презентацию 16 th July 2013 NCSLI 2013 Nashville 1

bacdb5bd2387ee12411d495853193978.ppt

  • Количество слайдов: 36

16 th July 2013 NCSLI 2013, Nashville 1 16 th July 2013 NCSLI 2013, Nashville 1

The Design of a Calculable AC voltage reference using digital waveform generation 16 th The Design of a Calculable AC voltage reference using digital waveform generation 16 th July 2013 NCSLI 2013, Nashville 2

Abstract The design of an AC voltage reference source using a digital to analogue Abstract The design of an AC voltage reference source using a digital to analogue converter controlled by a microcontroller to produce a calculable RMS AC voltage reference with accuracy suitable for calibrating high performance Digital multimeters. 16 th July 2013 NCSLI 2013, Nashville 3

Technical Objectives • To provide a method for secondary and below laboratories with the Technical Objectives • To provide a method for secondary and below laboratories with the ability to generate high accuracy AC Voltages • Investigate errors associated with digitally generated AC Voltages, as well as the practical application of digitally generated waveforms in commercial calibration 16 th July 2013 NCSLI 2013, Nashville 4

Learning Objectives • To investigate alternatives to traditional AC voltage verification methods • To Learning Objectives • To investigate alternatives to traditional AC voltage verification methods • To enhance calibration laboratories understanding of verification of high performance modern DMM’s 16 th July 2013 NCSLI 2013, Nashville 5

The Requirement The calibration of Today’s Modern high performance Digital Meters which offer ACV The Requirement The calibration of Today’s Modern high performance Digital Meters which offer ACV Accuracy in the order of 80 -100 ppm To achieve a stand off ratio of just 4 to 1 requires an accuracy of around 20 ppm. This is not possible with a multi-product calibrator 16 th July 2013 NCSLI 2013, Nashville 6

The Solutions 1) Multi–Junction Thermal Transfer Standard 2) Calculable AC voltage reference 16 th The Solutions 1) Multi–Junction Thermal Transfer Standard 2) Calculable AC voltage reference 16 th July 2013 NCSLI 2013, Nashville 7

The Solutions : Multi-Junction Thermal Transfer Standard Advantages Disadvantages Wide Frequency Range Requires both The Solutions : Multi-Junction Thermal Transfer Standard Advantages Disadvantages Wide Frequency Range Requires both a stable DC and AC voltage source Proven Accuracy Equipment is expensive Calibration is costly, especially when a wide range of points is required Even with protection, older Thermal Transfer devices are easy to damage by accidently over-ranging 16 th July 2013 NCSLI 2013, Nashville 8

The Solutions : Calculable AC Voltage Reference Advantages Disadvantages Due to the nature of The Solutions : Calculable AC Voltage Reference Advantages Disadvantages Due to the nature of the device, Limited frequency range does not require an AC or DC source Instrument can be internally verified with a DC volt meter Output voltages are limited Due to solid state design, instrument is hard to damage with incorrect connections Low cost due to ‘simple’ hardware 16 th July 2013 NCSLI 2013, Nashville 9

The Concept of the Calculable waveform • Digitally create an AC waveform using a The Concept of the Calculable waveform • Digitally create an AC waveform using a digital to analogue (DAC) converter. In our experiments we chose a waveform based on 256 steps. • Single step the waveform to allow the level of each step to be measured as a DC level. 16 th July 2013 NCSLI 2013, Nashville 10

The Concept of the Calculable waveform The RMS value of the waveform is then The Concept of the Calculable waveform The RMS value of the waveform is then calculated using the following formula Where : Vrms = RMS Ac voltage output V = measured DC voltage of ‘step’ n = number of steps 16 th July 2013 NCSLI 2013, Nashville 11

History This is not a new idea. This technique was pioneered as long ago History This is not a new idea. This technique was pioneered as long ago as 1988 where results against Thermal converters gave uncertainties at the 7 volt level of just 5 ppm. 16 th July 2013 NCSLI 2013, Nashville 12

What has changed There is now a much greater need for high accuracy ACV. What has changed There is now a much greater need for high accuracy ACV. In 1988 we were only seeing the first developments in High accuracy AC DMM’s. Now almost every laboratory has a high performance DMM 16 th July 2013 NCSLI 2013, Nashville 13

What has changed Digital Electronics along with the rapid development of digital to analogue What has changed Digital Electronics along with the rapid development of digital to analogue converters for high quality audio has now made it much easier to implement this concept. 16 th July 2013 NCSLI 2013, Nashville 14

The Circuit The design uses a PIC micro controller, programmed with the waveform to The Circuit The design uses a PIC micro controller, programmed with the waveform to directly drive a DAC. The PIC is directly controlled from an RS 232 interface. Clock Control Interface 10 V Reference Input Micro Controller PIC 16 F 84 DAC AC Voltage Output 16 th July 2013 NCSLI 2013, Nashville 15

The DC Reference The D to A converter needs a short term stable 10 The DC Reference The D to A converter needs a short term stable 10 Volt DC reference. The Linear Technology LTZ 100 reference can easily provide better than 1 ppm stability over the short term 2/3 hours required. As the LTZ 1000 is temperature stabilised temperature variations over the measurement period will not add any significant contributions to an uncertainty budget 16 th July 2013 NCSLI 2013, Nashville 16

Improvements Vs Performance The circuit used is extremely simple. Early design’s used two converters, Improvements Vs Performance The circuit used is extremely simple. Early design’s used two converters, with the output being switched between them to remove glitches. With modern converters this is not necessary. The PIC micro controller directly drives the converter, loading the code into the converter directly from it’s memory, also clocking the converter after each data load. This very simple approach requires some machine code for the PIC to make it run as fast as possible. All timing comes from the crystal oscillator driving the PIC. 16 th July 2013 NCSLI 2013, Nashville 17

The Software The AC reference is very simple to control, with just a few The Software The AC reference is very simple to control, with just a few commands needed. S - 1 - 15 Hz 2 - 60 Hz 3 - 200 Hz 4 - 1000 Hz Stop waveform and set to zero X - Advance 1 step Y - Advance 10 steps Z - 16 th July 2013 Advance 64 steps NCSLI 2013, Nashville 18

Taking the DCV Measurements The DCV measurements can be taken with a high performance Taking the DCV Measurements The DCV measurements can be taken with a high performance 8 digit DMM. This will typically give full scale & linearity uncertainties of less than 5 ppm. As there are over 256 measurements to be made it is recommend that the process be automated by using a PC & software. 16 th July 2013 NCSLI 2013, Nashville 19

Measuring Software A procedure to measure each step was written using Procal. As each Measuring Software A procedure to measure each step was written using Procal. As each step is measured it is squared and added to a running total. The procedure being 256 steps long, takes about 40 minutes to run automatically. Although the measurement procedure could be written in any language Procal has the advantage to work with any DMM without changing code. 16 th July 2013 NCSLI 2013, Nashville 20

Basic Measurement Diagram 16 th July 2013 NCSLI 2013, Nashville 21 Basic Measurement Diagram 16 th July 2013 NCSLI 2013, Nashville 21

Practical Set up 16 th July 2013 NCSLI 2013, Nashville 22 Practical Set up 16 th July 2013 NCSLI 2013, Nashville 22

Voltage Range With a 10 V input the peak output of the converter will Voltage Range With a 10 V input the peak output of the converter will be 10 Volts; giving an RMS voltage of 7. 07 V To get a 1 V output the 10 Volt output can be resistively divided down allowing the output to still be single stepped and measured on DC. Note an IVD cannot be used to divide down the DC output. However it can be used on the AC output 16 th July 2013 NCSLI 2013, Nashville 23

Errors Switching Glitches and transition This error is most difficult to evaluate. It is Errors Switching Glitches and transition This error is most difficult to evaluate. It is dependent on the DAC used. The DAC we have chosen is typically used for audio will be very ‘clean’. The approach used by Transmille has been to look at an individual step on a scope and estimate an error based on the size and duration of the transitions. 16 th July 2013 NCSLI 2013, Nashville 24

Errors Switching Glitches and transition Typical figures for this, at a frequency of 50 Errors Switching Glitches and transition Typical figures for this, at a frequency of 50 Hz, where there is a step every 80 u. S, worst case measurements give glitch times of around 8 n. S, with a glitch size of around 10% of the step. This gives transition errors around the 10 ppm level. In practice as some glitches will add and some will subtract the real error will be much less. 16 th July 2013 NCSLI 2013, Nashville 25

Errors from Rise / fall times & DC offsets Rise and Fall time Providing Errors from Rise / fall times & DC offsets Rise and Fall time Providing the rise and fall times are equal they do not contribute to errors. The error caused by a difference in rise/fall times can be mathematically calculated DC offsets Any DC offsets in the DAC will be measured and added to the calculated RMS figure. However if the device being calibrated is AC coupled DC offsets in the AC output will give an error and it may be best to trim any offsets out to avoid this problem. 16 th July 2013 NCSLI 2013, Nashville 26

Errors from Loading effects Output Loading When measuring the output with a thermal converter Errors from Loading effects Output Loading When measuring the output with a thermal converter the load effect on the 2 wire output will need to be considered. When using measuring the output with a typical DMM loading effects will be negligible 16 th July 2013 NCSLI 2013, Nashville 27

Errors from drift in DC Reference Source & DAC As the AC source is Errors from drift in DC Reference Source & DAC As the AC source is a transfer device it is only the short term drift between the DC measurement of the system and the subsequent use on AC that contributes to the error. Long term changes in the reference or the DAC linearity or gain do not contribute as the system is calibrated before use. 16 th July 2013 NCSLI 2013, Nashville 28

Uncertainties Typical laboratory figures for uncertainty Imported DC Voltage Measurement Resolution of DC voltage Uncertainties Typical laboratory figures for uncertainty Imported DC Voltage Measurement Resolution of DC voltage Measurement Switching transition errors Short term stability of DC reference and DAC* 5 ppm 0. 1 ppm 10 ppm 2 ppm Combining for 95% (K=2) gives 13 ppm * Includes effects of temperature for +/- 2’C from Tcal 16 th July 2013 NCSLI 2013, Nashville 29

Verifying the accuracy Measurements made by Transmille best AC measurement capability uses our Fluke Verifying the accuracy Measurements made by Transmille best AC measurement capability uses our Fluke 540 b thermal transfer, the accuracy of this Instrument is enhanced by measuring the output of thermal converter directly. Loading effects on both AC and DC measurements were compensated for. Tests performed against a Wavetek 4920, Transmille 8081, Agilent 3458 A and Fluke 8508 A provided similar results. Typical readings obtained Single step calculation = Measurement by thermal Transfer 540 B = Error = 14 ppm 16 th July 2013 NCSLI 2013, Nashville 7. 055681 7. 055779 30

Frequency Range To date Transmille’s interest has been In the frequency range 40 Hz Frequency Range To date Transmille’s interest has been In the frequency range 40 Hz to 1 k. Hz. The design could generate frequencies below 1 Hz with no additional errors. Frequencies up to 10 k. Hz could also be generated with increased error from switching. 16 th July 2013 NCSLI 2013, Nashville 31

Improvements…… Extending the Voltage range. A simple resistive divider to provide 1 V and Improvements…… Extending the Voltage range. A simple resistive divider to provide 1 V and 100 m. V outputs. A DC voltage amplifier to provide a 100 V output. Both could be again ‘calibrated’ at DC using a known DMM to provide the calculated AC RMS value 16 th July 2013 NCSLI 2013, Nashville 32

Other Functions…… Non sine wave outputs With this method it would be very easy Other Functions…… Non sine wave outputs With this method it would be very easy to generate other wave shapes to evaluate performance of converters including crest factor. 16 th July 2013 NCSLI 2013, Nashville 33

Power and Phase…. . Future development of this technique could be to additional converters Power and Phase…. . Future development of this technique could be to additional converters to allow for several outputs. Up to 6 outputs could easily be provided for 3 phase simulation of power, with an accurate phase, or delay between the outputs. If all the outputs were at the same 7 V level this would provide a very affordable solution for a phase and power reference. 16 th July 2013 NCSLI 2013, Nashville 34

References [1] R. Kinard, and L. A. Harris “ A digitally synthesized sine wave References [1] R. Kinard, and L. A. Harris “ A digitally synthesized sine wave source with +/- 1 ppm amplitude stability. “ in Euromeas ’ 77 Conf Digest. Institute Elect. Eng. Conf. Pub no 152, Sept. 1977 [2] Oldham N. , Hetrick P. , Zeng X. , “A Calculable, Transportable Audio Frequency AC Reference Standard”. IEEE Transactions on I & M, April 1989, Vol. 38. [3] Oldham N. , Bruce W. , FU C. , Cohee A. , Smith A. , “An Intercomparison of AC Voltage Using A Digitally Synthesized Source”. IEEE Transactions on I & M Vol. 39 No. 1, February 1990 16 th July 2013 NCSLI 2013, Nashville 35

Conclusions Digital synthesized waveforms can provide a low cost and easy to use solution Conclusions Digital synthesized waveforms can provide a low cost and easy to use solution for DC to AC voltage transfer. Copies of our paper and presentation are available on our booth on USB memory sticks 16 th July 2013 NCSLI 2013, Nashville 36