Author Topic: Questions Regarding Solartron 7150 Accuracy Specs  (Read 963 times)

0 Members and 1 Guest are viewing this topic.

Offline Chance92Topic starter

  • Regular Contributor
  • *
  • Posts: 52
  • Country: gb
Questions Regarding Solartron 7150 Accuracy Specs
« on: September 30, 2023, 08:30:07 pm »
The 7150 is a 5.5 digit multimeter that can display 6.5 digits by using a running average filter. I am a bit confused by the spec sheet. The two-year accuracy for dc volt with 5.5 digit is 0.01% of reading + 5 digits, and the 6.5 digit display will add 10 digits to that. Does that mean the accuracy for the 6.5 digit display is 0.01% of reading + 15 digits? 15 digits of 6.5d display is equivalent to 1.5 digits of 5.5d display. I thought averaging filter can't improve accuracy. Am I interpreting it wrong?
 

Offline alm

  • Super Contributor
  • ***
  • Posts: 2903
  • Country: 00
Re: Questions Regarding Solartron 7150 Accuracy Specs
« Reply #1 on: September 30, 2023, 09:00:33 pm »
The 7150 is a 5.5 digit multimeter that can display 6.5 digits by using a running average filter. I am a bit confused by the spec sheet. The two-year accuracy for dc volt with 5.5 digit is 0.01% of reading + 5 digits, and the 6.5 digit display will add 10 digits to that. Does that mean the accuracy for the 6.5 digit display is 0.01% of reading + 15 digits? 15 digits of 6.5d display is equivalent to 1.5 digits of 5.5d display. I thought averaging filter can't improve accuracy. Am I interpreting it wrong?
Yes, that would also be my interpretation. Averaging can improve accuracy if it is limited by either noise or quantization error. If you're measuring 10 uV on a range where the least significant digit is 1 uV, then you're always going to have an uncertainty of +/- 1 uV since even if the error would have only be 0.5 uV, it can not represent steps smaller than 1 uV so it will show up as 1 uV error. If the actual value is 10.4 uV, then even 0.2 uV error would push it to 10.6 uV which would be seen as going from 10 uV to 11 uV. When you add an extra digit, the error might be only 0.2 uV since the resolution is now 0.1 uV.

The other thing is noise. If that 5 digits of uncertainty is largely noise in the voltage reference, the front-end or ADC, then averaging over a longer period of time might reduce the noise (depending on the type of noise). Noise is one of the components in measurement uncertainty (together with other errors like systematic errors). Datron uses this trick in their 1080 series of multimeters that are 6.5 digit meters but through averaging (and injecting a bit of noise for reasons that are outside the scope of this post) they can get an extra digit of resolution that produces a lower uncertainty than in 6.5 digit mode.
 
The following users thanked this post: ch_scr

Offline ch_scr

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: de
Re: Questions Regarding Solartron 7150 Accuracy Specs
« Reply #2 on: September 30, 2023, 09:38:52 pm »
In the 7150 the frontend is mostly an divider (lower ranges direct) with an AD545 as buffer (Datasheet noise 3uV p-p in 0.1 to 10Hz). 7150 plus and 7151 use an discrete dual-FET buffer with OP14 instead. The ref is an IN829A zener diode in all of them, hardly know for it's noise. The 7150 plus manual says "for 6 1/2 digits add 3 to counts error + 1uV" in the same spot, the 7151 manual doesn't mention accuracy for 6 1/2 digits at all.
« Last Edit: September 30, 2023, 09:43:21 pm by ch_scr »
 

Offline Chance92Topic starter

  • Regular Contributor
  • *
  • Posts: 52
  • Country: gb
Re: Questions Regarding Solartron 7150 Accuracy Specs
« Reply #3 on: September 30, 2023, 09:42:04 pm »
The 7150 is a 5.5 digit multimeter that can display 6.5 digits by using a running average filter. I am a bit confused by the spec sheet. The two-year accuracy for dc volt with 5.5 digit is 0.01% of reading + 5 digits, and the 6.5 digit display will add 10 digits to that. Does that mean the accuracy for the 6.5 digit display is 0.01% of reading + 15 digits? 15 digits of 6.5d display is equivalent to 1.5 digits of 5.5d display. I thought averaging filter can't improve accuracy. Am I interpreting it wrong?
Yes, that would also be my interpretation. Averaging can improve accuracy if it is limited by either noise or quantization error. If you're measuring 10 uV on a range where the least significant digit is 1 uV, then you're always going to have an uncertainty of +/- 1 uV since even if the error would have only be 0.5 uV, it can not represent steps smaller than 1 uV so it will show up as 1 uV error. If the actual value is 10.4 uV, then even 0.2 uV error would push it to 10.6 uV which would be seen as going from 10 uV to 11 uV. When you add an extra digit, the error might be only 0.2 uV since the resolution is now 0.1 uV.

The other thing is noise. If that 5 digits of uncertainty is largely noise in the voltage reference, the front-end or ADC, then averaging over a longer period of time might reduce the noise (depending on the type of noise). Noise is one of the components in measurement uncertainty (together with other errors like systematic errors). Datron uses this trick in their 1080 series of multimeters that are 6.5 digit meters but through averaging (and injecting a bit of noise for reasons that are outside the scope of this post) they can get an extra digit of resolution that produces a lower uncertainty than in 6.5 digit mode.

Thank you the detailed explaination. It doesn't look like the 5 digits uncertainty is from noise, as the readings were almost constant when being hooked up to a LM399-based reference after sufficient warm-up.

I have to say the amount of accuracy improvement in this case doesn't sounds very logical to me. Imagine if I input a precise 1.000 000V to the 2V range, the meter may display a stable 1.000 05V under 5.5d, which is still within the 5.5 digit mode spec. However, no matter how much samples it takes to produce an averaged 6.5d reading, it will always be 1.000 050V, which will fail the 15 digit uncertainty for the 6.5 digit mode. In order to meet the 6.5 digit mode spec, the 5.5 digit mode must produce readings between 0.999 99V and 1.000 01V, but this will improve the 5.5 digit mode accuracy to +-1 digit. I don't think the manufacturer has any reason to underspec their products. Another possibility is that the meter's reading would jump all over between 0.999 95V and 1.000 05V with their mean sitting between 0.999 985V and 1.000 015V, but this is definitely not what I have seen with this meter.

In short, the specs for the 5.5d and the 6.5d don't seem compatible considering the stable reading of the meter. This is why I think I may misinterpreted the datasheet.
 

Offline Chance92Topic starter

  • Regular Contributor
  • *
  • Posts: 52
  • Country: gb
Re: Questions Regarding Solartron 7150 Accuracy Specs
« Reply #4 on: September 30, 2023, 09:52:12 pm »
Thanks for the additional info. I think the datasheet for 7150 plus has the same issue. I guess my question should be: does the counts error for the 5.5d mode need to be multiplied by 10 to apply to the 6.5d mode? I have to say Solartron doesn't seem to be very good at writing datasheets.
 

Offline ch_scr

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: de
Re: Questions Regarding Solartron 7150 Accuracy Specs
« Reply #5 on: September 30, 2023, 10:02:57 pm »
Thanks for the additional info. I think the datasheet for 7150 plus has the same issue. I guess my question should be: does the counts error for the 5.5d mode need to be multiplied by 10 to apply to the 6.5d mode? I have to say Solartron doesn't seem to be very good at writing datasheets.
That was my thought initially, "3 counts" in 5 1/2 digit land should equal "30 counts" in 6 1/2 digit land. Especially since these meter implement 6 1/2 digit purely by means of calculating the running average of 10 of the 5 1/2 digit readings! In the 7150plus, that was upped to 16 readings and even in 5 1/2 digit mode the result shown had a running average over 4 readings applied. They kept it at that for the 7151. Really interesting to compare the manuals to see how they evolved!
« Last Edit: September 30, 2023, 10:04:43 pm by ch_scr »
 

Offline Chance92Topic starter

  • Regular Contributor
  • *
  • Posts: 52
  • Country: gb
Re: Questions Regarding Solartron 7150 Accuracy Specs
« Reply #6 on: September 30, 2023, 10:11:53 pm »
Thanks for the additional info. I think the datasheet for 7150 plus has the same issue. I guess my question should be: does the counts error for the 5.5d mode need to be multiplied by 10 to apply to the 6.5d mode? I have to say Solartron doesn't seem to be very good at writing datasheets.
That was my thought initially, "3 counts" in 5 1/2 digit land should equal "30 counts" in 6 1/2 digit land. Especially since these meter implement 6 1/2 digit purely by means of calculating the running average of 10 of the 5 1/2 digit readings! In the 7150plus, that was upped to 16 readings and even in 5 1/2 digit mode the result shown had a running average over 4 readings applied. They kept it at that for the 7151. Really interesting to compare the manuals to see how they evolved!

Agree! It's funny to think how they tried to make sense of the 6.5d mode spec but eventually gave up.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf