Timeline for How to achieve analog zero-drift sample and hold for hours?
Current License: CC BY-SA 3.0
15 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 22, 2017 at 17:59 | history | edited | Ricardo | CC BY-SA 3.0 |
Dressed naked link and formatted scientific notation numbers.
|
| Nov 22, 2017 at 17:23 | comment | added | Synchrondyne | For those sort of time scales, I think you would be forced to look at a solution which used the magnetization state of a magnetic core as storage. By comparison, a 12-bit ADC, register, and DAC will cost less than $5. | |
| Nov 22, 2017 at 17:19 | comment | added | Synchrondyne | At 250ppm, you can only withstand about 0.00025 time constants, which means to hold this accuracy for five hours your RC time constant needs to be 7.2*10^7 seconds or somewhere in the neighborhood of two and half years. This seems dubious. At that sort of leakage requirement, the switch and the buffer come back into play - that 100 Tohm switch alone will drop your time constant down to the order of about a year. | |
| Nov 22, 2017 at 13:37 | comment | added | Johny Radio | Also, i added desired precision to the question, how does that affect this solution? | |
| Nov 22, 2017 at 12:30 | comment | added | pjc50 | All the parts cited are available at reasonable prices from normal distributors; the only catch in assembly is that you need to have a very clean PCB afterwards. | |
| Nov 22, 2017 at 12:15 | comment | added | Johny Radio | Bravo for showing how this can actually be accomplished in the analog domain. Are the components you suggested difficult to obtain? Prohibitively expensive? Hard to assemble? Also, i updated my desired resolution (ppm) in the Question-- what component values could achieve that? Many thx | |
| Nov 22, 2017 at 12:00 | review | Suggested edits | |||
| Nov 22, 2017 at 16:12 | |||||
| Nov 21, 2017 at 19:46 | comment | added | Synchrondyne | Also somewhat relevant, another Bob Pease article about femtoampere current measurement test setups: electronicdesign.com/test-amp-measurement/… | |
| Nov 21, 2017 at 18:34 | comment | added | τεκ | I found some others that have higher rated insulation resistance up to 30000 seconds, but definitely agree that real world results are much better. | |
| Nov 21, 2017 at 18:25 | comment | added | Synchrondyne | You're right though, that curve isn't a worst-case value, it's a typical. I've updated my answer accordingly. | |
| Nov 21, 2017 at 18:24 | history | edited | Synchrondyne | CC BY-SA 3.0 |
Corrected worst-case vs. typical.
|
| Nov 21, 2017 at 18:16 | comment | added | Synchrondyne | Page 11 of the datasheet has an RC vs. temperature curve: imgur.com/eHQjXej. Generally, manufacturers will be loathe to guarantee extremely high insulation resistance even if the cap can do it. As Pease found in the linked article, a polypropylene cap at room temperature after soaking can have an RC time constant on the order of ten years. That, of course, will be with the leads in air (since board surface leakage will dwarf that), and certainly can't be guaranteed with every cap that comes off the line. | |
| Nov 21, 2017 at 18:03 | comment | added | τεκ | I don't know where you got that RC time constant of 40000 s from. The datasheet says >5000 s. Film capacitor manufacturers seem to avoid rating anything to a higher insulation resistance than 5000 megaohm-microfarads (seconds). Here's a 0.47uF, 63V film cap with the same rating: digikey.com/product-detail/en/kemet/RSBDC3470AA10J/399-6045-ND/… | |
| Nov 21, 2017 at 17:27 | history | edited | Synchrondyne | CC BY-SA 3.0 |
Error is not depedent on voltage.
|
| Nov 21, 2017 at 2:23 | history | answered | Synchrondyne | CC BY-SA 3.0 |