Timeline for ADC input filter design for DC input signal
Current License: CC BY-SA 4.0
21 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 15, 2021 at 7:14 | comment | added | Marcus Müller | @dandavis the signal being DC doesn't mean noise + interference is, and considering a RC is basically free in all three aspects of cost, complexity and power consumption, there's really no reason not to at least foresee one. | |
| Jan 15, 2021 at 0:14 | vote | accept | James Pie | ||
| Jan 14, 2021 at 7:53 | answer | added | user57037 | timeline score: 3 | |
| Jan 13, 2021 at 23:26 | comment | added | dandavis | @MarcusMüller wait, i'm confused; we're talking about a DC signal here, right? I use the 1115 a lot and successive reads, like the batch of 64 OP mentions, are typically within a few bits of each other. When I add a 10uf cap across the differential inputs, outliers become rarer and less extreme. I suppose some harmonics could sneak in, if unlucky, do you think a random 1-50us delay between reads would curtail such co-incidences? | |
| Jan 13, 2021 at 20:29 | comment | added | Marcus Müller | @dandavis that's multiple orders of magnitude too high a cutoff frequency. Also, the argument "it's easy to add later" doesn't really work – the opposite is true: easier to omit later on; adding it implies a board re-spin. | |
| Jan 13, 2021 at 20:24 | comment | added | Marcus Müller | @dandavis that's incorrect. A signal at, say, 64.5 Hz will look like a signal at 0.5 Hz, and will be impossible to filter out digitally after sampling. At a sample rate this low, the analog bandwidths of the observing system allow for many Nyquist bandwidths worth of aliases, which is a terrible thing, because in any human environment, harmonics of grid frequency are rather common, and so are subharmonics of SMPS periods. Your "wire resistor" is negligible – say, the largest sensible capacity for a filter (he needs to keep things small) is 1µF, and generously assume 10Ω wire, then f_c = 15 kHz | |
| Jan 13, 2021 at 20:02 | comment | added | dandavis | @MarcusMüller a filter lets you get accurate/stable readings faster, but bandwidth is not a concern here, it sounds like OP has enough time to take thousands of samples to software filter, so long as his threshold detection is done on the MCU instead of the ADC's alarm. With such short leads and battery power it's hard to imagine not getting acceptable data as-is, which is why i suggest trying it before freaking out. What's the harm in it working acceptably without modification? He can always add parts later if needed... | |
| Jan 13, 2021 at 19:50 | comment | added | Marcus Müller | @dandavis an ADC without an AA filter is asking for trouble, unless you can guarantee a low-noise environment (which James so much failed to do). | |
| Jan 13, 2021 at 19:45 | answer | added | Marko Buršič | timeline score: 5 | |
| Jan 13, 2021 at 19:42 | comment | added | dandavis | you don't really need to filter the input for your application, unless you have a ton of noise, but since it's run from a battery, that shouldn't be the case. if you don't have a battery, why sleep? Anyway, you can screen out noise by taking several samples, sorting them, then taking the middle (median). That works WAY better then averaging because outliers are discarded instead of watered down. Try it in the real world before freaking out over datasheets. Remember too that wires are resistors, so with a big ass cap, you DO have an RC. | |
| Jan 13, 2021 at 19:37 | comment | added | Marcus Müller | James, again, a capacitor without a resistor simply is not a filter, no matter how much you wish for it: while the capacitor will indeed have a low resistance for high frequencies and "shunt" them to ground, but as long as you have a voltage source with no series resistance feeding that, you simply won't see a drop in voltage, and your capacitor will actually not filter. I've explained that four times now. | |
| Jan 13, 2021 at 19:02 | comment | added | user57037 | I get what you are saying. Let the AC noise alias all it wants. It is all noise and it will be filtered in software (by averaging). But, what if the noise is correlated with your sample rate? As far as the cap goes, you are correct that it will provide some filtering effect, but without a series resistor, the filtering ability of the cap is poorly defined. Theoretically, a zero impedance voltage source will not be filtered at all by a cap alone. Vcap = Vsource, period. | |
| Jan 13, 2021 at 18:42 | comment | added | James Pie | If it takes 10 minutes for my level to rise one inch, any voltage changes during a few millisecond sample will be meaningless, just noise. Therefore any AC signal during measurement, I'd want to ground. A cap on the input will take AC signal to ground depending upon the AC frequency, correct? What am I misunderstanding? I'm certainly not being argumentative, I just want to know where my thinking is incorrect so I can learn from it | |
| Jan 13, 2021 at 18:01 | comment | added | user57037 | I am going to break this down for you. You would be an absolute fool to have no RC filtering on an ADC input. There are many sources of noise including some you haven't thought of yet. Always filter out-of-band noise before sampling with an ADC. Remember, if you somehow determine no filter is needed, you can omit the caps during assembly, and put in zero Ohm resistors. | |
| Jan 13, 2021 at 17:54 | answer | added | Marcus Müller | timeline score: 1 | |
| Jan 13, 2021 at 17:39 | comment | added | Justme | The theory is the same regardless of frequency. Excess bandwidth must be filtered out before sampling, unless the transducer is incapable of generating more bandwidth than sampling is able to capture. And who says precision resistors and capacitors must be used? Some noise and RF filtering might be good to have, but depends on transducer output impedance and ADC input impedance. | |
| Jan 13, 2021 at 17:33 | comment | added | D.A.S. | You can expect lots of rework/scrap with no design specs | |
| Jan 13, 2021 at 17:23 | comment | added | user1850479 | What is the source driving the ADC inputs? Remember that they have finite input impedance. As for cost of an SMD capacitor, is that really important if you're using a several dollar ADC chip? Seems like cost optimization might be better focused elsewhere. | |
| Jan 13, 2021 at 17:13 | comment | added | BeB00 | why would you need precision resistors (as opposed to normal resistors) or class 1 capacitors? | |
| Jan 13, 2021 at 17:10 | comment | added | Andy aka | Cut and paste the circuit you are talking about. You can embed images. | |
| Jan 13, 2021 at 17:04 | history | asked | James Pie | CC BY-SA 4.0 |