If this is just for dc use, I would use a w.w. pot or ceramic type. How each mfr rates their parts would determine the life of the part. I would not run these liitle guys too close to their rating though. Contact (wiper) construction varies... Spectrol (now Vishey owned) uses a dual contact, gold plated for long life and stable contact resistance. They are cheap enough. NTE sells them under thier name/number. -RM
Depends, it usually says under the conditions. Most I think say +PSRR and -PSRR on seperate graphs which are as stated.
Nasa states that the backround noise is 10-6/(hz)raised to 1/2 . { ref. http://jila1.nickersonm.com/papers/Ellery_Thesis_d12.pdf } What equipment was used that can measure above about .2hz aprox. the noise floor at -200db . some old tests showed that a few could hear below a noise floor by 10db when the floor was about -60db . How then is you are able to measure 80db below noise floor that must be one very very impressive grounding system used. This fantastic grounding (earthing) system must be share with all of us .
99%+ of monolithic op amps will have PSSR to one supply be lower, it will follow the open loop gain gain curve - the rail the internal Ccomp is connected to
since the PSRR is asymmetric, the better rail often orders better at most frequencies, the worse rail rejection dominates the common mode power supply rejection
input CMRR will usually be higher than the worse rail PSRR too
many datasheets don't show this right - use a test circuit limited by the op amp's own open loop gain - pointed out by Pease
"hearing below the noise floor" is often due to misunderstanding/misstating the details of frequency dependence of both "noise" and our hearing in "critical bands" of few hundred Hz to few kHz BW - human hearing doesn't use very long integration time times, and masking applies all the way down to the noise floor
since the PSRR is asymmetric, the better rail often orders better at most frequencies, the worse rail rejection dominates the common mode power supply rejection
input CMRR will usually be higher than the worse rail PSRR too
many datasheets don't show this right - use a test circuit limited by the op amp's own open loop gain - pointed out by Pease
"hearing below the noise floor" is often due to misunderstanding/misstating the details of frequency dependence of both "noise" and our hearing in "critical bands" of few hundred Hz to few kHz BW - human hearing doesn't use very long integration time times, and masking applies all the way down to the noise floor
Last edited:
'
Jan:
14 mA will burn the contact points on anything but a larger wirewound. You will also get some instability from that. If you are using it as an offset trim (a guess) parallel some resistors to take most of the current and use the pot to make an offset tweak.
The best multiturn trimmer for that would be the Vishay bulk foil trimmer Vishay - Resistors, Variable - Trimmers - Bulk Metal® Foil Be prepared for sticker shock. They do work well for this. Datasheet: http://www.vishaypg.com/docs/63056/12801285.pdf (seems hard to find on the Vishay website).
Last edited:
This was 1/f noise at 10^-4 Hz of IC voltage references not exactly apples to apples.
Scott,
I want to get a bit into that you start with a clean DC supply, add isolation so you can inject a test signal into one terminal or the other.
JCX
You are correct one supply side is more critical than the other. The secondary issue is what happens when you draw in the "Stray" capacitors and then connect this circuit to another with it's own power supply.
George,
You asked about other effects. Did you know that from the same manufacturer, the same series of capacitors, even the same values will have different resonant frequencies for axial vs. radial?
Anybody care to explain why?
ESP
When you draft in ink, most folks start the vertical lines from the left and draw them in, then wait till they dry before starting at the top and placing the horizontal lines. (Resistors are usually done with a template by themselves!) So the + sign is easier to draw the closer it is to the bottom of the drafting! (Try bending over a drafting table for 8 hours, you'll get a feel for it.)
Now for architectural drawings the horizontal and vertical lines never close! They just come close enough to appear closed. If they are closed they then look thicker at the corners and the architects didn't like that!
Die size and process are about all that matters. Big power outputs probably matter in that case,
That was exactly my point- the noise that Ed wants us to worry about is ridiculously lower than anything that could possibly be imagined to be significant.
Because you (maybe deliberately) didn't take care of grounding or layout properly for your test, that would be my first guess. It certainly doesn't show up in any of my gear and doesn't seem to show up in any of Scott's high precision work.
Have you published measurements showing this on the output of a 797 fed by a reasonable raw supply and a regulator, or are you moving the goalposts again?
Have you published measurements showing this on the output of a 797 fed by a reasonable raw supply and a regulator, or are you moving the goalposts again?
Maybe inadvertent, maybe deliberate (to try to make a point), maybe he's switched gears and is talking about a different measurement of something else entirely. Nailing jello to a wall is not easy.
- Status
- Not open for further replies.