I was actually rigging that exact thing up... just got done testing it. I took a new board without the RC modification, completely severed the circuit between C8 and R4, rigged up a 7805 and a power adapter. Here are the results:
https://goo.gl/photos/U3rpHXU7YF95NMjt5

Ouch!

 

And here it is with the 9V battery going into the 7805, then into the PCB. Wow a lot cleaner. I'm confused about the occasional big dips that show up on the scope as well.
https://goo.gl/photos/gFQuckUUTys6Pyzo9

Yes, a lot cleaner. But now that we've got rid of incoming switcher noise by using the 9V battery and the 7805, I see something that looks suspiciously like oscillation in the LDO. I wonder if you add more capacitance (like another 1 uF ceramic or two on the LDO input), does that oscillation-like trace change? That would be a clue. LDOs tend to be very finicky in their requirements for stability, and it might be oscillating.

I suspect the occasional big dips are the nRF24L01 drawing current spikes. From my perusal of the MCP1700 spec sheet, it appeared that its transient load regulation didn't look very good (somewhat expected with a micropower device). A bigger cap on its output might help that.

 

Yes, a lot cleaner. But now that we've got rid of incoming switcher noise by using the 9V battery and the 7805, I see something that looks suspiciously like oscillation in the LDO. I wonder if you add more capacitance (like another 1 uF ceramic or two on the LDO input), does that oscillation-like trace change? That would be a clue. LDOs tend to be very finicky in their requirements for stability, and it might be oscillating.

I suspect the occasional big dips are the nRF24L01 drawing current spikes. From my perusal of the MCP1700 spec sheet, it appeared that its transient load regulation didn't look very good (somewhat expected with a micropower device). A bigger cap on its output might help that.

Ok I added 1uF capacitor directly ontop of another one of the MCP1700 input caps, and here are the results:
https://goo.gl/photos/qeKtQV8g7KQVawug7

I then added another 1uF capacitor for a total of 2 additional caps:
https://goo.gl/photos/tk3ednbELPPE59ez5

Thanks again for the help!

 

Hmmm... didn't seem to make much difference. Try adding some capacitance on the LDO output?

 

Hmmm... didn't seem to make much difference. Try adding some capacitance on the LDO output?

Ok I added a 1uF cap to the output of the MCP1700 now:
https://goo.gl/photos/vdYG2AcQQ3XQYNy29

 

Hmmm... not much difference there, either. What does the output of the 7805 look like? The input of the MCP1700? What does the output of the MCP1700 look like if you disconnect the nRF24L01 and substitute a load resistor (82Ω would give you ≈40 mA load)?

I'm close to being out of ideas, here. There's a lot of noise there, and I'm scratching my head trying to figure out where it's coming from...

 

Hmmm... not much difference there, either. What does the output of the 7805 look like? The input of the MCP1700? What does the output of the MCP1700 look like if you disconnect the nRF24L01 and substitute a load resistor (82Ω would give you ≈40 mA load)?

I'm close to being out of ideas, here. There's a lot of noise there, and I'm scratching my head trying to figure out where it's coming from...

I'll definitely try this out tomorrow when I get in.

One thing I thought was interesting was if I touch the wires that are between the 9 volt battery and the 7805, the amplitude of the wave on the scope decreases by about 25%. If I touch anything else on the PCB it does nothing, only the battery wires....Does that mean anything?

 

Hmmm... not much difference there, either. What does the output of the 7805 look like? The input of the MCP1700? What does the output of the MCP1700 look like if you disconnect the nRF24L01 and substitute a load resistor (82Ω would give you ≈40 mA load)?

I'm close to being out of ideas, here. There's a lot of noise there, and I'm scratching my head trying to figure out where it's coming from...

Ok so I took a scope reading of the input of the MCP1700, it looks identical only 5V instead of 3.3V (same magnitude).
I then disconnected the circuit completely so I just have a 9 volt battery and a 7805, and it still looks identical. So apparently the scope adds 30-80mV worth of noise to everything? Am I not using this thing right or something? I've seen people wrap the ground wire around the probe to prevent external noise... but this ground lead is so short I can't even do that.

This is the probe that came with the oscilloscope: http://www.tequipment.net/Rigol/RP3300A/Passive-Oscilloscope-Probes/
I tried changing the setting for coupling from "DC" to "AC" nothing changed.

 

I then disconnected the circuit completely so I just have a 9 volt battery and a 7805, and it still looks identical. So apparently the scope adds 30-80mV worth of noise to everything?

I'd say that about clinches it: I don't know whether the scope is adding the noise, but something sure is; with just the battery and the 7805, I would expect the noise to be less than a tenth of a millivolt.

Am I not using this thing right or something? I've seen people wrap the ground wire around the probe to prevent external noise... but this ground lead is so short I can't even do that.

No, it seems to me you're doing everything right, including using a short ground lead on the probe.

What happens if you turn off all lights and all other electronic equipment in the lab while you make these measurements? Does that improve things? I often have to douse the lights in my workshop at home when I'm playing with low-level analog stuff, because the CFL lights spew out an enormous amount of EM garbage in the 20 kHz - 200 kHz range.

Could you try another scope, from a different manufacturer? It might be interesting to compare readings.

That's about the end of my ideas, unless we want to get into Voodoo spells, animal sacrifices, and stuff like that.

EDIT: Did you try disconnecting the nRF24L01 and substituting an 80 ohm dummy load as I suggested?

 

En example:
Let's say 8 bit scope 5V range 1-bit, +- 1 bit or about 20 mV.

 

I'd say that about clinches it: I don't know whether the scope is adding the noise, but something sure is; with just the battery and the 7805, I would expect the noise to be less than a tenth of a millivolt.


No, it seems to me you're doing everything right, including using a short ground lead on the probe.

What happens if you turn off all lights and all other electronic equipment in the lab while you make these measurements? Does that improve things? I often have to douse the lights in my workshop at home when I'm playing with low-level analog stuff, because the CFL lights spew out an enormous amount of EM garbage in the 20 kHz - 200 kHz range.

Could you try another scope, from a different manufacturer? It might be interesting to compare readings.

That's about the end of my ideas, unless we want to get into Voodoo spells, animal sacrifices, and stuff like that.

EDIT: Did you try disconnecting the nRF24L01 and substituting an 80 ohm dummy load as I suggested?

Originally I had the scope plugged into a power strip. So I unplugged everything, turned off lights, and had scope plugged directly into wall and it looked the same. I also ran a automated calibration on the scope. The only thing that makes the noise/magnitude go down is if I grab the probe cables (makes it go down about 25% about).

I didn't try removing the RF module yet since I wouldn't be able to tell any difference with a 30mV resolution.

En example:
Let's say 8 bit scope 5V range 1-bit, +- 1 bit or about 20 mV.

The scope is a Rigol DS2202A which should be more than capable of making these measurements shouldn't it? It doesn't say anywhere of it being an 8 bit scope or anything like that.

 

I didn't try removing the RF module yet since I wouldn't be able to tell any difference with a 30mV resolution.

You can get a lot better resolution than that; just set the scope for AC coupling and increase the sensitivity. I think we REALLY need to see what this does with an inert, dummy load (80 ohms will give you 40 mA) in place of the RF module.

 

You can get a lot better resolution than that; just set the scope for AC coupling and increase the sensitivity. I think we REALLY need to see what this does with an inert, dummy load (80 ohms will give you 40 mA) in place of the RF module.

I tried setting the mode to AC coupling and it didn't change anything. I'm guessing because all of the noise? I guess I need to figure out how to increase the sensitivity so I can get proper readings.

Yeah I agree, being able to see whats going on will be important here.

 

I found a great video that explains this, its 25 minutes long but explains the oscilliscope noise thing perfectly:

 

And came across this that might explain my original oscilloscope readings that had the spikes in it with the original design and the power adaptor:

Ugh... so many variables here, seems impossible to get this thing figured out

 

You have a 12 bit scope. So on the 10 V range - 1 bit is about 10 mV

I didn't see your photo. Clearly switchmode spikes. You may have to use a faster regulator. See http://cds.linear.com/docs/en/application-note/an101f.pdf

http://www.ti.com/lit/an/slva549/slva549.pdf

http://www.aimtec.com/site/Aimtec/f...016e - reduction of output ripple & noise.pdf

https://www.digikey.com/-/media/Ima...en-US&ts=a968ec0c-bd59-4878-ba95-9a67ebbfd769

You can also look at devices eith Micrel's Ripple Blocker (TM) technology. See: http://www.farnell.com/datasheets/1563191.pdf

 

You have a 12 bit scope. So on the 10 V range - 1 bit is about 10 mV

I didn't see your photo. Clearly switchmode spikes. You may have to use a faster regulator. See http://cds.linear.com/docs/en/application-note/an101f.pdf

http://www.ti.com/lit/an/slva549/slva549.pdf

http://www.aimtec.com/site/Aimtec/files/documents/ApplicationNotes/a016e - reduction of output ripple & noise.pdf

https://www.digikey.com/-/media/Images/Product Highlights/M/Microchip Corporation/Ripple Blocker/APEC_Ripple_BlockerFINALWhitePaper.pdf?la=en-US&ts=a968ec0c-bd59-4878-ba95-9a67ebbfd769

You can also look at devices eith Micrel's Ripple Blocker (TM) technology. See: http://www.farnell.com/datasheets/1563191.pdf

I don't believe this to be accurate. It depends on the vertical division setting (e.g. how far in or out you are zoomed). For example if I am zoomed in to a 50mV division (which most of my oscilloscope readings are in), there are 8 vertical blocks so the entire vertical screen represents 400mV - you then factor in the bit resolution on this number -> e.g. 12 bits = 4096 numbers, so 400mV / 4096 = 0.097mV vertical resolution. You can zoom all the way down to 500uV per division if you want, for an extremely small resolution.

You can read page 2 of this document for reference: http://cp.literature.agilent.com/litweb/pdf/5991-1617EN.pdf

 

(Watching EEVblog videos always drives me nuts because of Dave's jumpy, hyper, manic, stream-of-consciousness style of presentation. The guy gives me a headache.)

While EEVblog #601 was interesting, I don't think it relates much to what we've been discussing: the noise you're experiencing in your circuit is MUCH greater than the noise he's talking about there, by more than an order of magnitude.

EEVblog #441 was much more relevant. And you were right when you said, "ugh"-- dealing with this stuff can be maddening. The takeaway from #441 is that it's very possible that a substantial part of the noise you've been seeing-- especially the spikey stuff-- is not being produced in your circuit, but could instead be a result of radiated interference from nearby devices or conducted into your circuit through the power line itself.

The RF module is another possible source of interference. We really need to take a look at the noise level with the RF module removed and replaced by a dummy load, and with the scope set for AC coupling and a higher sensitivity so we can see what's going on. We need to rule the nRF24L01 in or rule it out.

And finally, @KeepItSimpleStupid has a good point: you may need to replace the MCP1700 with a faster voltage regulator that does a better job of suppressing high frequency ripple and handling line and load transients. As I said before in post #8, low-dropout regulators by their very nature are usually poor in that respect, and micropower LDOs (including the MCP1700) are VERY poor because their control loops are slower. You've got plenty of voltage headroom available, so there's no need for an LDO regulator; and you've got oodles of power on hand, so there's no need for a micropower part. My advice would be to switch to an LM317 or something similar; and when selecting a part, pay close attention to its PSRR vs. frequency characteristics.

 

And came across this that might explain my original oscilloscope readings that had the spikes in it with the original design and the power adaptor:

Ugh... so many variables here, seems impossible to get this thing figured out

Just probe ground right next to the ground clip and see what the amplitude is. This will show you your measurement error.
No how to correct it is another question.

 

√2 C Er (p-p) F = I

Yep....brute force uber allis.