Why is my low-pass filter not filtering out this high-frequency noise?

[PfhorSlayer]

I've got one of those super cheap ($10) signal generator kits, and despite the fact that I'm almost positive the XR2206 chip it's using is a counterfeit knockoff (it gets hot, and I had to bodge some traces on the PCB to get it to output anything at all), it does produce a basic sine wave signal that I can use as an input as I'm playing around with guitar/bass effect pedal circuits. However, despite the fact that it is definitely producing the overall sine wave I expect, it is also outputting some very high frequency noise on top of the expected signal, at almost double the amplitude of the expected signal!



The signal I'm expecting is roughly 1kHz, and since the circuits I'm building are designed to operate in the audio range, I assumed that adding a low pass filter at roughly 20kHz should do something to reduce the noise. To this end, I added a 820 ohm resistor in series with a 0.1uF capacitor to ground. When the noise remained, I tried adding another low pass filter at ~1kHz (150ohm, 1uF). The noise yet remained!



Channel 1 is the original signal, channel 2 is after the 1kHz filter, channel 3 is after the 20kHz filter (I have the filters reversed - 1kHz first, followed by 20kHz - in the circuit; the order does not change the result).

Zooming in on one of the spikes, I see that there is a slight bit of attenuation on the noise, but not nearly as much as I would expect there to be:



Am I misunderstanding how an RC filter is supposed to work? Is there a better way to clean up the signal I'm getting out of this cheap-ass signal generator?

Also, the signal generator is able to output a pretty clean square wave with a ~100us rise time. Would I be better off trying to filter that square wave into a sine wave instead?

[Kleinstein]

The high frequency part does not look like noise, but more like massive interference from a digital circuit with poor supply decoupling. So this would be outside the XR2206. Such interference would be more like a thing with a DDS generator.
The usual way would be to look for the source of the interference and improve it there, not by filtering the signal.  With the very high frequencies a filter made from large parts may not be effective as the parasitic inductance may be too large. The junk may also come in after the filter, e.g. with the ground clip or the ground wire.

[Vovk_Z]

To TS:
At least to things:
1) It mustn't be hot. Those chips are quite stable usually.
2) Are you sure about 1 kHz output? Maybe it outputs on hundreds of kHz, and a picture you see - is just aliases?
So either it is faulty or have wrong settings and you measure something wrong.
You may use soundcard as sine wave oscillator (you can find appropriate software), and buy another oscillator module.

[Manul]

It may be noise on a ground loop. That is why it is not filtering. My guess is that you power it from SMPS or there is other serious noise source. As an experiment, try to power from a battery to fully float your circuit. I bet all the noise will disappear.

[PfhorSlayer]

The kit itself is incredibly simple - it's basically this, but with cheaper everything. The reason I think the XR2206 I've got is fake or counterfeit is that when I built up the circuit originally, it simply did not work and output nothing. After building it again on a breadboard, I discovered that the chip I have does not behave the same as the datasheet states: I had to swap the TR1 and TR2 inputs, as the pin that should select which one to use does not do anything, and the chip is stuck using the TR2 timing resistance input. On the PCB, I cut the trace to TR1 and bodge-wired the resistor over to TR2. Since it gets VERY hot, I'm guessing it's half-fried in some way (maybe a reject that was scooped up and resold?).

I am powering it from a SMPS wall-wart at 12V; despite the fact that the datasheet says it should work from 9V-12V, attempting to power it from a 9V battery does not work at all.

I'm fairly certain that the 1kHZ it's outputting is accurate and not just a much higher frequency aliased result, as there are multiple caps that can be selected to choose the frequency range, and I do get a fairly consistent linear increase as I move from the bottom to the top of each range, starting at the lowest (1-10Hz).

I'll see if I can set up a 12V battery source to power it without the SMPS in the way - if that fixes it, at least I'll know why it was happening in the first place (and maybe I could add some more decoupling caps across the supply?). If that doesn't fix it, I guess looking for a replacement XR2206 is in order.

Thanks, all!

[Vovk_Z]

Buy ready-made module, but not kit. Modules usually work without problems.
It it was a kit, then you may do something wrong when assembled it.

[magic]

Test the output voltage of the 9V battery when connected - chances are this thing just draws too much power for those small batteries.

It looks like switcher noise, repeating at some 30kHz or so. Is the switcher grounded to the mains socket? Try one that isn't / is.

I would see if star grounding makes any difference - ensure that the output connector and all the small passives around the chip are connected directly to the chip's ground pin, and then make another connection from the chip's GND to the power filtering capacitor and from there to the power brick.