Electronics > Projects, Designs, and Technical Stuff
EMF pickup from amplifier in I2C line causing glitches. (Now with scope trace!)
<< < (9/31) > >>
dmills:
"There may be a few circuit instances where it is necessary to add a complete LC reconstruction filter. These circumstances might occur if there are nearby circuits which are sensitive to noise. In these cases a classic second order Butterworth filter similar to those shown in the figures below can be used."

Noise sensitive circuit, sounds like the very definition of I2C to me......

Note also the essential requirement that the speaker leads be twisted, this makes a huge difference to radiated B field.

And the companies the FCC went after that time were folks doing boutique guitar pedals and amps, not just million selling multinationals.

How are you grounding your scope probes, that three inch long lead with a croc clip the probes come with is worse then useless for the fast stuff and will turn any kind of fast transient into the mess you are seeing on the screen.
SMT beads do snot all at 1MHz, you might be seeing some action by 10MHz, but really even a 1uH shielded inductor will do better.
 
Looking at the ring down on that scope trace, I would say that your snubbers would probably benefit from some attention.

Regards, Dan.
Starlord:

--- Quote from: tautech on July 04, 2016, 11:33:20 pm ---You can't explicitly trust what any scope displays unless the trigger level is set within the amplitude of the displayed waveform.

--- End quote ---

You you mean have the little T symbol within the region that includes the yellow waveform?  I'm aware of that, I just happened to not have it there when I took the screenshots. 

I found that depending on where I located it within the waveform, I got different results, and as I assume it triggers on every rising slope, I could see how I might get different results based on where I put it and that because the peaks were different heights and those were really what I wanted to know the frequency off, I couldn't be sure no matter where I put the trigger it would be reading those and not some higher frequency perturbations.


--- Quote ---Channel AC input coupling is also useful for viewing ripple/noise on a DC rail at low V/div settings without needing to adjust the offset/vertical position.

--- End quote ---

I don't know what that means.  I only got the scope a few days ago.
Starlord:

--- Quote from: dmills on July 04, 2016, 11:39:01 pm ---"There may be a few circuit instances where it is necessary to add a complete LC reconstruction filter. These circumstances might occur if there are nearby circuits which are sensitive to noise. In these cases a classic second order Butterworth filter similar to those shown in the figures below can be used."

Noise sensitive circuit, sounds like the very definition of I2C to me......

--- End quote ---

I agree.  And that is a concern.  But that's why I posted the scope traces.  They appear to show that the noise is in the MHz band, not the 400-600KHz switching frequency of the amplifier.  I'm not sure how it's generating noise in the MHz band though.


--- Quote ---Note also the essential requirement that the speaker leads be twisted, this makes a huge difference to radiated B field.

--- End quote ---

Twisting the speaker wires was one of the first things I tried, and it did not appear to affect the glitching, but I didn't have a scope on hand when I tested that to see what if anything it did.


--- Quote ---How are you grounding your scope probes, that three inch long lead with a croc clip the probes come with is worse then useless for the fast stuff and will turn any kind of fast transient into the mess you are seeing on the screen.

--- End quote ---

I'm using that three inch long croc clip.  I've never seen anyone, including Dave, use anything else, it's all I have, and I only got the scope a few days ago.  What do you suggest?



--- Quote ---SMT beads do snot all at 1MHz, you might be seeing some action by 10MHz, but really even a 1uH shielded inductor will do better.

--- End quote ---

I assume I would have to look up an LC filter calculator, plug in 1uH and, and specify the corner frequency I want to filter to get my capacitor size?  The datasheet only specified the right cap size for a 10uH inductor.



--- Quote ---Looking at the ring down on that scope trace, I would say that your snubbers would probably benefit from some attention.

--- End quote ---

Snubbers?
tautech:

--- Quote from: Starlord on July 05, 2016, 01:27:47 am ---
--- Quote from: tautech on July 04, 2016, 11:33:20 pm ---You can't explicitly trust what any scope displays unless the trigger level is set within the amplitude of the displayed waveform.

--- End quote ---

You you mean have the little T symbol within the region that includes the yellow waveform?  I'm aware of that, I just happened to not have it there when I took the screenshots. 

I found that depending on where I located it within the waveform, I got different results, and as I assume it triggers on every rising slope, I could see how I might get different results based on where I put it and that because the peaks were different heights and those were really what I wanted to know the frequency off, I couldn't be sure no matter where I put the trigger it would be reading those and not some higher frequency perturbations.


--- Quote ---Channel AC input coupling is also useful for viewing ripple/noise on a DC rail at low V/div settings without needing to adjust the offset/vertical position.

--- End quote ---

I don't know what that means.  I only got the scope a few days ago.

--- End quote ---
Select channel #, input coupling, AC.
This removes the DC component of the waveform and displays the AC component riding on the DC.
T3sl4co1l:

--- Quote from: Starlord on July 04, 2016, 09:29:13 pm ---Well the I2C communications are 400KHz, and the speaker cable isn't 7' long.  I don't know if the cable can re-radiate the noise from the speaker, but I already mentioned I would be dealing with that.
--- End quote ---

It will, as well as the power cord and signal input.

Likely, the waveforms you are measuring are common mode.  Try this: clip the probe tip to the ground clip, and poke that to circuit ground.  You'll probably see the same waveform.

In fact, almost all places you measure those blips, you'll be measuring the (common mode) ground error -- you have to mentally subtract that from the measurement to see the actual signal.

Again, like I said in the message -- this isn't something that can easily be solved by ferrite beads.

As an overall business decision, this seems to be the worst possible path, for someone struggling to make ends meet:
- You've spent a year in development (possibly discounting your labor to zero in the process?)
- You're creating a product which will earn a small amount of money immediately (past success, I guess, is a few hundred/mo?  Are you projecting more for this?)
- You're driving head first into a huge liability, likely worth more than the total sales.
- Can you absorb that cost, when (not if) it comes up?

If the answer is no, then as the movie said: "the best decision is not to play".

Your best business (and life!) decision should be, find a job, maybe some contract work -- anything to fill immediate demands.  Keep working on projects in what spare time you have, and when they're ready for the world, give it a try.

Regarding Chinese products:
Importers are liable for that cost.  Do they cheat by diversity?  (Does it matter? Would the FCC classify the products by type and levy a bigger fine for the collection?)  Do they turn enough profit anyway, from the sheer quantity and variety of imports (that presumably don't violate rules) that it's a net win?  (Remember, legal costs are just another operating cost -- whether the parties are rivals or the government!)  Repeated violations by a company might earn them more trouble, but so it goes.)


--- Quote ---Anyway, I said I'll be adding the ferrite bead filter to the amp.  And there's not much I can do about the I2C if you think that 400KHz data at extremely low power is an issue.  Do you think that's an issue?  It wasn't clear.  I was considering making a switch to shielded cable at some point, but only if I was still getting glitches.  Surely 3mA a 5V is not going to make a powerful enough transmitter to catch anyone's attention.

--- End quote ---

It can.  The falling edge has much more than 3mA peak at the driver, and the signal levels to meet Part 15 Unintentional Radiator are in the millivolts (and ~microamperes).

The distance between functional noise levels (low enough that the system doesn't interfere with itself) and regulatory noise levels (the system doesn't interfere with most radio services) is a huge gap.  60dB or more -- a factor of a thousand in voltage or current!

Tim
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod