Electronics > Projects, Designs, and Technical Stuff
EMF pickup from amplifier in I2C line causing glitches. (Now with scope trace!)
Starlord:
I found this one page where someone tested a Sure amplifier and made some improvements to it which includes a multisim file:
http://www.trevormarshall.com/class-d-tutorial/
I can see he's set it up quite differently from how I attempted to. I have no idea why two AC sources are used, but I'm guessing it has something to do with that ground in the center. I guess maybe it's simulating the push pull that the amplifier does, so the AC on the bottom has it's positive side reference to ground, and is pulling the voltage negative as a result? Sorry, I don't know how to state it properly.
Anyway, I see the pair of inductors and caps that would be on the amplifier output, and I assume R1 and R2 are the DC resistance of the inductors (why isn't that built into the model of the inductor and set as one of its parameters?) but what's C3 and R3 for? And why are there three bode plots? And how did he arrive at those models for those two speakers?
And why are the bode plotters connected to the circuit in such a strange manner? In other words, why do they have inputs and outputs, and why are both connected, and why to those locations?
Starlord:
Well I read back over the article again and I did find where he got those parameters for the speakers. The link was broken, but I found the datasheet for the 4 ohm driver:
https://www.parts-express.com/pedocs/specs/264-1042-peerless-830983-specifications.pdf
Delta:
Simply tweak your circuit so that all the EMI it spews forth is in the ISM bands.... Problem solved!
Starlord:
So I got that simulation from the audio site working. At least I think it's working. Obviously I have no way of knowing if the bode plot it's showing is correct.
However, I did try inputting both the suggested 1uH / 0.1uF and TI's 10uH / 1uF values to use as a reference, and then tried some a few other values for comparison.
For:
10uH / 1uF = 3.4db peak @ 71KHz, -52db @ 1MHz (TI suggestion)
1uH / .1uF = 10db peak @ 522KHz, -9db @ 1MHz (forum suggestion)
1uH / 1uF = 16db peak @ 182KHz, -32db at 1MHz (peak too high? may put amp into overcurrent protection?)
2.2uH / .22uF = 9db @ 255KHz, -25db @ 1MHz (better than the forum suggestion? and tiny 2.2uH inductors that can handle sufficient current exist)
2.2uH / .1uF = 6db @ 372KHz, -18db @ 1MHz (less peaking + uses capacitor value I use everywhere else, somewhat less filtering at 1Mhz range)
I'm not sure if -9db @ 1MHz is a comfortable margin, so I may just go for the 2.2uH inductor instead, possibly with the .1uF capacitor so I don't need to add another part to my BOM.
-9db would only reduce the noise by 1/8th, while -18db wouldr educe it by 1/63rd. That's a pretty major difference. Of course -25db would be 1/316th. But I think 1/63rd will be enough? That should take my +-400mV spikes, and reduce them to +-6mV.
Btw I forgot to mention it earlier, but Mouser offers a free version of Multisim called Multisim Blue, if anyone wants to check that out. It's not perfect, it seems to have only one level of undo which is crazy in this day and age, but aside from that it's leagues better than LTspice.
Starlord:
Ah crud, I just realized that that tiny resistance the inductors have seems to matter a lot. A 4.7uF inductor with 200mV of resistance shows a -1.5db drop in the portion of the spectrum that's supposed to be flat. That would mean a 30% reduction in volume. The 1uH inductor on the other hand with a 47mV resistance only shows a -0.2db drop, which means only a 5% reduction in volume.
Of course, again, I don't know how accurate these graphs are, so perhaps the drops aren't this severe. But I suspect they are.
Thankfully, I may be able to use a somewhat larger inductor without going too big to get that voltage drop down where I want it: less than 50mV.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version