USB supply filtering
Hi,
I've got a few different projects going on at the moment that are all USB powered so I thought I'd do some measurements of the USB supply noise and try some ways of improving it.
Initially I added bulk capacitance, 4400uF, to reduce the noise. This worked but obviously isn't really a solution particularly when the USB 2.0 standard specifies a limit of 10uF in parallel with the supply line to limit inrush current. The specification states that bulk capacitance can be used if a soft start method is implemented so that the inrush is not exceeded.
This information comes from an FTDI application note (found here) and in this they show an example circuit for filtering the supply line.
This circuit consists of a few capacitors and a ferrite bead, specifically BK0603HS330-T (Digikey link). I don't have any of these on hand but I thought I'd try a couple of inductors that I have. Specifically the B82141A 33uH(Digikey Link) and 77F1R0K 1uH (Digikey Link). I'll order some ferrites to try but for now I'm just going to duplicate this circuit with the two different inductors instead.
First things first, let's look at the noise on the USB supply line from a few different sources. First up is an old Samsung phone charger.
Next is the USB port on my Surface charger.
Next is the USB on the Surface itself running off the battery alone.
And finally the same USB port but with the Surface plugged into the mains.
One thing they all have in common is, they're quite noisy! For this experiment I'm going to be using the phone charger. The results are below.
Capacitors but no inductor
With the 1uH inductor
With the 33uH inductor
As I said earlier, I'll order some ferrites and do some more testing (with more consistency in the scope setup as well) but it is clear that this little LC circuit makes a huge difference to the noise on the supply line. The results were just as good as the bulk capacitance I tried earlier but in a much more elegant, and standard conforming, solution. I'll also try some more accurate probing techniques in future tests.
Justin
I've got a few different projects going on at the moment that are all USB powered so I thought I'd do some measurements of the USB supply noise and try some ways of improving it.
Initially I added bulk capacitance, 4400uF, to reduce the noise. This worked but obviously isn't really a solution particularly when the USB 2.0 standard specifies a limit of 10uF in parallel with the supply line to limit inrush current. The specification states that bulk capacitance can be used if a soft start method is implemented so that the inrush is not exceeded.
This information comes from an FTDI application note (found here) and in this they show an example circuit for filtering the supply line.
This circuit consists of a few capacitors and a ferrite bead, specifically BK0603HS330-T (Digikey link). I don't have any of these on hand but I thought I'd try a couple of inductors that I have. Specifically the B82141A 33uH(Digikey Link) and 77F1R0K 1uH (Digikey Link). I'll order some ferrites to try but for now I'm just going to duplicate this circuit with the two different inductors instead.
First things first, let's look at the noise on the USB supply line from a few different sources. First up is an old Samsung phone charger.
Next is the USB port on my Surface charger.
Next is the USB on the Surface itself running off the battery alone.
And finally the same USB port but with the Surface plugged into the mains.
One thing they all have in common is, they're quite noisy! For this experiment I'm going to be using the phone charger. The results are below.
Capacitors but no inductor
With the 1uH inductor
With the 33uH inductor
As I said earlier, I'll order some ferrites and do some more testing (with more consistency in the scope setup as well) but it is clear that this little LC circuit makes a huge difference to the noise on the supply line. The results were just as good as the bulk capacitance I tried earlier but in a much more elegant, and standard conforming, solution. I'll also try some more accurate probing techniques in future tests.
Justin
Comments
Post a Comment