LRC decoupling network on 5V USB rail.

Status
Not open for further replies.
Jump to Latest
M

MaverickPT

Member
Joined 2018
#1
Hello!

I'm building a DAC based on the PCM2912A (and have been nagging you guys because of it :shy: ). It's a learning project so I'm experimenting a bit in it's design. I mostly followed the schematic provided in the datasheet/evaluation module, but I still want to experiment a bit.

Since it will be powered by the USB port from my laptop, I thought it would be a good idea to have some filtering on the power rail. A RC low pass filter is out of the equation since the consequent voltage drop would most likely affect the IC. I thought of using a LC filter and came across this cool article.
http://www.designers-guide.org/Design/bypassing.pdf

It suggest the following circuit:

z71ZfQ9.png

For the damping resistor the article sets the following equations:
s4VNHf4.png


With these values in mind I arrived to the following values:
(I actually used a value of 12uH in an attempt of taking into account all parasitic inductances[pcb traces, leads, etc]. Is 2uH a big enough value?)

Cbyp: 0.1 uF (typical ceramic bypass cap)
Cdamp: 1000uF
L: 10uH
Rdamp: 1 Ohm


My biggest doubt lies in the inductor size. Is 10uH a fitting value? Should I use a different value? Should I use a LCR network at all?

Thanks you advance and Merry Christmas!!
:santa2:
 
#3
trobbins said:
Perhaps better to start with a known bench performance of your DAC equipment as it is.

Are you able to measure the output spectrum of that equipment?

Are you able to repeat that when connecting the DAC in to your audio system?
Click to expand...

Unfortunately right now I have neither the DAC (still designing it) or the equipment to do such measurements.

Although I already gave up the idea of using a LCR filter, because someone on my reddit post (I cross-posted) pointed out that on VBUS the max allowed capacitance is 10uF, due to the in-rush current. And since I don't want to bust the USB ports of my computer I'll have to make due.

I'm thinking on making a separate 5V power supply to feed the DAC later, but if I don't notice any noise at all I'll also skip that.
 
#4
How about an LR series network between USB and DAC and a capacitor for decoupling? TI recommends a circuit like that in Figure 37 of the datasheet, with L1 providing the inductance and the resistance. The spec for L1 doesn't make much sense, though; they only specify an upper limit to the resistance, but with too little resistance the Q of the LRC network becomes quite high.

TI recommends 1 uH, max. 0.6 ohm and 1 uF. When you change that to 0.47 uH, 0.6 ohm and 2.2 uF, you have a DC resistance that TI is still OK with, a reasonably low Q for the filter, a capacitance below the 10 uF limit (including the five 1 uF capacitors connected to locally regulated 3.3 V supplies) and similar filtering as with TI's values. The cut-off frequency is then 156.5 kHz, so audible frequencies will not be filtered, but HF rubbish that could produce audible noise by intermodulating with the out-of-band noise of the sigma-delta modulators is still filtered off.
 
#7
MarcelvdG said:
How about an LR series network between USB and DAC and a capacitor for decoupling? TI recommends a circuit like that in Figure 37 of the datasheet, with L1 providing the inductance and the resistance. The spec for L1 doesn't make much sense, though; they only specify an upper limit to the resistance, but with too little resistance the Q of the LRC network becomes quite high.

TI recommends 1 uH, max. 0.6 ohm and 1 uF. When you change that to 0.47 uH, 0.6 ohm and 2.2 uF, you have a DC resistance that TI is still OK with, a reasonably low Q for the filter, a capacitance below the 10 uF limit (including the five 1 uF capacitors connected to locally regulated 3.3 V supplies) and similar filtering as with TI's values. The cut-off frequency is then 156.5 kHz, so audible frequencies will not be filtered, but HF rubbish that could produce audible noise by intermodulating with the out-of-band noise of the sigma-delta modulators is still filtered off.
Click to expand...

I saw as much but the lack of an actual resistor left me wondering if a "proper" LCR network would be better. Ended up dimensioning something way exaggerated it seems. I'll try to measure the resistance in the inductor that I'll choose when eventually I order the components. If the resistance is too low I'll add the resistor in front of the capacitor like show in the above circuit
 
#8
I think the work 'bust' can have the meaning to damage. Yes the inrush is just related to possible hassles with other devices on the same USB port losing connection, or not operating correctly. Whether that is a problem for you would depend on if you connect other peripherals to the same USB bus, and if those peripherals were affected. The L in an LC may likely alleviate the USB bus disturbance.
 
#9
Sorry, I did meant damaging the USB port, but I guess exaggerated.

Since I'll have It connected to my laptop it will be a bit annoying if my external mouse drops out when I connect the DAC.

I'll still place the LCR network but complying with the requirements
 
#10
trobbins said:
I think the work 'bust' can have the meaning to damage. Yes the inrush is just related to possible hassles with other devices on the same USB port losing connection, or not operating correctly. Whether that is a problem for you would depend on if you connect other peripherals to the same USB bus, and if those peripherals were affected. The L in an LC may likely alleviate the USB bus disturbance.
Click to expand...

I use USB for many projects and have done some nasty things to it during that time. I shorted out a 3v3 regulator to zero volts and the PC shouted that there was a problem with the bus but it didnt break it.
I suspect there is some current limiting going on or you could start blowing up PC USB ports which could be expensive.
 
Status
Not open for further replies.
Top Bottom