[BeaverRat]

Hi everyone,
A few months back I designed and tested a DAC based around the PCM5102A by TI. Ever since I've been using it with my JDS Labs CmoyBB amp and it sounds quite wonderful, but I'm not a fan of carrying both DAC and AMP around constantly. Thus I've been working on expanding the DAC PCB to also include a cmoy amp section. The actual amp is complete, I just need to find a way to power it. I could go with the classic 9V battery and TLE rail splitter, but since this will always be used with a desktop computer, having a battery on-board would be an inconvenience. What is a nice way to get roughly +-4.5 V for use with an OPA2227 or similar amp off of the 5V USB VCC? Here is what I had in mind:
 
1) Inverting charge pump followed by a LDO reg for the (-) rail and then just an LDO reg for the (+) rail.
2) This part (LTC3260) http://www.linear.com/product/LTC3260
 
Thanks!

 

[Avro_Arrow]

Take a look at some of the concepts used here
 

 

[BeaverRat]

  Take a look at some of the concepts used here
 

Thanks - looked through that thread earlier...
 
I was really hoping someone could comment on the use of the LTC3260. Unlike that thread I only need +- 4.5-5 volts on the rails. Are the LDO outputs of this IC clean enough? If I remember correctly they are 100uV RMS. It would be pretty great if this could work as it's such a small footprint that houses the inverting charge-pump and both LDO's. With that said I wouldn't be opposed to adding more filtering if necessary.

 

[wakibaki]

What is audible is subject to constant dispute. There is a lot of discomfort about switching power supplies. It depends anyway on the PSRR of whatever the other circuits are. Anecdotes abound about problems with USB power.

You can build it and measure it or listen to it. If it looks OK measured using your soundcard as input, or if you simply think it sounds OK, then use it.

w

 

[wakibaki]

LTC3260 looks like an interesting device. No sooner will it be built into an audio device than it will be found to be 'audible' in some detrimental manner.

To avoid criticism in some quarters, you should take care that any USB device you design is carefully thought out to exclude any noise from the PC or other host device, particularly on the power lines, and prevent it passing through, as this is often held to be a cause of degraded sound quality in USB devices. Since this can be quite high frequency it needs to be recognised it may readily propagate in free space.

w

 

[BeaverRat]

LTC3260 looks like an interesting device. No sooner will it be built into an audio device than it will be found to be 'audible' in some detrimental manner.

To avoid criticism in some quarters, you should take care that any USB device you design is carefully thought out to exclude any noise from the PC or other host device, particularly on the power lines, and prevent it passing through, as this is often held to be a cause of degraded sound quality in USB devices. Since this can be quite high frequency it needs to be recognised it may readily propagate in free space.

w

Another option is to go the discrete part route: 

• TPS72301 for -4.5V rail
• LP5907 for +4.5V rail
• TPS60403 for Inverting Charge Pump

This would actually provide a system with even lower noise. What kind of filtering should be done on the USB 5V line before or after the regs/charge pump?
 
Thanks

 

[2JZ-fi]

I was wondering about this as well. Essentially, I wanted USB3 amperage without actually having a USB3 device. The specification for USB says that the host device must tell the computer how many unit loads (5x100mA for USB2 and 6x150mA for USB3) of current the port can actually send. I also read that most ports don't actually follow this. I performed a shorted output test on my work computer and found that the port will supply ~2A short circuited with no attachments. Jackpot. I designed my DAC/amp with 2 x http://www.digikey.com/product-detail/en/IR0509S/1470-1535-5-ND/4487916 for isolated left and right channels. I then filtered that well and used 2 very low noise LDOs. The USB -> I2C -> DAC stages I used were essentially the same as the pupdac. I diode ORed 2 power inputs (one from USB, the other from an external source) just in case I needed to run the DAC/amp on a system that couldn't supply the power. On extremely high gain, there is no hiss or noise noticeable. For its simplicity, the IR0509S worked very well, however, quiescent current is terrible at ~70mA no load. I plan on redoing the board with more efficient discrete components, but that's beside the point. If you filter effectively, you can not only create two voltage rails, you can completely isolate your power supply from the computer's noisy supplies by using those all in one switching converters.