My posts where not meant to derail the subject matter. Perhaps I should explain. It seams to me that with case real estate being diminished already by the components required to increase the voltage to the desired level as well as the DAC and the thought of only being able to use the amp while tethered to a PC, that the smaller the battery the better. With some type of internal battery in the design you would have to option of using the device as a portable amplifier as well.

I'm not really in the market for a portable amplifier. If I were, I'd instantly run to the Mini^3, but that's the reason I got the iPhone 2G. I could very well carry an amp around, but it's just not worth the trouble for my tastes. I'm usually stationed at a table when I study and do homework, so using it with my laptop would be more ideal than carrying the amp around. I was carrying around an iPod touch 1G and a Treo 680 in the same pocket for a month before I got the iPhone. I decided then and there that portable was meant to be portable and got an iPhone; each day's pockets were different from the last, and carrying two physically distinct devices in one pocket was tough on my sense of pragmatism. I'm not really looking for a portable amp, and if I were, I'd be using the Mini^3.

I'm now wondering about the merit of isolating the DC/DC converter. I'm looking at this TI TPS6734 and it seems to have almost everything I'm looking for besides being isolated. From what I understand, the isolation is meant to protect the USB from ... something. Mistakes? Couldn't a fuse do the same thing? With this TPS6734 regulator, it gives me 120mA output at a pretty solid 12V which I wanted to regulate down to 10V. At 12V with a full output current of 120mA at 85% efficiency, that draws 338.8mA.

I1 = (12V/5V)*(120mA)/85%
I1 = 338.8mA

That's plenty of room to play with. I know dsavitsk mentioned some reservation about inrush current in error401's USB-powered amp thread, and this one has a soft-start feature (see Figure 4). It seems almost too good to be true, so I'm wondering what I'm missing. The TPS6734 has the right input and output voltages with a respectable current capability and efficiency, it's in a smaller package at a quarter of the price, plus it's readily available. There is the bit about it not being isolated, but the HPDAC seemed to have survived well enough without the isolation. I am still planning on designing on a single supply with virtual ground, so I'm wondering what precautions I have to take if I do go this route. For prototyping, I could throw a fuse onto the board before applying power, which would at the same time give me a convenient place to check current draw on a DMM if it doesn't blow. What do you guys think?

Hm, I think that I would be shorting 5V to ground. *sigh* I will look into split supplies as an alternative. I can probably find other suitable isolated DC/DC converters, but I should do a little research thisaway before going ahead.

The isolation is useful because you'll be generating a positive voltage. With a virtual ground, you have to be very careful about what you connect your amp to. Most gear attaches ground to earth at some point, so you'd be effectively connecting your virtual ground (at 1/2Vcc in relation to earth ground) to earth. Not a good situation. With the isolation, the virtual ground (if you use one), or the centre-tap on those DCP02s will float, and the next stage can pull it where it needs to be. Much safer.

Most (I'd hazard to say all, but I'm sure some super cheap motherboards don't) USB ports will limit current. I've shorted mine many times, and the current limit on them is somewhere around 1.5A (on my board). You shouldn't need to protect the USB ports, and inrush current isn't a big deal because it will be limited by the USB port.

Isolation is also used to reduce the noise transmitted from the dc-dc converter. That is one reason why the high quality converters have transformers.
The Murata seems very good, but did cost $20 CND. I'll get the power supply built tomorrow and see how it is.

I'm wondering about the BantamDAC's output. Its ground will be sitting at USB V-, which is going to be the booster's negative rail. Will I not be shorting the 5V virtual ground to V- if I connect the separate grounds together, even if it is isolated? It can float to whatever value it needs to be, but would it still work if it floated right to the rail?

Though the DCP02 family of DC/DC converters doesn't have a 5V-to-12V model, there is still the readily available DCP01 that does have such a one. Again using the AD8397 operating at a regulated and filtered 10V as an example, there is only a 24 mA draw from the DCP010512B which can provide up to 83 mA. Full load through this particular booster means 234 mA to the USB.

I1 = (83mA)*(12V/5V)/85%
I1 = 234 mA

It was asserted earlier that current from the USB could be safely ignored, but this calculation is presented for the sake of knowing anyway. Furthermore, this means that a third channel can be safely added into the mix again. Huzzah!

Here is what my free time has produced in the past month.









I obtained permission from amb to use the Mini^3 amplification circuit. The voltage is boosted by the DCP010512 to 12V, filtered by a CLC LPF with a corner frequency of about 500 Hz and regulated by the TL750L10 to 10V. The virtual ground is up top. The isolated V- is on the bottom and separated from the case, but the USB ground is connected to the case.

Is there anything that I should try/keep in mind/consider? For example, what the best way of presenting the circuit without all the JPG artifacts? Please poke the holes in my design now before I get it prototyped. I have to rush off now, so I'll expand on the design maybe next week.

Nicely done!
I don't have the knowledge to help with your pcb layout, but the implementation looks good.
Maybe add some values for the power supply so we can see exactly what's going on there.

Something I noticed immediately after posting the pictures is that I can probably connect the virtual ground to the USB ground. The output voltage is isolated, so it float to whatever voltage no problem while I clamp down the virtual ground. That way the whole bottom can be routed as a ground plane and connected to the casing, much like the Mini^3. That should be okay, right?

Also, I can probably snake my way from the USB power input at the back to the front for a board-mounted LED like the Mini^3. It's only drawing a couple mA, but I figure it'd be best to save the precious boosted power for the amplifier itself.

There are lots of capacitors in the power supply just because there's room. I wonder about the inrush current of having all those capacitors. I might even be able to make them bigger speccing them for a lower voltage. For instance, the electrolytic C4+, C4-, C5+ and C5- only see half the regulated 10V, but I specced it for 16V in case of some sort of fault. It uses the same 470uF caps as the Mini^3. As you can see, I borrowed quite a bit from amb's design.

For the line filter, I used the calculator from,

Chebyshev Pi LC Low Pass Filter Calculator

Ripple occurs at 800 kHz, but the cutoff frequency of the CLC filter I've planned has a corner frequency of 580 Hz. Noise shouldn't be a problem as far as I know. Plug in 580 Hz, 1 ohm, 0.01 dB, and 3 components.

The 10V regulator is an LDO type with a dropout of 0.6V at 150mA. The DCP01 drops to 11.5V at 100% load, so that shouldn't be a problem either.

Thoughts? I think better when I do it out loud, so it helps when I can discuss with people. I have lots of little things to clean up, but I just wanted to get some of the technical stuff down pat now before it's too late. My biggest fear so far is about the isolated voltage somehow becoming +/- 5V relative to USB ground with the use of the TLE. That's what happens when you use an isolated DC booster, right?

Well, nobody else around here seems to like the idea of using USB voltage to power an amp, because they're all glass-half-empty naysayers , but I for one happen to think it's a spiffy idea.


Quote:



If I understand the science correctly, that's a bad idea, and the two grounds should remain fully isolated to avoid ground loops and other less pleasant events.

Quote:



I personally would put the LED on the isolated, boosted end of the DCP, as that way it tells you not that your computer is supplying voltage to the USB port, but that the power converter is working as it should, and hasn't gone into shutdown for whatever reason (overtemperature and undervoltage for sure; I'm not sure how it handles overcurrent.)

Quote:



That could be an issue; I'm not real clear how the DCP chips handle peak current draw. What are you using for C1 - 2,2uf, like in the datasheet? I wonder if adding a much larger cap in parallel with that, and moving the power switch to be in front of the DCP, would help; wouldn't the 5V reservoir cap, basically, then see at least part of the start-up current rush?

Quote:



I think it's a nice idea - I came up with an identical power design a day or two ago, not having seen this thread before. Have you considered skipping the TLE entirely, and using one of the split-voltage DCP chips? The DCP0112DP, I think it is, is 5V in, fully isolated +/-12V out. It'd simplify things a tiny bit, IMO. Otherwise, you should be fine so long as the USB ground never connects to anything on the "other side" of the isolator - +V and 0V in, and +V and 0V out, but the two 0V terminals shouldn't connect, otherwise you're defeating the whole purpose of the isolator... I think.

That is a lot of capacitance to have on the power supply. You're gonna need a resistor in there or something to keep that USB port from freaking out when you plug that in. Also, it hasn't been mentioned yet, but not all computers will run this thing flawlessly. Some USB hosts won't supply more than 100ma unless the device requests it and some others will supply 500ma but if the device is connected the computer won't go into and out of suspend smoothly if you're using more than 100ma.

Quote:



I really don't see the problem. The 2move can be completely powered using USB, and that means the internal DAC and amp powering 250 ohm headphones. Sounds very good.

Quote:



<DonaldSutherland>
Again with the negative waves!
</DonaldSutherland>

Not counting the initial inrush current, which I'm unsure how to model, the high performance version of this thing will only draw ca. 80mA from the USB port, if that; the better-performing (on paper, anyway) extended-runtime even less - 35mA or so. Hence the oh-so-scary USB 100mA "wall" that one could theoretically come up against is largely a non-issue. You can do the math yourself - the typical current draws for the Mini^3 are here; the DC-DC converter gets 5V from USB and produces 12V, at an efficiency of 80% or slightly better...

Quote:



There won't be ground loops. Isolation means exactly that - the supply is isolated.

Quote:



I agree with joneeboi - don't waste that current. Keep it for headroom.
Do you really need a LED to tell you that the power supply is working? Personally I'd be able to tell that something is wrong when the amp is turned on and no sound is coming out
Besides that there's no reason to think the dc-dc converter will suddenly die. By this logic you'd want LEDs for every chip to indicate whether they are working or not.