<<Curious about your rationale for some elements of the design? >>
The main goal was to fit a high quality balanced input, phantom powered 6x2 preamp/mixer in an altoids case. Almost every decision revolved around how to fit things into the case. As it stands now all I have to do is punch 6 holes in a 2$ mint tin and the casework is done.
Originally I was using That1510 but did not think through the pots carefully enough. There was some incorrect information on the net (hard to believe) about converting a linear pot to a reverse log, so the Bourns linear pots were a no go. There was one pot made by BI technologies that would have worked, but MOQ was too high and they are impossible to find. I reworked things a bit to fit alpha pots, but at that point I already had my feet wet in the That1580/5171 and I wanted at least two channels of well matched gain which pots can't do.
The Ltc3625 charge pumps worked well, and 2-3 of them would have been fine for the 1510 version. I had that whole thing about 80-90% done when I decided to switch gears.
<You don't say what the new chip is, but THAT1610 is 6mA/channel so that would be 36mA plus summing, say 40mA total, times 30V is 1.2W. Six-channel phantom at a typical 3mA >
Schoeps 8mA. That 1510 max current with no signal is 8mA plus wiggle room.
<<is another 0.9W. Plus conversion losses (that LT module has very low efficiency at the low loads you have, it's designed for loads in the 1A range), you're burning about 4W total? That's not an insignificant amount of heat for a box that small, and it will make a 9V battery cry.>>
at 225v-ish Ltm8049 is near 80% efficient.
I never intended this to run off a battery, certainly not a single 9V pp3. I was entertaining the possibility of the the first version to run from a usb rechargeable battery pack, but that wasn't a priority.
<<Using a bipolar supply is good insofar as it avoids the need for output capacitors, but sometimes an output capacitor can save the life of a circuit from an external mispatch, even if it's not strictly needed. My entire life is single supply. But if you stick with bipolar, I think you will find there are plenty of charge pump ICs that can do that job easily for the current output you need. They will hit close to 90% efficiency in the process and have almost no Iq, and much lower output ripple than the LT module.>>
At one point I was looking into running all of these ic's from single supply, but it was over my paygrade and I really did not want to create any design headaches. The ltc3265 was the only charge pump I found that did +-15v. I have a few of those working cards now in the ic graveyeard.
That1580 preamp: 10mA
That5171 gain control: 11mA
That1286 panpot: 6mA
That1206 balanced line receiver: 8mA
lme49720 summing amp: 12mA
222mA= too high current draw for the ltc charge pumps unless I had one on each channel - which I did consider and probably could have pulled it off.
One reason I didn't want to do that is to not bus supply voltage since it was coming from a wart and I'm not confident in layout or filtering.
I tried to challenge myself with this power supply just to see if I could get 4 rails in that small size. Plus noise and ripple should be lower with the better regulators - yea better spec but probably not audible. There is a very wide nominal input voltage on the ltm8049 and the lt8330 (phantom boost switcher), so theoretically this box can run anywhere from 5v-20v or higher. Heat might be a problem at 5v but I'll test it out later if it ever works at 12v.
<<You might design backwards from your output requirement instead. Chips like THAT1610 have low input noise but much higher output noise, so their noise performance isn't exceptional until you have gain of at least 30dB. There are plenty of chips that will equal or exceed that performance with less than 2mA draw. The second thing to think about is your operating level--what are you driving? Do you need 30Vpp peak output? Most converter chips are only running 5Vpp max input. So you could find yourself in a situation where you have to run higher gain than you otherwise would to overcome the output noise of your amp, only to attenuate that output to keep your ADC happy. That works, but it burns a lot of battery life.>>
It's not a battery powered box and never wanted to be. Yea you're right I don't need 30Vpp, but it was one of those just follow the nominal datasheet values. I don't care that much about wasting some power, it's a fairly low power device as is. The main thing was just to make something really small, very good sounding, and power it relatively easily with a wall wart. No actually I just wanted to use the That1510/1580 chips because they sound really good to me.
<<Instead, think about a topology that is designed around only needing 20-30dB of gain, and since you already have a digital circuit, why not go whole hog and use your own ADC? Now you can define your operating level and run the whole circuit on 5V. If you choose carefully you can find a DSP chip that will do everything you could dream of in that space (sum, pan, EQ, effects), controllable via UART or I2C via Bluetooth.>>
There's a big difference between spi gain control on the 5171 and rolling your own adc. The thought did cross my mind, but this project has taken up too much time as it is and I'll leave adc's to smarter people. If I can fit an adc into this box maybe in the future I'll give it a go - especially if it involves 5.6Mhz DSD stereo out. I have no use for eq, effects, digital summing, digital panning, or anything in the digital realm aside from adc. The last thing I want to do is get into dsp programming.
Just wanted to keep things as simple as possible, make a really small and good sounding analog box. I like the features now, proper 48v phantom supply, switchable panning, phase reverse, exact gain setting, switched gain values on the summing section, etc. Anywho it's a ways off from working. I finished soldering one of the preamp cards today. Had some problems with the smaller qfn20 attached to the ground plane - ground copper wicking too much heat away.