Stacking switch mode regulators for audio mixer/distribution

[Shadow351]

I'm working on a design for an audio mixing/distribution system, and I'm trying to figure out what to use for a power supply.
The layout uses a USB hub controller (requires +5V), a few PCM2705 USB DAC's (require +3.3V), and several TL074 Opamps for buffering & Mixing (requires a dual supply, thinking I will use +/- 9V). The Idea I had was to use a 19V laptop power brick and stack some LM2596 regulators to build a multiple voltage supply:

• set at 18V relative to the power supply -. (not sure if this is necessary, how symmetrical do the + and - rails of the op amp have to be? Can I just bring the 19V straight in as the + rail? (it's actually 19.5V OC)
• set at 14V, this will be the "5V rail"
• set at 12.3V, this will be the "3.3V rail"
• set at 9V, this will be the "0V rail"
• the power supply - output will be the "-9V rail" for the op amps

Will there be issues stacking switching regulators like this? I thought about just using linear regulators for each rail, but I'm concerned about heat dissipation, as I intend to build this in a 1u rackmount case. Since I can't seem to even buy the LM2596 chips (in the low qty I would need) for less than the off the shelf LM2596 modules, I figure I'll just get those and mount them in my enclosure. Will these cause issues with audio interference? Should I add some smoothing capacitors to the voltage rails?

Thanks

[inse]

Most probably, this won't work as the switching regulator is not capable of operating in the 4th quadrant (positive voltage, negative current), which would be required for the 9V ground reference.
Above that, I question whether those cheapish switching regulator modules are a good choice for audio application.
Estimate the current consumption of your circuit to decide whether switching regulators are worth it anyway.
I propose a research on 'rail splitter circuit'

[tooki]

A laptop power supply probably won’t be clean enough a power source to begin with. I’d avoid switching regulators if you can, especially if you’re not going to carefully design your own power supply. (I use switchers in an audio project, including a boost converter that generates ±5V for op-amps, but I wouldn’t use Chinese modules since they’re not optimized for that.)

It doesn’t sound like any of the parts are likely to need much power, so heat is almost guaranteed not to be a problem.

Can you get the 5V from the USB of the computer it connects to? Follow that with a precision 3.3V linear regulator (or a regular 3.3V and generous filtering) for the DACs. Then you only need to worry about the op-amp power supply, which you could do as a simple linear supply with a center-tap transformer and linear regulators, or as a single rail and rail-splitting. Check the op-amps you’re using for their happiest voltage range. ±15V may be the “traditional” values, but for many op-amps, you can get away with far less, like ±5V, and you can get some models that are happy with even less, like ±2.5V. For audio, you’re unlikely to be dealing with more than ±2V signals, and likely just 2Vp-p.

In another recent project, I used an external AC transformer, then dual half-wave rectification to produce two rails, then into linear regulators to produce ±12V. This is a common approach in audio gear. (My project wasn’t audio, but used op-amps and had to be extremely low noise, so we used expensive precision regulators. No need for that in audio, which is lucky for you, since the regulators I used are now out of stock everywhere with none expected for many, many months.)

[Siwastaja]

Your 3.3V and 5V loads will blow up when the regulators ramp up at the different rate.

[madires]

You can use the LM2596 in an inverting setup to create a negative rail (see the datasheet). And in any case you'll need a lot of filtering when using DC/DC converters to power audio circuits.

[Siwastaja]

You can use the LM2596 in an inverting setup to create a negative rail (see the datasheet).

Or any other buck regulator IC.

The topology becomes inverting buck-boost, then.

[jonpaul]

Avoid SMPS in any sensitive audio application: Powe is so low a well designed set of liner LDO regs is much beterr and dont need EMI filters!

Of course bulk power off line can be an external PC/Laptop/Monito SMPS eg star twith 14, 19, 48V bulk and the reg from that.

Get neg from bulk with an inverting SMPS.

Jon

[Shadow351]

Can you get the 5V from the USB of the computer it connects to?

So this was my original intention, but I'll need more than 500mA. I'm still digging through datasheets, some 'max' current ratings seem really high, and and it's tricky to estimate some other things.
My 5V bus has: an ATMega328 (?mA, looks like the chip itself only needs ~1mA?),  CH340G UART chip (30mA),2x TX50104E Logic Level Shifter (2x100mA this must depend on what its connected to, can't imagine its actually this high), Amp turn on signal (10mA), 2x FE2.1 USB Hub Controller (155mA each = 310mA), 2x CBT3244A bus switch (<1mA total), 10x PCM2705C DACs (10x 500mA, it does have a 100mA mode but you have to set the HOST pin high to write the descriptor ID and that puts it in 500mA mode0), 10x MCP4231-103E digital POT (10x 100mA = 1000mA This seems high?) so a total of 6.5A max, but I can't imagine those DAC's will actually pull 500mA I was hoping they would actually run closer to the 100mA each which puts me at 2.5A.

3.3V Bus only has: 2x CD74HC154 Demux (2x 50mA = 100mA), 2x TX50104E Logic Level Shifter (2x100mA, same as above)

+/- 9V powers 14 TL074 Quad Op amp chips

[H713]

Here's how I'd do it:

If you need to power everything off a DC input voltage:
-Use a potted SMPS isolated brick for the +/- 9 V rails. They're not overly quiet, but usually good enough. You could put a couple linear regulators on the output if you're really worried about it.

-Use another potted SMPS brick for the +5 V rail, then an LDO for your 3.3 V rail.

If you're running off mains:
Buy a MeanWell / TDK Lambda / Delta Electronics SMPS module for the +/- 9 V, then us a pair of linear regulators to get your 5 V and 3.3 V rails.

I wouldn't worry too much about a few linear regulators in a 1U chassis. It sounds like power dissipation is going to be <10W, in which case it really won't matter. I've got way more than 10W in some of my 1U line stages, and they run warm, but nowhere near hot enough for me to be even remotely concerned.

[inse]

Please re-check the datasheets carefully, the values you provided appear unrealistic high to me.
I cannot imagine how the 74HC could consume 50mA.
The values for the PCM2705C seem to be overrated as well.

[mariush]

They make PCB solderable transformers which have 5v and +/- 12v or +/-15v , here's a list : https://www.digikey.com/short/70tz9fpq

The 20VA (5v 750ma  and +/-12v 200mA ) or 30VA (5v 1250mA and +/- 12v 250mA) would probably be enough for your project.

The power consumption on those 5v devices seems really high.

[tooki]

Can you get the 5V from the USB of the computer it connects to?

So this was my original intention, but I'll need more than 500mA. I'm still digging through datasheets, some 'max' current ratings seem really high, and and it's tricky to estimate some other things.
My 5V bus has: an ATMega328 (?mA, looks like the chip itself only needs ~1mA?),
CH340G UART chip (30mA),
2x TX50104E Logic Level Shifter (2x100mA this must depend on what its connected to, can't imagine its actually this high),
Amp turn on signal (10mA),
2x FE2.1 USB Hub Controller (155mA each = 310mA),
2x CBT3244A bus switch (<1mA total),
10x PCM2705C DACs (10x 500mA, it does have a 100mA mode but you have to set the HOST pin high to write the descriptor ID and that puts it in 500mA mode0),
10x MCP4231-103E digital POT (10x 100mA = 1000mA This seems high?) so a total of 6.5A max, but I can't imagine those DAC's will actually pull 500mA I was hoping they would actually run closer to the 100mA each which puts me at 2.5A.

3.3V Bus only has: 2x CD74HC154 Demux (2x 50mA = 100mA), 2x TX50104E Logic Level Shifter (2x100mA, same as above)

+/- 9V powers 14 TL074 Quad Op amp chips

Oh boy, your numbers are way off. The absolute maximum ratings of a device are simply the values where you'll break the thing. They're not what the part is expected to draw during use.

With the logic chips, including MCU, it depends on how much you load the I/Os. If you keep the loads small, you'll be orders of magnitude away from the maximums.

10 DACs? Oof, I'd be very concerned about this even working, software-wise! Regardless, if you look at the actual consumption on the datasheet, it's a maximum of 30mA when driving line outputs. You could easily just have one DAC in bus-powered configuration to report the 500mA to the USB drivers, and all the others self-powered (so they're not reporting any current draw) from the 3.3V LDO. So 300mA, not 5A.

The pots are rated at 0.55mA typ/1mA max, while the serial interface is active. Literally 100 times lower than your numbers. The absolute maximums describe how much current through a pin will fry the device. The only output that contributes to the power dissipation of this thing is the SDO output, so unless you're loading that down insanely, the device is going to use practically no power. So 10mA total.

'HC154 and TX50104E: again, unless loading down the outputs insanely, just a few mA each. Let's call it 10mA for all of them.

The TL074's are specified to 5mA output current (V_OM/R_L = 10V/2kΩ = 5mA). So 5mA x4 x 14 = 280mA total.

So my grand total for your whole logic side is 300mA+10mA+10mA = 320mA, not 6.5A.

With that said, this whole concept seems sketchy to me. Is there any reason for not just using a proper multichannel USB codec IC and a DSP to accomplish everything?

[Shadow351]

I thought those numbers seemed high, but the data sheets make it difficult to determine realistic current draws.

I'm not sure about 10 DAC's, I've written software to stream 5 different stereo audio streams to 4 USB sound cards (cheap chip on board construction) and the sound card built into my motherboard. I don't foresee any issues, but the only way I can think to test is to build it and try. I did look for multiple output/channel USB DAC/CODECS but all of them seem to present the multiple outputs as "surround" audio devices instead of separate stereo audio output devices. I need something that presents itself to the USB host as separate Stereo Audio output devices. Are you aware of any off the top of your head?

I looked at some DSP's for like car audio applications, but I'm just not sure what to go with. Is there a 'jellybean'-esque DSP that is low cost and straightforward to use?

My project is an audio distribution and mixing system. It will present itself to a USB host as 10 USB Stereo audio output devices. 8 of the devices will be zone specific outputs, the next 1 will be a 'global' output that can be mixed into each zone (this allows playing the same audio stream in sync to any/all zones) and the last will also be a global audio output but will be for 'notifications' to all zones (ie the audio volume to all zones will be reduced/muted and the notification will play, then volume will be restored). I also want to have a set of 'AUX' audio inputs that can be mixed into any/all zones; so I can connect a turntable (with built in pre-amp) and TV(s) to play TV audio to any/all zones. The 8 zone mixes will then be output to multichannel/multi-zone amplifiers.

Thanks