• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

LDO vs Buck Converter for digital audio

MCH

Major Contributor
Joined
Apr 10, 2021
Messages
2,641
Likes
2,251
Hello,
I am in the process of reverse engineering a HDMI audio extractor to build a HDMI audio capture card. The extractor is USB powered and has two identical SOT 23-5 buck converters that provide 3.3 and 1.8V.

1707235877511.png

I could not identify the parts but the pinout and circuit match, for instance, Texas Instruments TLV62569 or this one:

I am not an experienced DIYer, but what i see most often in this sort of small audio devices are LDOs, that seem to be all advantages: low noise, few components, etc.
I am very intrigued of why they have used buck converters, that go with one inductor each (see picture) and a small army of ferrite beads and loooong arrays of capacitors, i assume, to do the necessary filtering. The device works with as little as 5V 0.5A input, so I don't think it is related to efficiency or high currents. As mentioned, the device is USB powered (5VDC), there isn't a large voltage drop to 3.3 and 1.8 Volts.
What are the reasons one would chose a buck converter over a LDO for a small USB powered digital Audio/video device?
And most importantly, as i don't have much space in my design, i wonder what are the chances to screw it if i forget about the buck converters and use a couple of LDOs?
thanks for any comments/advice.
 

wwenze

Major Contributor
Joined
May 22, 2018
Messages
1,310
Likes
1,871
Precisely because it has to work with as little as 5V 0.5A input.

A buck converter can generate 1.8V 1.38A from a 5V 0.5A input. With a linear regulator, it can only generate 0.5A
 
OP
M

MCH

Major Contributor
Joined
Apr 10, 2021
Messages
2,641
Likes
2,251
Precisely because it has to work with as little as 5V 0.5A input.

A buck converter can generate 1.8V 1.38A from a 5V 0.5A input. With a linear regulator, it can only generate 0.5A
Ah ok, that makes sense. Thanks.
If that was the only reason, as I will have 3A available , I guess I could do with LDOs? Or might be something else?
 

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
3,372
Likes
3,317
Location
.de
Power dissipation is another issue. If you're regulating 5 V to 1.8 V at 0.5 A, you have to dissipate 1.6 W already, which is a challenge for small surface-mounted regulators and results in heat buildup in a small enclosure. (That's an efficiency of 36%.) Not a big deal for powering small-fry circuitry, but complex digital stuff...
 
OP
M

MCH

Major Contributor
Joined
Apr 10, 2021
Messages
2,641
Likes
2,251
Power dissipation is another issue. If you're regulating 5 V to 1.8 V at 0.5 A, you have to dissipate 1.6 W already, which is a challenge for small surface-mounted regulators and results in heat buildup in a small enclosure. (That's an efficiency of 36%.) Not a big deal for powering small-fry circuitry, but complex digital stuff...
Makes sense as well. Actually the device gets relatively warm when in use even with the buck converters, and the enclosure is metallic. Maybe better stick to the original circuit, at least foe the 1.8V
 

DVDdoug

Major Contributor
Joined
May 27, 2021
Messages
3,011
Likes
3,956
If you don't have noise problems, it's OK. ;) USB power is often noisy but it's a bigger problem with high-gain preamps in audio interfaces and less of a problem with line level signals. (And it's only a problem with the analog electronics.)

If you are stepping-down DC, it's theoretically better to keep it DC and a linear regulator works GREAT for additional filtering.

I saw an amplifier schematic once where the phono preamp section had separate-dedicated linear regulators (in series following the main linear regulation).

If you are building an AC power supply you either have to filter the 50/60Hz AC or the high-frequency switching. The higher frequencies are easier to filter-out and usually above the audible range.
 
  • Like
Reactions: MCH

AnalogSteph

Major Contributor
Joined
Nov 6, 2018
Messages
3,372
Likes
3,317
Location
.de
The ranking in terms of efficiency goes like this, from best to worst / in order of descending typical load current:
Buck converter
Linear series regulator (LDO or not)
Linear shunt regulator

Shunt regulators can be useful for loads that draw a minuscule amount of current but are critical in terms of noise and stability, commonly oscillators and such (phono stages may qualify as well). At low single-digit mA, the regulator transistors in series regs may become quite sluggish, causing output impedance, regulation and bandwidth to suffer. You usually have the feedback network to keep things from going off the rails completely, but those may be drawing 5 mA and that's still not a whole lot (considering that e.g. a classic 78xx type is rated to 1.5 A and a 78Mxx still is a 500 mA affair, a 100 mA 78Lxx finally is in the ballpark).

People have certainly run power-hungry digital electronics on linear power supplies in the past (you are quite likely to find them in 1970s minicomputers, as well as a number of classic early to mid 1980s home computers including the popular Commodore 64). It is something that quickly went away by the late 1980s though, as aside from the added complexity / more tricky design, switch-mode operation had very obvious advantages to offer. The original Amiga 500 PSU for example had a transformer and 7812/7912 regulators for the +/-12V rails (1 A on +12 V was apparently deemed sufficient), but when it came to providing up to 2.5 A on the +5 V rail they resorted to using a buck converter.
 
  • Like
Reactions: MCH
OP
M

MCH

Major Contributor
Joined
Apr 10, 2021
Messages
2,641
Likes
2,251
Thanks a lot. I think I am decided to keep the buck for the 1.8V and switch to an LDO for the 3.3V.
My only limitation in this project is PCB space, and being able to get rid of at least one inductor is already something.
The project is a hat for a raspberry 5, that now comes with a 27W power supply that can deliver more than 5A at 5V. Power will definitely not be a limitation. Even if I wanted to power the hat through USB from the pi, the pi 5 can deliver 1.6 amps.
I am keeping the buck for the 1.8V in case the heat generated with a LDO was too much. But I also doubt this would ever be a problem as the size of the hat will be that of the pi and I believe it should be big enough + the enclosure will have a fan -my current enclosure with a pi4 plus other stuff inside has the smallest noctua 4x4 fan that working at the minimum speed keeps the temperature of the processor 20C less than without fan-

Good learnings from your comments, much appreciated.
 

Roland68

Major Contributor
Joined
Jan 31, 2020
Messages
1,449
Likes
1,271
Location
Cologne, Germany
Thanks a lot. I think I am decided to keep the buck for the 1.8V and switch to an LDO for the 3.3V.
My only limitation in this project is PCB space, and being able to get rid of at least one inductor is already something.
The project is a hat for a raspberry 5, that now comes with a 27W power supply that can deliver more than 5A at 5V. Power will definitely not be a limitation. Even if I wanted to power the hat through USB from the pi, the pi 5 can deliver 1.6 amps.
I am keeping the buck for the 1.8V in case the heat generated with a LDO was too much. But I also doubt this would ever be a problem as the size of the hat will be that of the pi and I believe it should be big enough + the enclosure will have a fan -my current enclosure with a pi4 plus other stuff inside has the smallest noctua 4x4 fan that working at the minimum speed keeps the temperature of the processor 20C less than without fan-

Good learnings from your comments, much appreciated.
There is another reason. The parts used are in the cents rather than the $ range. Using LDOs would be several dollars more expensive.
That doesn't make much of a difference with one device for do-it-yourselfers, but with 1,000 or 10,000 devices it's something different.
 
Top Bottom