Regulators for digital circuits

LT1762 at <20uV noise, a 17dB improvement over LT1085. 150mA.

LT1964 for negative. <30uV, 200mA.

These are so cheap, small and require so little in the way of support components that you can place a lot of them throughout a board, close to the chips you are powering.

Look through the rest of Linear's catalog; there are some higher power chips that still provide better noise performance than LT1085.
Multiples regulators ?

This is a little bit out of topic, but I was wondering if overall performance of digital audio boards would be improved if we assign one regulator for each IC on a board...

What I have in mind is a preregulator with heavy cap filtering (ripple and RF rejection) and ferrite-beaded on both input and output. This output is distributed all over the board. As close as possible to each IC on board, the preregulated voltage enters a local regulator through a ferrite bead (or an inductance) and a light bypassing ('lytic cap + SMT ceramic cap). Output connected to the supply pin(s) of the IC, with a "Goudreau triplet" (120uF/25V/Panasonic HFQ -or FC- // with stacked SMT 0612 ultra low inductance 270nF/Z5U and 10nF/X7R ceramic caps)...

Has someone here already tried this configuration ? Sould we use high quality local regs, or is the 78LXX series good enough in this config to do a nice job ?

Oh, for the "Goudreau triplet", see here
(Pete Goudreau has done a wonderful job in the 90's improving the performance of Parts Connection's Dacs. A worthwhile reading
That's what I did in my recent DAC. Ea. IC has it's own separate PS from the start: transformer, soft recovery diodes and filter caps. I have total 8 transformers with 16 separate winding altogether and 64 soft recovery diodes.
On the board there are 1000uF HFQ in parallel with .22u caps as close to ea. chip as possible. Active regulators are LT1085 and 1033. I went as far as providing separate regulators for dgital and analog section of PCM1704 DACs which are also paralleled.
This had a quite noticable improvement with better definition and tighter bass.
Overall the sound is exceptional.
"Goudreau triplet",

I urge everyone to read this link and try the decoupling stratagy outlined by Pete Goudreau. The guy is super sharp and a have a paper he wrote on digtal decoupling that is one one my references on the subject. While the regulators are important, the bypass caps are doing all the work above a few hundred KHz
The OSCON caps are great for electrolytics with COG ceramic bypass caps for the real high frequency stuff. We are talking currents in 10s of MHz here and a film cap is not going to cut it.

ftorres said:
WOW!!! Monster design !

Did you use the 1000uF HFQ in input or output of the regs ? Both ? Any ferrite/inductance ?

And did you use toroids for yout transformers? I was told they are too wideband for digital (Noise can pass through, back and forth between xformers)...

I used 1000u HFQ both before and after reg., ferrite beads everywhere. No toroids, split bobbin transformers.

Jocko, I have no idea about those parameters, I thought you might know better.
Harry makes a good point. All the regulator does is to keep the output voltage at a fixed point. It does so with lots of feedback, which means the impedance goes to pot at around 100 kHz or so.

At frequencies above this, the caps do all the work. Film caps work OK in non-critical applications. But if there is lots of switching transients, you need to switch to SM ceramics. Get those tweezers out.


You are the one who needs to decide how tight the regulation needs to be. With low current draw, and some leeway in regulation, you can use a simple discrete circuit. But you will still need caps. And lots of them.

That's what I'm talking about. What's better: using IC with tight regulation and maybe 5000u before and 1000u after reg., or using series transistor with zener and 80,000uF around it, somthing like that (check the DAC without oversampling schematic and power supply):
Again my only concern is the sound and I can only check it in the circuit, but maybe someone already made those comparisons and can share his findings. After all that what this Forum is all about.
You can build a simple follower that will have <10 ohms impedance without trouble. Might sound high, but how many times have you seen a circuit with a 10 ohm series resistor to isolate the stages? And what follows it: several caps. The follower should have better reverse isolation.

1000 uF sounds like overkill, especially when you look at the frequency it starts to become inductive. How high of an impedance the design can tolerate is your call.

Have designed commercial digital audio products, and they all had lots of 3-terminal regulators, popular film bypass caps (you know, the ones that make rotten coupling caps.......), all for "oooooo......aaahh!" effect. If anyone was smart enough to look on the underside of the board, there were axial-leaded moulded ceramic caps.