advertisement


5V Linear PS for "Higher Current" Devices

DoctorRad said:
Okay, so #controversial, but have you considered using a more modern SMPS or DC-DC converter?

For those manufacturers which are brave or open enough to publish them, I'm seeing output ripple waveforms much closer to low-harmonic distortion sine waves at the switching frequency - typically 2-300kHz - than the horrible slopes-and-spikes that you'd see 10-20 years ago, i.e. HF ripple rather than RF spikes.

I can dig out some waveforms from datasheets if you're interested to see them.
Click to expand...
My buddy is using an iFi SlientPower SMPS to power his Chord QUTEST, and he's itching to switch to a linear supply like the Allo Shanti. He sometimes is convinced he hears differences between gear that don't make any sense, and he's always tweaking something. I've told him to hold off while I play with linear supply design, but he keeps sending links to stuff like the BOTW P&P ECO MKII. I'm going to visit him in a couple of weeks, and I'm taking my own Shanti with me for us to compare, to see if he still thinks he needs a better supply.

FYI, this is the same guy for whom I built four different amps:
  • Stereo NCC200 that he justifiably rejected, due to a small ground loop buzz that was more noticeable with his speakers than with mine. Since getting it back, I've solved the buzz by re-routing the wiring.
  • Stereo NCC200 that didn't buzz, but he felt didn't boogie.
  • Dual mono NCC220/QUDOS that he still felt wasn't quite enough to make him happy.
  • Dual mono NCC300 build, which he absolutely loves!
I know... tough customer! :D He's a great guy, though, and really loves his music and his system. I'm a bit of a nutbar too, and had a great time trying to satisfy him, so I'm perfectly willing to cut him some slack. I still tease him about it, though.
 
S-Man said:
My preferred solution is one of these:
https://czh-labs.com/products/audio...-voltage-regulator-module-12520v-15-amp-p0688

This gave better sound than a much praised TPSA47 regulator, a coffin supperreg and a few LM723 based regs I lashed together.

Don't use a toroidal transforer, the EI types are much better. I also recommend a FW centre tap arrangement (2 diodes). Keep the i/o differential right and the LT1963 will do >2A.
Click to expand...
Thanks for the link. My "customer" is very fixated on specifications. Even though I suspect his DAC doesn't actually need 3A, there's no way he would let me get away with having something nominally rated at 1.5A. ;)
 
Mike Hanson said:
Thanks for the link. My "customer" is very fixated on specifications. Even though I suspect his DAC doesn't actually need 3A, there's no way he would let me get away with having something nominally rated at 1.5A. ;)
Click to expand...

It would get quite hot if it was burning off 10W.
 
337alant said:
Mike
Have you looked at the Linier LT1083 ser, at 5V you can get up to 7.5A with good heat sinking
See data sheet LT1083/LT1084/LT1085 - 7.5A, 5A, 3A Low Dropout Positive Adjustable Regulators (rs-online.com)
Alan
Click to expand...

I've used it also powering my audio PC and RME DAC. In an short A/B testing I had with another guy, we found LT1083 sweet sounding and Salas L-Adapter neutral. Both were lacking the tight bass which original switching supply of RME had. Switching supply was lacking in clarity though.

We even tried Salas BIB and UBIB regs on it with huuuuggeee heatsinks.

Mind that differences although they were noticeable, they weren't that huge.
 
ramona said:
I've used it also powering my audio PC and RME DAC. In an short A/B testing I had with another guy, we found LT1083 sweet sounding and Salas L-Adapter neutral. Both were lacking the tight bass which original switching supply of RME had. Switching supply was lacking in clarity though.

We even tried Salas BIB and UBIB regs on it with huuuuggeee heatsinks.

Mind that differences although they were noticeable, they weren't that huge.
Click to expand...
Coincidentally, my friend had read good things about both the RME and Chord QUTEST, so purchased both on the used market. He waited for me to visit to do the head-to-head, and both of us felt the RME sounded dry and lifeless compared to the Chord. Then we tried the iFi on the Chord instead of its standard PS, and found that it was more fluid and natural, with better body. He's since sold the RME, and is still quite happy with the Chord, even though he's itching for a linear PS.

I've ordered an LT1085 to play with (that's the 3A unit), but it's backordered at the moment.
 
You might get some ideas from the heater supply in this article. It's for a valve MC phono stage using the notoriously unstable 417a so the heater supplies had to be super quiet. I built a version of this and it worked very well. Obviously the current regulator stage wouldn't work in your application but the supply arriving at the regulators is already pretty clean so the capacitor banks are doing something right.
 
Mike Hanson said:
I've ordered an LT1085 to play with (that's the 3A unit), but it's backordered at the moment.
Click to expand...
That's a good place to start for such. have a read of the datasheet though - since these are a TO220 package, the heat dissipation is quite limited (15-20w with a big heatsink IIRC) so you will only get the full current output rating at relatively modest supply voltage overhead before it will go into thermal limiting - no real surprise there. As above, the raw supply input wants to be c. 9.0v max and even then, at full rated load (9-5) x 3 = 12w of heat to get rid of. The good news is these work down to 1v input-output differential - so a raw supply in the range 6-7vdc will be sufficient (and less wasteful.)
 
Mike Hanson said:
Coincidentally, my friend had read good things about both the RME and Chord QUTEST, so purchased both on the used market. He waited for me to visit to do the head-to-head, and both of us felt the RME sounded dry and lifeless compared to the Chord.
Click to expand...

Me too! I don't understand what the fuss is all about. Salas took it for awhile and had the same opinion.
Anyway, maybe because of the extended local regulation the main psu don't matter that much in this device or at least not worth spending hundreds of dollars for it.
LT108x has low dropout, decent noise specs and it is easy to implement (usually I'm using LM317 pcbs for it) and will give you a clue if linear psu will be worth for your case.
 
Afterthought on the LT108x series - same as LM3x7s - use decent-capacity electrolytics on the Vinput and Voutput pins, above datasheet min value recommendation: I'd suggest 100uF, connected at these pins - and connect their 0v ends to the point/plane you refer the output 0v return to. That's about the best you can do - and easy : )
 
The real test of any PSU is what you see on its output, under the prospective load.

With respect Mike, that linked post shows not a jot of useful info %)

PS:
to reiterate a/my theme here of >>last decade: 'low noise' is NOT the be-all & end-all! AC load regulation, probably is.

(it's very easy to achieve the former, to an arbitrary extent; the teddyregs are entirely-focused on it. But most of such approaches, inc the teddyregs - will not also do the latter!)
 
Last edited:
martin clark said:
The real test of any PSU is what you see on its output, under the prospective load.

With respect Mike, that linked post shows not a jot of useful info %)

PS:
to reiterate a/my theme here of >>last decade: 'low noise' is NOT the be-all & end-all! AC load regulation, probably is.

(it's very easy to achieve the former, to an arbitrary extent; the teddyregs are entirely-focused on it.

But most of such approaches, inc the teddyregs - will not also do the latter!)
Click to expand...

Very true! I've often thought there seems a preoccupation with low noise when regulators are discussed here and even when there is no need for very low noise for the application.
 
Further to that, personally I completely ignore noise re voltage regulators for everything but the front ends of phono stages. It really doesn't matter for pretty much anything else as even the likes of LM317 has low enough noise.

For phono stage front end with poor PSRR it very much does matter though! I have a balanced very low noise instrumentation amp (1nV rt Hz) I made which has x10 and x100 gain and which I use to extend the noise performance of my Bruel & Kjaer measuring amps so I can measure down to sub uV levels.

How low can I get the output impedance and over how wide a bandwidth are the important things technically.

Then the whole "sound of regulators" thang comes into it to muddy the waters... the very low impedance comes from loads of NFB...

Not that I would ever bother with changing PSU's for routers, or hubs, or DACs or streamers or anything like that anyway in a million years!!
 
I would suggest that a good read on this subject is the first part of Ben Duncan's AMP-01 constructional series, which covers the power supply design and methodology.

As well as making sure that the PSU itself doesn't chuck cr@p back onto the mains, the approach is that each op-amp has its own regulator pair, positioned as close as possible to the chip they're serving to keep the effective output impedance as low as possible*. If you have (say) 80dB of noise reduction from the regulator, and a similar PSRR from the op-amp, you've got 160dB of noise reduction from the power rails to the op-amp output, i.e. it's going to get lost in the weeds.

The stuff about power supplies starts on page 53 of the PDF (page 55 of the magazine).

*Which is why e.g. Naim's approach of hanging the active circuitry on the end of several feet of cable, distant from the regulators, just seems bonkers and ultimately limiting.
 
You must log in or register to reply here.


advertisement


Back
Top