Using an SMPS for audio amps

I actually already had a thread

http://www.superbestaudiofriends.org/index.php?threads/using-an-smps-for-audio-amps.2455/

But this might be a bit different

 

Well, my first build a was the regular Dynalo. Powered it with a Sigma22 supply and added a Joshua Tree attenuator for volume. It was a tough build for a greenhorn like me at the time. Sold it off to one of the HC higher ups after I built the QRV08.

All of the Dyna-family are well designed tried-and-tested designs working in Class-A with shallow NFB. What intrigues me is the CFA3. As far as I can tell, none have been built so far. My only beef is that KG likes to design super huge PCB's (and I can't draw pcb's for shit) so that they're DIY friendly. The thing's a 2U full size enclosure because you need to put two 191x155mm pieces in there.

 

Oh I don't mind criticizing designs. The attraction of the Dynalo SS to me is that it really looks like a high end design and it ticks off a bunch of boxes for what I'm looking for. There isn't a lot of specs or measurements which is why i want to send mine out. I think a closer comparison would be the Jotunheim or the Liquid Carbon since they are similar balanced designs as opposed to QVR08 or the HP-1.

As far as an SMPS for headphone amps, if it's under 50w I don't know if there is too many advantages over a linear supply. I do think it would work really well as a B+ supply for tube amps though. The noise is small compared to the large voltage swings that tubes are exposed to and this would be reduced even further after getting stepped down through an output transformer. Plus a significant amount of cost and real estate is devoted to the power supply where I think that having an SMPS starts to make more sense as long as it meets specs.

I have some ideas to make a 6sn7 + 6as7 OTL with separate power supply for experimental purposes. That way I can play around with unregulated, tube rectified, SMPS, tube regulated, batteries, etc.

 

Well, they're basically the same amp with minor differences to get higher power output or adapt to changes in part availability. That's why I'm looking into the CFA3. It's different from the rest and interesting due to lack of NFB. How it sounds remains to be heard. I'll take a stab tonight at redrawing that humongous PCB for SMD only.

 

I guess we know what product this is now.

 

how much of an issue is clean power for opamps? Most opamps I read about have a pretty high PSRR, so is obsessive filtering/regulation all that necessary?

 

Walt Jung (op amp design guru) published an enlightening article on distortion, pointing out that op-amps actually have (if I recall correctly) at least 5 separate inputs, all sources of distortion. The power supply inputs are 2 of these. As always, garbage in - garbage out. PSRR is specified as dB at DC and reduces, along with gain, over bandwidth. It can get very low where you may need it most. Try looking at a circuit as if YOU were a 1 MHz or 10 MHz signal. Tracks look like inductors. Capacitors look like tuned circuits. Opamps don't look so ideal anymore: low gain, low PSRR and high phase-shift.

I think Walt Jung's article referred to was "Op Amp Audio - Minimizing Input Errors" from Electronic Design, December 14, 1998, pp. 80-82. This can be found on his website: http://waltjung.org/

 

Holy crap! 1940s tube based SMPS!



http://www.righto.com/2018/09/glowing-mercury-thyratrons-inside-1940s.html

 

Burson and Benchmark make a case for a well designed SMPS. The key word is well designed, and pretty much the most important requirement looks to be being able to switch at very high frequency. And also some sort of RF noise filtering or something of that sort.

https://www.bursonaudio.com/about-us/max-current-power-supply/

Idk what they are doing at the black box "MCPS" and the diodes in the rectifier diagram looks incorrect lol (burson manual sucks). But the sound quality on the fun is great, so there is likely to be some truth to their claims. Also, I'm not able to see regulator in the schematic, but I think they have some 4 of them on the board.

https://benchmarkmedia.com/blogs/ap...audio-myth-switching-power-supplies-are-noisy

Both of them have interesting reasons, and both of them used to have LPS in their earlier products. So looks like they've found better performance in SMPS.

I'm interested to know what other hifi gear uses such kind of smps, and even more, if these could be diy'ed. Burson fun as a whole amp costs just 200$ so pretty sure it could be done in a cost effective manner.

 

I started this thread awhile ago and I've learned quite a bit since then.

From what I understand a properly designed LPS will ALWAYS be better than an SMPS. It's just whether the SMPS is "good enough" to where that doesn't matter. (Look at @atomicbob noise nuke to see how to get good performance from an SMPS)

Aside from performance, there are other reasons to favor an LPS. The circuit of an LPS is relatively simple and will last for decades before it needs to be serviced. When it does need to be serviced, the components are pretty simple and you can almost be assured that they will be available in the future. From personal experience I've had several SMPS units go bust on me and the values and sizes of certain components are very specific. It's usually better to replace the entire power supply than to attempt to fix an SMPS.

 

I can see the noise aspect as one very important part, but there is more to power delivery than just noise right?

There is power delivery transients, not only about being able to supply enough amount of current, but being able to supply that in quick transient time. A cap can discharge large current, but how much is the reaction time is something I'm not sure of (nothing is instantaneous in reality unfortunately). But I do know transistors can be very sensitive to these reaction times.

And also, really high frequency switching theoretically must be pushing the noise high enough to be filtered out without cutting down on the transients too much. Now I'm not directly advocating one over other, I'll have to listen more gear to know. Really fast switching could even push out electromagnetic radiation into the entire chassis and analog circuit, who knows? There is also the concept of line harmonics pushed back into transmission line,etc which are present switching devices. I think they are also there in LPS not sure, but level is quite high in switching devices.

In the end, I believe we are looking at a sum of products (or sum of aberrations), and it would be boiling down to implementation. Implementation >> buzzword.

Linn, burson, and benchmark used to have LPS and they have moved to SMPS, so I believe they found gains in their specific implementations. In fact, linn actually offered to replace their LPS to SMPS in some systems as an upgrade.

 

I see SMPS being worthwhile in high power applications for say power amps. In DACs and headphone amps the size/weight/heat difference won't be that pronounced. If anything, even when using SMPS, you'd still be inclined to use complex regulation.

As for companies going for SMPS... don't forget about the economic incentive. An SMPS takes all mains voltages, so you might be cutting your SKU database in half, as you don't need to stock different units for different mains voltages. They often are lighter than big ass toroids, so you shave off some logistics costs.

P.S. One thing - LPS doesn't have mains voltages anywhere after the trafo. It can't fail in a way that energizes parts of the device with deadly mains voltages. With SMPS all you need is a crappy electrolyte batch on the mains side.

 

At least in 99 audio qualities out of 100.

I agree. While I will buy an amp with a 15..20 kg toroid before one with any switcher, there is perhaps one feature of a well thought out switcher that will benefit audio quality over a linear psu.
It's the ability to be more immune to low frequency dynamic load fluctuations.
A linear psu, with as much capacitance one can fit into the chassis for reserve will still have relatively quite high source resistance for low frequencies.
Only cure is big fat toroid. And even there are limits to how much of it fit into the chassis, and it will break a weak person's back to move it.
Sure, one can stick in a regulation for linear psu and neg feedback of the amp section should take additional care of it, .. and hope for best.
A semi decent spice simulation will reveal that this will fall flat on it's face in some more demanding situations.
Amps with over sized transformers as a rule sound better in bass, even when not listening loud.

Browsing through top PA amps there are switchers being used more on the high-end and low-end series and linears in the mid-tiers. So the main driving decision is not even for mass produced stuff the price. A linear will sound better, but for a really high power unit it will be just too impractically big and heavy in the fear of sounding wimpy in bass.

Summary: for a beast sub or bass amp I see good potential to use a swithcer, when lifting heavy is not one's hobby.

 

A well protected and thought out switcher that could be relied upon costs as much as a LPS using premium capacitors and oversized transformer.
Only real win is less weight.
Perhaps one can omit the ceremonial crystals and special stones that are put on their dacs and amps for 'reasons' when you have heavy tranny inside.