Using an SMPS for audio amps

Discussion in 'DIY' started by Cspirou, Jul 17, 2016.

  1. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Lately I've been seeing more and more SMPS modules used in DIY projects from some respectable people. Here are some examples:

    http://www.pmillett.com/HVBFLY.html
    http://www.neurochrome.com/smps-86/
    http://www.hypex.nl/product/2012-11-23-13-42-14/smps400.html

    SMPS have a lot of advantages over linear supplies such as efficiency, cost and size that make it preferable in almost every application. One of the most important factors in the sound of an amp that I keep reading is the need for a very clean power supply. This clean power requirement typically disqualifies SMPS because of the high switching frequency required which would produce a lot of noise. However can't a lot of these noise issues be reduced with proper shielding and secondary power regulation? Plus the switching frequency is in the range of 100 khz, far above what is typically audible. Many amps include DACs which deal with digital data operating at high frequencies as well. Even the highly regarded Eddie Current amps use heaters that run at 40 khz as part of the design. On top of that, a lot of projects tend to encase their power supplies separately anyway, this should give a high amount of isolation from any interference that would come from an SMPS.

    Aside from noise the downsides of SMPS are complexity, high voltages and long term reliability. If you want an amp to last for 30 years you might want to stick with a linear PS. But seeing how SMPS already used in many professional and even HiFi settings, I think it's worth considering for some projects.

    Anyone else use them? Links to good modules to plans? thoughts?
     
  2. schiit

    schiit SchiitHead

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    9,937
    Trophy Points:
    93
    Location:
    Texas and California
    Home Page:
    There are several great reasons to use a switching supply:

    1. Space constraints.
    2. Heat dissipation constraints (usually allied with space constraints).
    3. Needs to be very efficient (such as when you're working from USB power and have limited current available.)
    4. Needs to be powerful and cheap (only when the supply is past a certain size.)

    Beyond those reasons, there is literally zero reason to use a switching supply, ever (in my opinion). One look at the hash from a (good, clean) switching supply should be enough to make you a believer. Yes, even switchers that work at 1+ MHz. Stuff above the audio band can easily cause problems in the audio band. It's better simply to not have the hash at all.

    So, unless you need the supply to be tiny and cool--or you can't afford a 50-lb transformer for your 300WPC stereo amp--don't use a switcher. At least IMO.

    Full disclosure: I just got done working on a new design that uses a 1.7MHz switcher. But it needs to be small and efficient. I'd never put a switching supply in any design if I didn't need to--and that includes $99 products like Magni.
     
  3. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    This seems to confirm my planned uses for a switchmode. Looking at the transformers needed for a 500W amp was daunting and the SMPS version was rather appealing. The other projects I have are a battery powered boom box type system (needs efficiency for battery life and doesn't need to be HiFi) and a transportable headphone amp (needs to be compact and powerful).

    For my stereo amp project I am going to keep the power supply separate. That way I am not tied to the SMPS if I ever find a good deal on some huge transformer.

    Edit: I can't wait to see this product!
     
    Last edited: Jul 18, 2016
  4. Rex Aeterna

    Rex Aeterna Friend

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    212
    Trophy Points:
    63
    Location:
    Cinnaminson, nj
    i only see issues with switchers if not filtered correctly. the amps i used mostly(from qsc and crown) are not noisy and actually sound pretty darn good and never found hf issues or noise from them. they actually were more quiet than some linear counterparts with large iron core transformers so, i just go with ''it all depends'' approach. space and heat is usually not the main issue when using a switcher usually. it's how robust and quiet it is that is most concern i think. you're pounding over several kw's hour after hour you're gonna have to have a reliable psu to do that besides robust outputs. efficiency is on the concern as well in lot of cases but, not top priority usually because there is some but, very little concern about wall power draw vs. actual amp output but, of course depends on situation and application.

    linear power supplies make noise too and not immune to it either. they just make noise at a low frequency instead of very high frequency like switchers do.
     
  5. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Another thing I wonder about is power variability. Switching supplies have a range from 90vac to 240vac and supply the same voltage without any need to mess with settings. Linear supplies need either 115v or 230v and you can't stray too far from either. If your power isn't perfect then I wouldn't be surprised if rail voltages aren't constant either. Sure, enough capacitance and regulators can smooth out a lot of issues but it isn't the range of what a SMPS can tolerate.
     
  6. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    I'd say it largely depends on how your circuit will react to the hash.
     
  7. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    I'm partial to switchers with a regulator. The trick is to use a regulator with decent bandwidth. LM317s just don't do much.

    Diodes make noise, too, and most linear supplies have at least two. I'll take the devil I know. :)

    [​IMG]

    [​IMG]

    Note different scales.
     
  8. firev1

    firev1 Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    44
    Trophy Points:
    18
    Location:
    Singapore
    Why not? I think for high power speaker amps its just better to use a SMPS for reasons like current demand and mostly heat. Also one can build a really high end speaker amp using a switched supply (Benchmark, Icepower, Hypex).

    Actually if we go by supply ripple is evil logic, ain't ripple from linear supplies uber difficult to filter? You need to afford the iron for the tranny, and size the insanely huge capacitance/ chokes needed depending on how much ripple rejection you want.

    In the end I guess it all comes down to needs I guess, whether you want 1kw of power without sagging or stressing your home power lines wasting it to heat or you have the best damn efficient speakers and only listen at lower volumes.

    From my readings and some use of the hypex modules, if you wish to use one for your speaker amp, do get a hypex one if you plan to keep things simple and vanilla. Many people though report getting very good sound out using Connexelectronics RxE psu series but only with modding them with boutique Rubycon caps. Also from the First One project, they recommend going the dual mono approach and Hypex seems to have taken a similar route when presenting the NC400 monoblocks.

    Of concern for either approach(linear or smps) would be the magnetics and EMI so either way both would require some degree of shielding from the circuit somehow. The great thing is that you can pay your way and order very good iron with shielding done for you. Notably these days emi signature of SMPS are really getting smaller though I can't say how much.

    If it interests you I will be building up a ncore based amp soon so will let you know how it goes. I have had a good experience with the UcD series amps, though with a linear supplies.

    Edit: For headphones I feel there is no good reason to deal with an SMPS really, just stick with linears for that and DACs. Speaker amps need it though.
     
    Last edited: Jul 19, 2016
  9. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,153
    Trophy Points:
    113
    Location:
    Estonia
    All sort of thoughts can be derived from how the gunk in SMPS (or even simple linear with LM317-ish regs) won't matter, yet all of this collapses in the face of actual listening tests.
    Clean and responsive power has always made a nice difference in my projects, sometimes more than tubes or the actual topology itself.
    Shunt for low power! Nuff said.
     
  10. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    Thanks for measuring! I wonder how it stacks up against discrete heavy hitters like Dynahi.

    P.S. Really surprising about the noise floor. Is your SMPS supply usable for headphone amps? Most likely it would need additional filtering, no?

    (Note from @ultrabike: merged further discussion about SMPS here)
     
    Last edited by a moderator: Dec 5, 2016
  11. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I had the same question. I dunno where, but I think Tom did mention in the past that he does some further additional filtering. In fact, I think he uses regulators after the SMPS (the HP-1 seems to have some heat sink thingys there).

    Based on what I've put together about Tom's headphone amp, I dunno about sound, but in terms of technical specs I have a feeling the HP-1 destroys the Dynahi in almost everything except output power. And then, relatively not by much.

    Based on Ti Kan's website, his Dynahi can put about 1W into 330 ohms. The HP-1 450mW into 300 ohms. That's a 3 dB advantage.

    For 33 ohms the advantage is also about 3 dB. Putting shit loads of power into 8 ohms is kind of pointless for headphones, unless one wants the headphone amp to drive speakers.

    I've yet to see numbers for the Dynahi.

    Wouldn't it be awesome to have it measured?
     
  12. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    As you might know, the Dynahi was measured here - http://www.ibiblio.org/tkan/audio/dynahi.html

    What I'd like to do would be an SMPS powered Dynahi/CFA3. So far it looks like I'd still be needing a beefy power supply to get proper filtering.
     
  13. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Well, those are RMAA. Sure, it's better than nothing. But wouldn't it be nice to measure it with an AP? I mean, even here we kind of give the cold shulder to RMAA:

    http://www.superbestaudiofriends.org/index.php?threads/amp-measurement-standards.426/

    Based on what I see at Ti-Kan's site, the measurements might have been limited by the measuring interface. The numbers on the Dynahi are not too impressive IMO. Not horrid though.

    I think I've seen in Head-Case some folks mentioning the IRM-24-20 in particular, and curious about it. For SMPS, you could parallel a few supplies to get more current. I think that's why you see two of those in Tom's design. Dunno for sure though.
     
  14. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    I agree with you about RMAA, it's pretty ok to see if something isn't wrong. As for absolute performance, there's a good chance you'd run into interface input stage distorting more than a good headphone amp.

    IRM modules are pretty versatile and it wouldn't hurt to do a discrete high current post-regulator a la Goldpoint to provide cleanup. So far I'd effectively be swapping a toroid for a SMPS, for most headamps. The space and cost improvement would be negligible with only international voltage being the only clear advantage.
     
  15. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I would also look into weight and likelihood of damage due to shipment when it comes to SMPS vs transformers. Availability of certain transformers seemed to be a problem for Tom in another design if I read his concerns correctly. I do not have ample experience on either one though.

    In the end, if it doesn't make sense to use SMPS for a particular solution, then it doesn't. Designs are not bounded to be the same. There are different paths to accomplish a goal. The goals may not even be the same.
     
  16. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    513
    Trophy Points:
    93
    Location:
    Calgary, Canada
    Home Page:
    The 1/f noise of the QRV08 is probably due to the input bias current of the bipolar input stage going into a high-impedance node (Zin = 1 MΩ) and an AC coupling cap. I'm a bit surprised it is as high as it is, but given the design it's what's to be expected. I had to design around a similar issue in the HP-1 and was able to solve it and get much lower noise.

    I use SMPSes in my differential preamplifier (DIFF PRE 8x2), HP-1, and SMPS-86 which I used to power the LM3886DR and Modulus-86. All these circuits provide state of the art performance and have very low noise floors. The switchers perform better than most linear supplies. I do add filtering on the output of the SMPS, but that's mostly a precaution. The switching noise is spread-spectrum, thus, low in amplitude and way outside the audio band (60-70 kHz), so it's not an issue for audio circuits.

    On the HP-1, I went one step further and added a set of low-noise regulators to the output of the filtered switching supplies. The main reason was to get the rail voltage I wanted so I could get 3 W into 20 Ω and 3 W into 32 Ω. The post-regulators also provide a nice marketing blurb about super filtered supplies, though, the filtering isn't all that necessary from a technical standpoint.

    There are ways to screw up the performance by adding filtering, by the way. If the designer isn't careful, it's possible to cause some severe peaking in the combined output impedance of the filter + SMPS. It should go without saying but I'll say it anyway: I spent considerable time and effort designing the SMPS post-filter to ensure good performance across the audio band.

    The DG300B uses a ClassicTone mains transformer. It's a one-pony race. If you're in Europe you either have to pay through the nose to get one of them shipped from the US to you, use multiple transformers, or have a custom wound transformer made. This means my DG300B design is basically "US only" unless you're willing to throw some cash at the shipper of your choice.

    Even with generic PCB mounted transformers, there are models which are popular and easy to source in Europe and models that are popular and easy to source in North America. Then you add the whole mess with the different mains voltages. Thankfully nobody has wired my DG300B for 120 V and plugged it into 240 V ..... yet. But that is a risk. It is so much easier with the switchers. International mains in. DC out. Done!

    The mistrust of switchers is mostly based on fear, uncertainty, and doubt ... and perhaps some really bad switching designs from the 1980ies. These issues have been solved long ago and fact is that if you choose a good switching supply (as I do) you'll get better performance than with a linear supply at a lower cost. On top the supplies are available internationally so my designs can be built for a reasonable cost across the world. They are also light so my products can ship across the world for a reasonable cost and arrive intact.

    Tom
     
  17. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,461
    Trophy Points:
    113
    Location:
    Winnipeg
    Benchmark makes a point that the radiated EMI of switchers is lower in magnitude than linear supplies (though possibly more complex).
     
  18. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    Cool! What I was getting at was if SMPS-86 is good enough for headamp duty as-is. If it's quiet on its own, then I might save good money on high current regulators.
     
  19. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    There's a head-case project where they use switchers and apply additional regulation with modules from AMB audio.

    I'll put the Dynalo SS on the loaner list when I finish and you can take all the measurements you want.
     
    Last edited: Dec 5, 2016
  20. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    I'm not the one with the fancy measurement equipment. You might want to propose the same to @tomchr or @atomicbob.
     

Share This Page