When I started this site, Elpac (now ICCNexergy*hack**urp**spew*) had a great product, the WM080. This linear-regulated power supply was widely available, was compatible with every circuit on this site, and performed well for the money. They must have sold many of them as a result of my site’s recommendations over the years it was avaiable. Then a few years ago, they discontinued it. Much sadness ensued.
For a while, I offered some DIY power supply designs of my own with even better performance than the ol’ WM080. In 2010, I decided to stop offering them for many reasons, some of which I will hint at below. (I am unlikely to be more discursive on the topic if you email me about it, so please don’t bother.) More sadness ensured. Sorry ’bout that.
These two events left a lot of my readers asking the question that is this article’s title. I had no simple answer to the question, and only now got around to writing an answer I felt comfortable putting on the site.
Sorry, I don’t know of one. If I did, it would be linked to from all over this site.
If you cannot be bothered to read the selection advice in the rest of this article, my best simple answer is to get part number 174879 from Jameco. Jameco says it is linear-regulated, a claim that seems justified in my measurements and brief listening tests. I tried it with several different amps built according to instructions on this site, with no failures or glitches. I have heard positive reports from others; nothing glowing, but nothing bad, either. It is cheap as linears go, so it’s probably not an expensive mistake if it doesn’t work out for you.
I have also tested an Elpac MW1224 to the same level of success except that its overload protection circuitry couldn’t cope with one of my amps, probably because the amp’s supply cap bank was too big for its own good. I heard no hum with the other amps I tested with, though I have found in other testing that it has considerably more noise than either the Jameco supply or the old Elpac WM080. This is because it is a switcher. It is good quality for a switcher, but not excellent. I mention it here mainly because it is more widely available than the Jameco unit. If you must get all your parts from a single distributor, you will be forced to make compromises of this sort.
Please tell me about it if you know of a reason either unit is unsuitable. Let me know also of better alternatives.
I might subtitle this section, “Why can’t I use this nasty old thing I saved in my junk box when I threw away my dot matrix printer/laptop/blender after it asplode?” (You laugh, but I see essentially the same question on the forums at least once a year.)
The short answer to that is that the amplifier designs on this site all require an isolated DC power supply. The CMoy pocket amp, META42, and MINT designs also require that the wall supply be regulated if you want decent performance. The newer amplifiers can get away with an unregulated supply, but it still isn’t recommended. Additionally, it is nice if the supply is linear-regulated, rather than switch-mode-regulated. Finally, it has to put out something within the voltage range each amp wants from its power supply, which though wide in all cases, does not cover the entire range of available power supplies. Current technically matters, too, but in practice it’s rarely actually a problem with the amps on this site.
The rest of this article/FAQ gives the long general answer. Bottom line, most people needing a wall supply for one of the headphone amps discussed here end up shopping for a new commercial power supply or DIYing one.
If you wanted a long specific answer instead, I must direct you to the documentation for each circuit. The best of these to date is the Power Supply Matters section of the PIMETA v2 amp’s documentation. It applies incompletely to the other headphone amplifier circuits on this site. It is quite possible to design or buy a power supply following its strictures that works for every amp on this site. It is also possible to end up with one that won’t work for one or more circuits for one reason or another. For example, it is possible to make a PIMETA v2 run from lower supply voltages than a CMoy with the same op-amp, due to the buffered outputs and stiff virtual ground.
Without getting too technical, isolation in terms of power supplies means that none of the AC power plug’s pins are connected to any of the DC-side contacts. Most commonly, non-isolated supplies have AC earth or AC neutral connected to V- on the DC output. This can provide safety or noise advantages to the power supply designer, but it fights with the virtual ground circuit used by every headphone amplifier on this site.
There are some uncommon corner cases where you can get away with using a non-isolated power supply but they’re not worth discussing here. Even if your situation happens to be one of those cases, it’s easy to forget, later, when reconfiguring your audio system that you’re relying on falling into a special case, then reconfiguring yourself into trouble.
The least bad failure mode stemming from missing power supply isolation is absolutely horrible sound quality. (In brief, the amp picks a fight with Planet Earth and loses, causing the amp to clip horribly.)
That in turn can damage headphones or ears if the volume level is high enough.
Sliding further down the Slope of Badness, the virtual ground circuit may die in its fight with the planet, possibly taking your headphones with it, as the drivers tear themselves apart in one loud violent *POP*. If the cans were on your head when this happens, you probably just damaged your hearing, too.
Don’t disregard the need for isolation. Just don’t.
The best way to tell whether you have an isolated power supply is to test it with an ohms or continuity meter. With it unplugged from the wall, test every combination of AC-side contact against every DC-side contact for continuity. If there is a 3-prong AC side cord and 2 DC output contacts, this means you will be making 6 continuity tests for a complete isolation check. If any combination of connections gives 0 Ω or near to it with an ohms meter or makes a continuity meter blink or buzz or whatever it does to tell you there’s a conductive path between the two points, the power supply is not isolated.
A non-isolated power supply is not flawed or broken, unless you know for a fact that it is supposed to be isolated. It merely tells you that that particular power supply is unsuitable for use with any of the amp designs on this site. A great many circuits work just fine without isolation; use it for one of those applications instead.
If you cannot test the power supply, you may be able to make a pretty good guess:
If you can assure yourself that a power supply is unregulated or linear-regulated, it’s a pretty good bet it is also isolated.
Most such supplies are transformer-coupled, which gives galvanic isolation between the AC and DC sides. It is possible to design an unregulated AC-DC power supply without using a transformer (dangerous, but possible) and it is possible to trade away galvanic isolation, but both are uncommon practices. Best to test anyway, of course.
If you believe a given power supply is switching-regulated but are unsure about whether it is isolated and cannot check, it is a good bet that it is not isolated unless the datasheet specifically says otherwise.
This is because it requires intentional design and extra parts to make a switcher isolated, so manufacturers are usually quick to point the feature out when present.
There is no technical reason why one could not design a switch-mode power supply with impeccable regulation, resulting in unimpeachable sound quality, at least as far as you can blame the power supply.
So much for theory. Reality differs.
The first rule of switchers is don’t use switchers. The second rule of switchers is don’t tell anyone you did if you do.
The fact is that it is not easy to find an audiophile-quality switcher. I mean “audiophile” in the ideal technical sense: one who values excellent sound reproduction, without prejudice about how that is accomplished. Because suitable switchers are very much the exception, you won’t go far wrong if you just blindly assume that Switcher = Bad when it comes to audiophile sound quality, even when faced with solid-looking datasheet claims to the contrary. Only bench and listening tests will prove a given switcher audiophile-quality, and then only conditionally.
Many who call themselves audiophiles make a false inference here and immediately discount a product when they learn it uses a switch-mode regulated power supply. I can immediately name three products from well respected manufacturers that have received unfair criticism rooted in this trope; I expect I could come up with more if I bothered to research it. If you were expecting impeccable logic in audiophile product reviews, well, welcome to the hobby, you must be new here. :) Thus the second rule.
The best way to tell that a power supply is unregulated when its datasheet doesn’t come out and say it is that it has a wide output voltage range that is dependent on the load current. It might say, for example, that the nominal voltage is 24 V but it is 29.4 V when under no load. Unregulated supplies’ output voltages drop in voltage approximately linearly as a function of output current; they drop to their rated output voltage at full output current.
Sometimes this fact is phrased obscurely in the datasheet, using terms like “load regulation,” with values in the 10 to 40% range. Our example supply above could be said to have a 22.5% output load current regulation, for example. This is not saying that it is truly a regulated power supply, just that its output may vary by up to 22.5% as a function of output current.
High percentages are good indicators of unregulated supplies in any case. True reguation results in lower percentage deviations, typically down in the single digits, and hopefully 5% or less.
Taking the above example power supply further, if it were rated for 400 mA and you wanted to draw 100 mA from it, its actual voltage would probably be around 1⁄4 the loaded-to-unloaded voltage range, or a bit over 25 V.
It is important to run such numbers, as it can cause problems with your power supply capacitors. If you had selected 25 V caps for your amp since you bought a “24 V” supply, you’d be at risk of blowing up your power supply capacitors.
In this section, I am going to be more definitive than I technically have a right to be, for reasons given elsewhere. For the purposes of this site’s circuit designs, it is better to rule out a suitable candidate as a nasty old switcher than to mistakenly identify a switcher as a good candidate through false assumptions or bad guesses. (Testing, now, that’s different.)
Switch-mode power supplies (SMPSes) are also called switching-regulated power supplies, or switchers for short.
The best way to identify a switcher from its datasheet or label is to look for a wide AC-side voltage range. If it claims to work from 90-260 VAC, it’s a switcher.
If it’s a wall wart and the plug contacts are swappable to allow for use on multiple world power systems, it’s a switcher. This goes hand-in-hand with the previous heuristic; it’s no good designing a supply that will run from any world AC wall voltage if you can’t swap out the plug contacts to match a given country’s wall outlets. Beware: I am not referring to things like user-replaceable IEC power cords here. That feature is found on power supplies of all types, not just switchers. Here I am only talking about the sort of wall wart where you can snap in different contacts. Often a selection of contact types comes with the power supply.
If the power supply is designed to be integrated within a larger device, all bets are off. But then, such supplies tend to have better datasheets than your average wall wart, so you can probably find out what kind it is from that.
Another ambiguous spec is when it says something like 90-130 and 210-260 VAC. The gap means it probably has a jumper or hard toggle switch to select the AC voltage standard, because it can’t work from the values between. There are both linear- and switching-regulated supplies with that feature, so it isn’t diagnostic either way.
If the datasheet claims efficiency of 80% or better, chances are excellent that it’s a switcher. It is possible under very careful design limits to make a linear with efficiency that high, but practically impossible to do so while also making it general-purpose. If the claimed efficiency is North of 90%, you can be virtually certain it’s a switcher.
Broadly speaking, if a power supply is linear-regulated, it will say so in its datasheet. Why? Because it’s uncommon and inefficient. To most of the electronics design community, it is a warning, not a feature. Those few areas of the electronics world that still really do want linears want to be told about this feature.
Okay, so your datasheet doesn’t say “linear” but you hope it might be anyway. How do you confirm this? The best heuristic — loose though it be — is that it gives an output voltage regulation spec in the 5% or less range and it is clearly not a switcher. Sorry, but that’s the best you can do without a lot of experience.
Be warned: I have seen a few datasheets that use the magic “linear” word, but they are in fact unregulated supplies. This is not fraud, just shady marketing. Technically speaking, unregulated supplies are linear electronic devices. They are just not linear-regulated devices. How to tell? The same way you identify any unregulated power supply. The best giveaway is the loose output voltage spec.
It can be as simple as following an unregulated wall wart with a 7824 3-pin regulator. If built correctly, it will give performance on par with most commercial linear-regulated power supply units, for the simple reason that many commercial linears are little more than this.
The trick is building it correctly. No, I will not advise you on how to do this. I got out of the DIY power supply business for a reason. More than one reason, in fact. All I will say is that you need to read and understand the datasheet of any regulator you use, and the docs of any project you build. If you cannot or will not do that, skip the DIY route. Power supplies are dangerous. You cannot expect to build one with the same sketchy knowledge base that many safely get by with in other areas of electronics.
When in doubt, buy a commercial power supply.
This space intentionally left blank. :)
This article is copyright © 2013 by Warren Young, all rights reserved.
|Updated Thu Dec 15 2011 12:30 MST||Go back to Audiologica||Go to my home page|