Should phantom power be supplied by a linear supply?

[Bo Deadly]

Using SMPS for audio is a relatively new concept and many mfgrs have taken that into account.

I don't think that's why. The MeanWell LED supplies are known to be very quiet. I think it's more likely that they're quiet because they want to be able to claim that their products satisfy the various EMI standards, EMC emissions, regulations, compliance and so on ...



And because they can. If an engineer can pick between two supplies for an industrial application, they'll pick the quieter one every time (as long as it not too expensive but SMPS are cheap).

 

[gswan]

Whatever topology of PSU is chosen should be in accordance with engineering requirements.
This includes heat, efficiency, noise, ripple, size, weight and cost.
A well designed ZVS resonant switch mode supply is preferable over a large, heavy and hot linear supply in most cases.

 

[dyamakuchi]

This is largely debatable.

Yeah, sure. You can choose less reliable parts. Lytic caps come to mind. But this is actually pretty well established science. How do you think the engineers in Detroit get your car to fail on average 24 months after the lease ends?

Moreover, the SMPSs will have at least as many if not _more_ lytics than a comparable linear supply. Unless you're thinking of some new topology I'm unaware of. Naturally, I'm open to the possibility of learning something new here. But think about it: if I roll my own linear, I can oversize the tranny, choose higher voltage caps and diodes, higher wattage R's, etc. What mass manufactured switcher has that? They'd be 'throwing away' money.

I submit, you might be comparing a well designed commercial switcher to a cheapo (or amateurish) linear, in which case, sure, the better design always lasts longer.

Everything else being equal, (an important qualifier) less parts is better. Always.

 

[Trouble Report]

Perhaps share the name of the "expensive" PS that didn't work for you so others here don't repeat the mistake. In my experience with studio gear, more expensive does not automatically correlate with good design.

I ASSume you contacted the PS manufacturer to confirm that it wasn't faulty.

As I have shared probably too many times I worked on switching PS back when they were new, and probably before some here were born (1960s). It is a mature technology now and properly implemented not the weakest link in those consoles.

JR

Oh OK - you've twisted my arm enough. The company was Atomic, and this was just a few years ago. There was sort of an industry rush at that time to replace the PS on many Neve and SSL desks. And they made several attempts to correct the condition. Those attempts moved the offending frequencies around but did not eradicate them. I've seen this same issue with other "made for audio" SMPSs.

Although we didn't get way into it (we had a studio to maintain), from our rough examination I believe the artifacts wound up imposed on the audio ground bus - then it's really problematic. I like to think I have an open mind about advances in technology, but so far, I haven't seen an SMPS that outputs dead, flat, nothing else but DC.

The suggestion, a few posts back, of adding external filtering, says to me that we're trying to effect a kluge, to compensate for a power supply shortfall. Unless there's a limit imposed by weight or space, why not stick to a linear supply approach?

 

[musgrave]

On my VR frames I spread a significant amount of copper across the frame to seriously drop the impedance and distribute about 170,000uf across each rail across the frame. With the Kingshill supply or the Atomic I get the same great noise sweeps. experience has shown poor distribution inside the units are the first cause of power related noise (unless your cables etc truly suck). On Robs MCI I did the 'over copper, and cap', but kept the power distribution on the modules anemic. Get great noise floor, but when you push signal you can see the 2nd order rise significantly around -5 on up till +15 where the 3rd order stuff starts to take over. The R then C then R then C of the distro sorta creates a filter and as Paul says it filters the hashy stuff. With the Atomic, my frame owners save a pile of cash and grief...

 

[Matador]

Since PRR isn't around anymore to provide an elaborate tractor analogy, I'll try one close: carburetors versus fuel injection. Carburetors are remarkably simple devices, but they are demonstrably worse in every measurable way: fuel efficiency, air-fuel mixture consistency across different loads and RPM levels, tune-ability for different purposes, etc. However EFI requires vastly more complicated control loops, many more sensors, and miniature computers to run everything. However the number of mass-produced consumer automobiles running carburetors today could probably be counted on a single hand.

The only advantage a linear supply has is simplicity *for the DIY'er*. If I were building a car myself I would likely use a carburetor, because I know I can get it running with a few jet selections, setting the fuel bowl level, and adding a simple filter. And I would be ok with it being worse in every possible way because I wouldn't need to learn complicated software control loops or debug some manufacturers problems.

All of that being said, manufacturers of switch mode controller IC's have gotten better educating their customers about topologies and many even have parametric building applications (TI comes to mind) where you simply put in your requirements and it spits out a complete topology even with component values picked for you.

 

[abbey road d enfer]

. I've seen this same issue with other "made for audio" SMPSs.

AFAIK, there are no "made for audio" SMPSs commonly available.

The suggestion, a few posts back, of adding external filtering, says to me that we're trying to effect a kluge, to compensate for a power supply shortfall. Unless there's a limit imposed by weight or space, why not stick to a linear supply approach?

Many applications do not require the ultra-low noise level we are used and entitled to. Audio represent 1% of SMPS sales, so mfgrs build for the other 99%. As it has always been, we use whatever the mainstream industry produces, and we need to take them as rough components taht need to be integrated in a system. You call it a kluge, I call it building. If the end result is good, it's a success, not a Band-Aid.

 

[abbey road d enfer]

Yeah, sure. You can choose less reliable parts.

Do you think I'm stupid?

How do you think the engineers in Detroit get your car to fail on average 24 months after the lease ends?

The same way Korean engineers offer 7 year warranty. Different goals, same method.

Moreover, the SMPSs will have at least as many if not _more_ lytics than a comparable linear supply. Unless you're thinking of some new topology I'm unaware of.

Not topology. Linear PSU's have an enormous issue: heat. Life span of lytics is severely compromised by heat. One could put a fan, but that would probably become their first cause of failure.
Just a question: how often did you have to replace the PSU in your desktop computer?

if I roll my own linear, I can oversize the tranny, choose higher voltage caps and diodes, higher wattage R's, etc.

You can't compare a boutique product with an industrial one. The former can be way better or way worse.

What mass manufactured switcher has that? They'd be 'throwing away' money.

What mass manufacturer of linear supplies does that?

I submit, you might be comparing a well designed commercial switcher to a cheapo (or amateurish) linear, in which case, sure,

Again, am I stupid?

the better design always lasts longer.

I couldn't agree more. Who's the judge of the "better design"?

Everything else being equal, (an important qualifier) less parts is better. Always.

You say that like you never heard of MTBF...

 

[JohnRoberts]

if we haven't beat this topic to death, heres another design anecdote that might confuse the topic.

Early this century I was supporting a specialized DSP vocal processor. It already used a switching supply. I needed to come up with an additional +48V supply for its one mic input. I decided to KISS and think outside of the box at the same time.

I decided to make a cap doubler, that usually grabs AC from the transformer secondary. Instead I grabbed the HF square wave from the DC-DC switching supply. Switching at 100kHz instead of 60Hz allowed me to use tiny SMD caps for the doubling circuit. The parts were cheap and the PC real estate used was modest. I used simple zener pass regulator for the 48V regulated.

Worked like a charm.

JR

 

[dyamakuchi]

In a related anecdote: I recently dug two 48V (75W & 150W iirc) TDK Lambda off line switching supplies out of storage and plugged them in. Now, many of you may have heard of Lambda supplies, they were kinda higher end, at least a while back. IDK about after the TDK acquisition, but whatever. These (TDK LS series supplies) look very similar to the popular Meanwell (LRS series,) actually.

Anyway, I plugged them in, the light came on...and then went right back out. Both of them. No output voltage. After a number of years in storage of course, but still. I was not super impressed. You can find aluminum lytics that last in storage, if you like. One of them seemed to have some corrosion problem from not cleaning the board after assembly, and time of course. The other one? No reason AFAIK. Literally failed out of the box new old stock.

IDK. I'm going to need some more convincing here.

 

[JohnRoberts]

Don't use them...

JR

 

[dyamakuchi]

am I stupid?

I don't think you're stupid AT ALL. Quite the contrary. I just think you're mistaken on this particular issue.

Maybe it just boils down to, as you point out, who is deciding the criteria for what's 'better.'

I like my gear to live for years in hibernation, run in Arizona in the summer, Alaska in the winter, and last until the next ice age. Plus it has to sound pristine.

Of course, I never get all of that but, you know, ...goals.

P.S. BTW, I just recalled another issue re: SMPSs no one has brought up here yet: starting up in the cold. Most of the particular ones I've really investigated don't like it. It's...and whoda thunk it... the 'bulk' lytics on the input, as I understand it. I've done enough New Years shows over the years, plus we used to do Chicago's Thanksgiving Day parade outdoors. I just can't have it. Those clients don't want to hear "my gear won't work because it's freezing."

 

[gswan]

For those who have little understanding of switching topologies, look up ZVS resonant mode. This is used in high quality low noise power supplies for many industries, including medical, military and AV. The power density is high and the noise is low. Reliability is as good as any linear supply as the efficiency is around 90% (much less heat). I've designed class D amplifiers and used ZVS resonant switchers for the supply which allow high power output without supply sag, very low heat generation and low noise. Also, they are lighter and take up much less room in an enclosure.

 

[abbey road d enfer]

I've done enough New Years shows over the years, plus we used to do Chicago's Thanksgiving Day parade outdoors. I just can't have it. Those clients don't want to hear "my gear won't work because it's freezing."

What about amps? Most current amps use SMPS. I've never seen a Powersoft or Lab Gruppen amp not starting because freezing.

 

[Bo Deadly]

P.S. BTW, I just recalled another issue re: SMPSs no one has brought up here yet: starting up in the cold. Most of the particular ones I've really investigated don't like it. It's...and whoda thunk it... the 'bulk' lytics on the input, as I understand it. I've done enough New Years shows over the years, plus we used to do Chicago's Thanksgiving Day parade outdoors. I just can't have it. Those clients don't want to hear "my gear won't work because it's freezing."

I have no experience with that but if I had to guess I would say that the problem is more likely due to high inrush currents. SMPS do have very high inrush currents and more so when starting cold. The line and switch and everything has to be able to handle it. This is yet another reason to have a capacitance multiplier. And probably a line filter helps too. Definitely a solvable problem.

You will flip dyamakuchi. SMPS are an inevitability and they do genuinely work fantastic if you know what you're doing.

Incidentally someone mentioned something about SMPS polluting the ground bus in a console, it can important that there is a cap (or better still an LC) to shunt common mode noise on the analog 0V to the input neutral. Common mode noise can go right through the SMPS transformer. But I have never had a problem so I think the better SMPS have that C or LC builtin already or the topology has advanced to a point were the switching currents just don't make it through as much. Not sure. But I can imagine that when stacking SMPS such that the positive output is connected to analog 0V (for the negative rail supply), it won't have that cap and so it might be good to add an extra C to give that noise a path back to earth ground (and thus not downstream into your analog stuff). Although presumably the negative output of the positive rail supply will have that cap and so it will be shared by the two so maybe that's also why I've never seen a problem with it.

 

[dogears]

The basic switchers I use have a rated operating range temp of -30 to +70 C. New Years should be ok, I think.

 

[dyamakuchi]

it can important that there is a cap (or better still an LC) to shunt common mode noise on the analog 0V to the input neutral.

We needed a 5V rail on one box and derived it from a higher (18V) one thru a little switchy regulator bit. Really user friendly. OUR circuit liked it fine. One part. (plus a filter cap.) And a post regulator...with it's bits...but I digress... 5V 1A max. Nothing, right?
The 100s of kHz backfeed...multiple, multiple, frequencies...back thru the 18V winding, back thru the primaries (of the linear transformer,) back out the connector, all the way back down the power line. Violated FCC 'conducted' requirements. Straight thru that giant toroid plus caps. We had no broadcast license for those frequencies, basically.
The compliance testing lab guy says "oh yeah, a couple caps (L+N) to ground will clear that right up" ...which they did. Kinda pricey big ole film caps, but whatever. I don't mind film caps. 'Self healing' and all that jazz.

Then we threw it on a GFI receptacle. You know, just to see...

It turned out smaller filter caps got us FCC cert too. And if we made them small enough, you can even use it on a GFI receptacle. It's still bullshit doing it like that tho. What's wrong with maybe just not broadcasting in the first place? IDK. Seems a lot simpler to me. Plus / minus the big film caps.

I kinda gotta wonder too. How many SMPS enthusiasts advocating on here are concerned, or even _aware_ about complying with regulatory agency emissions requirements? This is a question of target audience then. If you want to build _one,_ the FCC probably ain't pulling up with the cat detector van to razz you about it. 'More than one,' maybe different story. But I'm still following here. Meanwell LED supplies, capaciatnce multipliers, desks that run electric bills up. Things change. We'll see.

Plus there's this guy too though:

https://groupdiy.com/threads/51xaudio-by-silent-arts.39570/
He seems to have went linear. Kinda maybe might know a bit about this particular subject too... perhaps? IDK.

 

[ruffrecords]

Back in the 1970s, Neve used a dc/dc convertor of their own design to raise the internal 24V to an isolated 48V phantom supply.

Cheers

Ian

 

[abbey road d enfer]

Plus there's this guy too though:

https://groupdiy.com/threads/51xaudio-by-silent-arts.39570/
He seems to have went linear. Kinda maybe might know a bit about this particular subject too... perhaps? IDK.

He never had anything else than linear. I have one of these. It's a very well made piece, that I use to power an 11-slot 500/51X chassis.
At the same time I wanted to make a 6-slot lunchbox. Hauling a big and heavy 2RU box was not an option, so I put 4 Sun Power 15W smps and a floating 24/24 DC converter (there are 5 rails in a 51X system) in Hammond box, with a PCB including CLC filters for each rail. The first thing I did was check the rails, then the system noise, with mic preamps cranked up. I couldn't measure any difference in a 20Hz-1MHz range. I didn't go to have it CE certified, so I don't know, there may be monster radiations, but I have yet to experience any consequence.
That was my first time incorporating smps i a build; I always used to build my PSU's.
I recently designed a product where I chose PC-mount smps. The ones that looked right were of the RS brand. I had all sorts of noise issues, like hiccup, ripple and RF pollution, so I replaced them with closed-frame Meanwells and everyting was fine and dandy. So yes, there are bad smps, but there are also good ones.
In an other build, I made two versions, one with a conventional linear PS, and one with cheapo (<$3) switchers from Ebay replacing the power xfmr and rectifiers, so I still had the linear regulators taking the 24V down to 18V rails. After comparison, and weighting the pros & cons, I kept the smps. Actually I just had to do the layout in such a way that dirty nodes don't pollute clean ones. I was worried that the two un-sync'ed clocks woul produce beating at audible frequencies; I could not detect any.

 

[abbey road d enfer]

Back in the 1970s, Neve used a dc/dc convertor of their own design to raise the internal 24V to an isolated 48V phantom supply.

It was not so uncommon in these days. I assume they used sine-wave converters. Audio Developments used them in their battery-powered location mixers.
Efficiency is not as good as new switchers, but the single clock frequency is easy to filter out.