• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Swapping class AB amp linear supply with a SMPS - incl measurements

dlaloum

Major Contributor
Joined
Oct 4, 2021
Messages
3,092
Likes
2,352
I believe they are more efficient in most circumstances... but they do not have "headroom" - pretty similar to class-D amps!

From my reading (I am no expert) - a linear power supply with big iron transformer, can provide short peaks of output above its continuous specs, which provides some headroom for musical peaks potentially - with SMPS, the max out is the max out, there is no headroom - so if the design was in any way dependent on headroom, replacing linear with SMPS, requires "oversizing", to provide the headroom. (the end result will no doubt be an improvement, as a result)

It would be nice to be able to take a vintage Quad 606, and replace the trannie with an SMPS having a higher current rating... current was always the 606's one weakness IMO.
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
903
Likes
1,861
Location
NZ
I guess Its a bit like saying your car has 200hp max but sometimes it will do 300hp when you stomp on it.

No. You have 200hp and no headroom.

Just oversize your SMPS to get your headroom.
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
Most SMPS's designed for audio applications will state their power levels for different peak durations in the literature, a Hypex 1200 for example is 1200 watts but only 325W continuous. This comes down to cooling, for short peak durations the power can be very high but there will be lot of heat build up if that is required all the time and forced cooling isn't used.

It's not just the transformer though, if there is not enough capacitance in the Linear supply the voltage rails will sag as increasing loads are required which actually limits the power the amplifier can put out. It isn't very simple to decide which is the best overall without considering a lot of different variables.

Benchmark has a good article if a little self serving.
https://benchmarkmedia.com/blogs/ap...audio-myth-switching-power-supplies-are-noisy
 

fpitas

Master Contributor
Forum Donor
Joined
Jul 7, 2022
Messages
9,885
Likes
14,191
Location
Northern Virginia, USA
I believe they are more efficient in most circumstances... but they do not have "headroom" - pretty similar to class-D amps!

From my reading (I am no expert) - a linear power supply with big iron transformer, can provide short peaks of output above its continuous specs, which provides some headroom for musical peaks potentially - with SMPS, the max out is the max out, there is no headroom - so if the design was in any way dependent on headroom, replacing linear with SMPS, requires "oversizing", to provide the headroom. (the end result will no doubt be an improvement, as a result)

It would be nice to be able to take a vintage Quad 606, and replace the trannie with an SMPS having a higher current rating... current was always the 606's one weakness IMO.
It's true, you can do like some people and have massive banks of capacitors on a linear supply. That's a lot of fun, and impressive looking, but the more straightforward way is just get a larger switcher instead.
 

Gorgonzola

Major Contributor
Forum Donor
Joined
Jan 27, 2021
Messages
1,016
Likes
1,398
Location
Southern Ontario
@McFly, thanks very much indeed for your reporting. I have a linear-to-SMPS swap I reported on ASR months ago. My DIY amp is a Class D Audio SDS 258; the power supplies, both linear and SMPS are Connexelectronic, the latter being the SMPS800RE, 60 volt.

My case was trivial because (a) the amp driven is class D and (b) I have no measurements. However personally I think my swap was a success: the weight of the amp was hugely reduced which was my first objective, and the sound quality was at least good. Shown are before and after pictures. Note the wiring was tidied up as well.

gi.mpl


gi.mpl
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
903
Likes
1,861
Location
NZ
Most SMPS's designed for audio applications will state their power levels for different peak durations in the literature, a Hypex 1200 for example is 1200 watts but only 325W continuous. This comes down to cooling, for short peak durations the power can be very high but there will be lot of heat build up if that is required all the time and forced cooling isn't used.

It's not just the transformer though, if there is not enough capacitance in the Linear supply the voltage rails will sag as increasing loads are required which actually limits the power the amplifier can put out. It isn't very simple to decide which is the best overall without considering a lot of different variables.

Benchmark has a good article if a little self serving.
https://benchmarkmedia.com/blogs/ap...audio-myth-switching-power-supplies-are-noisy
Ah, so the smps do have headroom, and they are marketed/sold on this spec, relative to continuous use. So the marketing for the hypex SMPS should be more like "A 325W power supply, capable of 1200W peaks"

Do linear tx/rec/cap supplies have 'headroom' because of the tank/filter caps? because I thought the VA rating of a transformer would be the absolute maximum, i.e. magnetic saturation.

Hadn't seen that benchmark app note. Yes a little trumpet blowing, but has some good points. Tell you also what you don't see on their SMPS - massive output caps.
 
Last edited:
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
903
Likes
1,861
Location
NZ
@McFly, thanks very much indeed for your reporting. I have a linear-to-SMPS swap I reported on ASR months ago. My DIY amp is a Class D Audio SDS 258; the power supplies, both linear and SMPS are Connexelectronic, the latter being the SMPS800RE, 60 volt.

My case was trivial because (a) the amp driven is class D and (b) I have no measurements. However personally I think my swap was a success: the weight of the amp was hugely reduced which was my first objective, and the sound quality was at least good. Shown are before and after pictures. Note the wiring was tidied up as well.

Nice work! Did you take any measurements? Interesting, because your using class D whereas mine was class AB.

What would be nice - We don't have any data of any nCore or Purifi being fed from its conventional hypex supply vs a linear supply. But this would be difficult to do because the hypex SMPS has the auxiliary and driver rails to accommodate the ncore and purifi driver and control sections.
 

fluid

Addicted to Fun and Learning
Joined
Apr 19, 2021
Messages
691
Likes
1,196
Ah, so the smps do have headroom, and they are marketed/sold on this spec, relative to continuous use. So the marketing for the hypex SMPS should be more like "A 325W power supply, capable of 1200W peaks"
The rating is there in the datasheet but they have chosen the higher number to go with as technically it is true and most people will not use forced cooling.

The supply could produce 1200W continuously if enough cooling was provided to prevent thermal shutdown. This graph shows how much power is put out as heat, at 1200W it is 150W, that is a lot if it is continuous but easy enough to absorb if it is only short in duration.

Power Loss.PNG


They have this section in the datasheet to explain further
"The SMPS1200 is designed for music reproduction and is therefore not able to deliver its maximum output power long-term. The RMS value of any common music signal generally doesn’t exceed 1/8th of the maximum peak power. The SMPS1200 is therefore perfectly capable of driving the connected amplifier in clipping continuously with a music signal without the need of forced cooling."


Do linear tx/rec/cap supplies have 'headroom' because of the tank/filter caps? because I thought the VA rating of a transformer would be the absolute maximum, i.e. magnetic saturation.
Transformers can pull huge current, just like they do when you first turn them on and the cap's aren't charged you get inrush which without a soft start can be enough to dim your lights with a large high VA torroid. It is easy to be comparing apples to oranges with these things unless you consider all aspects of the power supply and not just one section of it.
Hadn't seen that benchmark app note. Yes a little trumpet blowing, but has some good points. Tell you also what you don't see on their SMPS - massive output caps.
There is no need for large output capacitors with SMPS power supplies because the capacitors are charged so frequently due to the high frequency that many of them run at that they don't get the chance to discharge. 50/60Hz vs 100KHz for the Hypex.

It depends a lot on the design of the SMPS as to how they behave. The Hypex is a little bit like a linear supply in that the rail voltage sags a touch when it is loaded down.

I have other SMPS supplies that I use with my Neurochrome amp and they get stiffer as the load goes up and sag if the load is too low.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,273
Location
Gold Coast, Queensland, Australia
and response out to 100k looked like this, you can see the SMPS switching noise

You need to crack out the scope and look up higher. Much higher. The 2150B has a wide bandwidth- unity gain a 3MHz...

1661986469623.png
 
OP
McFly

McFly

Addicted to Fun and Learning
Joined
Mar 12, 2019
Messages
903
Likes
1,861
Location
NZ
Good call John. Haha, anyone know of an audio ADC that's capable of 6,000kHz sample rate? I shall have a look at the output with the scope I have here sometime soon. Has an FFT function too but rather limited.

There's bound to be 100khz+ switching harmonics all the way up - I did wonder if this might perhaps effect the feedback loop or affect amplifier stability, and maybe the switching noise would fold down into the audible band but it doesn't show in the measurements I can get (currently 192khz max). I think this is a limitation of ASIO4ALL.

I'm currently daily driving it, I will report back any quirks or faults.
 

levimax

Major Contributor
Joined
Dec 28, 2018
Messages
2,348
Likes
3,462
Location
San Diego
Just oversize your SMPS to get your headroom.
That is not a good idea.... linear supplies can supply higher voltage for a very limited amount of time for musical peaks and then sag down to safe levels very quickly. An "oversize" SMPS will deliver it's full rated voltage all day which unless your power amp section has some type of protection circuity can easily destroy the amps output stage. Especially for 4 ohm loads a few extra volts on the rails can make a big difference.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,273
Location
Gold Coast, Queensland, Australia
Good call John. Haha, anyone know of an audio ADC that's capable of 6,000kHz sample rate? I shall have a look at the output with the scope I have here sometime soon. Has an FFT function too but rather limited.

There's bound to be 100khz+ switching harmonics all the way up - I did wonder if this might perhaps effect the feedback loop or affect amplifier stability, and maybe the switching noise would fold down into the audible band but it doesn't show in the measurements I can get (currently 192khz max). I think this is a limitation of ASIO4ALL.

I'm currently daily driving it, I will report back any quirks or faults.

Just see on the 'scope if anything is riding on the sine at high powers. Of if your 'scope is a DSO, use the math functions (FFT). It'll probably be ok, but I'd be looking anyway- just don't trust SMPSs spewing out garbage high up...
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,273
Location
Gold Coast, Queensland, Australia
That is not a good idea.... linear supplies can supply higher voltage for a very limited amount of time for musical peaks and then sag down to safe levels very quickly. An "oversize" SMPS will deliver it's full rated voltage all day which unless your power amp section has some type of protection circuity can easily destroy the amps output stage. Especially for 4 ohm loads a few extra volts on the rails can make a big difference.

Not the case with this Perreaux.

The PMF2150B has 2x3 pairs of Hitachi K135/J50 MOSFETS for each channel. They are a 160V 7A each device. There's no problem with slightly higher rail voltages- in fact the next model up, the PMF3150 uses exactly the same output compliment with a higher voltage for a 50% increase (300wpc) in rated power output.
 

levimax

Major Contributor
Joined
Dec 28, 2018
Messages
2,348
Likes
3,462
Location
San Diego
Thats old school... enough output devices and big enough heat sinks to run flat out limited only by PS output. I guess to pass the old FTC test thats how you had to do it. Over kill but kind of cool.
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,579
Likes
38,273
Location
Gold Coast, Queensland, Australia
Over kill but kind of cool.

Perreaux basically made two types of amplifiers. PA amplifiers and Audiophile amplifiers. They were basically the same. IIRC, the PA amps had some bandwidth limiting, came with XLRs, cool LED meters and sometimes fans.

They were/are virtually indestructible. (apart from the power switches which exploded or welded themselves on due to the massive current draw on startup).

Go check out the big brother, the 5150B:

1661992869909.jpeg


5150b01 (Medium).jpg

The bargains are old 8000c amps. 500wpc all day.

1661993164606.png
 

Cars-N-Cans

Addicted to Fun and Learning
Joined
May 19, 2022
Messages
819
Likes
1,009
Location
Dirty Jerzey
Does the smps noise leak through to any other components, pre amp, sources?
It shouldn't if its a good design. Even my simple bipolar flyback supply based amp w/ 20 wpc is quiet with no discernable hiss, which is surprising given how simple the design was at the time (first amp I ever built myself, make of it what you will :)). With my ear over the tweeter all that is really audible is some faint ticking in one channel from the IC going into burst mode with nothing playing due to there being no load. But more sophisticated supplies should have fairly quiet outputs. Just have to be sure you allow some separation so any local switching transients cant couple into the rest of the circuitry. Been many moons since I built it, but I do recall that the class-Y cap can be a path for 60Hz to make its way back into the signal chain if there are multiple grounds (which normally you should not have, anyhow). Still it beats having to use a transformer in a small cabinet, which is at best problematic with respect to hum. Don't have any fancy REW plots, just an old scope FFT w/ a few averages to unmask any potential harmonics, but still it looked ok and I would guess the ICs meet their datasheet specs. ICs were LM3886s for the amps and a PI TOP250 integrated controller, FWIW. I did make a linear version of it as well as a breadboard amp, and overall it had less white noise than the one with the SMPS. Still, it won me over vs. using transformers (not that they cant work well with careful design). I used to hate SMPS technology, esp. when they first started coming with PCs. Those cheap chinesium ones would just blow one right after another, whereas the old linears would just go and go and go. But now they are much better.
 

Attachments

  • 20220706_062322.jpg
    20220706_062322.jpg
    204.8 KB · Views: 75
  • LM3886 Amp 1 kHz w_ averaging.jpg
    LM3886 Amp 1 kHz w_ averaging.jpg
    238.5 KB · Views: 79

Waxx

Major Contributor
Joined
Dec 12, 2021
Messages
1,931
Likes
7,688
Location
Wodecq, Hainaut, Belgium
A remark from a non electric engineer.

I remember the days that SMPS's were noisy, too noisy for amplifiers. But those days are gone since more than a few years and it's nice to see that finally those psu's arre also used in class AB and even class A and tube amps. The big transformer psu's are heavy and often not that noisefree also (but were better than the oldskool smps's). I worked in pro audio untill 2007 and my back did suffer a lot of lifting those heavy amp racks with big class AB amps and linear psu's. These days a high power class D amp with psu (like a Powersoft X8) weights nothing anymore and smoke any class AB amp on power while still sounding good.

I like tube amps (if you did not know that already) but i hate that they always weight a lot. I would love to have tube amps with SMPS's to reduce the weight. It will still weight a lot because of the output transformers, but each transformer removed can help. Manley Labs is doing it with pro audio (studio) gear with a high voltage SMPS designed by Bruno Putzeys, and their gear became more stable and less noisy. They don't do it for the hifi stuff yet because their clients don't want its says the owner. But i think that is audiphoolery, and they should also do it with their hifi gear

And my Pass ACA's, very coloured class A amps, already run on SMPS's made by Meanwell (who make very high quality) and work flawless in their intended way. The noise from the psu is very little, and the colouring comes due to the amp (Harmonics), not the psu. And that is how i like it. I also had a old Sugden A21 (the original, not the remake that is sold today) and that psu is noisy as hell. A friend bought it from me and recently removed the old linear psu and put a suited meanwell smps and the noisefloor also dropped a lot. He is an electric engineer, and did measure it but i don't have the data down here.

So amp builders, remove the linear psu and use smps please. Amps get better with it in most cases when it's done right.
 

Piere

Active Member
Joined
Feb 24, 2022
Messages
195
Likes
190
SMPS sure have their advantages and nowadays you will not see any PA amplifier without it. But their reliable life time is limited, usually to about 10 years. For Pro PA not a problem at al, as it is their usual service life too. But for private home use, recapping after 25 years is normal. Regularly swapping SMPS? I doubt.

Some posts earlier the Perraux amplifiers came along as a kind of reference example. These are beautiful amps. Despite the design is rather simple, they have precise and smooth mids and highs and a huge bass drive. Nevertheless they are never fatiguing and almost indestructible. Recently I did restore a over 25 years old 3370 model and a 3400 model Perreaux, rated at 300W_rms continuous. I considered also putting a SMPS into it. But as they are, these beasts can spit out over 25 amps at 80 V_peak for short burst periods, thanks to its large rail buffer caps. That is more than 2 kW peak! Hard to find a suitable SMPS of plus and minus 90V that can deliver that peak power and fit. So I left the heavy tranny at its place for another happy 25 years :)
 
Top Bottom