Thanks
-
Welcome to ASR. 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
- Thread starter McFly
- Start date
"signal", not "sign".
By the way, is it not a problem the constant current of CLASS AB with these SMPS?
For example DH220 draws about 270mA per rail, per channel. Maybe active colling is needed?
DVDdoug
Master Contributor
- Joined
- May 27, 2021
- Messages
- 5,689
- Likes
- 8,266
It's fine as long as the new power supply is capable of the same (or higher) current.
And you might be confusing class A with A/B. Class A draws high nearly-constant current, even when idle. (A dumb design these days.
) Class A/B draws a smaller bias current and more when feeding actual power to the speakers.Glad I ran into this, glad someone did a measured comparison. Back when I was DIYdesigning amps some years back, I used SMPS supplies and could find no negatives at all with that. It always makes me shake my head when I read someone talking about and praising old 50/60Hz transformer, rectifier, and bigass capacitors as "linear supplies". There is absolutely nothing about those that is "linear", just antiquated. Changing AC power into DC is about as nonlinear as it gets, and having line frequency related residual in the audio band is way worse than at supersonic frequecies in every imaginable way
OK. I thought that even with 300mA rail (e.g. class AB Hafler DH220) there will be a problem.
I saw that for some SMPS it is specifically mentioned that it fits CLASS AB (or more than almost zero idle like Class D). They can put peak powers but much less average (some of them).
So, for class AB, 300mA idle per channel, 65V and about 150W/channel. What is your suggestions?
Thanks
I saw that for some SMPS it is specifically mentioned that it fits CLASS AB (or more than almost zero idle like Class D). They can put peak powers but much less average (some of them).
So, for class AB, 300mA idle per channel, 65V and about 150W/channel. What is your suggestions?
Thanks
I have found the term "linear power supply" rather strange for the traditional format - but language/society/linguistics often follows fashions and mythologies rather than straight logic.... so I comply with the term rather than swim against the current...
I think it comes from mystical dislike for the idea of switching. As if switching at 10s of times per second, then trying to clean up all the junk is somehow better than at 100s of kHz. Probably many don't realize that a tradional rectifer arrangement is switching also, and in a pretty bad way..
Miercoles2022
Member
That DH-220C served me very well for several years. I set its bias a bit high and it ran very warm (~50C), but the SMPS never had any trouble with it.
I actually just today finished a new class AB build with a Connexelectronics SMPS500RXE and it is quite nice. The distortion numbers are designed to be at the diyAudio Wolverine level (similar, if not simpler EF3 design). This was a bit of a gamble, but it has paid off.
I actually just today finished a new class AB build with a Connexelectronics SMPS500RXE and it is quite nice. The distortion numbers are designed to be at the diyAudio Wolverine level (similar, if not simpler EF3 design). This was a bit of a gamble, but it has paid off.
Attachments
Thanks. Good to know a specific model number of SMPS. Actually at connex they are writing if the PS fits AB class with some normal idle or only class D. Also, most of their PS are regulated. I don't see such a description it in Microaudio, in general don't see see datasheet file in the website but maybe it is by private request. Anyway, the owner Sami is very kind and responsive. On the Cobra ps it is written it fits Wolverine.
I see in the picture you sent you installed post filtering. How much capacitance per rail?
Did you measure ability to put ful or high powers for low frequency, like 100hz or 50hz?
Thanks.
I see in the picture you sent you installed post filtering. How much capacitance per rail?
Did you measure ability to put ful or high powers for low frequency, like 100hz or 50hz?
Thanks.
Hi, I'm no expert, so could you elaborate on this a little?
SMPS has switching noise with fundamental of 60-100k, depends on switching frequency.
On the other hand diode bridge has also an untrasonic noise that comes from diode switching, even if your are using fast-soft recovery. These noise has a wide spread.
So, it doesn't innocent of switching noise.
In addition, it is hard to mitigate the main frequency (50 or 60 Hz) and for that capacitance is not helping, it is a leakage noise. The PSRR of the amplifier doesn't help much with this too.
On the other hand diode bridge has also an untrasonic noise that comes from diode switching, even if your are using fast-soft recovery. These noise has a wide spread.
So, it doesn't innocent of switching noise.
In addition, it is hard to mitigate the main frequency (50 or 60 Hz) and for that capacitance is not helping, it is a leakage noise. The PSRR of the amplifier doesn't help much with this too.
Miercoles2022
Member
2x4700uF per rail. The filter board is described in the SMPS500Rxe pdf guide and can be ordered/customized from the connex shop.
No.
I followed the setup guide as part of the DH-220C package from Bob Cordell. I do not recall the exact figure, but I believe it was ~200mA per output FET. See the 2-part article in AudioXpress.
The idle current is much less in the EF3 design I pictured. It's like 95mA per output BJT. The 3U300 sinks barely get warm.
Last edited:
Thanks.
By the way, in connex websites it is not written specifically that SMPS500RXE fits AB mode, for some other models it is written.
It is nice to realize it is working, I guess the relatively low idle consumption is fine.
I will give a try. I like the fact that the voltage is regulated which is not always the case.
By the way, in connex websites it is not written specifically that SMPS500RXE fits AB mode, for some other models it is written.
It is nice to realize it is working, I guess the relatively low idle consumption is fine.
I will give a try. I like the fact that the voltage is regulated which is not always the case.
Class AB for a switching supply might be because, for some supplies, a minimum current drain from the load is needed for it to regulate. A very low idle current of (some) class D might cause the supply voltage to rise high enough to damage components.
HF garbage from the fast rise & fall times of waveforms in switch mode supplies, which switch at high frequencies, is difficult to clean up, since it sprays everywhere like RF. That's not such a big problem in 50/60Hz supplies, especially when soft recovery diodes are used.
Yes, but at which frequency this HF noise spread? Above the switching, so very high.
Diodes switching noise start at lower frequency,
Tweeters can't possibly reproduce the hf noise you mention, midranges or woofers can most definitely reproduce the 60, 120 and 180Hz residual from a rectifier and capacitor supply. HF *might* affect some poorly designed electronics, but orders of magnitude lower from what a cell phone or wifi nearby would (and the fix there IMO is to use electronics that won't be bothered by such)
Different switch mode supplies may use different switching frequencies. The sharp edges of the switching waveform generate a spectrum of noise frequencies. I agree that these frequencies are vastly higher than humans can hear, but they could cause misbehavior in amplifiers, especially those that use negative feedback, which the majority of them do.
