SMPS currently in use with amps: ACA, few small JLH, F1J and few F1. Seems good match to class A amps.
I have experimented with two smps connected to symmetrical +/- voltage for powering chip amplifiers. Its a little tricky but can be done.
In order to filter smps with big cap, and to prevent protect trigger, i treat smps as high frequency rather than dc. I put rectifier bridge then big cap, and no protection is tripped.
One thing which bothers me with smps, it interferes with short wave radio reception.
I have experimented with two smps connected to symmetrical +/- voltage for powering chip amplifiers. Its a little tricky but can be done.
In order to filter smps with big cap, and to prevent protect trigger, i treat smps as high frequency rather than dc. I put rectifier bridge then big cap, and no protection is tripped.
One thing which bothers me with smps, it interferes with short wave radio reception.
Just looking at a modular smps, a 300W pfc then using about 390V dc link and then using separate step down / step up as needed. The amp is a class a valve amp so the heaters should keep it loaded.
Trying to model a Cuk as it supposedly has good noise characteristics. I’m also trying to keep the inductor in CCM as it modelled nicely with the boost.
Trying to model a Cuk as it supposedly has good noise characteristics. I’m also trying to keep the inductor in CCM as it modelled nicely with the boost.
I use 2x Meanwell RS 100 15v in series for my class A amp. Can be adjusted to give +/- 18V.
In my own opinion, SMPS supplies have no business in a good audio amplifier.
These noisy, touchy, often cheaply made buggers will never go into an amp that I'd build.
Sorry, I'm Old School, and a nice qualified power transformer owns its place pushing a well designed power supply.
Since amplifiers were built, why do you think they always had a transformer?
These noisy, touchy, often cheaply made buggers will never go into an amp that I'd build.
Sorry, I'm Old School, and a nice qualified power transformer owns its place pushing a well designed power supply.
Since amplifiers were built, why do you think they always had a transformer?
And they're welcome to, of course.
By the way, that Emotiva guy talks a good tale, obviously for sales purposes.
But I'm as I said, Old School, and not impressed easily.
Last edited:
For preamps, there is no more doubt about the choice of modern SMPS (with higher switching frequency out of audio band). With good filtering and LDO (Linear REG), you can achieve only a few µV of ripple and noise easily. So cost, size and efficiency are better than the old topology with a big transformer, diodes and RC/LC filtering.
For power amps, it is a little bit more complicated because of high currents. However, if SMPS has a post linear regulator, it could be better I think (but Linear reg shall be very rugged). The big advantage is to avoid very big and expensive capacitors (you could divide at least by 10 electro caps values).
It is the same think for amp class choice. Some people don't want to use a SMPS then uses a Class A/B for their amp. Why using a Class A/B? for efficiency of course but sound is less better than class A. It is the same think for PS. A SMPS has a better efficiency but sound can be less better than a linear PS (depends on design).
For power amps, it is a little bit more complicated because of high currents. However, if SMPS has a post linear regulator, it could be better I think (but Linear reg shall be very rugged). The big advantage is to avoid very big and expensive capacitors (you could divide at least by 10 electro caps values).
It is the same think for amp class choice. Some people don't want to use a SMPS then uses a Class A/B for their amp. Why using a Class A/B? for efficiency of course but sound is less better than class A. It is the same think for PS. A SMPS has a better efficiency but sound can be less better than a linear PS (depends on design).
It's quite surprising how little noise you get from an SMPS. However even rectifier noise can appear on the output - it's transparent, and will simply chop up your input noise..
The machine, in the background, is modelling an isolated Cuk step-down - taking 390V and running 12 tube heaters (12.6V 400mA each). With it being isolated, I can adjust the ground reference as I wish.. It has a 10,000uF cap after a 1ohm resistor simply to provide a little passive LPFing.. low cost given the voltage for cap don't get above the set amount.
Seems happy enough without a separate regulator. The inductors in this are small - only about 1.2cm across and operating at 100kHz. And yes I need to revisit the voltage divider for that extra 0.2V
I'm still experimenting but the concept is good so far.
I've also experimented with bi-polar secondaries for the Cuk and it works, although the initial version did give me both +600V and -600V rails.. oops! I need to revisit that module. It's acting as a booster from the same 390Vdc PFC rail.
Lastly I'll explore using the 390Vdc rail directly I think with the direct-tube modulated 'class tube-D' output stage.. Only issue is transient response time, so I may put a shared boost and then run both channels off the same supply. Not got to that yet.
Yes it's complex, but in reality the pieces are like building blocks. The front PFC is shared, the sepecialist front ends are the same module but deployed with a secondary that's different. So it's not too complex - just need to design the right cap and inductors for the required purpose.
Cuk as it uses more inductors seems to be far more effective in an audio setting given the lower EMI.
SMPS does stress components more - no doubt. However given the right design, I suspect it's probably not as bad as people are expecting.
Btw - that Cuk is running 0.2uF caps in the main power switching area. High freq = low capacity caps although the mosfet switching losses can be high.
The machine, in the background, is modelling an isolated Cuk step-down - taking 390V and running 12 tube heaters (12.6V 400mA each). With it being isolated, I can adjust the ground reference as I wish.. It has a 10,000uF cap after a 1ohm resistor simply to provide a little passive LPFing.. low cost given the voltage for cap don't get above the set amount.
Seems happy enough without a separate regulator. The inductors in this are small - only about 1.2cm across and operating at 100kHz. And yes I need to revisit the voltage divider for that extra 0.2V
I'm still experimenting but the concept is good so far.
I've also experimented with bi-polar secondaries for the Cuk and it works, although the initial version did give me both +600V and -600V rails.. oops! I need to revisit that module. It's acting as a booster from the same 390Vdc PFC rail.
Lastly I'll explore using the 390Vdc rail directly I think with the direct-tube modulated 'class tube-D' output stage.. Only issue is transient response time, so I may put a shared boost and then run both channels off the same supply. Not got to that yet.
Yes it's complex, but in reality the pieces are like building blocks. The front PFC is shared, the sepecialist front ends are the same module but deployed with a secondary that's different. So it's not too complex - just need to design the right cap and inductors for the required purpose.
Cuk as it uses more inductors seems to be far more effective in an audio setting given the lower EMI.
SMPS does stress components more - no doubt. However given the right design, I suspect it's probably not as bad as people are expecting.
Btw - that Cuk is running 0.2uF caps in the main power switching area. High freq = low capacity caps
although the mosfet switching losses can be high.
Last edited:
As always it depends.
There are high quality SMPS that do not have/need a post linear regulator.
These can be very low noise and have regulated output voltage and would, in a power amp, better any linear power supply unless the latter is regulated too.
And of course chances are, that the little noise you get is a couple of mV with f>100kHz.
I personally would do it the other way round and build a linear (regulated) power supply for preamps and use a regulated SMPS for power amps.
(assuming I don´t consider the much higher price of a good SMPS ;-))
Can't say it better myself, wise in old tech, probably not so much otherwise.not so wise old tech
No need to bash other people.Can't say it better myself, wise in old tech, probably not so much otherwise.
If you don´t have anything constructive to add, just keep quiet.
The diyAudio Store has sold out of its initial one hundred units (PCB + kit of all parts) of my little SMPS filter board store sales link , Forum discussion thread link. I'm told they're preparing another batch for sale, but this time several hundred kits. If you don't feel like waiting for the Store, the Gerbers are freely downloadable from the Forum. Send em off to a PCB fab, order parts, and build the filter on your kitchen table in two hours.
It's a modest little thing: max voltage 48V, max current 3A. Intended for line level applications: preamps, headphone amps, line stages, DACs, stuff like that. Wall wart powered audio gear.
Builders have reported great sonic improvements when using the SMPS filter with the B1 Korg NuTube preamp, the ACP+ preamp and HPA, DACs, electronic crossovers, even Raspberry Pi equipment. Here's a Forum posting which includes the test result attached below: post #515
The peak-to-peak noise at the input to the SMPS filter (yellow) is 5.0 vertical divisions, which equals 250 millivolts. Peak-to-peak noise at the output of the SMPS filter (green) is 1.0 vertical division, which equals 10 millivolts. So the filter output has less noise than the filter input, by a factor of (250 / 10) = 25X. I am pleased with this result.
Answers to frequently asked questions:
1. Has anybody cascaded a number "N" of thess filters (N>1) in a daisy chain, to get even greater filtration? Yes. They report the sonic result IS perceptable and IS improved, but only slightly. Diminishing returns even at N=2.
2. Do I have any plans to experiment with higher current filters, suitable to 30WRMS per channel, Class A power amps? Plans, yes. Success is not guaranteed and completion date is unknown. Neither one of us knows; not you, not me.
Click on the image to see it full size and undistorted:
_
It's a modest little thing: max voltage 48V, max current 3A. Intended for line level applications: preamps, headphone amps, line stages, DACs, stuff like that. Wall wart powered audio gear.
Builders have reported great sonic improvements when using the SMPS filter with the B1 Korg NuTube preamp, the ACP+ preamp and HPA, DACs, electronic crossovers, even Raspberry Pi equipment. Here's a Forum posting which includes the test result attached below: post #515
The peak-to-peak noise at the input to the SMPS filter (yellow) is 5.0 vertical divisions, which equals 250 millivolts. Peak-to-peak noise at the output of the SMPS filter (green) is 1.0 vertical division, which equals 10 millivolts. So the filter output has less noise than the filter input, by a factor of (250 / 10) = 25X. I am pleased with this result.
Answers to frequently asked questions:
1. Has anybody cascaded a number "N" of thess filters (N>1) in a daisy chain, to get even greater filtration? Yes. They report the sonic result IS perceptable and IS improved, but only slightly. Diminishing returns even at N=2.
2. Do I have any plans to experiment with higher current filters, suitable to 30WRMS per channel, Class A power amps? Plans, yes. Success is not guaranteed and completion date is unknown. Neither one of us knows; not you, not me.
Click on the image to see it full size and undistorted:
_