Linear transformers are big, heavy and expensive, therefore it is obvious that they must be better. Never trust anything that is lighter and cheaper, it is clearly inferior and down that road lies the madness of liking digital.
-
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!
Why does the audio industry like linear transformers so much over SMPS?
- Thread starter ZeDestructor
- Start date
- Joined
- Sep 21, 2018
- Messages
- 507
- Likes
- 747
Whoa, is that an airplane sized anechoic chamber?
Yep ... one of the largest chambers in the world from what I was told. They move pallets of the cones around on large fork lifts.
We have the thermal fuses in transformers in Canada too.
Let me add some measurements on output spectrum of linear power supplies
1) PSU 10VA transformer, uA7815 regulator, +15V/26mA, dV (voltage difference between regulator input and output DC voltage) = 9V
2) PSU 14VA transformer, uA7812 regulator, +12V/21mA, dV = 3V
We can see how big difference makes a dV DC voltage between regulator input and output, in (1) we have quite nice output spectrum, (2) becomes unacceptable. So it is difficult to make any general statements, as always, the results depend on the implementation.
1) PSU 10VA transformer, uA7815 regulator, +15V/26mA, dV (voltage difference between regulator input and output DC voltage) = 9V
2) PSU 14VA transformer, uA7812 regulator, +12V/21mA, dV = 3V
We can see how big difference makes a dV DC voltage between regulator input and output, in (1) we have quite nice output spectrum, (2) becomes unacceptable. So it is difficult to make any general statements, as always, the results depend on the implementation.
solderdude
Grand Contributor
How big is the ripple on the dV ?
For 12V regs 2.5V dV is minimum voltage.
When there is a 0.5V sag on 3V average one is already awfully close to the 2.5V.
For 12V regs 2.5V dV is minimum voltage.
When there is a 0.5V sag on 3V average one is already awfully close to the 2.5V.
As you know, I prefer plots and proofs to words and speculations , so please find it below.
solderdude
Grand Contributor
How does the 7912 react at the same levels ?
When that is quite different the PSRR of the circuit it feeds may become important.
When that is quite different the PSRR of the circuit it feeds may become important.
My experience is that those 78XX/79XX regulator behaviour depends on the manufacturer of the part. Especially the start-up behaviour of the 79XX line, when they put high overvoltage at their output if not loaded at least of 5mA and sometimes more. Similar re regulation behaviour. We tend to look at those parts as if they were some kind of a "standard", however they are not.
solderdude
Grand Contributor
I have noticed different performance as well between different brands as well as the LM340 series.
When one has enough dV and low enough ripple on the primary and uses at least a decent PSRR circuit it is feeding there were no consequences in the audio performance (unless something was defective in some way)
When one has enough dV and low enough ripple on the primary and uses at least a decent PSRR circuit it is feeding there were no consequences in the audio performance (unless something was defective in some way)
Well. I may just have an alternative answer.
If you are able to measure thx 789 with unbalanced input, you will see horrible mains leakage. So for the most part cheap SMPS are bad, simply. You have to have medical grade with 2MOPP with 80uA leakage or lower. I haven't tested the improvement yet, that's at least the minimum requirements.
Also for linear supplies. The transformer needs to be far far away from the circuit. It's impossible to have a low interference when they are in the same housing. Thick shielding might help but so far I haven't seen any useful solution if the transformer is close enough. Solution from ATOM and O2 is actually an accidentally amazing solution. They were about cheap solution that still generates real ground instead of the mains leakage and transformer interference. This way one should be able to get -150dbr or lower supply noise on single end which is difficult. Other solutions can vary from -140dbr(rare) to around -110dbr(common) (2V as reference voltage).
Does anyone have better cheaper more universal and versatile solution?
If you are able to measure thx 789 with unbalanced input, you will see horrible mains leakage. So for the most part cheap SMPS are bad, simply. You have to have medical grade with 2MOPP with 80uA leakage or lower. I haven't tested the improvement yet, that's at least the minimum requirements.
Also for linear supplies. The transformer needs to be far far away from the circuit. It's impossible to have a low interference when they are in the same housing. Thick shielding might help but so far I haven't seen any useful solution if the transformer is close enough. Solution from ATOM and O2 is actually an accidentally amazing solution. They were about cheap solution that still generates real ground instead of the mains leakage and transformer interference. This way one should be able to get -150dbr or lower supply noise on single end which is difficult. Other solutions can vary from -140dbr(rare) to around -110dbr(common) (2V as reference voltage).
Does anyone have better cheaper more universal and versatile solution?
audio2design
Major Contributor
- Joined
- Nov 29, 2020
- Messages
- 1,769
- Likes
- 1,832
Looking at some power supplies in audio equipment over the years, I would argue many so called designers can't design a proper linear supply either regulated or not.
Now those people are designing things like DACs.
Now add in some poor grounding practices and lack of knowledge of how ground currents can flow.
It is probably better that many stick to linear supplies.
Nice bench picture. I like the microscope front and center.
Now those people are designing things like DACs.
Now add in some poor grounding practices and lack of knowledge of how ground currents can flow.
It is probably better that many stick to linear supplies.
Nice bench picture. I like the microscope front and center.
audio2design
Major Contributor
- Joined
- Nov 29, 2020
- Messages
- 1,769
- Likes
- 1,832
A single ended cable with a pseudo differential input can negate most ground noise but this goes back to my comment about audio designers.
Most off the shelf non medical supplies have high leakage to GND to address EMI as it is cheaper than doing it in the transformer and usually yields better performance.
Whether that ground leakage impacts your audio is a matter of system design, but a copper strap between components can fix a lot of design ills
On the other hand a linear supply has very high current spikes that can cause neutral/GND voltage to rapidly change inducing noise through EMI capacitors.
Most off the shelf non medical supplies have high leakage to GND to address EMI as it is cheaper than doing it in the transformer and usually yields better performance.
Whether that ground leakage impacts your audio is a matter of system design, but a copper strap between components can fix a lot of design ills
On the other hand a linear supply has very high current spikes that can cause neutral/GND voltage to rapidly change inducing noise through EMI capacitors.
Accuphase and TAD (Pioneer, EXCLUSIVE) have been manufacturing amplifiers for 50 years.
Although it is a power amplifier, it will be a linear power supply even in 2021.
I can't think of any benefits.
Material cost
Packing cost
Transportation costs
Can't be lifted
Magnetic flux leakage
Fever
electric bill
Looks heavy and troublesome both during manufacturing and inspection
Need more time
It's full of flaws.
That cost and effort should be spent on SMPS development, but are Accuphase and TAD incompetent?
Although it is a power amplifier, it will be a linear power supply even in 2021.
I can't think of any benefits.
Material cost
Packing cost
Transportation costs
Can't be lifted
Magnetic flux leakage
Fever
electric bill
Looks heavy and troublesome both during manufacturing and inspection
Need more time
It's full of flaws.
That cost and effort should be spent on SMPS development, but are Accuphase and TAD incompetent?
AnalogSteph
Major Contributor
The implications of a power supply go far beyond what you might think. If you want to build a traditional hi-fi power amp as an IEC Class II device with RCA inputs, you are going to have one hell of a time pulling it off with anything other than a (probably custom-wound) mains transformer with shield winding, as manufacturer have been doing for 40-odd years. At the same time, relatively high-voltage, unregulated linear supplies actually can be made decently efficient as-is, while at the same time being robust, long-lived and not bothered by short-term overload, which is critical in audio amplifier use.
Likewise, if you only need to supply a little DAC (5 V at 200 mA or so), a linear supply is the easiest way of keeping mains leakage to a minimum. There are relatively few small SMPS that are addressing this issue, and they're not necessarily the most reliable, though I guess this may be unrelated.
Large-scale adoption of SMPS would require dumping unbalanced (RCA) audio connections in all but a few places, and even then care in construction and parts choice is required. It would also be beneficial to have outlets with protective earth available, which still is not standard in a fair few corners of the world.
Realistically, we're already well underway... future repair statistics will show how things are doing long-term.
Likewise, if you only need to supply a little DAC (5 V at 200 mA or so), a linear supply is the easiest way of keeping mains leakage to a minimum. There are relatively few small SMPS that are addressing this issue, and they're not necessarily the most reliable, though I guess this may be unrelated.
Large-scale adoption of SMPS would require dumping unbalanced (RCA) audio connections in all but a few places, and even then care in construction and parts choice is required. It would also be beneficial to have outlets with protective earth available, which still is not standard in a fair few corners of the world.
Realistically, we're already well underway... future repair statistics will show how things are doing long-term.
Jan Didden’s Silent Switcher begs to disagree.
How do you reach that conclusion? It's much easier to filter out high-frequency switching noise than mains hum. Also, since it's inaudible, it doesn't matter if you don't get rid of it all.
In a small signal setting like a preamp, inexpensive linear voltage regulators do an excellent job of removing 60Hz or 120Hz ripple noise. Large split-supply power amps have high PSRR and don't care about the LF ripple if it's not stupidly large. It's easy to correct that out on the fly. It's true that HF noise does not need large amounts of C to clean it up, but it's more pernicious because it contains RF hash that can be radiated and get picked up elsewhere. LF ripple does not have that radiation problem. With good design, though switchmode noise can be minimized to negligible levels. Also, it is usually so high in frequency that it won't be audible if it does not disturb circuit performance. It's just not as easy to get it right (removed) as it is with LF noise. Beginner hobbyists are likely to have more success with linear designs. As for linear post regulation after a switcher, it's a mixed bag. Slow error amps in the regulators may not be able to follow the HF noise and correct it out, and the noise will ride right through them Yee-ha! style.
