A properly designed SMPS for a power amplifier brings a lot more to the table than just possibly reducing the need for reservoir capacitor size. The fact that an SMPS may need a 10,000uF reservoir capacitor to handle infrequent peak currents is not a bad reflection on the SMPS. Just because an SMPS needs a reservoir capacitor of 10,000uF or more to make the best amplifier is no reason to consider reverting to a conventional supply.
It is very important to consider both the small-signal and large-signal aspects of the SMPS and the role of the reservoir capacitor. Confusion here can lead to wrong conclusions.
A typical SMPS will do everything it can to maintain tight regulation at its output. But if it can't supply the peak current, the reservoir capacitor will supply it, the output will begin to fall a bit, the SMPS will fall out of regulation, and it will probably go into constant current mode or some semblance thereof. This is OK. The reservoir capacitor is doing its job. In such a scenario, we do not want an SMPS that shuts down under such a brief event.
A semi-regulated SMPS, where the control loop is not closed from the output side, will naturally sag under the high peak current conditions and allow the reservoir capacitor to supply the current. This is probably the best choice for an audio power amplifier SMPS.
Cheers,
Bob
It is very important to consider both the small-signal and large-signal aspects of the SMPS and the role of the reservoir capacitor. Confusion here can lead to wrong conclusions.
A typical SMPS will do everything it can to maintain tight regulation at its output. But if it can't supply the peak current, the reservoir capacitor will supply it, the output will begin to fall a bit, the SMPS will fall out of regulation, and it will probably go into constant current mode or some semblance thereof. This is OK. The reservoir capacitor is doing its job. In such a scenario, we do not want an SMPS that shuts down under such a brief event.
A semi-regulated SMPS, where the control loop is not closed from the output side, will naturally sag under the high peak current conditions and allow the reservoir capacitor to supply the current. This is probably the best choice for an audio power amplifier SMPS.
Cheers,
Bob
I agree with you on the PSU performance aspects.
My problem is with noise. I am heavily involved in this field in my professional life (100's of millions of controllers and associated mosfets anually).
They are great for computers, laptops, mobile chargers or anywhere where you need high power density and low cost.
But for high end audio, we need to do some more work in my view. Conducted, radisted etc. and it's not just on the secondary side but also the gunk getting pumped back onto the mains.
Naturally, I am no fan of class D either for similar reasons.
My problem is with noise. I am heavily involved in this field in my professional life (100's of millions of controllers and associated mosfets anually).
They are great for computers, laptops, mobile chargers or anywhere where you need high power density and low cost.
But for high end audio, we need to do some more work in my view. Conducted, radisted etc. and it's not just on the secondary side but also the gunk getting pumped back onto the mains.
Naturally, I am no fan of class D either for similar reasons.
Noise can always be filtered out. It just costs money, and requires design skill and effort, which is just another kind of money.
I recently worked on an ultrasonic preamp for industrial testing. It had to resolve signals of a few microvolts in a 3kHz to 500kHz bandwidth, while sharing a board with a half dozen DC-DC converters and a small computer. It took us two board revisions (money) to get the grounding right, but we did it. (Relevance to this thread: I tried several circuits for the front end and the winner was something not unlike an old Douglas Self phono preamp.
)
I recently worked on an ultrasonic preamp for industrial testing. It had to resolve signals of a few microvolts in a 3kHz to 500kHz bandwidth, while sharing a board with a half dozen DC-DC converters and a small computer. It took us two board revisions (money) to get the grounding right, but we did it. (Relevance to this thread: I tried several circuits for the front end and the winner was something not unlike an old Douglas Self phono preamp.
)Nigel, what I asked was how did you calculate your resistor & cap values for your filters.
Was it "Practical high order Bessel & Butterworth filters for Dummies"? It must be quite an advanced text as it includes a spec for the OPAs too.
Could you scan the relevant pages & post for the benefit of us unwashed masses? If it was a link .. even better.
Sorry . I skipped past it . It is a program I used in DOS on a 286 years ago ( trying to find that as I loved the DOS version , like NASA almost when Moon shots ) . Hit F1 and wait 30 seconds . Now as freeware from Texas . It was for a bespoke filter chip that had precision capacitors built in . Texas ( Burr Brown ) realized all of us used standard op amps instead . The Chebishev are very useful for single frequency filtering . I have built filters of frightful complexity using this program . Nearly all worked .
Active Filter Design Application - FILTERPRO - TI Software Folder
Hi Bonsai,
I agree, there is more work to do on the noise/EMI front. Some of the cost savings of SMPS will be reduced by the increased need for attention to, and mitigation of, noise. As far as gunk getting pumped back into the mains, I think that SMPS with PFC wins IF proper attention is paid to that aspect in an SMPS designed for audio applications. Nevertheless, it all adds cost. The shielding that is often needed is a potentially significant cost.
The mains we have to work with are getting worse every day, so we need to take as much advantage of the mains-isolating capabilities of SMPS as possible.
Cheers,
Bob
I noticed noise coming out of the inputs to the Hypex modules into the test gear . Surprised and will investigate . Even if a measuring error real life can be the same . That is how one measures is often how one connects . Perhaps that's why a differential input is offered apart from it being useful ? I note Hypex gives examples of best practice and tell of safety requirements which conflict , they tell with such zeal as to make e think they are saying more . Perhaps these are warnings ?
As someone said , as long as the oscillator frequency is stable is probably will be OK and gross effects should show up ( sub harmonics ) ? Maybe not as my test apparatus I built had a very subtle problem which took time to spot . Even the Hypex itself might do better with an OPA 2604 input rather than NE 5532 etc . Maybe those who say J FET's sound better are right ? Maybe it has nothing to do with the usual . More to do with a RFI polluted environment ? My Hypex has the world most powerful long-wave transmitter in it's locality . I bet 1 MW at 1 mile is less ( Daventry 198 kHz ) ?
I will warm to class D when I can build one from scratch . Today I changed the gain on the Ucd 180 . Not easy . For this one component a standard resistor should have been used .
As someone said , as long as the oscillator frequency is stable is probably will be OK and gross effects should show up ( sub harmonics ) ? Maybe not as my test apparatus I built had a very subtle problem which took time to spot . Even the Hypex itself might do better with an OPA 2604 input rather than NE 5532 etc . Maybe those who say J FET's sound better are right ? Maybe it has nothing to do with the usual . More to do with a RFI polluted environment ? My Hypex has the world most powerful long-wave transmitter in it's locality . I bet 1 MW at 1 mile is less ( Daventry 198 kHz ) ?
I will warm to class D when I can build one from scratch . Today I changed the gain on the Ucd 180 . Not easy . For this one component a standard resistor should have been used .
Bob, perversely, the mains is getting worse IMV because of the SMPS, despite the greater use of PFC.
Scope boy, I too once was the project leader on a high end panel,indicator and used an SMPS that was just cms away from a TC front end amplifier and conditioner - so uV signals.
The biggest issue was preventing mains conducted noise from finding its way onto the input.
Scope boy, I too once was the project leader on a high end panel,indicator and used an SMPS that was just cms away from a TC front end amplifier and conditioner - so uV signals.
The biggest issue was preventing mains conducted noise from finding its way onto the input.
Yes, you are right. Most SMPS out there are designed as cheaply as possible to get away with the governmental requirements. And they are now everywhere! TVs, computers, even CFL light bulbs and LED light bulbs.
But that, of course, does not mean our audiophile equipment should make the problem bad locally near our other equipment, for our own good.
For what its worth for our amplifiers, SMPS solves our problems of inrush (if using a PFC boost converter up front) and DC on the mains.
Cheers,
Bob
Hello Bob,
Is it just for the speaker load or also including a passive cross-over network, if any?
Since i have made various measurements with drivers ranging from 6 inch to 21 inch [mostly woofers] with nominal impedance of 8 ohms and Z' dipping to anywhere from 4 to 3 ohms at max. With passive XO, the things change considerably, where you have huge current peak demands itself posed by the passive XO alone.
Also i would like to add another interesting fact is that having 2 amplifiers say Class-AB and Class-D of equal power rating, driving same reactive network[Passive XO + Driver], the current demands posed by Class-AB would be huge in comparison to what Class-D can take from the power supply. In other words Class-D recycles the reactive energy[True Power Converter] whereas Class-AB dissipates it as heat.
Kanwar
Last edited:
Hypex SMPS is without PFC. Also their EMI filter is not good in preventing the mains contamination. They only use single PI stage EMI filter.
Very interesting to know . PFC making the mains worse ? I guess that is one big problem spread smaller and wider ?
I will be working on the Hypex SMPS 400 ( ? ) today . It anything of note comes up I will post it . If a reasonable device I plan to try it with some Hitachi derived class AB to see what comes of it . It is the PSU for 2 x UCD 180 I have under test .
PFC makes the mains healthy by not sucking up charging peaks. PFC draws the current from mains for almost whole of the sinewave cycle.
A SMPS without PFC is more prone to contaminate mains.
An externally hosted image should be here but it was not working when we last tested it.
Non proper implementation of EMI filters in SMPS also cause mains contamination by injecting it with RF junk.
Last edited:
I would believer that . When the US perfected nuclear subs the Soviets rushed to have their own . Not able to get the risk factors low they went for carefully pumped air to spread the problem . A contamination democracy ( PFC ) . Also it is said all areas were below unsafe levels thus . Now I know more about these things I totally doubt that story . The radiation not a substance in the way air is . However mains filtering is much like that . Ideal filtering is like a harbour . Storms outside and calm inside . Michale Gerzon said much about this . Gilbert Briggs also gave anecdotal evidence in the 1950/60 's . In his book of musical electronic instruments he said a friend seemed to be able to accurately detect the advantages of ideal filters ( Briggs was correct in his assumption about filter type , friend of Peter Baxendale ) . When his gifted friend's hearing was tested he had very noticeable hearing loss . Contrary to popular belief ( now even ) Briggs felt the person with reduced hearing could be more aware of any loss . Comparing with eyesight that makes sense . What Briggs was saying is the ringing of filters is audible . Gerzon went further and gave the maths to cover it ( I think he called one quality Echo Flutter ) . Michael would have been a friend if I had taken more trouble . Ringing mains filters ? I think that would be very possible .
There are standards for conducted emissions ("junk into the mains") that any SMPS has to meet to be legal for sale. A competently designed SMPS will only just meet the standard and no more, because pushing the emissions down any further would increase the cost with no tangible benefit. My own experience in EMC testing bears this out, I have seen all the little harmonics from the SMPS snuggle up just under the limit line leaving hardly any room for the crud coming out of my own circuitry. :-o
There's nothing stopping you from taking a SMPS and throwing it in a metal box with additional filtering. Say you added one of those filtered IEC inlets to the mains input, and another stage of LC filtering to the output, with feed through capacitors as the final C. That would cut down the noise quite a bit, at least at ultrasonic and radio frequencies.
Another more sinister issue is when a SMPS shows instability in its control loop at audio frequencies. This often happens at low currents around the transition to continuous current mode. It is completely harmless, until you put it in an audio amp and become plagued by whistles, buzzes and bacon frying sounds injected into the audio.
Obviously if there is no control loop it can't happen, and I think this partly explains the popularity of unregulated and "semi-regulated" designs for audio.
There's nothing stopping you from taking a SMPS and throwing it in a metal box with additional filtering. Say you added one of those filtered IEC inlets to the mains input, and another stage of LC filtering to the output, with feed through capacitors as the final C. That would cut down the noise quite a bit, at least at ultrasonic and radio frequencies.
Another more sinister issue is when a SMPS shows instability in its control loop at audio frequencies. This often happens at low currents around the transition to continuous current mode. It is completely harmless, until you put it in an audio amp and become plagued by whistles, buzzes and bacon frying sounds injected into the audio.
Obviously if there is no control loop it can't happen, and I think this partly explains the popularity of unregulated and "semi-regulated" designs for audio.
Last edited:
The metal box usually does nothing or makes it worse . Close to an inductor for example . Mu metal would often be used . Null testing is best .
Dumping into the mains is the question . Looking at many designs it seems common-mode is the culprit as the inductors seem to be of the common-mode type . That is not intuitive as the source of noise seems to be differential ? My late brother maintained that differential filters create common-mode noise and vice-verse . Ringing and harmonic distortion asymmetry ?
I dare say a very stable SMPS oscillator is not too harmful as far as proximity effects are concerned . If subharmonics are only the result of modulation a fixed frequency might not be so bad ? If bad it should show up if a conventional PSU was used as a reference ?
Dumping into the mains is the question . Looking at many designs it seems common-mode is the culprit as the inductors seem to be of the common-mode type . That is not intuitive as the source of noise seems to be differential ? My late brother maintained that differential filters create common-mode noise and vice-verse . Ringing and harmonic distortion asymmetry ?
I dare say a very stable SMPS oscillator is not too harmful as far as proximity effects are concerned . If subharmonics are only the result of modulation a fixed frequency might not be so bad ? If bad it should show up if a conventional PSU was used as a reference ?
H.Ott tells us that the mains input filter must be inside it's own box and that box must be HF connected to the enclosing Chassis.
The better the connection between the Box and the Chassis the better the filter performs.
The plated finish of an IEC filter can make electrical contact around the whole perimeter of the hole in the chassis that the boxed filter is inserted through.
I have never used the electrical gasket that H.Ott refers to, but he does say this gasket is important to the HF impedance, that is necessarily low.
The better the connection between the Box and the Chassis the better the filter performs.
The plated finish of an IEC filter can make electrical contact around the whole perimeter of the hole in the chassis that the boxed filter is inserted through.
I have never used the electrical gasket that H.Ott refers to, but he does say this gasket is important to the HF impedance, that is necessarily low.
Yup. We hope that our SMPS won't produce any "junk" in the VHF and UHF bands (although this isn't necessarily the case as switching frequencies go up) so we don't need to go to the lengths of finger stock and conductive gaskets. Holes in shielding are OK as long as their size is only a small fraction of the wavelength we're trying to shield.
A metal enclosure will rarely make matters worse. It gives electrostatic shielding (Faraday cage) and some magnetic shielding, though not really enough to be useful at lower audio frequencies, hence my comment about instability in control loops. When shopping for SMPSs for audio you have to be really picky about this.
I've heard it said that a steel chassis can channel hum fields from transformers and couple them into other components, but I've never experienced it in practice.
A metal enclosure will rarely make matters worse. It gives electrostatic shielding (Faraday cage) and some magnetic shielding, though not really enough to be useful at lower audio frequencies, hence my comment about instability in control loops. When shopping for SMPSs for audio you have to be really picky about this.
I've heard it said that a steel chassis can channel hum fields from transformers and couple them into other components, but I've never experienced it in practice.
Hi Kanwar,
What I was referring to is the loudspeaker system as a whole, including the passive crossovers. And you are correct: the passive crossovers can also create quite high currents.
In Class-D amplifiers the recycling of reactive energy results in what is called "bus pumping". Indeed, bus pumping can occur even when there is no reactive load, due to the energy storage in the output inductor. Bus pumping is explained in my book in Section 29.3 and Figure 29.4. Bus pumping can increase the rail voltage above its nominal level. A large reservoir capacitor mitigates this effect.
Cheers,
Bob
Hi Bob,
Yes the "Bus pumping" is the resultant effect from recycling energy. Also if the output stage is of Full-Bridge topology[in my experience], you can eliminate the pumping completely.
An externally hosted image should be here but it was not working when we last tested it.
regards,
Kanwar
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.