Hello,
I am confused that there are so many Bridge SMPS's being built on the diyaudio forum, for powering Class D amplifiers.
Since we are diy'ers, and since we are powering our audio equipment, and since our main goal is Audio Amplifier performance, why are so many people choosing to vastly over-engineer their SMPS?
Below 500W, for an audio application , there is absolutely no need to do anything more than a Flyback.
The average power of a Class D amplifier is Peak Power /8
So for a 500W amplifier, that means its basically 62W.....and that is flyback territory.......do we want a half-bridge or full bridge for 62W?
Also, a flyback is the simplest SMPS.....it gives you well-coupled split rail outputs without needing external coupled inductors.
A split output flyback has just one FET and two diodes.
The 500W Single switch flyback will give you about 4W or so of dissipation in the primary RCD clamp even when on no load......but we in the audio world are not required to conform to "Blue Angel" type efficiency requirements..........just like we are not required to conform to PFC requirements.....so its no problem........if it is a problem, then simply do a two switch flyback instead.
Doing a Bridge SMPS for a 500W amplifier which is going to be on 62W average power is massive overkill.
Anybody would think that this was a Switch Mode Power Supply website.........rather than an audio website.
Another thing is that Class D supplys need fast transient response, and so they need current mode control, with duty cycle of 0.4 maximum.
-That means that the flyback primary RMS current need be no bigger than any other topology.
-Admittedly a full bridge will have less rms current *per FET*.....but for 62W power, who cares?
The Half-Bridge is a real bogus smps......it may have two fets switching alternateley, but it scores a MASSIVE own goal in that only half the bus is switched...what a waste of time that is for audio!
So , does any reader know why the SMPS's being built on this site are miles over-engineered?
I am confused that there are so many Bridge SMPS's being built on the diyaudio forum, for powering Class D amplifiers.
Since we are diy'ers, and since we are powering our audio equipment, and since our main goal is Audio Amplifier performance, why are so many people choosing to vastly over-engineer their SMPS?
Below 500W, for an audio application , there is absolutely no need to do anything more than a Flyback.
The average power of a Class D amplifier is Peak Power /8
So for a 500W amplifier, that means its basically 62W.....and that is flyback territory.......do we want a half-bridge or full bridge for 62W?
Also, a flyback is the simplest SMPS.....it gives you well-coupled split rail outputs without needing external coupled inductors.
A split output flyback has just one FET and two diodes.
The 500W Single switch flyback will give you about 4W or so of dissipation in the primary RCD clamp even when on no load......but we in the audio world are not required to conform to "Blue Angel" type efficiency requirements..........just like we are not required to conform to PFC requirements.....so its no problem........if it is a problem, then simply do a two switch flyback instead.
Doing a Bridge SMPS for a 500W amplifier which is going to be on 62W average power is massive overkill.
Anybody would think that this was a Switch Mode Power Supply website.........rather than an audio website.
Another thing is that Class D supplys need fast transient response, and so they need current mode control, with duty cycle of 0.4 maximum.
-That means that the flyback primary RMS current need be no bigger than any other topology.
-Admittedly a full bridge will have less rms current *per FET*.....but for 62W power, who cares?
The Half-Bridge is a real bogus smps......it may have two fets switching alternateley, but it scores a MASSIVE own goal in that only half the bus is switched...what a waste of time that is for audio!
So , does any reader know why the SMPS's being built on this site are miles over-engineered?
Last edited:
I see what you mean and somewhat agree.
But we need to be able to deliver 500W instantly. This can be solved with reservoir capacitors.
Otherwhise we need to design a power supply with 62W average and 500W peak. This means that we can use less cooling. We can use FETs with higher RDSon.
But the transfomer, here we have two criterias, it shall not overheat and not saturate. A transformer must never saturate, this gives a current rush, since the inductance disappears.
But the PSU must be able to handle a sine output from the amplifier at full power.
But we need to be able to deliver 500W instantly. This can be solved with reservoir capacitors.
Otherwhise we need to design a power supply with 62W average and 500W peak. This means that we can use less cooling. We can use FETs with higher RDSon.
But the transfomer, here we have two criterias, it shall not overheat and not saturate. A transformer must never saturate, this gives a current rush, since the inductance disappears.
But the PSU must be able to handle a sine output from the amplifier at full power.
A 62 watt flyback supply that has to provide 500 watts peak is a mess. Let me try to explain before you resopnd. Any supply has to be designed to provide the maximum "rated" power no matter long it lasts. Otherwise it will sag significantly when the high power load is applied. Therefore, you must design for 500 watts. The net result will be that at 62 watts the duty cycle will be very small and the currents will be relatively large at this power and duty cycle. The efficiency will suffer. In a "forward" type converter the duty cycle is a function of the output voltage, not out power. Therefore it will have a higher efficiency.
Remember i'm speaking of Continuous conduction mode at 500W for flyback, forward or bridge.....
The primary current profile, whether flyback, forward or bridge, will be the same.
This is because we cannot allow the duty cycle to be more than 0.4......if we allow more than 0.4 then we have slower transient response and need slope compensation.
So we are looking at trapezoidal shaped primary current for flyback, forward or bridge.................
So the duty cycle will not vary with load for flyback, forward or bridge......that is, until the load gets really light and we go into discontinuous mode.....in which caase the duty cycle will then decrease in flyback, forward or bridge.
The shape of the primary current will be the same whether flyback forward or bridge.
Regarding bridges.....
The textbooks tell us that , for example , a full-bridge is superior to a two transistor forward because it more greatly utilises the transformer....in that it can have duty cycle greater then 0.5.
...but in supply for class D audio, we want to be in current mode, and we do not want a duty cycle >0.4 because it just means a slower transient response...so the bridges ability to have duty cycle > 0.5 is useless to us.
I will however, admit that the flyback would be less efficient than bridge or 2-transistor-forward..................but many amplifiers use mains transformers these days and are very inefficient...............................in audio, efficiency isnt really bothersome.......no government officials are going to crack open an amplifier to test it for blue angel conformity.
The flyback would be more inefficient than bridge or 2-tran-forward because of switching losses and the use of ~800V fets which have higher rdson.............but with 62W average power and for audio use, are we concerned about shaving the efficiency to hell and back.?
I would certainly rate the components for the 500W, and it still wouldnt be that large, and in any case for diyaudio, its absolutely fine, leaves more time to get on with designing the class D audio amplifier
The primary current profile, whether flyback, forward or bridge, will be the same.
This is because we cannot allow the duty cycle to be more than 0.4......if we allow more than 0.4 then we have slower transient response and need slope compensation.
So we are looking at trapezoidal shaped primary current for flyback, forward or bridge.................
So the duty cycle will not vary with load for flyback, forward or bridge......that is, until the load gets really light and we go into discontinuous mode.....in which caase the duty cycle will then decrease in flyback, forward or bridge.
The shape of the primary current will be the same whether flyback forward or bridge.
Regarding bridges.....
The textbooks tell us that , for example , a full-bridge is superior to a two transistor forward because it more greatly utilises the transformer....in that it can have duty cycle greater then 0.5.
...but in supply for class D audio, we want to be in current mode, and we do not want a duty cycle >0.4 because it just means a slower transient response...so the bridges ability to have duty cycle > 0.5 is useless to us.
I will however, admit that the flyback would be less efficient than bridge or 2-transistor-forward..................but many amplifiers use mains transformers these days and are very inefficient...............................in audio, efficiency isnt really bothersome.......no government officials are going to crack open an amplifier to test it for blue angel conformity.
The flyback would be more inefficient than bridge or 2-tran-forward because of switching losses and the use of ~800V fets which have higher rdson.............but with 62W average power and for audio use, are we concerned about shaving the efficiency to hell and back.?
I would certainly rate the components for the 500W, and it still wouldnt be that large, and in any case for diyaudio, its absolutely fine, leaves more time to get on with designing the class D audio amplifier
topology choices are not quite so straight forward...
To each there own...
I have done multi-kilowatt flybacks and I have done 65W HB designs (in fact I have even designed a 9W full-bridge before). In each case the end application ripple requirements,EMI performance,and cost determined my topology.
Most of the time the driving consideration for me is output ripple at peak load. Usually a flyback has a duty cycle less than 75%. The secondary capacitance has to support the output the entire time that the primary side switch is conducting. For HB/FB applications this means that on average my output capacitance has to be larger for a flyback than a HB/FB topology. If my end application can tolerate this, then there really isn't an issue.
EMI is the other portion of this that seems to be overlooked a lot in the design considerations. In general I tend to use the resonant topologies for that reason.
The rule of thumb I have usually used is that for average power to about 65W I use standard flybacks. When I have a PFC front end I can stretch the Flyback to 130-180W. Above that I am using HB/FB topologies. When I have high peak to average power ratios such as you describe, I tend to end up with slightly higher complexity designs to make sure that I can handle the corner cases (low line and peak current) in order to ensure design margin.
One final thought... I design power supplies for a living so what is easy for me is not so easy for others. Thus... I leave the topology choice recommendation to what the designer is comfortable with as long as the cost and performance criteria are met.
Tony
To each there own...
I have done multi-kilowatt flybacks and I have done 65W HB designs (in fact I have even designed a 9W full-bridge before). In each case the end application ripple requirements,EMI performance,and cost determined my topology.
Most of the time the driving consideration for me is output ripple at peak load. Usually a flyback has a duty cycle less than 75%. The secondary capacitance has to support the output the entire time that the primary side switch is conducting. For HB/FB applications this means that on average my output capacitance has to be larger for a flyback than a HB/FB topology. If my end application can tolerate this, then there really isn't an issue.
EMI is the other portion of this that seems to be overlooked a lot in the design considerations. In general I tend to use the resonant topologies for that reason.
The rule of thumb I have usually used is that for average power to about 65W I use standard flybacks. When I have a PFC front end I can stretch the Flyback to 130-180W. Above that I am using HB/FB topologies. When I have high peak to average power ratios such as you describe, I tend to end up with slightly higher complexity designs to make sure that I can handle the corner cases (low line and peak current) in order to ensure design margin.
One final thought... I design power supplies for a living so what is easy for me is not so easy for others. Thus... I leave the topology choice recommendation to what the designer is comfortable with as long as the cost and performance criteria are met.
Tony
Regarding EMC, a 500W flyback for Audio Class D supply would be EMC tested at 1/8th power....ie 62W.....this does ease the situation with EMC a bit.
For Class D amplifier Power supplies, the Class D output has a huge LC filter and so switching frequency ripple on the power supply output isn't too much of an issue, as switching frequency is typically well above audio frequency and so gets filtered out by the LC filter of the Class D amplifier.
Also, An audio offline flyback solution at 500W isnt necessarily all that physically big.
....also, the flyback means you avoid coupled output inductors if its a split rail supply.
(pse see pg 3 part 5.............
http://www.ti.com/lit/an/slua119/slua119.pdf
)
Also, its pointless to use a high side drive if theres no need.
The single switch flyback does give primary clamp resistor losses, though the "RCDQ" clamp can mitigate this at light load.
I think Class D audio amplifier design, as well as all the tweaking of the pre-amplifier circuitry so that the best sound can be achieved, is hard enough, and more than a lifetime's work for anybody, if you really go into audiophile territory.......the same for other amplifiers too.........
...given this fact, its surprising that there's so much over-engineering of the SMPS being done...........Definetely for 500W and below, flyback is all you need in the audio world.
Some would say some type of resonant converter should be used as it can give small solution size at high switching frequency...........but small solution size, certainly for guitar amplifiers, is not wanted............the guitar amplifier and speaker cabinet is part of the stage "furniture"......no self-respecting guitar player is going to stand there with some tiny cube of a guitar amplifier head........on the contrary , its the bigger the better.
For Class D amplifier Power supplies, the Class D output has a huge LC filter and so switching frequency ripple on the power supply output isn't too much of an issue, as switching frequency is typically well above audio frequency and so gets filtered out by the LC filter of the Class D amplifier.
Also, An audio offline flyback solution at 500W isnt necessarily all that physically big.
....also, the flyback means you avoid coupled output inductors if its a split rail supply.
(pse see pg 3 part 5.............
http://www.ti.com/lit/an/slua119/slua119.pdf
)
Also, its pointless to use a high side drive if theres no need.
The single switch flyback does give primary clamp resistor losses, though the "RCDQ" clamp can mitigate this at light load.
I think Class D audio amplifier design, as well as all the tweaking of the pre-amplifier circuitry so that the best sound can be achieved, is hard enough, and more than a lifetime's work for anybody, if you really go into audiophile territory.......the same for other amplifiers too.........
...given this fact, its surprising that there's so much over-engineering of the SMPS being done...........Definetely for 500W and below, flyback is all you need in the audio world.
Some would say some type of resonant converter should be used as it can give small solution size at high switching frequency...........but small solution size, certainly for guitar amplifiers, is not wanted............the guitar amplifier and speaker cabinet is part of the stage "furniture"......no self-respecting guitar player is going to stand there with some tiny cube of a guitar amplifier head........on the contrary , its the bigger the better.
My reasoning for resonant is several fold. First I am more efficient thus heat sinking, airflow, etc... are minimized. It has very low harmonic content on the secondary side so that the noise is lower (less chance of the noise coupling into the amplifiers). I can run higher power levels with better core utilization, EMI is better almost everywhere, the Sync Fet drive is easy unlike CCM Flyback and finally, there is no need for a couple inductor since I use the leakage of the transformer as the resonant element.
But again, to each their own. There are a number of ways to arrive at the same endpoint. I mention resonant because I am fond of them for the upper power ranges. Just as I am for a phase shifted full bridge or interleaved resonant converters. They all have their place and I won't denigrate anyones choice on the issue.
It all comes down to what you are comfortable with and what design goals you are trying to achieve. That doesn't mean that any one topology is the only way or even the "right" way.
Tony
But again, to each their own. There are a number of ways to arrive at the same endpoint. I mention resonant because I am fond of them for the upper power ranges. Just as I am for a phase shifted full bridge or interleaved resonant converters. They all have their place and I won't denigrate anyones choice on the issue.
It all comes down to what you are comfortable with and what design goals you are trying to achieve. That doesn't mean that any one topology is the only way or even the "right" way.
Tony
One limitation of LLC resonant converters is the dynamic range of the gain......this means that universal operation (even with a changeover link) is not really possible.
.....ie please see under "frequency and gain" section of the following......
http://www.eetimes.com/design/power.../Using-quasi-resonant-and-resonant-converters
Also, with LLC resonant converters, they are voltage mode controlled.....and so the transient response is not as quick as a current mode controlled topology......transient response is very important for class D power supplys.
There is a way round this, ie you can run the LLC resonant converter at a very high switching frequency and end up enabling a high loop bandwidth like that, but high switching frequencys cause problems even for LLC converters..........the main problem is the gate drive transformer, because at high frequencies, it is particularaly sensitive to leakage inductance, so it has to be manufactured with a very strict tolerance.
Another problem for LLC converters is multiple outputs, including bias windings........The modus operandi of an LLC converter with an integrated transformer is that there is *not* good coupling between primary and secondary............this means problems for multiple windings, where you have to worry over the way in which each coil is coupled in to the core..............with an integrated transformer, the secondary is often 'scrunched' into a tight space, and this is not a good starting point to get an extra output winding to couple up to it.
The same problems arise for the bias winding, .....and where llc converters are supplied by PFC, a coil is often used on the pfc inductor to supply the LLC resonant converter.
Although it might be worth thinking about it with an external "leakage" inductor......although it may end up being rather large.
So LLC converters for Class D amplifier supply at least, is not necessarily the best way.
The thing about audio is that the equipment is going to be big and chunky anyway..........speaker cabinets are big pieces of furniture......you cant make them sleek and flat, so the name of the game in audio is basically having the amp and speaker as big blocks of furniture, and making them part of the stage set up.
This means that small 30W/inch^3 SMPS's are not really all that required in the audio world.
The one application that actually screamed for an LLC resonant converter is the LED streetlight........theres a PFC front end, so no line regulation problems, and the load is constant load......so the LLC converter is basically the only realistic way forward there due to the superb cost and efficiency for that particular use. (being streetlights, very numerous, the authorites insist on efficiency requirements, unlike the audio world, where the authorities turn a blind eye even to very inefficienct mains transformer based power supplies)
.....ie please see under "frequency and gain" section of the following......
http://www.eetimes.com/design/power.../Using-quasi-resonant-and-resonant-converters
Also, with LLC resonant converters, they are voltage mode controlled.....and so the transient response is not as quick as a current mode controlled topology......transient response is very important for class D power supplys.
There is a way round this, ie you can run the LLC resonant converter at a very high switching frequency and end up enabling a high loop bandwidth like that, but high switching frequencys cause problems even for LLC converters..........the main problem is the gate drive transformer, because at high frequencies, it is particularaly sensitive to leakage inductance, so it has to be manufactured with a very strict tolerance.
Another problem for LLC converters is multiple outputs, including bias windings........The modus operandi of an LLC converter with an integrated transformer is that there is *not* good coupling between primary and secondary............this means problems for multiple windings, where you have to worry over the way in which each coil is coupled in to the core..............with an integrated transformer, the secondary is often 'scrunched' into a tight space, and this is not a good starting point to get an extra output winding to couple up to it.
The same problems arise for the bias winding, .....and where llc converters are supplied by PFC, a coil is often used on the pfc inductor to supply the LLC resonant converter.
Although it might be worth thinking about it with an external "leakage" inductor......although it may end up being rather large.
So LLC converters for Class D amplifier supply at least, is not necessarily the best way.
The thing about audio is that the equipment is going to be big and chunky anyway..........speaker cabinets are big pieces of furniture......you cant make them sleek and flat, so the name of the game in audio is basically having the amp and speaker as big blocks of furniture, and making them part of the stage set up.
This means that small 30W/inch^3 SMPS's are not really all that required in the audio world.
The one application that actually screamed for an LLC resonant converter is the LED streetlight........theres a PFC front end, so no line regulation problems, and the load is constant load......so the LLC converter is basically the only realistic way forward there due to the superb cost and efficiency for that particular use. (being streetlights, very numerous, the authorites insist on efficiency requirements, unlike the audio world, where the authorities turn a blind eye even to very inefficienct mains transformer based power supplies)
eem2am
I will have to disagree with you with regards to response times. A resonant converter running at resonance in the voltage domain is only limited by the tank impedance on the primary. I can do a 0-12A (50KHz resonant tank using integrated Lk/Lm) stepped squarewave at 10KHz using a loop crossover frequency of 1KHz. The only sag in the output is due to the IR losses through the power train. Remember, at resonance the transformer acts as an ideal transformer (strictly the tVoltage divided by the turns ratio minus IR losses). Your current automatically adjusts without having to change your operating frequency. If you try doing this with a flyback you will have a serious sag/overshoot issue as you are outside the loop crossover frequency.
As to harmonic content, obviously the PFC front end is the dominant source of of EMI as the downstream LLC has little harmonic content. The PFC will be the same for both models while the flyback will have common mode/differential noise depending on its operating modes (CCM/DCM).
The last item is that I can run my LLC at 240W continuous, at 14W.in^3 and no fan and get 90% efficiency. Audible noise in the room is also important to me.
eem2ram, I am not arguing the relative merits of different topologies with the flyback. In fact I am a huge flyback fan, all I am saying is that I pick and choose when and where I use them and carefully consider the end applications requirements.
Tony
I will have to disagree with you with regards to response times. A resonant converter running at resonance in the voltage domain is only limited by the tank impedance on the primary. I can do a 0-12A (50KHz resonant tank using integrated Lk/Lm) stepped squarewave at 10KHz using a loop crossover frequency of 1KHz. The only sag in the output is due to the IR losses through the power train. Remember, at resonance the transformer acts as an ideal transformer (strictly the tVoltage divided by the turns ratio minus IR losses). Your current automatically adjusts without having to change your operating frequency. If you try doing this with a flyback you will have a serious sag/overshoot issue as you are outside the loop crossover frequency.
As to harmonic content, obviously the PFC front end is the dominant source of of EMI as the downstream LLC has little harmonic content. The PFC will be the same for both models while the flyback will have common mode/differential noise depending on its operating modes (CCM/DCM).
The last item is that I can run my LLC at 240W continuous, at 14W.in^3 and no fan and get 90% efficiency. Audible noise in the room is also important to me.
eem2ram, I am not arguing the relative merits of different topologies with the flyback. In fact I am a huge flyback fan, all I am saying is that I pick and choose when and where I use them and carefully consider the end applications requirements.
Tony
Dtproff:
It sounds to me that your transient response testing is being done at too high frequency.....it sounds like you are switching the load 0-12A at 10KHz.........thats too high frequency......transient response testing is literally that......looking at a single transient.
Regarding this idea of great transient response of an LLC converter thats switching at the upper resonant frequency........sorry but i cannot agree, none of the semiconductor company LLC datasheets or app notes, and no text book anywhere, advertises the LLC converter as "a converter for good transient response"
...the salesmen would have hit on this one if it was the case....they would have been plugging the transient reponse of their LLC chips if its really as good as you say.
If you can proove that the LLC is better in terms of enabling an improved audio performance of the amplifier then i can take my hat off to you....especially with this being an audio forum.
-You did claim earlier that the LLC has improved EMC, and i can see where you're coming from, but coldamp.com dont have any difficulties with their hard switched converters....
coldamp.com dont need fans for 240W either.
And also as to the subject of integrated transformers............supposing that you have a split output voltage...........now those split output coils need to be well coupled with each other, and each output coil needs to be coupled into the core with the same coefficient..........but thats not easy with an integrated transformer as the secondarys tend to be "scrunched" up into the bobbin chamber, and thats not a good recipe for well coupled output coils.
If this was a forum for Engineers who are designing products for the mass market such as flat screen TV's etc etc, then you would literally be blowing me out of the water with the LLC converters for <500W........you would be blasting me off the walls and watching me drip back down off the ceiling.
But how interested are audio diy'ers in SMPS features that dont actually make their amplifiers sound any better?
-most of the guys on this forum are happy with mains transformers.
-those who want improvements that an smps offers at <500W can get all those improvements with a flyback smps...and then have more time to get on with their amplifier tweaking.
I appreciate that anybody can pick and choose what they want, but if you turn up at the local skateboard park with a formula one ferrari racecar, people might ask why you need that for there........as great as it is.
It sounds to me that your transient response testing is being done at too high frequency.....it sounds like you are switching the load 0-12A at 10KHz.........thats too high frequency......transient response testing is literally that......looking at a single transient.
Regarding this idea of great transient response of an LLC converter thats switching at the upper resonant frequency........sorry but i cannot agree, none of the semiconductor company LLC datasheets or app notes, and no text book anywhere, advertises the LLC converter as "a converter for good transient response"
...the salesmen would have hit on this one if it was the case....they would have been plugging the transient reponse of their LLC chips if its really as good as you say.
If you can proove that the LLC is better in terms of enabling an improved audio performance of the amplifier then i can take my hat off to you....especially with this being an audio forum.
-You did claim earlier that the LLC has improved EMC, and i can see where you're coming from, but coldamp.com dont have any difficulties with their hard switched converters....
coldamp.com dont need fans for 240W either.
And also as to the subject of integrated transformers............supposing that you have a split output voltage...........now those split output coils need to be well coupled with each other, and each output coil needs to be coupled into the core with the same coefficient..........but thats not easy with an integrated transformer as the secondarys tend to be "scrunched" up into the bobbin chamber, and thats not a good recipe for well coupled output coils.
If this was a forum for Engineers who are designing products for the mass market such as flat screen TV's etc etc, then you would literally be blowing me out of the water with the LLC converters for <500W........you would be blasting me off the walls and watching me drip back down off the ceiling.
But how interested are audio diy'ers in SMPS features that dont actually make their amplifiers sound any better?
-most of the guys on this forum are happy with mains transformers.
-those who want improvements that an smps offers at <500W can get all those improvements with a flyback smps...and then have more time to get on with their amplifier tweaking.
I appreciate that anybody can pick and choose what they want, but if you turn up at the local skateboard park with a formula one ferrari racecar, people might ask why you need that for there........as great as it is.
Anyway why bother so much for the transient response for audio use?
Most of the people use linear supply compltely unregulated or bridges running at plain 50% duty cycle and they are happy with it.
You are not powering someting like a cpu core than needs 1V and more than 50A where a voltage sag of only 100mV can be critcal for operation.
Suppose that we have an smps, flyback, forward, llc or whaterver that gives +/-50V 10A (1kW).
Suppose that everything is done as perfect as possible and the transient response gives you only 0.1v of undershoot when suddenly going from 0A to 10A and it is able to recover in 10usec.
Now if I change the topology or the loop phase-frequency compensation of that smps I will have let's say 1V undershoot and 1msec recovery time.
Who cares for audio?
Try to do the same on a 50Hz transformer or on an unregulated brigde and see what is your undershoot (the recovery time is not applicable since you are unregulated).
The amplifier (class D, A, AB, H or whatever else) has lots of PSRR and your 1V undershoot or 0.0001V is basically the same for the amplifier.
I am not a big fan of LLCs (no enough experience in this) but anyway explain me why LLC should have a transient response worse than a CM flyback provided that they have the same loop crossover frequency and phase margin. Not using marketing ******** articles used to sell something but with proven facts and measurements on real and comparable stuff.
Why current mode should be better than voltage mode? It is true that current mode removes a pole and make the compensation easier but a properly compensated voltage mode is not different from a current mode. If current mode is the cure for every problem why there are still some new ICs using voltage mode control? Is every IC manufacturer so stupid that can not understand why current mode is the only way to go? I don't think so....Check your loved Power Integration stuff: TOP switch, tiny switch, etc, they are all voltage mode controlled.
Check Fairchild TM PFC ICs, they are all voltage mode
Ciao
-marco
Most of the people use linear supply compltely unregulated or bridges running at plain 50% duty cycle and they are happy with it.
You are not powering someting like a cpu core than needs 1V and more than 50A where a voltage sag of only 100mV can be critcal for operation.
Suppose that we have an smps, flyback, forward, llc or whaterver that gives +/-50V 10A (1kW).
Suppose that everything is done as perfect as possible and the transient response gives you only 0.1v of undershoot when suddenly going from 0A to 10A and it is able to recover in 10usec.
Now if I change the topology or the loop phase-frequency compensation of that smps I will have let's say 1V undershoot and 1msec recovery time.
Who cares for audio?
Try to do the same on a 50Hz transformer or on an unregulated brigde and see what is your undershoot (the recovery time is not applicable since you are unregulated).
The amplifier (class D, A, AB, H or whatever else) has lots of PSRR and your 1V undershoot or 0.0001V is basically the same for the amplifier.
I am not a big fan of LLCs (no enough experience in this) but anyway explain me why LLC should have a transient response worse than a CM flyback provided that they have the same loop crossover frequency and phase margin. Not using marketing ******** articles used to sell something but with proven facts and measurements on real and comparable stuff.
Why current mode should be better than voltage mode? It is true that current mode removes a pole and make the compensation easier but a properly compensated voltage mode is not different from a current mode. If current mode is the cure for every problem why there are still some new ICs using voltage mode control? Is every IC manufacturer so stupid that can not understand why current mode is the only way to go? I don't think so....Check your loved Power Integration stuff: TOP switch, tiny switch, etc, they are all voltage mode controlled.
Check Fairchild TM PFC ICs, they are all voltage mode
Ciao
-marco
If its voltage mode or current mode but the same Xover frequency and phase margin, then obviously the transient response is the same.
Voltage mode is much more difficult to get good transient response.
Please see page 297 of book "switch mode power supplies" by Christophe Basso.
.....here it confesses that type 3 compensation is needed for voltage mode...and that when using TL431....."TL431 does not lend itself to the type 3 implementation..........etc etc"
So why would you choose voltage mode when current mode is easier for your nice transient response.?
TOPswitch is voltage mode, but thats because they are for simple designs and they dont want users to need slope compensation, so they use voltage mode.......not only that, , but current mode , is known sometimes to give noisy in light load, especially if sense resistor too small.....and powerint is for beginners so they make it easy voltage mode, and hope they dont want better transient response
If you supply Class D amplifier, then you definetely need great transient response as they have poor PSRR.
If you have current mode and duty cycle is >0.5 then you need slope compensation, and so sometimes you pick voltage mode instead because its easyer to make stable at high duty cycle.
With audio smps for class d , you will choose duty cycle <0.4, and its just a whole lot easier to do good transient response with current mode at low duty cycle, so why not use that?
PFC IC's are a different case altogether.......they definetly dont need good transient response...............many have a slow voltage loop, and a faster current loop.
Another reason for current mode is the one you mention of removing a pole.
Voltage mode is much more difficult to get good transient response.
Please see page 297 of book "switch mode power supplies" by Christophe Basso.
.....here it confesses that type 3 compensation is needed for voltage mode...and that when using TL431....."TL431 does not lend itself to the type 3 implementation..........etc etc"
So why would you choose voltage mode when current mode is easier for your nice transient response.?
TOPswitch is voltage mode, but thats because they are for simple designs and they dont want users to need slope compensation, so they use voltage mode.......not only that, , but current mode , is known sometimes to give noisy in light load, especially if sense resistor too small.....and powerint is for beginners so they make it easy voltage mode, and hope they dont want better transient response
If you supply Class D amplifier, then you definetely need great transient response as they have poor PSRR.
If you have current mode and duty cycle is >0.5 then you need slope compensation, and so sometimes you pick voltage mode instead because its easyer to make stable at high duty cycle.
With audio smps for class d , you will choose duty cycle <0.4, and its just a whole lot easier to do good transient response with current mode at low duty cycle, so why not use that?
PFC IC's are a different case altogether.......they definetly dont need good transient response...............many have a slow voltage loop, and a faster current loop.
Another reason for current mode is the one you mention of removing a pole.
Gentlemen;
I feel we are missing the point here. The question was why we favored the HB/FB topologies for the higher power designs. I believe we have responded to the best of our ability as to why we use them.
I am not here to advise you one way or another as to your choice of topologies.
Sometimes I use a flyback, sometimes I use other topologies. It all depends on the price/performance tradeoff that every engineer makes on a daily basis.
I am not here to denigrate anyone's experience, only to offer some kind and helpful hints that may help those in the DIY community.
Tony
I feel we are missing the point here. The question was why we favored the HB/FB topologies for the higher power designs. I believe we have responded to the best of our ability as to why we use them.
I am not here to advise you one way or another as to your choice of topologies.
Sometimes I use a flyback, sometimes I use other topologies. It all depends on the price/performance tradeoff that every engineer makes on a daily basis.
I am not here to denigrate anyone's experience, only to offer some kind and helpful hints that may help those in the DIY community.
Tony
Hi dtproff,
I completely agree with you.
There is no magic topology that has only advantages and no drawbacks.
It all depend on the performance you need, the price you can pay and also from the personal experience you have with a given topology.
There are many ways to get to a final product that meets the specs and every way is right.
To get to the same result probably I will choose one topology and you will choose another one but we are both right provided that we both come to a working product.
@eem2am
so you are saying that current mode has better response than voltage mode just because it is not easy to make a type 3 with tl431.....
TL431 is not the only way to close the loop and generalizing it to the statement that "a current mode flyback has better tranient response compared to an llc" sounds sounds strange.....
Ciao
I completely agree with you.
There is no magic topology that has only advantages and no drawbacks.
It all depend on the performance you need, the price you can pay and also from the personal experience you have with a given topology.
There are many ways to get to a final product that meets the specs and every way is right.
To get to the same result probably I will choose one topology and you will choose another one but we are both right provided that we both come to a working product.
@eem2am
so you are saying that current mode has better response than voltage mode just because it is not easy to make a type 3 with tl431.....
TL431 is not the only way to close the loop and generalizing it to the statement that "a current mode flyback has better tranient response compared to an llc" sounds sounds strange.....
Ciao
OK , your responses are appreciated.
-With my direct approach i am (malheuresement) at risk of seeming as if i am denigrating....but in fact, i am trying to inspire somebody out there to literally blow my initial post out of the water and proove to me that flybacks for <500W audio applications are out of the looney bin.
I am grateful to you because it seems that there really is no reason against 0-500W flybacks for audio purpose.........apart from minor reasons such as smaller heatsinks, efficiency etc etc....that is , reasons that dont relate to the musical quality of the downstream amplifier.
For audio usage then, i will re-iterate my point that flyback really is king for <500W split-rail , Class D power supplies..............because.................
1. You get well-coupled split rail outputs without needing coupled output inductors
2. You can get a very good transient response.
The LLC-with-integrated-transformer is a bad-guy for this purpose because it has scrunched up secondaries , which makes it difficult for you to get good coupling with each other (each of the split rail secondaries) , and makes it awkward to get "identical" coupling into the core of these secondaries..........So these are points which will hit our precious musical quality and so surely "we" declare the LLC-with-integrated-transformer as a bad-guy for music a la split rail.?
Anyway, back to the original post.......where i pointed out that there are a very large number of bridge converters being used for <500W audio purpose......we appear to agree that this can easily be handled by a flyback, which i would think would be cheaper , quicker and simpler.
.....casting the LLC aside for one minute, the full-bridge with its two high side drives..........i cant think why anyone would want to put themselves through that for <500W.
The Half-bridge, hard-switched for <500W audio.........please dont get me back onto that point......the Half-Bridge is a waste of time in my book..........you only switch half the bus, and need a current sense transformer even at the low end of the power scale.
......so half-bridge and full bridge hard-switched for audio with split rail outputs and <500W Class D supply is a waster of time....there, ive said it, its 62W average power.
I am avoiding discussing the two transistor forward or 2 tran flyback, becasue i couldnt think of big reasons against them for audio <500W, though the forward will have you winding coupled output inductors.
One disadvantage for flyback is when output current is high (>6A)..........thankyou for pointing this out, becasue as you say, synchronous output diode(FETs) are awkward in flyback.
----so high current outputs are going to heat the output diodes with flyback......of course, the LLC suffers this point too......but its sync diodes are easier to control.
So audio <500W, with split rail for Class D is flyback territory, unless you want greater efficiency because you like the sound of Blue Angel regulations.......also, unless you want a particularly small solution size and you want your speaker cabinet to look even bigger than it is, sitting next to a tiny PSU.
The fact is that non-musical people, with no interest in audio, are coming onto this forum pretending to be audio enthusiasts, and discussing doing smps's which they in fact intend for use at work in TV's, computers, washing machines etc etc..............genuine audio enthusiasts are then seeing these posts, and copying them, having totally lost site of the fact that the guy was NEVER designing for music in the first place............................probably would like to see music banned as things like guitar amplifiers can't realistically conform to the latest <100mW standby regulations.......because poor Ozzy Osborne would find his guitar amp constantly turning off in between songs, and his first few axe-like strums would be as silent as a mouse, BUT VERY EFFICIENCT!
-With my direct approach i am (malheuresement) at risk of seeming as if i am denigrating....but in fact, i am trying to inspire somebody out there to literally blow my initial post out of the water and proove to me that flybacks for <500W audio applications are out of the looney bin.
I am grateful to you because it seems that there really is no reason against 0-500W flybacks for audio purpose.........apart from minor reasons such as smaller heatsinks, efficiency etc etc....that is , reasons that dont relate to the musical quality of the downstream amplifier.
For audio usage then, i will re-iterate my point that flyback really is king for <500W split-rail , Class D power supplies..............because.................
1. You get well-coupled split rail outputs without needing coupled output inductors
2. You can get a very good transient response.
The LLC-with-integrated-transformer is a bad-guy for this purpose because it has scrunched up secondaries , which makes it difficult for you to get good coupling with each other (each of the split rail secondaries) , and makes it awkward to get "identical" coupling into the core of these secondaries..........So these are points which will hit our precious musical quality and so surely "we" declare the LLC-with-integrated-transformer as a bad-guy for music a la split rail.?
Anyway, back to the original post.......where i pointed out that there are a very large number of bridge converters being used for <500W audio purpose......we appear to agree that this can easily be handled by a flyback, which i would think would be cheaper , quicker and simpler.
.....casting the LLC aside for one minute, the full-bridge with its two high side drives..........i cant think why anyone would want to put themselves through that for <500W.
The Half-bridge, hard-switched for <500W audio.........please dont get me back onto that point......the Half-Bridge is a waste of time in my book..........you only switch half the bus, and need a current sense transformer even at the low end of the power scale.
......so half-bridge and full bridge hard-switched for audio with split rail outputs and <500W Class D supply is a waster of time....there, ive said it, its 62W average power.
I am avoiding discussing the two transistor forward or 2 tran flyback, becasue i couldnt think of big reasons against them for audio <500W, though the forward will have you winding coupled output inductors.
One disadvantage for flyback is when output current is high (>6A)..........thankyou for pointing this out, becasue as you say, synchronous output diode(FETs) are awkward in flyback.
----so high current outputs are going to heat the output diodes with flyback......of course, the LLC suffers this point too......but its sync diodes are easier to control.
So audio <500W, with split rail for Class D is flyback territory, unless you want greater efficiency because you like the sound of Blue Angel regulations.......also, unless you want a particularly small solution size and you want your speaker cabinet to look even bigger than it is, sitting next to a tiny PSU.
The fact is that non-musical people, with no interest in audio, are coming onto this forum pretending to be audio enthusiasts, and discussing doing smps's which they in fact intend for use at work in TV's, computers, washing machines etc etc..............genuine audio enthusiasts are then seeing these posts, and copying them, having totally lost site of the fact that the guy was NEVER designing for music in the first place............................probably would like to see music banned as things like guitar amplifiers can't realistically conform to the latest <100mW standby regulations.......because poor Ozzy Osborne would find his guitar amp constantly turning off in between songs, and his first few axe-like strums would be as silent as a mouse, BUT VERY EFFICIENCT!
Last edited:
Mag:
You yourself have already answered your own question regarding the potential for improved transient response of a current mode flyback.
...you spoke of the removed pole etc etc.
I dont just mean TL431, i am referring to opto feedback, which is the cheap way, and means its slow too, because of the opto pole.
For an LLC with a 80KHz upper resonant frequency, and a current mode flyback with a 67KHz switching frequency, then i would say that you will ALWAYS be able to get a faster transient response with the current mode flyback, providing that you keep duty cycle <0.4.
The LLC is an SMPS for a LED streetlight, or an LCD TV, etc
The designers of which regularly masquerade as audio enthusiasts on this forum.
-I am just hoping that Amplifier enthusiasts know this, and are not getting sucked in to unecessary , over expensive, and over complicated designs.......and never getting time to do their Class D amps etc.
You yourself have already answered your own question regarding the potential for improved transient response of a current mode flyback.
...you spoke of the removed pole etc etc.
I dont just mean TL431, i am referring to opto feedback, which is the cheap way, and means its slow too, because of the opto pole.
For an LLC with a 80KHz upper resonant frequency, and a current mode flyback with a 67KHz switching frequency, then i would say that you will ALWAYS be able to get a faster transient response with the current mode flyback, providing that you keep duty cycle <0.4.
The LLC is an SMPS for a LED streetlight, or an LCD TV, etc
The designers of which regularly masquerade as audio enthusiasts on this forum.
-I am just hoping that Amplifier enthusiasts know this, and are not getting sucked in to unecessary , over expensive, and over complicated designs.......and never getting time to do their Class D amps etc.
Dear eem2am, is admirable the effort you make to document and further information on the SMPS and amplifiers.
but I feel that (in reality) maybe do not know well,anyone of the two.
sorry for this.
it is folly to use a flyback of an amplifier.
not only see the simplicity, but the dirt is proportional to power.(this is in side of concept)
and, just because he has to have a fast response (peak power) one MOSFET topology is not reliable. all i know .. 200w blow ever.
typical flyback chips are unstable with impulsive loading. (in fact are used in constant load at TV etc.)
I do not know of a company of chip for SMPS who presented a scheme or application note for audio amplifier...think about it.
Sorry if i not explain all reason ,but is long about smps and a bit far from datasheet.
regards
but I feel that (in reality) maybe do not know well,anyone of the two.
sorry for this.
it is folly to use a flyback of an amplifier.
not only see the simplicity, but the dirt is proportional to power.(this is in side of concept)
and, just because he has to have a fast response (peak power) one MOSFET topology is not reliable. all i know .. 200w blow ever.
typical flyback chips are unstable with impulsive loading. (in fact are used in constant load at TV etc.)
I do not know of a company of chip for SMPS who presented a scheme or application note for audio amplifier...think about it.
Sorry if i not explain all reason ,but is long about smps and a bit far from datasheet.
regards
Thankyou AP2:
I am not sure whether you are for or against use of flyback SMPS's to power Class D amplifiers at powers less then 500W?
I also dont know of a company who make PWM controllers especially for Audio Amplifiers..............all i've seen is this....
http://www.powerint.com/sites/default/files/PDFFiles/rdr203.pdf
You are saying that all that i've said is folly?...i am not sure?
I've certainly heard many engineers telling that flybacks are not possible for supplying Class D amplifiers <500W......but i've never heard any reason why not, and the Labgruppen company actually have an 8KW flyback, that another poster mentioned.
By "dirt" are you referring to Conducted emissions?
Any diy audio hobbyist can use a flyback up to 500W without problem for supplying class D amplifiers...........mag has already doen so in a different thread of his.
I wouldnt have thought DIY audio people want to be spending amplifier time building over-compliciated SMPS's?....specially when they offer no real audio improvement over a flyback.
I am not sure whether you are for or against use of flyback SMPS's to power Class D amplifiers at powers less then 500W?
I also dont know of a company who make PWM controllers especially for Audio Amplifiers..............all i've seen is this....
http://www.powerint.com/sites/default/files/PDFFiles/rdr203.pdf
You are saying that all that i've said is folly?...i am not sure?
I've certainly heard many engineers telling that flybacks are not possible for supplying Class D amplifiers <500W......but i've never heard any reason why not, and the Labgruppen company actually have an 8KW flyback, that another poster mentioned.
By "dirt" are you referring to Conducted emissions?
Any diy audio hobbyist can use a flyback up to 500W without problem for supplying class D amplifiers...........mag has already doen so in a different thread of his.
I wouldnt have thought DIY audio people want to be spending amplifier time building over-compliciated SMPS's?....specially when they offer no real audio improvement over a flyback.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.