[+ cap] -- [+ inductor winding] -- MOSFET -- [+ xformer winding] -- GND -- [- xformer winding] -- MOSFET -- [- inductor winding] -- [- cap]
Since + and - inductor windings are wound on the same core, they have the same voltage. Likewise, both transformer windings have the same voltage. MOSFETs allow bidirectional current flow. End result is that both caps are forced to the same voltage.
The leakage inductance of the inductor and transformer windings limits this effect at higher frequencies, but you don't really need it at higher frequencies.
1. Why do you think switching freq is "not high"?
2. Simply replacing diode by MOSFET is not the proper way to make synchro-rectifier, if current flow is bidirectional.
3. I often shorting secondary winding while primary is still on nominal voltage. I wouldn't say same core quaranty same voltage (per turn). Leakage is more significant than you think.
2. Simply replacing diode by MOSFET is not the proper way to make synchro-rectifier, if current flow is bidirectional.
3. I often shorting secondary winding while primary is still on nominal voltage. I wouldn't say same core quaranty same voltage (per turn). Leakage is more significant than you think.
Depeand on wave. if it is simple fixed pwm (square wave 50%/50%), it have ratio depeand of turn's. (ferroxube can not change this) reason is simple..
Re #2: It's done in endless power supplies. I have a 50V/50A DC supply under my belt where I drove the sync rectifier gates directly with the main power transformer, and IIRC I stole the idea from an Intersil appnote.
Re #1/#3, if the leakage inductance of the power transformer secondaries is low enough that you can transfer power from primary to secondary at rated power (hence, the supply working in the first place) at the switching frequency, then theoretically you can move at least half the rated power between the two windings, this lower limit assuming that the leakage inductance is entirely on the secondary.
This doesn't include the leakage inductance of the coupled inductor, but that can be calculated. Generally the power capability of a switching power supply isn't limited by the leakage inductance of the magnetics.
Of course you could always build a current-fed supply without secondary side inductance if your design absolutely depended on this behavior...
AP2:
to answer thsi you need to run the simulation in post 1.....you will then see that the amplifier dissipates 350W, but that for several ms at the top of the 82Hz sine, the power draw is around 750W.
gmarsh
...too true.
But we have started to talk about split rail power supplies to supply half bridge class d amplifiers..............and the cross regulation for the power supplys split rails is so troublesome that it'd be better to just have a single output smps and use a full bridge class d amplifier.
The half bridge class d amplifier is a waste of time....just use a full bridge and have an easier, better , cheaper power supply.
to answer thsi you need to run the simulation in post 1.....you will then see that the amplifier dissipates 350W, but that for several ms at the top of the 82Hz sine, the power draw is around 750W.
gmarsh
...too true.
But we have started to talk about split rail power supplies to supply half bridge class d amplifiers..............and the cross regulation for the power supplys split rails is so troublesome that it'd be better to just have a single output smps and use a full bridge class d amplifier.
The half bridge class d amplifier is a waste of time....just use a full bridge and have an easier, better , cheaper power supply.
My thinking is also the same....when a producer say the amp has an efficiency of 90% then it will dissipate only 10% as heat....assuming a power supply of 100 watts regardless of voltage.. If amp is 90% efficient then it will give output of 90 watts on full load and dissipate only 10watts then why on earth will the amp require twice of its rated output....talking about peak output ,its always measured for 10ms or a little more...so why to worry...and no one is interested with peaks everybody asks ''WHATS THE RATED OUTPUT''??? Then why are you all buddy wrestling on peaks??? This one is my decent question to all!!!🙁🙁
sweetperfume.
Please run the simulation in post 1, then you will change your thinking, i promise you of this.
Please run the simulation and view the overall avergae power consumed in the 8R (speaker) load......you will see that its 350W.
....now look at the supply current when the class d amplifier is "playing" the 82hz sinewave near its peak......and you will see that the class d is drawing around 8 Amps for a few milliseconds.
Since the rail voltage is 80V......then 8A from 80V is 640W.....and that is obviously the power that the smps must be rated to supply.
Please , if you do not run the simulation, then you will never know.
I gaurantee you....the simulator is not lying.
I plead with you to run the simulation, and all others who disagree
Please run the simulation in post 1, then you will change your thinking, i promise you of this.
Please run the simulation and view the overall avergae power consumed in the 8R (speaker) load......you will see that its 350W.
....now look at the supply current when the class d amplifier is "playing" the 82hz sinewave near its peak......and you will see that the class d is drawing around 8 Amps for a few milliseconds.
Since the rail voltage is 80V......then 8A from 80V is 640W.....and that is obviously the power that the smps must be rated to supply.
Please , if you do not run the simulation, then you will never know.
I gaurantee you....the simulator is not lying.
I plead with you to run the simulation, and all others who disagree
Sekhar.
I tell you i tell the truth.
If you want the class d to sound out those peaks of those low frequency waveforms like the low E on a guitar (82Hz) then you're gonna need a power supply that can give you twice the average maximum power of the class d .
The simulation of the first post prooves beyond any doubt that i speak accurately and in perfect truth.
This is not so...i beg you to run the simulation of an 82Hz sine wave being "played" by a class d amplifier...it is in the first post.
I tell you i tell the truth.
If you want the class d to sound out those peaks of those low frequency waveforms like the low E on a guitar (82Hz) then you're gonna need a power supply that can give you twice the average maximum power of the class d .
The simulation of the first post prooves beyond any doubt that i speak accurately and in perfect truth.
eem2am
For repeated periods. Not constantly. Not thermally. Just like in case of a ClassAB.
gmarsh!
Really? Is it a conventional PWM supply otherwise? Did you try to reverse the current flow? Doubtfully!
(The secondary? There are 2 of them!)
Yes, it can move power. But with a hugh drop! And this is what must have been avoided!
Except (for example) in case of integrated series resonant supplies, eg. LLC converter, what is the most common, highly efficient PSU nowadays.
Did anybody asked this?
You told general statements, without specifying any details!
AP2!
It's impossible for me to decode your so-called english sentences. But I'm afraid after correcting them it would be still a puzzle, involving more questions then answers.
For repeated periods. Not constantly. Not thermally. Just like in case of a ClassAB.
gmarsh!
Really? Is it a conventional PWM supply otherwise? Did you try to reverse the current flow? Doubtfully!
(The secondary? There are 2 of them!)
Yes, it can move power. But with a hugh drop! And this is what must have been avoided!
Except (for example) in case of integrated series resonant supplies, eg. LLC converter, what is the most common, highly efficient PSU nowadays.
Did anybody asked this?
You told general statements, without specifying any details!
AP2!
It's impossible for me to decode your so-called english sentences. But I'm afraid after correcting them it would be still a puzzle, involving more questions then answers.
Last edited:
eem2am said:
Funny! More then 99.999999 % of the electrical engineers on Earth doesn't run that simulation. So won't they be able to design PSU for ClassD amps? 😀
Electrons know physics! Experiment is not lying! Simulators do! Many times! OK, probably not in this case, but understanding the consequences of the result is not trivial (apparently).
Last edited:
...and a harder, less efficient, more expensive class D amplifier.
We've already had this discussion; it makes the supply slightly more complicated (doubled output capacitance and rectifiers) but the added cost is pretty minimal. Making a full bridge class D amplifier adds far more parts - doubled transistors, doubled gate drive for those, doubled output filtering, doubled current sensing... and you have to do this for every channel in your amplifier.
It was a phase shifting, full bridge forward converter. Allowing current reversal in the output inductance means that discontinuous mode doesn't exist, allowing the control loop to be better optimized for the application. Saves a good bit of power rectifying the output too. Lots of things are built the same way, allowing negative inductor current (VRMs for powering CPUs, etc)
Secondary refers to the secondary (DC output) side, not to a specific winding. Yes, there are two of them.
"Huge drop" is all relative, and depends on your design and what it's intended to do.
gmarsh
...the full bridge class d isnt the great burden that you appear to label it.
...."it isnt really double the output filtering....there might be two class d inductors but really ...so what........and you can do current sensing with 1 current sense resistor.
A full bridge class d with an 80v rail is no sweat.......80V is low and you can get bootstrap brdieg fet driver chips these days to make the fet driving easy.
Full bridge and single output smps to supply it is the way to go........The half bridge class d harps back to the days when mosfets and mosfet drivers were very expensive and couldnt be switched at high frequency.
Today, full bridge is the one to go for.
Pafi
...if they dont heed what that simulation tells, then they wont properly be able to design an smps for a class d.
...the full bridge class d isnt the great burden that you appear to label it.
...."it isnt really double the output filtering....there might be two class d inductors but really ...so what........and you can do current sensing with 1 current sense resistor.
A full bridge class d with an 80v rail is no sweat.......80V is low and you can get bootstrap brdieg fet driver chips these days to make the fet driving easy.
Full bridge and single output smps to supply it is the way to go........The half bridge class d harps back to the days when mosfets and mosfet drivers were very expensive and couldnt be switched at high frequency.
Today, full bridge is the one to go for.
Pafi
...if they dont heed what that simulation tells, then they wont properly be able to design an smps for a class d.
maybe the problem is reversed. many engineers use the simulator, then see and know only the theoretical part of a smps, such as class D, a lot of formulas in these years without an amplifier that has really solved the definition at high audio frequencies. it measures seem that the amplifier is fantastic. This is because the measures are far away from the signal that will then amplify.
I also use the simulator, but only when I'm trying new circuitry. I prefer the actual measurements of a prototype with a multichannel analyzer to see a lot of parameters while it works.
is admirable that you want to know many things, but it is impossible in a short time.
for your guitar, if it is 60w, 150w smps is sufficient. but your problem is other at this point. right on the guitar, the dirt on the supply rails will be mixed with the residue class of d, get a sound very bad especially on the medium range. a professional guitar amp, it is not easy even with class AB.
I also use the simulator, but only when I'm trying new circuitry. I prefer the actual measurements of a prototype with a multichannel analyzer to see a lot of parameters while it works.
is admirable that you want to know many things, but it is impossible in a short time.
for your guitar, if it is 60w, 150w smps is sufficient. but your problem is other at this point. right on the guitar, the dirt on the supply rails will be mixed with the residue class of d, get a sound very bad especially on the medium range. a professional guitar amp, it is not easy even with class AB.
AP2
Of course you are right but this is not really relevant to the point here.
The particular case being examined here is that of a class d amplifier "playing" a 82Hz sine wave.........and that the average power in the 8R speaker as a result of this is 350W.........................and the point is that a 700W power supply will be needed if the class d amplifier is to be able to produce this 82Hz sine wave.
........In this case the simulator would show the same as a real circuit.
The situation being played out on the simulator here is a very general , simple situation and there is no way that greatly different behaviour would be seen in an actual circuit.
Questioning the simulator "for the point in question here" is irrelevant.
If you are going to speak to me about some simulation circuits not having the parasitics etc , that are present in real circuits, then that would be right......but it is not relevant to the general situation of a class d amp "playing" a 82Hz sine wave, and needing a power supply of double the average power.
I agree that any development always needs a physical prototype.....but getting a simulation together first is always a great idea....all of the best engineers that ive ever worked with are avid pro-simulation guys.
So you are saying that ripple and ringing on the class d amplifiers power supply rails will worsen the quality of the class d amplifier.................this is a very good point, and i am very grateful to you for making it, but it is not relevant to the subject of these posts here.
Of course you are right but this is not really relevant to the point here.
The particular case being examined here is that of a class d amplifier "playing" a 82Hz sine wave.........and that the average power in the 8R speaker as a result of this is 350W.........................and the point is that a 700W power supply will be needed if the class d amplifier is to be able to produce this 82Hz sine wave.
........In this case the simulator would show the same as a real circuit.
The situation being played out on the simulator here is a very general , simple situation and there is no way that greatly different behaviour would be seen in an actual circuit.
Questioning the simulator "for the point in question here" is irrelevant.
If you are going to speak to me about some simulation circuits not having the parasitics etc , that are present in real circuits, then that would be right......but it is not relevant to the general situation of a class d amp "playing" a 82Hz sine wave, and needing a power supply of double the average power.
I agree that any development always needs a physical prototype.....but getting a simulation together first is always a great idea....all of the best engineers that ive ever worked with are avid pro-simulation guys.
So you are saying that ripple and ringing on the class d amplifiers power supply rails will worsen the quality of the class d amplifier.................this is a very good point, and i am very grateful to you for making it, but it is not relevant to the subject of these posts here.
AP2
Thanks...are you saying that the point i am trying to make is correct, and that it is common knowledge?
I can attach a waveform for you, but it really would be very easy if you want to download the free LTspiceIV simulator from linear.com....and then you can see everything in front of you....just open the .asc file and hit the running man icon......then click whatever node you want to monitor...............it will all just open out before you.
Thanks...are you saying that the point i am trying to make is correct, and that it is common knowledge?
I can attach a waveform for you, but it really would be very easy if you want to download the free LTspiceIV simulator from linear.com....and then you can see everything in front of you....just open the .asc file and hit the running man icon......then click whatever node you want to monitor...............it will all just open out before you.
The .asc file is called CLASS D _80V 82 Hz.asc
It is attached at the bottom of the post.....i think i also attached it as a .txt file, and if you use this you have to change it to .asc.
There are two .txt files that i attached in the first post..."that.txt" and "that1.txt".
i say "attached"..but they are uploaded to the 2shared.com website and you can get them from the link to that site (in the post)
These two txt files must be in the same folder as the .asc file, because they control the switching of the fets.
It is attached at the bottom of the post.....i think i also attached it as a .txt file, and if you use this you have to change it to .asc.
There are two .txt files that i attached in the first post..."that.txt" and "that1.txt".
i say "attached"..but they are uploaded to the 2shared.com website and you can get them from the link to that site (in the post)
These two txt files must be in the same folder as the .asc file, because they control the switching of the fets.
Last edited:
simple...not?
please put only this file with extention changed
i want ceck well your simulation
- Status
- Not open for further replies.