Here is my simplified D amp, sourced from the IR AUDAMP.
There is no protections on it. Its the prototype.
No components values on this schematic yet...
It will be supplied with a 37v rail, on a 4 ohm load. IR papers suggest as high as 95% eff with this kind of amp, so,
(0.707x37)²/4=171w...-5%=162w rms...that will be good enough as a first real D amp project.
Board will be able to accept 60v rails to go close to the 500w claimed by IR.
Any comments?
thanks group.
There is no protections on it. Its the prototype.
No components values on this schematic yet...
It will be supplied with a 37v rail, on a 4 ohm load. IR papers suggest as high as 95% eff with this kind of amp, so,
(0.707x37)²/4=171w...-5%=162w rms...that will be good enough as a first real D amp project.
Board will be able to accept 60v rails to go close to the 500w claimed by IR.
Any comments?
thanks group.
pat allen said:Here is my simplified D amp, sourced from the IR AUDAMP.
There is no protections on it. Its the prototype.
No components values on this schematic yet...
It will be supplied with a 37v rail, on a 4 ohm load. IR papers suggest as high as 95% eff with this kind of amp, so,
(0.707x37)²/4=171w...-5%=162w rms...that will be good enough as a first real D amp project.
Board will be able to accept 60v rails to go close to the 500w claimed by IR.
Any comments?
thanks group.
Hello,
Very nice
It's easier to follow if you use VCC and VEE for the op amp stuff, VSS for positive high voltage (output stage) and VDD for negative output stage.
Might consider removing those resistors in the bootstrap circuit. Especially the one between the diode and bootstrap capacitor.
Check out the datasheet for the IR2011, they don't show any resistors. You want it to be able to refresh the cap really fast and hold a charge as long as possible.
Re: Re: schematic of my future D amp for the pb12
in my use of electrical terms for my job, i am used to see VDD as the + source while VSS is the return or ground...
Microchip, Ti, Atmel use it that way...(microcontrolers that i work with all day)
They were labeled with the voltage before, i tougt it was easier like that, it is a personal choice i think.
The resistor you talked about is a very small value, 10 ohm.
classd4sure said:
Hello,
Very nice
It's easier to follow if you use VCC and VEE for the op amp stuff, VSS for positive high voltage (output stage) and VDD for negative output stage.
Might consider removing those resistors in the bootstrap circuit. Especially the one between the diode and bootstrap capacitor.
Check out the datasheet for the IR2011, they don't show any resistors. You want it to be able to refresh the cap really fast and hold a charge as long as possible.
in my use of electrical terms for my job, i am used to see VDD as the + source while VSS is the return or ground...
Microchip, Ti, Atmel use it that way...(microcontrolers that i work with all day)
They were labeled with the voltage before, i tougt it was easier like that, it is a personal choice i think.
The resistor you talked about is a very small value, 10 ohm.
Re: Re: Re: schematic of my future D amp for the pb12
You're right I meant to say VDD+ VSS-.
I just found it confusing due to there being multiple sources.. meant no insult of course.
Please read page 9:
http://www.powerdesigners.com/InfoWeb/design_center/Appnotes_Archive/an-978.pdf
A bootstrap cap even so big as 100uF would be better too.
pat allen said:
You're right I meant to say VDD+ VSS-.
I just found it confusing due to there being multiple sources.. meant no insult of course.
Please read page 9:
http://www.powerdesigners.com/InfoWeb/design_center/Appnotes_Archive/an-978.pdf
A bootstrap cap even so big as 100uF would be better too.
Hi, please look at the .pdf, if not done. There is the original schematic of the IRAUDAMP. I have not invent anything, it is a copy-paste of theirs desing.
Please, look at it, maybe they are wrong??
They sell this amp for around 600$ (2 channel+protections on board.)
thanks.
IRAUDAMP ORIGINAL PAPERS
Please, look at it, maybe they are wrong??
They sell this amp for around 600$ (2 channel+protections on board.)
thanks.
IRAUDAMP ORIGINAL PAPERS
Hi Pat,
I know.. I know, it's there demo board, I've seen it.
The app note I linked to are discussing these very drivers though.
Guess who the app note is by?
Here's what's likely an updated version of that same app note:
http://www.irf.com/technical-info/appnotes/an-978.pdf
They took that line out, yet show the same IC and everything, changed their minds? I think so, all the same..... let's think about.
Some of their diagrams (most of them) do NOT show this resistor, the odd one does.
This app note discusses it more:
http://www.irf.com/technical-info/designtp/dt04-4.pdf
600$..... Doesn't make it right.
I'd use a cheap electrolytic 100uF bypassed by a small low ESR... ceramic?
It's only ever going to be charging from 12Volts there's no great need to limite the current and the ESR should do something for it.
If you really feel the need for a resistor though, I'd try a few ohms but placed before, the diode, not after it. (R49?) removed.
But you know.. you can experiment.
They selection of cap size is pathetically small though, minimum calculated ? It should be 10X that at least.
I know.. I know, it's there demo board, I've seen it.
The app note I linked to are discussing these very drivers though.
Guess who the app note is by?
Here's what's likely an updated version of that same app note:
http://www.irf.com/technical-info/appnotes/an-978.pdf
They took that line out, yet show the same IC and everything, changed their minds? I think so, all the same..... let's think about.
Some of their diagrams (most of them) do NOT show this resistor, the odd one does.
This app note discusses it more:
http://www.irf.com/technical-info/designtp/dt04-4.pdf
600$..... Doesn't make it right.
I'd use a cheap electrolytic 100uF bypassed by a small low ESR... ceramic?
It's only ever going to be charging from 12Volts there's no great need to limite the current and the ESR should do something for it.
If you really feel the need for a resistor though, I'd try a few ohms but placed before, the diode, not after it. (R49?) removed.
But you know.. you can experiment.
They selection of cap size is pathetically small though, minimum calculated ? It should be 10X that at least.
Hi, thanks for the links, they are very helpfull. I didnt realisez how far they gives equations to optimise every aspects of this topology.
I have done some average calculation with the formulas for the bootstrap cap, it always came very small, below 1uf, considering the esr and frequency of the overall switching gizmo...anyway, i build it as a demo test prototype to eventualy make it burn on its limit...who knows, maybe it works good as is ??
I realy plan to do several test on this board, as well as many components swaping, i will try to optimise it myslef with the help of the app notes. Scope snap shots are on call to show them to all of you.
As for the caps on the board, yes they are small, but they claims that they are not to store or to filter the ac from a transfo, and one have to build its own regulated power supply with enough caps to power up theses boards..
I do have 6x 34 800uf caps and a 60v ct 12 amp toroidal transfo for the task...it should be good enough for the tests...
thanks.
I have done some average calculation with the formulas for the bootstrap cap, it always came very small, below 1uf, considering the esr and frequency of the overall switching gizmo...anyway, i build it as a demo test prototype to eventualy make it burn on its limit...who knows, maybe it works good as is ??
I realy plan to do several test on this board, as well as many components swaping, i will try to optimise it myslef with the help of the app notes. Scope snap shots are on call to show them to all of you.
As for the caps on the board, yes they are small, but they claims that they are not to store or to filter the ac from a transfo, and one have to build its own regulated power supply with enough caps to power up theses boards..
I do have 6x 34 800uf caps and a 60v ct 12 amp toroidal transfo for the task...it should be good enough for the tests...
thanks.
Hi,
You know how those OEM's like to pinch every penny right?
Those calculations give you an estimate of the bare minimum requiremed value, good design practice say's take that value X10, other app notes say the bigger the better (less odds of overcharging, less ripple((less ripple.... better drive to the mosfet, don't want your Rdson varying with duty cycle right)))...
I'm using 100uF electrolytic in my little amp, paralleled with a 100nF ceramic across the 12V zener. Works fine. Up to you.
You could also make the driver /output stage on a seperate PCB, that way you can try other kinds of amps out with it too.
You know how those OEM's like to pinch every penny right?
Those calculations give you an estimate of the bare minimum requiremed value, good design practice say's take that value X10, other app notes say the bigger the better (less odds of overcharging, less ripple((less ripple.... better drive to the mosfet, don't want your Rdson varying with duty cycle right)))...
I'm using 100uF electrolytic in my little amp, paralleled with a 100nF ceramic across the 12V zener. Works fine. Up to you.
You could also make the driver /output stage on a seperate PCB, that way you can try other kinds of amps out with it too.
pat allen said:
IR have two different version of IRAUDAMP1 on its site.
one have a capacitor paralled at the input of the integrator:
http://www.irf.com/product-info/audio/classdtutorial.pdf
one not:
http://www.irf.com/technical-info/refdesigns/iraudamp1.pdf
the appearence is also slightly different:
http://www.irf.com/graphics/nr/nr040826.jpg
v.s.
http://www.irf.com/technical-info/refdesigns/images/IRAUDAMP1.jpg
Another thing is, there seems to be no hystresis,so it may be using 180 deg phase shift to oscillate like UCD's. (though the feedback is taken from switching stage, not filtered output)
Hi group, i have in my hands all the components to build one channel of this amp.
One question remains in my head. i want to use this amp, if it works, on some subwoofers for my home theatre. So, overall sound quality is not that much important, as long as is plays imho.
I would like to be able to drive 2 channels bridged and under 2 ohm load. Yes, thats call for a lot of power. If not ok, then 8 ohm because the woofers i got are 4 ohm and i want to achieve the more bang for the buck, you know...
Lets asume that i have the proper power supply, and that the woofers are rated at 375wrms each.
The question is: can i parrallel output devices so the amp could drive lower ohm load? They rate it "ok" at 4 ohm load, and going lower could affect the circuit, but they are not precise.
Could the circuit be influenced by adding more capacitance, the free runing frequency seems to be related to all the components, so i am wondering.
Lets start to build one first, as one could suggest me
Thanks.
One question remains in my head. i want to use this amp, if it works, on some subwoofers for my home theatre. So, overall sound quality is not that much important, as long as is plays imho.
I would like to be able to drive 2 channels bridged and under 2 ohm load. Yes, thats call for a lot of power. If not ok, then 8 ohm because the woofers i got are 4 ohm and i want to achieve the more bang for the buck, you know...
Lets asume that i have the proper power supply, and that the woofers are rated at 375wrms each.
The question is: can i parrallel output devices so the amp could drive lower ohm load? They rate it "ok" at 4 ohm load, and going lower could affect the circuit, but they are not precise.
Could the circuit be influenced by adding more capacitance, the free runing frequency seems to be related to all the components, so i am wondering.
Lets start to build one first, as one could suggest me
Thanks.
I'd recommend to build it first, using a reliable PCB layout. Then you can start fiddling around.
Since you are using it for a subwoofer only, you might reduce the switching frequency and use "beefier" output devices for higher current load. But do such experiments only after you have a reliably working one, according to the original design.
One remark:
The reason for some members to recommend the use of a larger bootstrap cap is achieving a proper clipping behaviour. While clipping is to be avoided of course, one still wants a proper behaviour during clipping. If the cap is too small the circuit may show erratic behaviour during clipping of the positive half-wave or even latch-up depending upon topology. Since you use it as a sub amp it is likely that it is overdriven by a LOW frequency signal (if it happens at all) an thus asking for a large cap.
Regards
Charles
Since you are using it for a subwoofer only, you might reduce the switching frequency and use "beefier" output devices for higher current load. But do such experiments only after you have a reliably working one, according to the original design.
One remark:
The reason for some members to recommend the use of a larger bootstrap cap is achieving a proper clipping behaviour. While clipping is to be avoided of course, one still wants a proper behaviour during clipping. If the cap is too small the circuit may show erratic behaviour during clipping of the positive half-wave or even latch-up depending upon topology. Since you use it as a sub amp it is likely that it is overdriven by a LOW frequency signal (if it happens at all) an thus asking for a large cap.
Regards
Charles
I agree with Charles.
Let's see what happens during clipping: the output of the comparator sticks to high or low during a long time, so there is no switching in the low side mosfet, so no bootstrap charge in the bootstrap capacitor, that starts discharging. The bigger the capacitor, the longer the time it can provide supply to the HS gate driver with no switching.
I use 47uF tantalum in parallel with 100nF ceramic SMD and it clips quite well down to 20Hz (and less).
BTW: I have also studied this schematic. One thing that worries me is that it doesn't have any dead-time control circuitry apart from the resistor and parallel diode in series with the gates of the mosfets. Keep in mind also that the mosfets they use have a very low gate charge, so the switching speed should be very fast. The dead time must be veeeery small. Is that reliable enough?
They also don't use zeners in parallel with the gates, so any spike higher than, say, 20v, can destroy the gate oxide. They don't use any snubber also. Is that acceptable from your experience, Charles?
Best regards
Let's see what happens during clipping: the output of the comparator sticks to high or low during a long time, so there is no switching in the low side mosfet, so no bootstrap charge in the bootstrap capacitor, that starts discharging. The bigger the capacitor, the longer the time it can provide supply to the HS gate driver with no switching.
I use 47uF tantalum in parallel with 100nF ceramic SMD and it clips quite well down to 20Hz (and less).
BTW: I have also studied this schematic. One thing that worries me is that it doesn't have any dead-time control circuitry apart from the resistor and parallel diode in series with the gates of the mosfets. Keep in mind also that the mosfets they use have a very low gate charge, so the switching speed should be very fast. The dead time must be veeeery small. Is that reliable enough?
They also don't use zeners in parallel with the gates, so any spike higher than, say, 20v, can destroy the gate oxide. They don't use any snubber also. Is that acceptable from your experience, Charles?
Best regards
IME it is much easier to get a clean signal by the use of snubbers than without, though they "kill" some efficiency.
Using a proper layout overshoots etc can be greatly minimised. If you look at Tripath's application notes you will see that they don't generally call for snubbers.
Bruno once discussed the MOSFET's susceptibility to overshoots within this forum (have to search for the thread) and he doesn't use any snubbers on UcD AFAIK.
I think (or do at least hope so) that IR have tested their design for reliability.
It is always a differnt thing to design a circuit for oneself or one that has to work reliably in thousands including component tolerances etc.
The dead-time question is indeed a tricky one. If there is too much, the body-diode gets into conduction (so we have crossconduction between body-diode and opposite fet due to reverse-recovery), if too less we have cross-conduction between the two mosfets. I think a variable dead-time that lies within a control-loop might be advisible. Or a topology where the body diode doesn't get into conduction at all like the following one uses (fig 5 & 6):
http://v3.espacenet.com/origdoc?DB=...600891&DOC=dcb65d04ab6623da915010c89048efbf66
There are also topologies that don't deal with dead-time at all and simply accept shoot-through as a fact:
http://www.onsemi.com/pub/Collateral/AN1042-D.PDF
Regards
Charles
Using a proper layout overshoots etc can be greatly minimised. If you look at Tripath's application notes you will see that they don't generally call for snubbers.
Bruno once discussed the MOSFET's susceptibility to overshoots within this forum (have to search for the thread) and he doesn't use any snubbers on UcD AFAIK.
I think (or do at least hope so) that IR have tested their design for reliability.
It is always a differnt thing to design a circuit for oneself or one that has to work reliably in thousands including component tolerances etc.
The dead-time question is indeed a tricky one. If there is too much, the body-diode gets into conduction (so we have crossconduction between body-diode and opposite fet due to reverse-recovery), if too less we have cross-conduction between the two mosfets. I think a variable dead-time that lies within a control-loop might be advisible. Or a topology where the body diode doesn't get into conduction at all like the following one uses (fig 5 & 6):
http://v3.espacenet.com/origdoc?DB=...600891&DOC=dcb65d04ab6623da915010c89048efbf66
There are also topologies that don't deal with dead-time at all and simply accept shoot-through as a fact:
http://www.onsemi.com/pub/Collateral/AN1042-D.PDF
Regards
Charles
Well, IRF has very good THD and noise figures with no dead-time control apart from the resistors || diodes in the gates.
Do you think that mosfets can be killed by allowing some shoot-though if it lasts in the order of 100ns? What are other possible causes of instantaneous mosfet failure? (I am not talking about overcurrent/overtemp)
Could you point to the mentioned Bruno's discussion, Charles?
Best regards,
Pierre
Do you think that mosfets can be killed by allowing some shoot-though if it lasts in the order of 100ns? What are other possible causes of instantaneous mosfet failure? (I am not talking about overcurrent/overtemp)
Could you point to the mentioned Bruno's discussion, Charles?
Best regards,
Pierre
The discussion was about overshhot, and not shhot through (refering to the snubber subject). I will try to find it.
I think that a little shoot-through might be acceptable as long as:
1.) is is not too detrimental to efficiency and
2.) the associated power loss isn't too high for the mosfets
The big problem is to set these things reliably.
Regards
Charles
I think that a little shoot-through might be acceptable as long as:
1.) is is not too detrimental to efficiency and
2.) the associated power loss isn't too high for the mosfets
The big problem is to set these things reliably.
Regards
Charles
Thanks for those links Charles, much nicer patent search site, original documents aaaaahhhh.. thank you.
From what I remember Bruno saying on snubbers, something along the lines of "they can be a big help ... if you have ringing"
So they aren't ruled out but aren't exactly a necessity.
If you do increase the size of the bootstrap capacitor here it might be a real good idea to remove those series resistors to decrease the charge time, thoughts? Also, my thoughts weren't only towards clipping behavior, though that's the crucial one, but since it will be a sub amp it will have longer on times and I thought there may be excessive ripple on the driver supply, or the possibility of.
As far as any kind of a reference design I think this is a poor example, can't believe they sell this for 600.
Anyway it's likely to be more of a reference with respect to mass production and not high fidelity DIY so every corner comes pre cut.
Doesn't hurt to make a few simple changes at this stage of the game if they can improve anything.
From what I remember Bruno saying on snubbers, something along the lines of "they can be a big help ... if you have ringing"
So they aren't ruled out but aren't exactly a necessity.
If you do increase the size of the bootstrap capacitor here it might be a real good idea to remove those series resistors to decrease the charge time, thoughts? Also, my thoughts weren't only towards clipping behavior, though that's the crucial one, but since it will be a sub amp it will have longer on times and I thought there may be excessive ripple on the driver supply, or the possibility of.
As far as any kind of a reference design I think this is a poor example, can't believe they sell this for 600.
Anyway it's likely to be more of a reference with respect to mass production and not high fidelity DIY so every corner comes pre cut.
Doesn't hurt to make a few simple changes at this stage of the game if they can improve anything.
Pierre said:
"Gate Driver
The IRAUDAMP1 uses the IR2011S gate driver IC which is suitable for high speed, high speed
switching applications up to 200V. In this design, the difference between ton and toff is used to
generate a dead-time (a blanking time in between the on state of the two MOSFETs). Because of
this, there is no gate timing adjustment on the board."
Charles I know the exact post you're referring to and I can't find it either.
I've read someplace the possible causes of mosfet failure are relatively few, but the number of ways in which they can happen are many.
and that is not in the document !!- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.