Has any one any links/ schematics for a full bridge self oscillating amplifier for single rail use?
I have done a couple of comercial class d designs using fixed frequency crystal locked triangle generators, mostly based on phase shift amplification, mainly to get predictable emc designs.
see:http://www.current-thinking.com/perma4.htm
I run a version of this circuit at higher voltage rails for my home HI FI and results are good at 50W 8R I get 0.05% thd (0.01% AES17) using a linear PSU at 42V /40,000uF caps (measurements done with AP cascade 2 DSP analyser)
I tried earlier self oscillating designs, but the only stable always kick off designs were split rail systems, which i don't like due to the PSU pumping, especially at higher powers.
Ideally I would like to develop this into a self oscillating phase shift amplifier, where a common mode choke in the output will remove virtually all switching components regardless of frequency, a modest low order balanced filter could then be used after the cm choke to futher demodulate the output, lowering the amplifiers reaction to varying loads:~)
The simplicity of self oscillating appeals if the emc can be contained (don't want the amp to blot out classic FM in the front room!!)
I have done a couple of comercial class d designs using fixed frequency crystal locked triangle generators, mostly based on phase shift amplification, mainly to get predictable emc designs.
see:http://www.current-thinking.com/perma4.htm
I run a version of this circuit at higher voltage rails for my home HI FI and results are good at 50W 8R I get 0.05% thd (0.01% AES17) using a linear PSU at 42V /40,000uF caps (measurements done with AP cascade 2 DSP analyser)
I tried earlier self oscillating designs, but the only stable always kick off designs were split rail systems, which i don't like due to the PSU pumping, especially at higher powers.
Ideally I would like to develop this into a self oscillating phase shift amplifier, where a common mode choke in the output will remove virtually all switching components regardless of frequency, a modest low order balanced filter could then be used after the cm choke to futher demodulate the output, lowering the amplifiers reaction to varying loads:~)
The simplicity of self oscillating appeals if the emc can be contained (don't want the amp to blot out classic FM in the front room!!)
The single rail is just a cost saving bi-product of using the full bridge, as long as the closed loop inductance of the legs on the full bridge is low(and legs are well{i mean well} decoupled), then the pumping effect is minimal, no point in paying for a dual rail supply if you put the money into the silicon for a full bridge 
however if you lay it out wrong then you may as well not bother IMHO.
I am interested in persuing the self oscillating phase shift amplifier, mainly for the simplicity and to see if it gives the sonic benifits it proports, from other threads peoiple here are pretty set on self oscillating, I am not so sure, I like the advantages of natural sampled phase shift pwm, especillay in op filter design, but then I think rain is wet, so who am I to Judge??
regards
however if you lay it out wrong then you may as well not bother IMHO.
I am interested in persuing the self oscillating phase shift amplifier, mainly for the simplicity and to see if it gives the sonic benifits it proports, from other threads peoiple here are pretty set on self oscillating, I am not so sure, I like the advantages of natural sampled phase shift pwm, especillay in op filter design, but then I think rain is wet, so who am I to Judge??
regards
Thanks, good answer.
Checked out your website real quick, are you still currently thinking of a name? I'm sure that's an old one, I need more sleep. Your monitoring circuit using a 22khz test tone is...... intriguing. Not something I've heard of in an audio amp, or anything else. Can you tell me a little about it? Why 22kHz as well, seems like an odd choice.
Your post seemed confusing because pumping would occure as usual with split rails or single rail if using a half bridge, I see now it's the cancellation of full bridge you're after though.
You have to think Hypex has had much sucess with their 400 watt module, and now have released a 700-900W module, still half bridge, and aren't concerned at all with pumping. They do feature overvoltage protection though. What kind of power level are you interested in?
It's good you're open minded enough to want to test the waters. Sorry I'm unable to direct you to a schematic, have you considered contacting Hypex directly?
Regards,
Chris
Checked out your website real quick, are you still currently thinking of a name? I'm sure that's an old one, I need more sleep. Your monitoring circuit using a 22khz test tone is...... intriguing. Not something I've heard of in an audio amp, or anything else. Can you tell me a little about it? Why 22kHz as well, seems like an odd choice.
Your post seemed confusing because pumping would occure as usual with split rails or single rail if using a half bridge, I see now it's the cancellation of full bridge you're after though.
You have to think Hypex has had much sucess with their 400 watt module, and now have released a 700-900W module, still half bridge, and aren't concerned at all with pumping. They do feature overvoltage protection though. What kind of power level are you interested in?
It's good you're open minded enough to want to test the waters. Sorry I'm unable to direct you to a schematic, have you considered contacting Hypex directly?
Regards,
Chris
Hi,
Is the pumping effect really a nightmare? I did not see that problem as really bad.
For DIYer, I thought that it safe if we have spare working voltage. I think the issue just important for commercial builder that prevent shooted by their competitor.
But no heared is actually how many percent increase caused by pumping effect?
IRAUDAMP1 set overvoltage till 131.6V from -50+50V=100V, or 31.6%.
We know that amp power output depent on gain not depent on how high of supply rails. Most of input circuit already have current constant. Drivers already regulated. Some circuit use voltage follower. So what are we worry about pumping effect?
I think just the CAPs.
Regards,
Kartino
Is the pumping effect really a nightmare? I did not see that problem as really bad.
For DIYer, I thought that it safe if we have spare working voltage. I think the issue just important for commercial builder that prevent shooted by their competitor.
But no heared is actually how many percent increase caused by pumping effect?
IRAUDAMP1 set overvoltage till 131.6V from -50+50V=100V, or 31.6%.
We know that amp power output depent on gain not depent on how high of supply rails. Most of input circuit already have current constant. Drivers already regulated. Some circuit use voltage follower. So what are we worry about pumping effect?
I think just the CAPs.
Regards,
Kartino
Thanks for that Chris,
In reply, the amplifier is primarily aimed at the voice alarm/life safety market, and the monitoring of the load is crucial, 22K as the pilot frequency is just a handy value, (actually its 22.05KHz i.e. 44.1KHz sampling clock/2) and is high enough so that 99.9% of the people can't hear it.
As for PSU pumping, I personally hate it, given the almost nil PSVRR of the output stage in any class d amplifier, pumping of the rails leaves nasty products in the sound, so you have to damp with unrealistically large capacitors and hope they don't overheat :~(
still the quest continues.
regards
In reply, the amplifier is primarily aimed at the voice alarm/life safety market, and the monitoring of the load is crucial, 22K as the pilot frequency is just a handy value, (actually its 22.05KHz i.e. 44.1KHz sampling clock/2) and is high enough so that 99.9% of the people can't hear it.
As for PSU pumping, I personally hate it, given the almost nil PSVRR of the output stage in any class d amplifier, pumping of the rails leaves nasty products in the sound, so you have to damp with unrealistically large capacitors and hope they don't overheat :~(
still the quest continues.
regards
Anthony C Smith said:
Hi,
Bruno has said in the past it's not really that big an issue provided you've enough capacitance in your supply which is required to make a good amp anyway, that's almost verbatim. As Kartino states a standard "good practice" margin of safety on the caps is wise also.
At very high sustained power as Jan-Peter pointed out since music is not a perfectly symmetrical sine wave it can act as having some level of DC bias which can pump the rail.
I think it really only becomes a serious issue if you're say a commercial builder that wants to get away with a cheap SMPS that doesn't have enough capacitance to absorbe the pumping.
In that case you can recycle the power in other ways.
It wouldn't seem to be enough of a real issue to allow it to dictate the topology. I'd go with full bridge for more power and better carrier suppression though.
In my case I ensure it's a non issue by having two amps sharing the same supply in bi-phase, kind of like a full bridge too I guess, so each amps pumping cancels that of the other. I also have enough capacitance and a good margin of safety built in. I actually welcome what little pumping there is.
You make an interesting point I think with the output stage having zero PSRR, but the amp itself does, and through self oscillating corrects itself long before any supply artifacts can become audible. I dont' think the output itself would get pumped or modulated by the supply.
Were you wanting to use a SMPS?
What sort of deviation do you test for in your tone, and how does that differentiate one fault from another, or can it, perhaps it's all inclusive?? That's what really intrigued me the most, even if it can't be of use in audiophile land, it sounds like a neat idea. Mainly I guess I'd lile to know how a test tone can act as a protection mechanism.
Thanks,
Chris
Hi Chris-
the test tone is not for protection, just for measuring the load; thus
I measure the voltage of the 22K tone with a high q notch, then measure the current in a series resistor using the same shape notch; take the log of both and add; this gives me the power the speakers absorb at 22KHz, I pass this into a window comparator and hey presto is someone removes 5% of the load or increases the load then a fault is logged.
with pumping; I tend to approach these things with my commercial builder hat on, 'cause first & foremost thats my job;
the problem is not what you measure with a multimeter, the sustained pulses are in most cases only a few 10's of nans wide, and yet they pulse the rail by 10-15% this means the peaks can be in the order of 150% rail. the capacitors smooth this out at an expense- they act like a shock absorber in a car- sure you can ride the rocky road for a month before the shock fails, where as the same shock will last a year on the freeway.
what I think I am saying is I leave my home amps on 24-7 and there's nothing worse than coming into the front room to the smell of expired electronics- it puts a crimp on the whole day.
As for switchmode PSU's I hate them with vengence. IMHO switch mode supplies are good when the load is constant and efficiency is paramount- from my tests in the life safety market forget green energy efficiency stamps for the psu- the quiecent draw of a smps is usually far worse than a well designed torroidal psu and easier to get emc tick boxes to as you don't have to worry about IM products from the two switching frequencies (3 with 2 async amp modules)
In HIFI the weight is not a problem, and the dangers of DIY smps are a real problem, so for me its torroids all the way! (commercially torroids can be cost effective, if the right core material is chosen, you can run the torroid at a nominal 80% of load knowing as long as the peaks are short the torroid core will provide coverage (as long as the ambient temp is not excessive) and you do not continuously sine wave test of course!)
Back to the single ended/ bridge configurations- for the cost/ benifits I still like the bridge configuration for any power above 25w for the engineering of it.
I tend to design around the HIP4081 as I know the beast; but from reading posts here it may be possible to impliment a complete amp around its brother the HIP4080 which has an comparator in it- that would be neat!
Ho hum still in search-
regards
the test tone is not for protection, just for measuring the load; thus
I measure the voltage of the 22K tone with a high q notch, then measure the current in a series resistor using the same shape notch; take the log of both and add; this gives me the power the speakers absorb at 22KHz, I pass this into a window comparator and hey presto is someone removes 5% of the load or increases the load then a fault is logged.
with pumping; I tend to approach these things with my commercial builder hat on, 'cause first & foremost thats my job;
the problem is not what you measure with a multimeter, the sustained pulses are in most cases only a few 10's of nans wide, and yet they pulse the rail by 10-15% this means the peaks can be in the order of 150% rail. the capacitors smooth this out at an expense- they act like a shock absorber in a car- sure you can ride the rocky road for a month before the shock fails, where as the same shock will last a year on the freeway.
what I think I am saying is I leave my home amps on 24-7 and there's nothing worse than coming into the front room to the smell of expired electronics- it puts a crimp on the whole day.
As for switchmode PSU's I hate them with vengence. IMHO switch mode supplies are good when the load is constant and efficiency is paramount- from my tests in the life safety market forget green energy efficiency stamps for the psu- the quiecent draw of a smps is usually far worse than a well designed torroidal psu and easier to get emc tick boxes to as you don't have to worry about IM products from the two switching frequencies (3 with 2 async amp modules)
In HIFI the weight is not a problem, and the dangers of DIY smps are a real problem, so for me its torroids all the way! (commercially torroids can be cost effective, if the right core material is chosen, you can run the torroid at a nominal 80% of load knowing as long as the peaks are short the torroid core will provide coverage (as long as the ambient temp is not excessive) and you do not continuously sine wave test of course!)
Back to the single ended/ bridge configurations- for the cost/ benifits I still like the bridge configuration for any power above 25w for the engineering of it.
I tend to design around the HIP4081 as I know the beast; but from reading posts here it may be possible to impliment a complete amp around its brother the HIP4080 which has an comparator in it- that would be neat!
Ho hum still in search-
regards
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.