Then this is not a conventional amplifier, it's class G or class H. Where are the rail switching points and how much does the rails sag under load?
Anyway, if we assume 2100Wrms per channel with 85% efficiency (class H), we have 2100*2/.85=4940W average power drawn during each 33ms out of 100ms period, so the power supply is providing 4940*33/100=1650W total average power during that test, not 4200W. That makes a great difference in terms of dissipation.
Anyway, if we assume 2100Wrms per channel with 85% efficiency (class H), we have 2100*2/.85=4940W average power drawn during each 33ms out of 100ms period, so the power supply is providing 4940*33/100=1650W total average power during that test, not 4200W. That makes a great difference in terms of dissipation.
Can someone help me out understanding the feedback loop? I'm still a tad lost.
Is the idea to bias the pwm at some % duty cycle NOT at cutoff? For example, say my desired output voltage is 20. Does it get biased for 20V with the pwm at, say, 25% (just an example, assuming my pwm goes from 0% - 50% in a full bridge arangement). This gives me +/- 25% variation in duty cycle, and some latitude to adjust for component tolerance.
With the circuit that Eva provided, the feedback (error) signal is max when the input is less than threshold. Therefore, the pwm runs at max duty cycle. Once Vout crosses the threshold, the feedback circuit is drawing current, the opto transistor conducts. But, does it jump to a value that causes the pwm to be close to max duty cycle, or does it jump to some value - perhaps 25% duty cycle.
There's a webcast on techonline.com (presented by fairchild, actually) that deals with opto coupled feedback loops. I got some useful info from it, but still need some more.
gene
Is the idea to bias the pwm at some % duty cycle NOT at cutoff? For example, say my desired output voltage is 20. Does it get biased for 20V with the pwm at, say, 25% (just an example, assuming my pwm goes from 0% - 50% in a full bridge arangement). This gives me +/- 25% variation in duty cycle, and some latitude to adjust for component tolerance.
With the circuit that Eva provided, the feedback (error) signal is max when the input is less than threshold. Therefore, the pwm runs at max duty cycle. Once Vout crosses the threshold, the feedback circuit is drawing current, the opto transistor conducts. But, does it jump to a value that causes the pwm to be close to max duty cycle, or does it jump to some value - perhaps 25% duty cycle.
There's a webcast on techonline.com (presented by fairchild, actually) that deals with opto coupled feedback loops. I got some useful info from it, but still need some more.
gene
Excuse me for sounding stupid, but my understanding was that in symmetrical drive topologies (like push-pull, 1/2 bridge, full bridge) the net duty cycle ( the ratio between the positive and negative pulse widths) always had to be 50% so as not to saturate the core.
The variation in output voltage came from the amount of dead-time between the cycles, which both resets the core (from any assymetric magnetization), and lowers the outout voltage below some theoretical maximum.
Please correct me if this is wrong.
The variation in output voltage came from the amount of dead-time between the cycles, which both resets the core (from any assymetric magnetization), and lowers the outout voltage below some theoretical maximum.
Please correct me if this is wrong.
I figure you can't sound any stupider than me 
Although I am just a beginner on this subject, here's my take on the symetrical topologies. Basically, you drive current into the transformer primary from pin1 to pin2 (made up names just for discussion purposes) for up to 50% duty cycle. Then, you turn things around and drive current from pin2 to pin1 for up to 50% duty cycle. There's a short dead time between the two, so in reality you drive for slightly less than 50%. To put it in terms of time, say the switching period is 10 uS, then first drive current for 5uS from pin1 to pin2, then for the remaining 5uS drive current from pin2 to pin1.
It's kind of cool because the tranformer is actually driven for nearly the entire cycle time (10 uS, in this example).
I've been playing with this in simulation. Seems ok. Any of you experts can certainly correct me if I am wrong.
gene
Although I am just a beginner on this subject, here's my take on the symetrical topologies. Basically, you drive current into the transformer primary from pin1 to pin2 (made up names just for discussion purposes) for up to 50% duty cycle. Then, you turn things around and drive current from pin2 to pin1 for up to 50% duty cycle. There's a short dead time between the two, so in reality you drive for slightly less than 50%. To put it in terms of time, say the switching period is 10 uS, then first drive current for 5uS from pin1 to pin2, then for the remaining 5uS drive current from pin2 to pin1.
It's kind of cool because the tranformer is actually driven for nearly the entire cycle time (10 uS, in this example).
I've been playing with this in simulation. Seems ok. Any of you experts can certainly correct me if I am wrong.
gene
During transient conditios duty cycle may rise to 100% (50% if you want to consider it this way) or may drop to 0%. You can limit maximun duty cycle if desired, but this will reduce the slope at with inductor current rises when a sudden load increase is applied. Also, duty cycle will be very small at light loads due to discontinuous mode, inductor current will drop to 0 during dead time before the next cycle starts.
Actually, quick load changes in push-pull converters may give rise to temporal asymmetry and transient transformer saturation.
Actually, quick load changes in push-pull converters may give rise to temporal asymmetry and transient transformer saturation.
Eva said:
Are you talking about the bias point of the system? You just made me think about the discontinuous / continuous aspect of these things. I guess if the output rises enough, the pwm shuts off, so to speak, and stops generating any pulses. Same thing, sort of, for the output going too low where the pwm max's out trying to keep the output voltage stable.
Getting back to the bias point and your circuit - If the design does not want to bias the circuit at say 25% for Vout-nominal, then does the system have an rough response when the opto begins to conduct? That is, for voltage less than threshold, the opto is off and pwm is at max. When the opto turns on, there may be an instantaneous drop in the error signal, and a big change in the output voltage.
Maybe what's bothering me, and what I am doing wrong in simulation, is how the things starts up. Is it proper to leave the output without a load until it reaches it's nominal level? I've been simulating with a load attached from T=0, and it runs all wrong.
Although I think nobody is listening to this thread any longer, here's a question regarding this power supply design and how it pertains to UL safety (IEC 60950, for example). If nobody replies, it'll repost as a new thread . . .
So far, this thread discusses a power supply capable of 500 watts - it's for an amplifier that runs 6 amplifier channels at 100 watts of audio each. Current plans are for the supply to run +/-40V or alternatively 80V to ground, in order to get 100Wrms output power per channel into 8-ohms.
So how does 60950 apply to this situation? They claim SELV equipment has output of less than 42.5V or 60V peak. Certainly +80 violates that and becomes a TNV cicrcuit. Does +/40V violate the letter of that law, or is it still considered 80V because of the +/- peak-to-peak thingy.
OK, supposing it does violate SELV limits (42.5V). Now the equipment runs in one flavor of TNV or another. My understanding of this spec is really limited, so correct me if I make mistakes. But, as I read it, the TNV circuit must be electrically isolated from earth ground - makes sense, I believe, letting the output float relative to earth so the user cannot get shocked if in contact with the outputs! Applying this notion to the supply, the output amplifier circuit would then be running on an earth isolated supply. Therefore, the rest of the circuitry (i.e. the entire secondary side of the power supply) would need to run on an isolated supply, and the circuit ground could never be chassis/ground ground.
Does this sound correct? Since the commercial companies are no doubt facing this same issue, how do they handle the safety requirements?
So far, this thread discusses a power supply capable of 500 watts - it's for an amplifier that runs 6 amplifier channels at 100 watts of audio each. Current plans are for the supply to run +/-40V or alternatively 80V to ground, in order to get 100Wrms output power per channel into 8-ohms.
So how does 60950 apply to this situation? They claim SELV equipment has output of less than 42.5V or 60V peak. Certainly +80 violates that and becomes a TNV cicrcuit. Does +/40V violate the letter of that law, or is it still considered 80V because of the +/- peak-to-peak thingy.
OK, supposing it does violate SELV limits (42.5V). Now the equipment runs in one flavor of TNV or another. My understanding of this spec is really limited, so correct me if I make mistakes. But, as I read it, the TNV circuit must be electrically isolated from earth ground - makes sense, I believe, letting the output float relative to earth so the user cannot get shocked if in contact with the outputs! Applying this notion to the supply, the output amplifier circuit would then be running on an earth isolated supply. Therefore, the rest of the circuitry (i.e. the entire secondary side of the power supply) would need to run on an isolated supply, and the circuit ground could never be chassis/ground ground.
Does this sound correct? Since the commercial companies are no doubt facing this same issue, how do they handle the safety requirements?
I don't know how big companies manage to get through this nonsense safety standard mess, but I can assure you that all the amplifiers with over +-60V rails that I've had in my bench had either their internal ground connected directly to earth or indirectly through a "ground lift" switch. Most of them also included a high voltage warning message printed near the speaker terminals, though.
Maybe these limitations do not apply for professional equipment?
Maybe these limitations do not apply for professional equipment?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.