EMC considerations apply to any design not just commercial, especially when doing a SMPS, to ignore EMC during the design cycle is stupidity. EMC and signal integrity are two sides of the same coin.....
I presume you will know about creepage and clearance distances required for your design, so it is safe.....
Funny I have just done a 50Hz design where the transformer was wound in litzt wire... No myth, you could also use copper ribbon transformers....
What's the typical duty cycle of a transformer-isolated converter?
Some ICs have a fixed 50% duty cycle, but some can go to 100%.
What really happens at 100% duty cycle?
Some ICs have a fixed 50% duty cycle, but some can go to 100%.
What really happens at 100% duty cycle?
What do you mean? Fly-back? 2 transistor forward? Push-pull? LLC?
Neither 50 nor 100% are typical. These are maximum values.
LLC is the easiest to tell, it is almost 50+50%, with some dead-time.
Push-pull is 0...49+49%. Unless it is current driven, resonant topology (I can't recall its name), because then more than 50+50%, fixed.
2 transistor forward has an upper limit of 50 %, 1 transistor forward can go a little higher, flyback may go up to 90%, and these has no lower limit.
But knowing these values without understanding the operation principles is useless...
It depends on the topology, and the meaning of "100%". (50+50% in push-pull is the maximum output, but in flyback 100% means damage.)
I was refering to flyback, half bridge and push pull, the ones that I've worked with. Some controllers have a fixed duty cycle of 50%, but some can go from 0 to 100%. What I didn't really understand is if it's the controller's job to regulate the duty cycle based on the feedback, or it needs to be considered when actually designing the converter.
Both.
A designer must know how to calculate duty cycle in any operating point.
The topologies you mentioned are different, each requires a different answer.
"Some controllers." There are many, with totally different operation. Be specific!
UCx84x family has both 50% and 100% capable chips for the same flyback topology. So in which situation would one or another be used?
In fly-back operation I can mention 2 cases where limited duty cycle is beneficial/neccessary:
- in continuous conduction operation mode (CCM) above 50% duty cycle there can be a phenomenon called subharmonic oscillation. This can be avoided by so called slope compensation or by limiting duty cycle.
- at high duty cycle the converter can draw high input current at low input voltage while trying to maintain the neccessary output power. If the input power source is limited to a lower current, then the system can stuck in a state where maximal duty cycle is set, the input voltage drops to a very low value, high input current is flowing, but there is not enough power transfer to maintain desired output power, although at a lower duty cycle the higher normal input voltage with normal input current would be perfectly enough. In this case limiting may solve the problem.
But there can be reasons in a specific design for duty cycle above 50% also.
- in continuous conduction operation mode (CCM) above 50% duty cycle there can be a phenomenon called subharmonic oscillation. This can be avoided by so called slope compensation or by limiting duty cycle.
- at high duty cycle the converter can draw high input current at low input voltage while trying to maintain the neccessary output power. If the input power source is limited to a lower current, then the system can stuck in a state where maximal duty cycle is set, the input voltage drops to a very low value, high input current is flowing, but there is not enough power transfer to maintain desired output power, although at a lower duty cycle the higher normal input voltage with normal input current would be perfectly enough. In this case limiting may solve the problem.
But there can be reasons in a specific design for duty cycle above 50% also.
This is usually called "right half plane zero" in compensation techniques. A hard task.
I explain your "This" if you don't mind: If you want to stabilize this nonlinear system in this instable-by-nature-state, then it is a hard task. But in many cases you can simply restrict the operation in the problematic area. And this second solution is more relevant to the original question, however not so interesting, state-of-art topic as controlling instable systems in frequency-domain.
I continue:
Close to 100% duty cycle capability (not fixed high, but widely variable) is required if input/output voltage range varies much, while high power is needed in every operation points. For example:
- Creating and maintaining stable electric arc. For creating arc a high voltage is required at lower current, but then arc resistance drops very much, and low voltage but high current is demanded. The first state requires high duty cycle (~90%), because otherwise the stabilised state would require extremely low duty cycle where efficiency is very bad.
- Recuperating brake of electric vehicle. At low speed the induced voltage is low, therefore to create braking torque as strong as possible the switching element must reach 100% duty cycle. I guess this kind of operation is the only one where exactly 100% is useful in fly-back topology.
But using UCx84x is not limited to fly-back, it is a simple, but universal single-ended PWM controller family. For examle I built also a ClassD audio amplifier from it (=synchonous buck converter). A quite bad and small one, to be honest, but it worked, and 100 % duty cycle was required...
Thanks for the answers!
I have one more question regarding 3525's feedback loop.
As far as I've seen in schematics, one of the onboard op-amp's inputs is tied to the reference pin through a resistor and the other input to the reference through the opto. So both inputs must be at half the reference voltage and the opto biased, if I'm right.
I have one more question regarding 3525's feedback loop.
As far as I've seen in schematics, one of the onboard op-amp's inputs is tied to the reference pin through a resistor and the other input to the reference through the opto. So both inputs must be at half the reference voltage and the opto biased, if I'm right.
Youre welcome.
I guess you mean a divider.
I see no question.
But otherwise yes, the inputs of OPA (better saying OTA) tracks the divided reference voltage as long as feedback loop is stable and the operating point is within the limits.
I guess you mean a divider.
I see no question.
But otherwise yes, the inputs of OPA (better saying OTA) tracks the divided reference voltage as long as feedback loop is stable and the operating point is within the limits.
Is there any way to estimate the maximum output power that I can get from a transformer given that I know its core size and frequency? I know it's not the right way to do it, but I have a bunch of them salvaged from old PSUs and I have no idea what model they are.
The topology will be either flyback or push-pull.
The topology will be either flyback or push-pull.
Yes. It is 0 with +10kW/-0W tolerance. 🙂
Yes, you should compare it to a similar size, known power rating transformer. The inaccuracy is about +300/-75%
The best way is to measure. Very easy to build a half-bridge fixed duty cycle PSU.
Yes, you should compare it to a similar size, known power rating transformer. The inaccuracy is about +300/-75%
The best way is to measure. Very easy to build a half-bridge fixed duty cycle PSU.
I suggest to take a read to a good book about switch mode power supplies, like the Abraham Pressman's one. To "cut and try" may be expensive and very risky.
- Status
- Not open for further replies.