built 5V/100mA LNK306 from AN-37 -> need help from PI
Hi there,
actually we are planning a new device. Until now capacitive power supplies were used and they wasted a lot of wattless power. If the design is successful, we could sell ~10,000 devices a year. I read through the designs on this site but I did not find anything suitable. Here are the facts:
- 85-265V AC input where 230V is the normal operation voltage (European mains)
- 5V DC output at approx. 50-100mA
- 90% of the time the design "idles" so the efficiency must be optimized to that state. It doesn't matter if the efficiency factor drops during heavy operation of the CPU - important is that we save energy in the "idle" state. Relays will be switched and we could take that power from capacitors
- no galvanic isolation needed/wanted
I figured that a Linkswitch design is close but not close enough to what we need: epr48
In the "no load" or almost "no load" state we need a way better efficiency of let's say >=80% if possible? Is there a way to design such a power supply with Power Int or shall we stay with the capacitor PSU?
I appreciate every help and thank you for your support.
Josef Dean Butler
Comments
Thank you very much for your reply. If I get figure 5 in chapter 7.2 of the EPR48 on page 9 right, the no-load input power is around 150mW at 230VAC. Could this be improved let's say by adding a kind of charge pump, a Delon circuit or something similar instead of a bias winding (due to the lack of a transformer) at the output and feedback >5,6V to the BYPASS pin in the "CPU idle mode"? 150mW is still better than a cap-dropper I assume, isn't it?
I built a prototype of the 5V power supply based on AN-37. I attached the schematic. I tried to figure out the efficiency of that circuit and it does not really match the datasheets so I must do something wrong. At no load my output voltage Vout is 4,9V. At 10mA load I got 4,77V Vout and 2,25mA input current. At 250mA Vout is 4,44V and the input current is 30 mA. As you can see the efficiency is far away from 80% even at 250mA.
I tried the version with the optocoupler feedback as well but the no load consumption is even higher and at load the efficiency is more or less equal to the other circuit.
Am I missing an important thing here? I appreciate any hint ord help.
Regards,
Josef Dean Butler
Hi
I do not see any wrong thing with your design. let me duplicate in the lab your circuit and measure efficiency. your specs are 85-265V AC input 230V is normal operation voltage.
5V DC 100mA. Right?
thank you
Pi Worf
Hello PI Worf,
thank you for your response. Yes the specs are correct. I appreciate your help and await your results.
Regards,
Josef Dean Butler
Hi Worf,
did you succeed with testing the schematic? I'll attach the PCB in a PDF file as well. Ignore the parts not shown in the schematic, we tried to increase the output voltage and feed it back to the BP pin to turn off the internal linear regulator to improve the no-load consumption. I disabled that "feature" with the jumper because it did exactly the opposite...
With the disconnected jumper the circuit is exactly taken from the AN-37, I just changed the values of the inductors, capacitors and resistors to get the desired output voltage.
I appreciate your support and await your answer.
Regards,
Josef
Since I did not receive any answer neither here nor by email I tried to find a solution for my problem. I found a PCB guideline for switching power supplies and I realized that my old PCB design was not very well. So I tried to get the board EP48 from EPR-48 but unfortunately the German PI sales office could not help me with that, they need a "RDK number" which does not exist for an EPR, right? So I tried to "copy" it as good as possible and transferred it to SMD. I am waiting for the PCB manufacturer to deliver.
I really need some help here. We are pressed for time and I'd really like to get rid of cap droppers to make the device a little more "green". If it helps: we are planning a new ripple control receiver. Until now every manufacturer used cap dropper power supplies because they are cheap and worked since ages. Since one has to filter the ripple control signal from the mains line any noise has to be avoided. Anyway I'd like to give it a try.
Back to the topic: my problem is that the efficiency is not even close to what the datasheet promises even with heavy load (>200mA). So I assume that I made something wrong. PI-Worf wanted to test my circuit - any results yet? I even tried the optocoupler feedback but the efficiency was not better, is was just about the same as the galvanic feedback.
I really appreciate and need any help. Thank you for reading.
Regards
Josef
According to your input and output, our LinkSwitch-TN may be your choice and EPR48 is a good reference. You only need to change the feedback for 5V. However, it is almost impossible for 80% efficiency at very light load. At no load, the converter still needs to switch to maintain the output and control circuit is still working. There is always some no load consumption ranging from tens of mW to hundreds of mW. So the efficiency at no load is zero. At very light load, the efficiency is also very low.