Popular Discussion
- 6 comments
- 1 comment
Legend
Community Member
Power Integrations Engineer Power Integrations ModeratorDPA425 slow response/overshoot on step load change
I have a DC-DC converter design based on the DPA425GN, generated using PI Expert 7.1. It's performing reasonably well, but when I subject it to a step change in load from 500mA to no-load, the DPA-switch does not reduce the PWM duty cycle fast enough to prevent a significant output voltage overshoot, which causes the DPA-switch then to stop switching altogether for several milliseconds while the output settles. Adding a small preload as suggested by AN-31 only reduces the length of time this phenomenon occurs, it does not prevent it from occurring. Since this SMPS is required to supply power to a circuit that drives a communication waveform at 1200 Hz into a 500mA load, and reapplying the load within the 4ms while the DPA-switch has ceased switching causes the output voltage to collapse, this behaviour is unacceptable.
Attached is a schematic and oscilloscope traces that I hope describe the problem in sufficient detail.
Oscillograph 1 shows the response at 1ms/div, and you can see DPA-switch MOSFET turns off for almost 4 ms. The overshoot in the output voltage causes a significant rise in the phototransistor collector current, and causes the voltage on the Control pin of the DPA-switch (not shown) to rise to 6.3V. The DPA-switch does not resume normal operation until the voltage at the Control pin falls back below 5.8V.
Oscillograph 2 shows the response at 100us/div, this time instead displaying the primary current in the transformer. The red line I added to demonstrate the problem; the DPA-switch, in conjunction with my feedback circuit, takes over 200us (60 switching cycles at 300kHz) to enter cycle skipping, during which time the output voltage rises nearly 200 mV.
What do I need to do to eliminate this behaviour?
Thanks in advance for your help,
Tim Bates
Electronics Design Engineer
Installation Systems & Control (ISC) Business
Clipsal Australia Pty Ltd
| Attachment | Size |
|---|---|
| DPA425GN Schematic.pdf | 330.22 KB |
| Oscillograph 1.PNG | 47.91 KB |
| Oscillograph 2.PNG | 47.86 KB |
tim.bates
Hi,
I finally found a solution to the problem described above. I noticed that the application note and several of the reference designs used an R and C in series between the cathode and the reference terminal of the TL431. Replacing C205 (1nF) in my circuit with 1uF and 200R in series fixed my problem.
Tim Bates
tim.bates | 4 posts I am sorry but I don't understand the problem.
The nominal output voltage is 15V, and the overshoot is 0.2V which is around 1.3% with 0-500mA transient load.
Unless you have a high bandwidth post regulator in your design, you will have some degree of undershoot and overshoot with a transient load.
Usually it is acceptable if the overshoot and undershoot is within allowed voltage tolerance range, which is around 5%
Skywalker | 39 posts Posted: Mar 11, 2010 | 19:26 PM In response to Transient load overshoot
Sorry, in my original post I said "reapplying the load within the 4ms while the DPA-switch has ceased switching causes the output voltage to collapse" but I didn't make it clear what I meant. I've attached another oscilloscope trace, Oscillograph 3, which shows what happens if the load is reapplied within the 4ms while the DPA-switch is recovering. Since my application overlays a communication waveform (at 1200 bps) on the power supply, the load will be switching in and out within less than 1ms. This causes similar behaviour to what is shown here but I have reduced the frequency to 250 Hz to make the problem clear.
| Attachment | Size |
|---|---|
| Oscillograph 3.PNG | 56.37 KB |
tim.bates | 4 posts I've attached another oscilloscope trace that shows the problem more precisely. Oscillograph 4 shows that the control pin voltage rises significantly above 5.8 V, and when the feedback current is shut off it drops ~150 mV in the first 100 us and then decays slowly from there. Normal operation doesn't resume until the control pin voltage is back down to 5.8 V.
The question for me then is, why is the control pin voltage allowed to overshoot so dramatically? Is my loop gain too high? How do I reduce it without causing instability?
Thanks,
Tim Bates
| Attachment | Size |
|---|---|
| Oscillograph 4.PNG | 31.35 KB |
tim.bates | 4 posts 
