Solution Finder Get Tech Support

Negative Drain Current of TOP265 in continuous operation

Posted by: Prafulla S on

Hello Support,

In one of my circuit of SMPS power design (3W), I am using TOP265 JX switch where I am facing negative drain current going upto some -300mA. I went thorugh datasheet of TOP265 but could not find any information regarding this phenomenon. Output side has two windings. 24V,100mA & 6V,50mA. Both outputs have separate isolated grounds. This -ve drain current increases when i am adding RC snubber (33Ohms, 47pF) on 24V winding diode. Below is the drain current waveform during continuous operation with and without RC snubber. RC snubber is required to pass my design in RE test.
Power Supply: 240 VAC, 50Hz. How I can reduce this -ve current? or what are the detail consequences of this -ve drain current?

Files

Attachment Size
Waveforms.png 111.1 KB

Comments

Submitted by PI-Wrench on 05/26/2020

It would be helpful if I could see the schematic. Having some HV bypass capacitance(if not already there) from B+ to source close to the TOPSwitch and transformer is useful to help reduce undershoot. Also, the RC snubber you used on the 24V output appears very underdamped. Higher series resistance would be better. Try starting at 150 ohms and work down from there.

Submitted by Prafulla S on 05/26/2020

unfortunately i can not share the schematic. Can you elaborate more on HV bypass capacitance from B+ to source node. Accordingly i can check in my design. Also in absence of snubber across secondary diode, i am seeing -ve drain current of -150mA approx, which increases with adding snubber.

Submitted by PI-Wrench on 05/27/2020

Using a poorly damped snubber as mentioned previously will make secondary ringing worse, and reflect back to primary. Please try tuning the snubber with higher resistance as previouly mentioned. If the converter is located far from bulk capacitor, a small bypass capacitor (1-2 nF for such a small supply) is useful to reduce effect of long traces. Capacitor should be located close to converter to do the most good.
Also, what kind of diode are you using for your 24V output? There seems to be a big initial current spike for a supply that by all rights should be running in discontinuous mode given the high input voltage and low output power.

Submitted by PI-Wrench on 05/28/2020

On another note, I tried running a spreadsheet for your basic design (24V, 3W, 180-265 VAC) using PI-XLS, and I'm mystified that such a large device was used for only 3W. I ran the design again using the smallest Tiny-4 device and obtained results more in line with what I think is needed for a flyback design at high line and low power output.

Submitted by Prafulla S on 05/28/2020

1. I will try out with higher value of snubber resistor and will check -ve drain current.
2. In parallel to bulk cap of 22uF, small bypass cap of 10nF is added. 24V winding diode is ultra fast diode STTH108A. I am getting drain current up to some 2.6A in start up. In continous operation , maximum drain current is 1.6A.
3. As of now i can not change PWM switch. Power supply is designed to work from 20 VAC TO 265 VAC. -ve drain current starts coming into picture after some 110VAC. One more point i want to add is along with secondary snubber, i have RCD snubber added between D-S to pass RE test. Please go through attached image.

Attachment Size
Snubber.png 154.35 KB
Submitted by PI-Wrench on 05/29/2020

Aha! Now I see your problem. Trying to run over such an extreme input voltage range is an extremely ambitious undertaking. I ran a spread sheet using the input voltage range you mentioned, and found out a few things. First off, your input filter capacitor will have to be greatly oversized for proper function at the low end of the range, since energy storage is so poor at such a low AC input voltage. This is a consequence of the 20 VAC low end. If the converter is run from straight DC, that requirement is relaxed..I chose 100uF for the spreadsheet. You will need to run at a low reflected voltage (VOR) to accommodate the low voltage requirement. I chose 25V in my spreadsheet. As you no doubt have found out, this results in a very high voltage across the output rectifier at the high end of your input voltage range, as evidenced by your use of an 800V diode for 24V output. The part you used does not specify a junction capacitance, which makes it difficult to choose snubber capacitance. I tried a US1J, which is an ultrafast part with a 10 pF junction capacitance. With that in mind, I would try a 33pF snubber cap, with a 150 ohm series resistance. I decided on a KP of 0.8 at low line, and a KI of 0.4, and ran the TOPSwitch at 66 kHz. 132 kHz operation may offer the option of a smaller transformer with fewer turns, reducing the stray capacitance.

The clamp you showed in your previous post appears to have the diode oriented the wrong way round. I would suggest using a diode plus TVS clamp from the TOPSwitch drain to the B+ rail, with a TVS value of ~39V (~1.5 X VOR) to keep the clamp voltage away from the reflected voltage.

The TOPSwitch uses an integrated power mosfet, so excessive ring below ground is not a good idea. If all else fails, an ultrafast diode in series with the drain would block the reverse ring, but I would rather we come up with a more optimal solution, as I worry about the extra voltage drop at low input voltage.

Submitted by Prafulla S on 05/30/2020

Hello,

May be attached image help you to understand rough connection diagram. The RCD snubber i showed in previous post is between drain & source not between drain & transformer. With attached connection diagram I am seeing -ve drain current reduced upto - 250mA. Please share your expertise on improving -ve drain current.

Attachment Size
Rough_Connections.png 54.24 KB
Submitted by PI-Wrench on 06/01/2020

What is the purpose of snubber D4, R2, C9 from drain to source?. This configuration looks like it would result in a large slug of current at turn-on, making matters worse for reverse current. This would especially be true at high line, when you have a long off time, with lots more time to charge C9. Adding a damper resistor of a few tens of ohms in series with C9 may help if that snubber really needs to be there. What happens to the reverse current if you lift C9?
If that snubber is there for radiated EMI, another alternative might be to add a snubber across you primary bias rectifier, which in our experience can be a hot spot for radiated EMI. This particular application might pose EMI challenges due to the unusually high voltage stress on all rectifiers at high line.