Today we replaced our EE42 transformer with the PC power supply ATX transfomer changing the switching frequency to 31.25KHz as it is in the ATX SMPS to test if the problem is in the transformer or somewhere else. Now, when we apply load to the ATX transformer on our SMPS there is a significant voltage drop and high voltage spikes at the output! But on the ATX board this doesn't happens! Even without snubbers!
The output rectifier and filter is almost the same so we are now thinking the problem may be in the GDT or with the IR2110 not being able to properly drive the MOSFET when we start applying load.
The output rectifier and filter is almost the same so we are now thinking the problem may be in the GDT or with the IR2110 not being able to properly drive the MOSFET when we start applying load.
Ok! We changed the MOSFET gate resistor from 10 Ohm to 330 ohm to increase the rise time. The MOSFET has 1200pF on the gate. The fall time is still very short because of the PNP transistor discharging the gate. Anyway, the problem persists
Not only the spikes, but the large voltage drop Something we noticed, when we add a 5nF cap between GND on the 300V side and GND on the output the output ripple decreases significativelly, is still there, but it is reduced! Maybe we have some problem with the GND connections?
I think C6 and C10 are fine. But I hope T1 is a misprint. There should really not be a current transformer here.
-The current here has a DC component that will just saturate it.
-It adds a ton of stray inductance in the drain of Q1.
Q1 and Q3 need to be as close as possible to C6, C10 to minimise stray inductance in this loop.
-The current here has a DC component that will just saturate it.
-It adds a ton of stray inductance in the drain of Q1.
Q1 and Q3 need to be as close as possible to C6, C10 to minimise stray inductance in this loop.
Last edited:
T1 is not a misprint :/ It's really there, AFAIK it's not saturating, the current protection works great, it is a small EI core from a PC power supply. We can remove it if you think that could be the problem!
Yes, Q1 and Q3 are very close to C6 and C10 and very close to the transformer too.
Well, the dead-time is generated by the micro-controller. We tried small dead-times like 500nS up to 5uS at 31.25KHz with no difference. The only difference is that the voltage drop is even bigger when adding more dead-time but it seems reasonable.
We will try to remove the current sense transformer, we don't know what else to test. We can also try to change the GDT with bigger toroids to test, maybe there is a voltage drop in the gates when applying load
Ok, so today we did some more tests with the ATX transformer. Here you can clearly see that this output spikes are being produce on the rising and falling edge of the waveform. The blue signal is measured between GND and a secondary hot-point and the yellow signal is the output voltage:
So we replaced the 10 Ohm gate resistors with 2.2k resistors to increase the rise time of the MOSFETs. Here is the waveform on the secondary side under 1A load:
The spikes are much lower now, and we noticed it even more when we tested with our transformer, but the output voltage is drastically affected too dropping even more under load with our transformer.
Also we added 1500pF cap between mains GND and the output GND and we noticed the output spikes are reduced significantly. Here is the secondary and output voltage without the cap:
And this is with the cap:
I don't know why, but that happens. Maybe adding some snubbers to the MOSFET can help to reduce the spikes?
So we replaced the 10 Ohm gate resistors with 2.2k resistors to increase the rise time of the MOSFETs. Here is the waveform on the secondary side under 1A load:
The spikes are much lower now, and we noticed it even more when we tested with our transformer, but the output voltage is drastically affected too dropping even more under load with our transformer.
Also we added 1500pF cap between mains GND and the output GND and we noticed the output spikes are reduced significantly. Here is the secondary and output voltage without the cap:
And this is with the cap:
I don't know why, but that happens. Maybe adding some snubbers to the MOSFET can help to reduce the spikes?
We solved it! Here is what we did:
Now there is a perfect DC output voltage! Now we have about 5V of 50Hz ripple but just adding more input capacitance can solve it! Thank you very much to everybody for your help we really appreciate it. Without your help this couldn't be possible, it's our first SMPS Now we have to finish the software of the micro-controller and we will post everything once we have all finished! Here is the schematic: https://dl.dropboxusercontent.com/u/70561817/Esquema fuente.pdf
Thank you very much!!!
- Changed frequency to 200KHz so now we have 9 turns on primary and 4 on secondary, we wound half primary, secondary and the other half primary. Now we had more space so we wound the primary with 16 wires of 0.3mm and the secondary with 0.1mm copper foil.
- Added snubbers in parallel to each diode in the output rectifier, this significantly reduced the oscillations
- Added an snubber on the primary side
- Output choke value 330uH
- Added 100nF and 10nF capacitors on the output
- New PCB with shorter tracks than the previous version
- Measure the output voltage as shown here (Very important)
Now there is a perfect DC output voltage! Now we have about 5V of 50Hz ripple but just adding more input capacitance can solve it! Thank you very much to everybody for your help we really appreciate it. Without your help this couldn't be possible, it's our first SMPS
Now we have to finish the software of the micro-controller and we will post everything once we have all finished! Here is the schematic: https://dl.dropboxusercontent.com/u/70561817/Esquema fuente.pdfThank you very much!!!
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.