Ok, we finally could buy some copper foil! So we winded as you told us:
1) 16 turns of the primary side, we used 3 wire of 0.3mm in parallel (not litz)
2) Isolation and copper foil (not closing it) to ground
3) 6+6 secondary turns using copper foil of 0.1mm and 25mm of width winded at the same time
4) Copper foil around all the transformer to ground
We tested not connecting the shields to ground yet. We got 120uH from ground point to each side of the secondary and 0H between the extreme points (is this correct?).
So, we connected everything and put some load, there is some vibration on the transformer now, not that high but it's noticeable. With the first load with about 2A we got 45V, second load about 4.5A and 42V (we got about 38V before) and with the last load with almost 10A we got 33V (huge drop again
)
1) 16 turns of the primary side, we used 3 wire of 0.3mm in parallel (not litz)
2) Isolation and copper foil (not closing it) to ground
3) 6+6 secondary turns using copper foil of 0.1mm and 25mm of width winded at the same time
4) Copper foil around all the transformer to ground
We tested not connecting the shields to ground yet. We got 120uH from ground point to each side of the secondary and 0H between the extreme points (is this correct?).
So, we connected everything and put some load, there is some vibration on the transformer now, not that high but it's noticeable. With the first load with about 2A we got 45V, second load about 4.5A and 42V (we got about 38V before) and with the last load with almost 10A we got 33V (huge drop again
)
Not enough primary.
More parallel wires.
Let me show you something.
This is an ETD34, i.e. much smaller than yours.
It has 54 turns 2x0.6mm on primary, i.e. wire area on primary = 31mm2
Secondary: 6x0.6mm, 28 turns; i.e. wire area = 48mm2
Shield between layers.
Switching at 50kHz.
Easily throws out over 400W, easily... I have not even tested the flat out power.
Your transformer has 3mm2 on primary.
More parallel wires.
Let me show you something.
This is an ETD34, i.e. much smaller than yours.
It has 54 turns 2x0.6mm on primary, i.e. wire area on primary = 31mm2
Secondary: 6x0.6mm, 28 turns; i.e. wire area = 48mm2
Shield between layers.
Switching at 50kHz.
Easily throws out over 400W, easily... I have not even tested the flat out power.
Your transformer has 3mm2 on primary.
Attachments
Ok now we have 4 wires of 0.5mm on the primary side and 0.1mm foil on the secondary side. Where do we have to connect the ground shields? Output GND or GND from mains? The transformer behaves better! How do you calculate the wire diameter and how many do you have to use in parallel?
Now we got 45V at 2.3A, 42V at 5.5A and 33V at 10.5A, is a big voltage drop again but we can use 33V. Maybe if we add more copper on the primary side we could get more voltage but there is no more room on the transformer, maybe we could try doubling the frequency to 200Khz to reduce the number of turns, but we will see later, we want to finish this.
The transformer does a strange noise, not that loud but is still there. We added a snubber network on the primary side using a 66Ohm resistor and 2.2nF capacitor because we noticed huge peaks of about 150V. The noise is not there anymore but as soon as we start draining 10A in just 3 seconds the 5A fuse blows up and MOSFETs burns instantly, drain, source and gate all shortcircuited!
Thank you very much for your help we really appreciate it, we have learned a lot
Now we got 45V at 2.3A, 42V at 5.5A and 33V at 10.5A, is a big voltage drop again but we can use 33V. Maybe if we add more copper on the primary side we could get more voltage but there is no more room on the transformer, maybe we could try doubling the frequency to 200Khz to reduce the number of turns, but we will see later, we want to finish this.
The transformer does a strange noise, not that loud but is still there. We added a snubber network on the primary side using a 66Ohm resistor and 2.2nF capacitor because we noticed huge peaks of about 150V. The noise is not there anymore but as soon as we start draining 10A in just 3 seconds the 5A fuse blows up and MOSFETs burns instantly, drain, source and gate all shortcircuited!
Thank you very much for your help we really appreciate it, we have learned a lot
Blowing a 5A fuse at the primary is good. That means you are finally getting some serious power to the transformer. Blowing the mosfets is not.
Do not use a bigger fuse.
Lets see your waveforms before and after the output inductor at closer to the 10A max you are getting out at the moment (waveforms just before the china syndrome that is)...
You do have heatsinks on your fets, right?
You connect the x-former shield to the chassis.
Do not use a bigger fuse.
Lets see your waveforms before and after the output inductor at closer to the 10A max you are getting out at the moment (waveforms just before the china syndrome that is)...
You do have heatsinks on your fets, right?
You connect the x-former shield to the chassis.
Last edited:
We will measure the waveform and post the results! Yes the MOSFETs are on heatsinks. What do you mean by x-former shield? The ground shields of the transformer are not connected, should we connect it? We have to connect it to output GND or mains GND?
We will buy some bigger MOSFETs but that will take some days!
Connect to the "mains ground" (safety earth).
X-former is the tranfo...
If you start looking for for a new pair of switches, check also what is available from the IGBT range.
IR2110 is actually not that powerful. Try keep the gate capacitances under 1nF in the switches...
Also, everything you have between the IR2110 and the switches creates losses to your circuit and reduces the gate turn on/off power.
You can direct drive the switches with the IR2110. Or/and you could change that to a FAN7390 or something similar, a more powerful device...
You might also need more dead time than just 625ns, now when the fets are seriously been put in to work...
X-former is the tranfo...
If you start looking for for a new pair of switches, check also what is available from the IGBT range.
IR2110 is actually not that powerful. Try keep the gate capacitances under 1nF in the switches...
Also, everything you have between the IR2110 and the switches creates losses to your circuit and reduces the gate turn on/off power.
You can direct drive the switches with the IR2110. Or/and you could change that to a FAN7390 or something similar, a more powerful device...
You might also need more dead time than just 625ns, now when the fets are seriously been put in to work...
Ok! I will take a look to IGBT. We didn't drive the MOSFET directly from the IR2110 in order to isolate the control circuit from the mains. We can however change it for a more powerful driver.
1uS dead-time is ok or we need more? Should we add some 1A high speed diodes in parallel to MOSFETs?
Ok, as requested here are some waveforms! All were taken using a 100W incandescent lamp in series with main voltage (220V RMS) with a load of about 1A so the output voltage is lower than without the lamp. We don't want to try without the lamp until we buy more MOSFETs in case we blow up everything again!
Secondary
Secondary - After rectifier and before DC Choke
After DC Choke
Primary using AC trigger (measuring in differential mode)
Primary using DC trigger (measuring in differential mode)
No ground shields connected. We are planning to buy some FDP12N50NZ to replace IRF840. IGBTs are really expensive here and only a few models available.
Secondary
Secondary - After rectifier and before DC Choke
After DC Choke
Primary using AC trigger (measuring in differential mode)
Primary using DC trigger (measuring in differential mode)
No ground shields connected. We are planning to buy some FDP12N50NZ to replace IRF840. IGBTs are really expensive here and only a few models available.
Some useful informations, at last.
Some details are missing however: is the secondary waveform actually that: ie GND of the probe on one end of the winding and hot side on the other? If it is the case, there is a serious problem with the control logic.
The primary waveforms are also seriously questionnable, but with the differential mode, one cannot be sure of anything: it would be preferable to use an isolation transformer and measure the waveforms directly.
You should measure the gate drive voltages directly, but without applying the main power, just the controller, both of them on the same screen
Yes, we are measuring between GND (mid point) and hot side. We are using a PIC18F14K50 with the ECCP module that generates the Half-Bridge signal. However we know that is not the same signal that should be applied to Half-Bridge SMPS topology as you can see here where they also provide a solution. But we are using 50% duty cycle so as we understand the waveform should be pretty much the same generated by a TL494 or SG3525.
Here is a measurement of the primary side without differential mode:
And this is the signal of both MOSFETs gates:
Definitely something wrong with those waveforms. I don't think one MOSFET is failing to turn on though. If this was the case, there would be no power output at all, as the primary is AC coupled.
I suspect wrong phasing of the secondaries, so that both diodes conduct on the same half cycle. Try reversing the phase of one half of the secondary.
The primary voltage should look like this: http://scopeboy.com/elec/pshack/scopetrace.jpg Note that both diodes conduct during the deadtime periods and when a MOSFET turns on, one of the diodes turns off.
I suspect wrong phasing of the secondaries, so that both diodes conduct on the same half cycle. Try reversing the phase of one half of the secondary.
The primary voltage should look like this: http://scopeboy.com/elec/pshack/scopetrace.jpg Note that both diodes conduct during the deadtime periods and when a MOSFET turns on, one of the diodes turns off.
Last edited:
Wrong phasing the secondaries? Both are winded at the same time. What do you mean by reversing the phase? Winding one half clockwise and the other half counter-clockwise?
Yes, although, you do not need to rewind anything, just reverse the connection in one of the secondaries...
Hi,
Are you sure about the drive circuits?
This is what I found on eletronicdesign.com
The phasing of your gate drive transformers looks strange to me. Unless you take care of that with the programming?
It seems that one transistor is not allowed to conduct for the full 4us.
Cheers,
Serge
Are you sure about the drive circuits?
This is what I found on eletronicdesign.com
An externally hosted image should be here but it was not working when we last tested it.
The phasing of your gate drive transformers looks strange to me. Unless you take care of that with the programming?
It seems that one transistor is not allowed to conduct for the full 4us.
Cheers,
Serge
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.