High voltage > Transformer (Ferrite Core)

SG3525 push pull snubber calculation

(1/4) > >>

PowerTech:
First hi all, and sorry for my english  :)

I built test SG3525 push pull inverter from chinese SG3525 LM358 Inverter Driver Board module and pair of IRFP260N, drain-source is protected by transil 1.5KE200CA, now waiting parcel with R-C snubber parts.
HV transformer based on UF86A ferrite core, homemade primary & secondary - 5T+5T/700T.
Driver has separate 15V power supply.
Main power supply is 600W transformer, under load(450W) after rectifying -35V.
Transistor heating is moderate, after 1min case temperatue is 45*C.(one large heatsink for rectifier and transistors)
In future i add EMI filter to prevent high frequency going in home mains line.



Connected oscilloscope and get ugly waveforms  :-\
After looking on drain-source waveform i think that 1.5KE200CA is not working- voltage spike more than 200V, or it is normal for a short period ?

Drain-Source


Gate, driver only working


Gate, 450W power consumption, load - hv arc


Main DC bus, on load


220/32V transformer, secondary, on load


For RC snubber calculation i used method from google -" For a RC snubber, you want the C to be about three times the parasitic Cs and R to be equal to sqrt(Ls/Cs)", where Cs - transistor output capacitance, Ls - internal source inductance, this is correct ?
Any things to remove spikes ?

Weston:
Ls here is going to be predominately the leakage inductance of the transformer, which you can measure by shorting the secondary and measuring the inductance of the primary.

A RC snubber or the TVS is both going to dissipate energy. Unless it's dissipating too much energy the TVS snubber should work by itself. The RC snubber will dampen the ringing and likely have lower EMI. The TVS snubber will just clamp the peak voltage and likely have a bit lower total power dissipation than the RC snubber.

The issue with the current TVS snubber is that the 1.5KE200CA has a clamping voltage above 200V. If you look at the datasheet the breakdown voltage, where 1mA is conducted, is 190-210V, while the clamping voltage, when the device sinks 5.5A is 274V. Based on the current it's going to be somewhere between those two values.

Your DC bus is also spiking up with the switching transitions, you could also benefit from better layout between the DC filter capacitor to the transformer and the return from the FET source to the capacitor / added film snubber capacitors. This would reduce the stray inductance.

PowerTech:
Thanks for the reply, need revise dc bus connections for optimal way, also I leave TVS and RC snubber together.


--- Quote from: Weston on November 13, 2021, 09:10:07 PM ---Your DC bus is also spiking up with the switching transitions, you could also benefit from better layout between the DC filter capacitor to the transformer and the return from the FET source to the capacitor / added film snubber capacitors. This would reduce the stray inductance.

--- End quote ---
Not shown in schematics, 3uF film capacitor is connected to dc bus electrolyte.

davekni:
As Weston mentioned, transformer leakage inductance is the source of energy that must be clamped.  For this topology, leakage inductance between one primary 5T winding and the other 5T winding is most important.  (To measure, short one primary and measure inductance of the other primary winding.)  Energy stored in leakage inductance between primary and secondary is mostly returned to the DC supply through the conducting FET, so not of as much concern.

To minimize primary leakage inductance, wind the 5 turns bifilar.  Wind two parallel wires for 5 turns.  Even better, use many pairs of smaller wire, wind 5 turns of the bundle, then split out wires of each pair, one for each primary winding.  Also, lead length from transformer to inverter board should be as short as is reasonable, with leads of each winding paired together (twisted or taped together or ...) to minimize lead inductance.  Lead inductance can be a significant portion of leakage inductance.  (Even though GDTs are low-power, looking at GDT designs may help, as low leakage inductance is important there too.)

Also as Weston said, 270V is normal for 1.5KE200CA clamping high current.  Chances are that the IRFP260 FETs are handling half or more of the current during clamping.  IRFP260 is rated for 28mJ repetitive clamping energy, and 46A clamping current, so should be fine clamping even without TVS diodes.  However, coming from China, the IRFP260 parts are very likely counterfeit.  IRFP260 counterfeit parts are widely available at much lower cost than genuine parts.  Probably a range of qualities of counterfeit parts, from junk to somewhat-close to real IRFP260 capabilities.  Obviously yours are working at least reasonably.

PowerTech:

--- Quote ---For this topology, leakage inductance between one primary 5T winding and the other 5T winding is most important.  (To measure, short one primary and measure inductance of the other primary winding.)

--- End quote ---
With this method Ls=0.014mH, so calculate RC snubber -
Cs = 603pF = 0.603nF
Csnubber = 603*3.14=1809pf=1.893nF
Ls = 0.014mH = 14000nH
Rsnubber = sqrt(Ls/Cs) = sqrt(14000/0.603) = 152R
My calculations are absolutely incorrect, so when parcel arrives, I try with 10R+1nF, 10R+2nF and chose best by waveform.

Transformer don’t have any gaps between core halves, removed them after using in "ZVS driver".
Primary winding is bifilar, made from thick Cu wire, approx. 6mm*2, diy litz wire is almost ready - difficulties to qualitative remove enamel isolation from multiple wires.
IRFP260 is purchased long time ago from trusted local shop, IOR original, they worked perfectly in "ZVS driver" at ~1kW with many unwanted inductances - long wires in assembly.

Test inverter assembly

Navigation

[0] Message Index

[#] Next page