ZVS for 5 KV output
Neukyhm, Sun Jul 23 2017, 01:03AMHi guys, I'm still designing the PSU for my x-ray machine. I have decided to go with a ZVS and run it with 48V (I know it's a high voltage to use with a ZVS).
I will drive the MOSFET gates separately with an external 12v PSU.
The idea is to build a transformer using the ETD59 N87 core, 3T+3T (3.45uH+3.45uH) primary and ~250T secondary, and then 6 stages multiplier to reach ~60kV.
I have simulated the entire circuit in OrCAD and everything seems to be fine. Current in primary is a bit high (>30A) but using ETD59 with 1.5mm gap gives a saturation current of 49A for the same primary, so I'm below saturation.
Capacitor value is the typical 600n and frequency is around 50kHz.
As I said, everything in OrCAD seems normal and within limits, but I know that ZVS is an oscillator that tends to draw A LOT of current when you increase input voltage, so I'm here to look for advice. Do you have experienced with 48V input in a ZVS?
Edit: I forgot to say: my 48V PSU is 300W, but OrCAD says that the ZVS will only draw 4-5 amps. That's why I'm confused, because I have seen people running a ZVS with 1000W and a very high current input, my machine is intended to be only 200W or less.
Re: ZVS for 5 KV output
Sulaiman, Sun Jul 23 2017, 07:54AM
Random thoughts;
To maintain zvs operation the VAR in the resonant circuit must be greater than 2.3x the power output,
I use at least 5x for a decent sinewave.
For 200W output that equates to about 1 kVAR.
Your zvs should be operating at about 100V x 20A = 2 kVAR ... more than adequate.
Skin depth in copper at 50 kHz means that wire diameter should be <1mm od.
preferably use 'Litz' or copper tape.
With no load the power consumption should be about 2 kVA / Q
e.g. if Q = 40, power dissipated in the primary windings and core of the transformer plus primary capacitor and wiring = 50 W
... the dc supply current with no output load should be about 1 A dc.
In practice, almost all of the losses are in the transformer core at no load
so the dc current under no load gives a clear indication of transformer heating.
i.e. if your zvs draws 5 Adc at 48 Vdc .
.. almost all of that 240 W is being dissipated in the core of the transformer.
2 kVAR seems high, even for an ETD59 set,
consider 4 + 4 turns primary with 330 nF etc.
(or even 5 + 5 turns with 220 nF for significantly cooler running)
The number of secondary turns will need to be scaled accordingly for a similar output voltage.
It is rarely magnetic saturation of the core that is the limiting factor, it is usually core loss.
You need to know core volume and ferrite grade, flux density and frequency, then look up the loss from the datasheet.
ALWAYS apply gate drive BEFORE the main power to the output stage,
and remove the main power input before the gate drive ... to avoid
Sulaiman, Sun Jul 23 2017, 07:54AM
Random thoughts;
To maintain zvs operation the VAR in the resonant circuit must be greater than 2.3x the power output,
I use at least 5x for a decent sinewave.
For 200W output that equates to about 1 kVAR.
Your zvs should be operating at about 100V x 20A = 2 kVAR ... more than adequate.
Skin depth in copper at 50 kHz means that wire diameter should be <1mm od.
preferably use 'Litz' or copper tape.
With no load the power consumption should be about 2 kVA / Q
e.g. if Q = 40, power dissipated in the primary windings and core of the transformer plus primary capacitor and wiring = 50 W
... the dc supply current with no output load should be about 1 A dc.
In practice, almost all of the losses are in the transformer core at no load
so the dc current under no load gives a clear indication of transformer heating.
i.e. if your zvs draws 5 Adc at 48 Vdc .
.. almost all of that 240 W is being dissipated in the core of the transformer.
2 kVAR seems high, even for an ETD59 set,
consider 4 + 4 turns primary with 330 nF etc.
(or even 5 + 5 turns with 220 nF for significantly cooler running)
The number of secondary turns will need to be scaled accordingly for a similar output voltage.
It is rarely magnetic saturation of the core that is the limiting factor, it is usually core loss.
You need to know core volume and ferrite grade, flux density and frequency, then look up the loss from the datasheet.
ALWAYS apply gate drive BEFORE the main power to the output stage,
and remove the main power input before the gate drive ... to avoid
Re: ZVS for 5 KV output
Neukyhm, Sun Jul 23 2017, 12:56PM
OrCAD says it draws 4-5 amps from my 48V psu WITH load (x-ray tube). I don't have a Pspice model for my x-ray tube, so I have replaced it with a big resistor in my circuit. The value of that resistor has been obtained from the datasheet (it says ~3mA when 65kV is applied, so 65000/0.003=21.6Mohms)
For the core, I'm using the k_linear model of Pspice, I don't know how to use the ETD59 model.
Here you can download my schematic Pspice files https://drive.google.com/file/d/0B4jpJVxcLvSEa0RvMlZUTzAyZFU/view?usp=sharing
Neukyhm, Sun Jul 23 2017, 12:56PM
OrCAD says it draws 4-5 amps from my 48V psu WITH load (x-ray tube). I don't have a Pspice model for my x-ray tube, so I have replaced it with a big resistor in my circuit. The value of that resistor has been obtained from the datasheet (it says ~3mA when 65kV is applied, so 65000/0.003=21.6Mohms)
For the core, I'm using the k_linear model of Pspice, I don't know how to use the ETD59 model.
Here you can download my schematic Pspice files https://drive.google.com/file/d/0B4jpJVxcLvSEa0RvMlZUTzAyZFU/view?usp=sharing
Print this page