Some questions about my SMPS TIG Welder build...

legg

New member
I'm not sure how your original asc file was intended to function, but I've made some simplifications and adjustments to the magnetics and spice directives in order to get a reasonable current flow and some kind of characteristic start-up behavior.

At this frequency and output current, I don't think UF rectifiers are recommended. Schottkys are used here. A phase modulator really can only benefit from capacitive snubbing on the lagging phase.

RL
 

Attachments

  • UC3879.zip
    31.1 KB · Views: 25

kees54

kees57
I'm not sure how your original asc file was intended to function, but I've made some simplifications and adjustments to the magnetics and spice directives in order to get a reasonable current flow and some kind of characteristic start-up behavior.

At this frequency and output current, I don't think UF rectifiers are recommended. Schottkys are used here. A phase modulator really can only benefit from capacitive snubbing on the lagging phase.

RL

Hi Thanks for your effort to make the model better, maybe now I get a better respons in the feedback path who is very slow, I can not make this models, not enough knowledge about LTspice I am searching also for the UC1865 or 1861, 0r UC3861 etc resonant controller, because as I read this are the best way to make a inverter with above resonance control, this controller uses a vco and the transformer sees a sinoidal current..

The softswitching of the igbt is with snubber caps and with mosfets the body diode and capacitance is be used and so no caps are there, but failing of mosfets in the past was cause special mosfets are be made now with a fast body diode.

The models I have are also on LTwiki, and are in the http://valvol.ru/topic62.html

http://www.ltwiki.org/?title=LTspiceIV-library_Library_Listing_Expanded

http://valvolodin.narod.ru/models.html

I do use schottky diodes fgor rectifier 120EBU02. For mine inverter I use the UC3875 controller because I have these in stock, is the UC3879 a better choise? did read it is a better update of the UC3875 but has only one feedback input, but this is not a problem because a simple change of fedback is enough, I do use extern feedback with opamps to make it more rugged..

regards
 
Last edited:

kees54

kees57
I'm not sure how your original asc file was intended to function, but I've made some simplifications and adjustments to the magnetics and spice directives in order to get a reasonable current flow and some kind of characteristic start-up behavior.

At this frequency and output current, I don't think UF rectifiers are recommended. Schottkys are used here. A phase modulator really can only benefit from capacitive snubbing on the lagging phase.

RL

I have some sims done with your model, however the UC3879 has only one input to the error amplifier and the other is on 2.4 volts reference, I have used your filter on that input and invert mine feedback, normally the comp pin can be overruled by a other opamp, but I did get this not to work, the UC3875 did work fine here.

I have a more complicated feedback network with more opamps and a sawtooth generator so it regulate better, normally I do input that straight into the com pin and ground the - pin, when I do just use the error amplifier of the UC3879 with a resistor as a 1 x amplier it do work with that.

I do struggle a little how fast need a volatge source to be when using as a welder, later on I go make a version for the hybrid ampllfier with 300 volts, 2 x 65 volts 15 amps and 15 volts 5 amp, also here for a audioamp i need a fast feedback, so need to look at stability then, I have already the EDT59 core, think that is big enough.

View attachment testen-31-UC3879-1.zip
 
Last edited:

legg

New member
Hi Thanks for your effort to make the model better, maybe now I get a better respons in the feedback path who is very slow, I can not make this models, not enough knowledge about LTspice I am searching also for the UC1865 or 1861, 0r UC3861 etc resonant controller, because as I read this are the best way to make a inverter with above resonance control, this controller uses a vco and the transformer sees a sinoidal current..

A converter with series resonant output might be desirable for welding, as it is inherently self-limiting, but maintaining a safe open circuit terminal voltage might be an issue. If you're operating above resonance, then the converter is series resonant.
Models of integrated controllers are rare (I've not seen one offered), but the modulation method alone isn't necessarily complicated.

Where are you pulling your IGBT models from? The Valvol package covers a few Infineon parts - but not the IR (?) numbers you're calling up. I haven't used them in practical work since 1995, for various reasons.
 

kees54

kees57
here are these, also from the russian site I did already mention.

View attachment extra.rar


For the welding question, I did read that much inverters are made quasy resonant or series resonant, I have a LCC version who looks
promissing, but for savety we need a UC3861 controller because we have to stay above resonance, when go low on resonance the voltages
rise to Kilovolts what do distroy the igbt and dc blocking cap, wo design of these need attention.

For mine welding inverter, I need really a voltage source, so a constant voltage inverter because it is for a MIG but I have a feedback schematic
with some opamps who can do both, when putting voltage pot max, I can adjust current, and when needs voltage I set current max en adjust
with voltage pot as a normal supply, I inject the output straigt into the UC3875 comp pin as is allowed by datasheet, maybe it is also better to
use only chip fase shift and oscillator.

regards
 

kees54

kees57
here are these, also from the russian site I did already mention.

View attachment 6187


For the welding question, I did read that much inverters are made quasy resonant or series resonant, I have a LCC version who looks
promissing, but for savety we need a UC3861 controller because we have to stay above resonance, when go low on resonance the voltages
rise to Kilovolts what do distroy the igbt and dc blocking cap, wo design of these need attention. Series resonance also demand a different feedback
setup becaause of the resonance, I have ordered two big toroids for LIVI system with faseshift, nice to try, if I still am alive because it is not
playing tools.

For mine welding inverter, I need really a voltage source, so a constant voltage inverter because it is for a MIG but I have a feedback schematic
with some opamps who can do both, when putting voltage pot max, I can adjust current, and when needs voltage I set current max en adjust
with voltage pot as a normal supply, I inject the output straigt into the UC3875 comp pin as is allowed by datasheet, maybe it is also better to
use only chip fase shift and oscillator.

regards
 

kees54

kees57
I have sim the UC3879, I see the low amperage do cause a big swing in the feedback, but think the output cap making a low pass
do cause it and the slow simulation, in real time I think it is different, in high current it is quite nice feedbacked and stable.

ScreenHunter_151 May. 15 19.20.jpg

ScreenHunter_150 May. 15 19.10.jpg

Yes I need to setup the drivers on output of UC3879 who has not much drive power using extern transistors or mosfets.

regards
 

kees54

kees57
I have received the NCP1395AP resonant controller, this was for a supply for a audio amp hybrid, however i do not
now if it is suitable for a welding inverter also, looks like the feedback part is not usable for this.

It is a tv \pc\ supply controller.

For as you say about resonance has better behavior in shorted mode I think a faseshift do also well, because current protection
is easy to implement for all chips, however the arc is much more stable with a resonance version of welder it regulates much
better obove resonance.

regards
 
Last edited:

kees54

kees57
Here the version with fase shift uc3875 chip, these I go make pcb for, how do yopu like the feedback action in simulation? looks good to me.
sim do make the ripple because it do slow, in real world I think have a nice feedback.

ScreenHunter_193 May. 31 17.44.jpg

Only did get broken toroid for the LVI coil, but these big toroids has to high u anyway, but can be used as a nice output transformer for future tests like a smps for the audio hybrid amp.

DSC_0712.jpg
 
Top