Some questions about my SMPS TIG Welder build...

kees54

kees57
I have do a new sim with again a other zvs idea.

These looks good, even with extra induction on the emmitors of the igbt I do not see
strange things, looks very clean and it do zvs.

But without see photo two, now things go less nice even dramatic, zo zvs do avoid ringing and such.

ScreenHunter_668 Jul. 20 12.09.jpgScreenHunter_667 Jul. 20 12.07.jpg

These is a LVI coil who adjust himselfs with current and so get extended soft switching.

As you can see on the bleu line the igbt,s do switch or with zero current or zero voltage.leading or lagging leg.

ScreenHunter_669 Jul. 20 12.22.jpg
 
Last edited:

twenglish1

New member
I have been doing alot of reading about the subject, all types of topologies, i havent really had the time to prototype and play around with this kind of stuff lately, last work i did on the subject was i was working on the driver circuitry, i had the sg3525 circuit build, and working, and was working on a gate drive transformer and trying to use salvaged ferrites, i never did have a working gate drive transformer, that problem was found to be that i didnt use some sort of buffer between the sg3525 and the transformer, but once i get back into this i may go with an all silicon drive and leave out the gate drive transformers, i have some large EE80 ferrites and a whole box of 150amp, 1200v(i believe) igbt half bridge modules
 

kees54

kees57
I do not now if sillicon drivers do work fine in welding inverters because of the very wide current reach and fastness
to do so.

I have a schematic who do use them like this pdf.

ScreenHunter_673 Jul. 21 12.36.jpg

When driving gating transformer do aware about saturation, use a series capacitor with it, dependent on kind of drive circuit.

and a explaning how to make them.

http://www.ferroxcube.com/FerroxcubeCorporateReception/datasheet/gate drive trafo.pdf

regards

kees
 
Last edited:

twenglish1

New member
I have read that page from ferroxcube, lots of good information there and that schematic is similar to what i was planning on trying. I had the idea of using an SG3525 to drive two paralleled IR2110s(obviously ensuring they are phased correctly), is it possible to do it that way? i am not sure if this is a proper way to drive a full bridge, or if other means of driving it will need to be designed?
 

kees54

kees57
I do use gate transformers for welder, it is more save.

Oke, I have done some work and need advise, for welding I need fast respons of feedback but to fast will give trouble with stability, someone has some ideas?

I have a feedback system with current and voltage, so I can use both for mig mag tig and mma, just by put current pot max and adjust voltage or voltage pot max
and adjust with current.


I have a ripple of 10 amps on output when make feedback slow, but when faster it do give better respons until it go get unstable, I use comparators for it.

Last pictures let see soft switching nicely even with low amperage. controller is a fase shift version UC3875. the EA+ input gets some kind of variable duty cycled pulses
from feedback, and maybe this is not the right way and need really a analog version of feedback.

thanks

regards

ScreenHunter_879 Dec. 30 18.35.jpg

ScreenHunter_880 Dec. 30 18.36.jpg

ScreenHunter_881 Dec. 30 18.38.jpg
 
Last edited:

kees54

kees57
After some simming the past months I get to a conclusion that a serie parallel resonant circuit with a uc1861 works the best, even with low and high
load, and has high voltage when open output with also a stable behavior, these when keep inside the min and max frequentie of pwm.

I did the test with a normal 3875 chip, I have not yet found a uc1861 simulation model.

regards

kees

ScreenHunter_923 Jan. 24 13.53.jpg

ScreenHunter_922 Jan. 24 13.32.jpg

ScreenHunter_924 Jan. 24 13.54.jpg
 

kees54

kees57
Any update ?

Hi There


I the meanstime I did designing a amplifier circlotron and past months cancer with my brother who died a week ago.


I have to go on I do now but after searching for a resonance controller sim model like UC18** series who I can not find
for Ltspice I did not so much anymore.

Buit my old car is waiting with a rust devil inside so I must go on soon.

Thanks for your interests.

regards
 

picgak

New member
Something I wanted to add. Since the transformer calculator put the peak primary winding current at ~48A.
Hi,
As indicated by you the primary current is 48 A , but actually the current will be below 10A because your welder operates like in a short circuit condition at that time the output voltage will be down near say 5V @ 300A so the power will be 1500W max
 

kees54

kees57
Is it possible to construct an spotwelder with 1000 amps or more with a smps? because the voltage is just 1 or 2 volts but uge amps.

The best will then use a low switch frequentie for less stress, like 20 Khz and so it is more easy to get a lower voltage per winding.
 

blasphemy000

New member
Hi,
As indicated by you the primary current is 48 A , but actually the current will be below 10A because your welder operates like in a short circuit condition at that time the output voltage will be down near say 5V @ 300A so the power will be 1500W max

No. That is not at all true in any way, shape, or form. Arc welding machines just simply don't work like that. I've welded with many machines, both transformer and inverter, and no arc welders work like this.

TIG and MMA/Stick welders operate on a Constant Current control system where the output current remains rock solid while welding but the arc voltage will fluctuate depending on the length of the arc and what type of shielding gas is being used or type of flux powder on the rods(stick).
For example, when TIG welding on DC, with pure argon shielding gas, and using a standard arc length: 100A output will result in ~14V across the arc plasma. At 325A the output will be about 23V. Changing the shielding gas will change this, for example, using an argon/helium mix, or pure helium, will result in a higher arc voltage, and a hotter arc for the same output amperage. The hotter helium arc comes in handy for TIG welding thick aluminum on AC.
Stick welding on the other hand, will generally result in an arc voltage approximately 10V higher than TIG welding with argon at the same amperage. Again, this is dependent on the type of rods used(the different fluxes vaporize into different shielding gasses) and the length of the arc(different rods also require the use of different arc lengths). This is why if you go and look at the specs of any production switch-mode welding machine(Miller, Lincoln, and Everlast are some examples) the rated output current for welding in stick mode will always be less than the rated output current for TIG, while the voltage for stick will be higher than the voltage for TIG. With a little bit of math done on these ratings, the output wattage will be pretty close when maxed out in either mode. For example, the Everlast 325EXT is rated for 325A@22.9V in TIG mode while only rated for 250A@30V in stick mode. Both of these are right around 7.5kW.

As a general rule, the voltage when TIG welding with argon will be approximately: Vweld = Iweld * 0.04 + 10
For stick welding it is generally Vweld = Iweld * 0.04 + 20

Obviously these will only be approximations and since these welding modes are constant current, the voltage will fluctuate in order for the system to maintain a steady current output. These rules also don't work at very low currents because there is a minimum voltage require to maintain the arc across the gap through the shielding gas. These equations are pretty accurate in the amperage range where most welding is done.

On a separate note, I don't exactly follow your logic either. Short circuiting the secondary side of a transformer will cause the primary current to hit it's maximum amount. This maximum will be limited by stray resistance in the windings, how well the primary and secondary windings are coupled together, and whether or not the primary switching devices and input source(usually the mains) are able to supply that amount of current. It will essentially cause extremely fast saturation of the transformer core if the energy from the secondary cannot be dissipated somehow and things will go wrong very quickly. The current-controlled switching methods in a SMPS welding machine prevent meltdowns under short circuit conditions by monitoring the primary winding's current ramp when the switches are active. In the case of a short on the secondary, the maximum allowed primary current is hit very quickly and that switching cycle is terminated. The resistance of a TIG welding arc in argon is approximately 65-75mOhms, and while that might be considered a dead short in lower power systems, it's just normal operating conditions for a welder.


Is it possible to construct a spot welder with 1000 amps or more with a SMPS? Because the voltage is just 1 or 2 volts but huge amps.

The best will then use a low switch frequency for less stress, like 20 Khz and so it is more easy to get a lower voltage per winding.

You know kees, I'm not really sure if that's possible or not. I've never personally seen a SMPS spot welder, only transformer based ones. I can't think of any reason why it wouldn't be possible though.

All of the ones that I've ever seen run the AC output current directly from the transformer's secondary into the two electrodes. It is just resistance welding so there is no need to rectify the output. I've seen them built with amperage selection being done using either a tapped transformer or some sort of SCR based Phase-Angle Control on the primary to regulate the output current. You are correct about the voltage in that it is very low. Often the transformers only have a couple to a handful of turns for the secondary, even being large iron 50/60Hz units. Quite a few of the ones I've seen though have had magnetic shunts placed between the primary and secondary windings to limit the magnetic coupling slightly. I would assume these are there to help allow the secondary voltage to drop very low and draw a large amount of current without overloading the primary. The shunts would also limit the output current as well since they are limiting the magnetic coupling between the windings. I'm not sure if this would be needed in a SMPS style machine or how that would work though. Like I said, I can't think of any reason it wouldn't work in theory. Since the output can be AC, that would help simplify things quite a lot, although I'm not sure if the high frequency would have any effect on how the weld formed. The biggest problem I've seen with trying to build very low voltage power supplies is the rectification of such low voltages. Since no secondary rectification would be required for a spot welder, I would imagine getting the transformer correct would be the most difficult part.




On an note unrelated to these posts. I have finally gotten some free time between work and family and life in general that I have gotten back into researching and planning for this welder project. I know that it has been multiple years since I started this thread and I have essentially zero to show for it so far, but I'm hoping that this time will be different. The big thing now is that being able to afford the components to build such a machine is now within my reach so I'm hoping that will help me get this off the ground.
 

kees54

kees57
I have seen some rapports about resistance welding with a smps in resonance technology for low switch stresses.

I have being done a zvs inverter for mine migg welder who has a burnt transformer, however I not ready yet for
building it because I am busy with a circlotron amplifier.

I have found a russian program and pcb where it is programmed for differend welding actions but use a normal chip
with half bridge, and no zvs or resonance, I has plan to use mine with that controller because the feedback
of that chip is just like all are, a simple opamp in it with a ref voltage, maybe the speed of regulation is
different in a fase shift resonance bridge, and has to be altered, maybe the russian designer will help.


this is a sim of the inverter with quite a high speed of feedback regulation for constant voltage.

regards
 

Attachments

  • ScreenHunter_920 Jan. 20 20.27.jpg
    ScreenHunter_920 Jan. 20 20.27.jpg
    266.5 KB · Views: 27
  • ScreenHunter_898 Jan. 07 18.13.jpg
    ScreenHunter_898 Jan. 07 18.13.jpg
    321.4 KB · Views: 23

blasphemy000

New member
I don't know very much about resonant supplies. I understand the concept and the benefits of ZVS or ZCS switching techniques to reduce switching losses and reduce stress on the components, but I've never built a supply where it was worth it to go about the added complexity of a resonant topology.

I have a question about your simulation pictures. The snubber capacitors across the IGBTs (C5, C7, C8, C9 in the first picture) are not in series with low value resistors as is normal for snubbers. In the scope view their are current pulses through C5(red trace) that are over 10A when that side of the bridge switches. I'm guessing that this would cause a large amount of unwanted heat dissipation and energy loss in those caps. Is there a reason that you've left out the resistor that is normally in series with each of the snubber capacitors? Normally the snubbers are there to help reduce or eliminate high-frequency ringing through that part of the circuit.
 

kees54

kees57
I don't know very much about resonant supplies. I understand the concept and the benefits of ZVS or ZCS switching techniques to reduce switching losses and reduce stress on the components, but I've never built a supply where it was worth it to go about the added complexity of a resonant topology.

I have a question about your simulation pictures. The snubber capacitors across the IGBTs (C5, C7, C8, C9 in the first picture) are not in series with low value resistors as is normal for snubbers. In the scope view their are current pulses through C5(red trace) that are over 10A when that side of the bridge switches. I'm guessing that this would cause a large amount of unwanted heat dissipation and energy loss in those caps. Is there a reason that you've left out the resistor that is normally in series with each of the snubber capacitors? Normally the snubbers are there to help reduce or eliminate high-frequency ringing through that part of the circuit.

Hi

The capacitors over the igbt,s are because of resonant used as temperary energy storage so the igbt can switch zvs, or zcs.

I have test it a time ago, for what concerns the feedback these are not much different between normal hard switching and resonant or
fase shift resonant where capacitors over igbt do act as part of whole resonat system, because it is not hard switch there are no
normal snubber present.

L1 and L4 on secondairy of output transformer refelct this energy back during death time to discharge these capacitors so igbt can switch on without
energy over it.

There is a resonant topology who do even better, using a VCO and min and max osc frequenty but I have not a model for the UC1861 controller and
I can also do not find one.

regards
 
Last edited:

blasphemy000

New member
After doing a little bit of reading and research, I've found a few of ZVS Phase-Shift schematics without any snubbers across the switches. On the other hand I've only been able to find a couple that had capacitors directly across the switches. With the low impedance path through the switches, the midpoint between them will always either be high or low. I just don't understand how those capacitors could be doing anything other than generating heat from the constant charging and discharging they are suffering as the IGBTs switch back and forth.

In your first schematic, I guess I could see where they could possibly serve a purpose since the UC3825 is a standard(non-resonant) PWM controller, and there is dead-time that is inversely proportional to the duty-cycle of the outputs, and both switches are off during this time.

With the UC3875 or UC3895 though, each leg of the bridge is always being switched at 50% DC so there is only enough dead-time to prevent shoot-through and the duty-cycle of the transformer is realized due to the phase-shift. In this case, I don't see where those capacitors would be of any use at all.

If that is the case with the first schematic, I'd venture to say that there should be a better way to achieve the resonance in the circuit. As I said though, I'm not very knowledgeable on resonant topologies, as I have never had a need to use one.
 

kees54

kees57
Hi All

Some time ago here but my ****** of 85 years old was sick and in hospital for 7 weeks.

I have done some sim for a feedback for making the inverter usable for a mig, but I did see that when sim it the output current do swing between 120 and 180 amps, things
are independent of type of opamp, fast or not for good respons, however I think this is because of the simulation, I need just to sim and when work to build and test then
what happens.

see the pictures for some info. the middle picture is what I want to see, because I think that is the proper feedback and delay in poles.

The capacitors who are over the mosfets or the igbt is for zvs switching, the leakage of the transformer or the resonant coil do discharge caps for ZVS or charge (lagging leg) for ZCS while freewheeling and in dead time, but with low load cap do not discharge completely and the it discharge over the transistor switches, new topologies do get better but more complicated.

I build a weld inverter so soft switching is nice, the topology I go try is with a resonant LVI coil who I did find info on in the internet. LVI coil do lower iduction when current rise, that way we get a very broad soft switching behavior, with fase shift controller it get a lot more easy, the UC 1861 was a resonant controller who is still used but seems not easy to feedback it stable.

regards
ScreenHunter_139 May. 13 15.14.jpg

ScreenHunter_143 May. 13 16.31.jpg

ScreenHunter_141 May. 13 15.39.jpg
 
Last edited:

kees54

kees57
The schematics you did see without snubber capacitors do use the body capacitance of mosfets, for igbt this is not the case and using snubber caps,
sometimes only for leading leg, but that is is tuned resonance bridge for a UC1861.

ScreenHunter_144 May. 13 17.01.jpg
 

legg

New member
kees54, your LTspice file, included in phase shift-UC3879.zip, refers to a spice model for UC3879, for which a subciruit and assy file is needed to run.
Do you still have these? AFAIK, unencrypted spice models for TI chips are not commonly available.
 

kees54

kees57
Hi There


Here is the file contains smps controllers of all kind.

View attachment ValVol.rar

Is these days the UC3865 resonance controller not more used as faseshift? or is faseshift the new way of making smps inverters,
because the resonant types are quite nice because of sinusiodal excise of output transformer with frequentie modulation giving
very low EMI. I did see a lot of welding inverters who did use that topologie but most are old designs from end 1999 and older
some somewhat newer, I do see now in recent designs the fase shift is more used.

regards.

kees
 
Last edited:

legg

New member
kees, I'm more interested in the modulator's performance - whether or not the model predicts certain issues with later versions of TI/Unitrode's phase modulators. The UC3879 hasn't been recommended for new designs for some time.
Do you have any idea where this model came from, or who Valentyn Volodin was/is ?
 
Top