Arc welder triac controler

[oldway]

I do not agree: 14V is too low voltage for MMA / stick welding.
You should use at least a 24V alternator, not a 12V one.
TIG and MMA DC inverters have an open voltage of about 80V to iniciate the arc easily.

MAG need a welding machine with constant voltage of low value and it should work well with this 12V alternator.

[moya034]

I do not agree: 14V is too low voltage for MMA / stick welding.
You should use at least a 24V alternator, not a 12V one.
TIG and MMA DC inverters have an open voltage of about 80V to iniciate the arc easily.

MAG need a welding machine with constant voltage of low value and it should work well with this 12V alternator.

I'm not so sure you actually read the site... he has a circuit which controls the signal going to the primary in the alternator. Being a constant current power supply when no current is flowing open circuit voltage will be higher until you strike the arc. He stated his open circuit voltage is right around 50v.

Commercial versions of these welders based on alterantors are sold and installed in lots of farm and off road equipment.

If you can strike an arc on TIG, you can strike an arc on stick. The PSU's are interchangeable (assuming you are doing scratch start TIG on steel, and no HF start or aluminum work )

[aburastas]

Rodrigopires: good to hear from you! If your circuit is good for control welding machines will be great! Because it's simplicity...

Oldway: uhmmm since you don't like my graphics capabilities... I'll try with my english grammar... Although it isn't much better... My big first question is: it's possible substitute the principal triac with a diode bridge and a scr triggered by an optotriac in order to switch always with the same current in both ways?

[oldway]

It is not a problem of graphics capabilities at all.

I hoped that you would at least have the decency to do new diagrams of a neater way in respect for yourself and for the readers of this forum.
You made a drawing filled with erasures and almost unreadable, you do not give a good impression of yourself.

Answering to your question: Schematic used by rodrigopires is very simple and I don't see how you could replace the triac by a diode bridge and an scr in such a diagram.
For beeing a very simple schematic, it has a lot of shortcomings.
one of them is the problem of safety.
Potentiometers isolation is not intended to withstand full mains voltage.
Use only full plastic potentiometers (axle and case) as PIHER PC16.

http://www.tme.eu/en/details/6pmi-2k2/carbon-single-turn-axial-potentiometers/piher/pc16sh10ipo6-2k2a/

[aburastas]

Well.. I hope I haven't offended nobody... That wasn't my intention. I thought that the previus sketch will suffice for a general idea. I hope the new schematic will be more clear... Without values. Only general idea. And sorry. I haven't near the pc so I haven't any cad program....

[oldway]

Yes, YOU CAN DO IT......      
(I knew it   )

You have only ofended yourself but it is allright now..."errare humanum est, perseverare diabolicum"

About your schematic: The original schematic of Rodrigopires is from an application note of SGS-THOMSON.
http://www.iascaled.com/docs/psu-400w/an308.pdf
It is intended to be a non isolated low cost circuit without transformer and without optocoupler.
If you choose the new 3 quadrants snubberless triacs, this circuit works very well with low power motors and transformers.
You must remember that Rodrigopires is making a welding machine with free MOT transformers and he is looking for the lower cost possible.

If you choose a more expansive solution with isolated control, there are hundreds, perhaps thousands of possible solutions.
To trigger the triac, you can use a triac optocoupler as a MOC3023 or you can use a pulse transformer
In this case, with inductive load, you must have a pulse train to trigger the triac.

In your schematic, you can't use a bridge rectifier with a SCR instead of a triac.
The SCR will probably stay fully conducting with inductive load without any control.
Choose a new 3 quadrants triac, they have a high dV/dt caracteristic adequate for high inductive load.
The photo-triac of the MOC3023 has a lower dV/dt, read the application note to see the right schematic to protect it from high dV/dt.
SCR need also a 470R resistance between gate and cathode.
All the schematics are generally comparing a dc adjustable voltage with a 100 or 120 Hz sawtooth sincronised with mains to generate the triggering pulses.
This may be done with µc, dedicated IC as TCA785 (absolete), OP amp, 555, ....
If you use bipol 555, you can't use 3 output pin to drive directly a transistor as you did, you need a resistance between base and emitter of this transistor.
You should also provide a softstart to prevent saturating the transformer.

[aburastas]

Ok... Once I'm exorcisted from my diabolic possesion we can continue with the discussion...

Really the scr with diodes bridge... Seemed to me a bit strange but... Always is better to know others opinion. Following your advices I have readed the literature you purpose and I've arrived to the following conclusions..

About the second 555 seem to be inefective since choping the triggering signal it isn't neccesary for opto trigger. Perhaps it's better to do a simple trigger signal until final of semiperiod. Don't you think so?

Another question is if you think it's neccesary to put the opto triac inside a diodes-bridge???

In the motorola application note an780 explain how to make the snubber for the optotriac depending of the cosphi of the load. I haven't this number of my welder... It's a cheap ac welder of 160A max so I'll suppose may be 0.5

About the soft start... I think that I have a solution for the monostable to start softly as for change the capacitor rate of charge softly also.. But I'll wait for your opinion before I'll make another "Super Clear" schematic... Jejeje

Really I have seen a lot of such schematics but the most utilize uC and I prefer the analogic solution and the others don't convince me.. So I prefer to do my own... With your help of course and if it isn't funtional i have learned somthing inbthe way.

[oldway]

For controlling AC with inductive load, you can use a triac or 2 SCR's connected in opposite.(anti - parallel)

Phase control can be very simple....you can do this with only 4 transistors.
(NB: in the schematic, OK1 is a triac optocoupler...I had no one in the EAGLE library)

[rodrigopires]

Hi, today i went trying the controller with two MOT's and a 100W LB. The standby amperage was from 0 to 5A. Using a 2.5mm electrode i adjusted the controller to the lowest power possible while still able to weld. But them i found a problem! No matter how low i put it, when the arc started the amperage jumped to 10 A! At 5A, 3A, 1A, it all jumped to 10A  What is happening here? 

[oldway]

You probably loose the control of the triac...it goes to full conduction by too high dV/dt or too high temperature.
What triac are you using ?
What temperature does it reach ?
What are the values of your rc snubber ?

[rodrigopires]

I'm using a BTA40-700B with a 220 Ohm and a 0.1uF cap. The temperature i dont know because i have it inside the box, but with two fans i think it wouldn't get too hot.

[oldway]

BTA40 is a standart triac, I recommand using a snubberless 3 quadrants triac with highly inductive load.
A 25A 3 quadrants triac is enough, you could choose a snubberless BTA25-600W

Your snubber is not enough for such a high current inductive load.
Try 22R 10W(or 15W) in serie with 1.5µF (3 x 0.47µF in //) 275Vac X2.

[aburastas]

Many thanks for your help! Who need uc??? As an old proverb says... "the more you know, the less you need" however studing your schematic... I think that the trigger mechanism (T1-T2) may fail to reset the capacitor C5 when noise is present in the line. The discharge time may not be enough. The 555 solution perhaps is better because it has an flip flop inside... When I Have a litte time I'll post my schematic idea.... I hope you like...

[oldway]

If you worry about the reset time of C5, you can increase it lowering the value of R3 as 1K5 or 1K instead of 2K2.
I used this circuit in low cost battery charger and never had any issue with that.

I made several trigger circuits using LMC555 (Cmos version of LM555), but the simpliest circuit is often the best and the more reliable.

One of the commun error of beginners is that they want to make very complicated design...and there is a good reason for this: it is easier to design an elaborate circuit than a simple one doing the same function because you must use more imagination and trick to succeed high performance with few simple components.

NB: as you can see with latest post of rodrigopires, you can't focus only on phase control circuit....power circuit is also critical....
For instance, photo-triac of the MOC2031 is a "weak" point because max dV/dt of this triac is low.
For controlling high power transformers, I prefer to use a pulse transformer with 2 secundaries and two thyristors head to tail.
Even when it is possible, avoid to make phase control in the primary of the transformer and make it in the secondary instead.

[aburastas]

Hi! I've been searching information about triggering scrs with pulse transformers but.... No luck.. Do you know any application note or similar that teach scr triggering with pulse transformers from begining? That would be geat!

[oldway]

Pulse transformers for triggering SCR's or triac:
http://fusiblesysemiconductores.com.mx/semikron/14access/Accessor.pdf
http://www.vacuumschmelze.com/fileadmin/Medienbiliothek_2010/Produkte/Kerne_und_Bauelemente/Anwendungen/Uebertrager/Ansteueruebertrager/Ansteueruebertrager_en_2012.pdf
You can also make it yourself with 10 turns isolated wire (primary) and 10 turns isolated wire (secondary) wound on a ferrite toroid.

ratio 2/1 with 24Vdc and 1/1 with 12Vdc power supply.
Pulses of 10µs with 20µs delay between pulses.
For high power SCR's, it is recomended to have no load pulses of 10V peak and short circuit pulses of 1A peak.
With low power and sensitive triacs and SCR's, this can be reduced a lot.

[aburastas]

Well... Here we go again! this is the power control unit I'm thinking. But before do the maths do you think that it can be funtional? PErhaps to put a 470 resistor between scr gate and catode has you sugested in an post before? And put a diode in serie with the scr's catode to protect them from reverse voltage conduction?... Or to put an diode in antiparalell to the mosfet?... What's your opinion?

[oldway]

Very good, that's allright.
Why a mosfet ? use a BC337 ! (low power mosfet are very sensitive to ESD, bipolar are more secure)
Zener diode is not necessary, you have 20µs to desmagnetize the transformer...with losses in 0.6V of the forward diode voltage, that's enough.
470R needed between G and K as you said.

R1 = 22R
C1 = 0.1µF
R2 = R3 = 10R
Diodes in the secondaries are not needed. (no high reverse voltage possible because you have a diode in the primary.)

NB: with triacs, use negative pulses for triggering (quadrants II and III) . Triacs don't like to work in quadrant IV...
Snubberless can only work in I, II and III quadrants, never in IV !

[aburastas]

Jeje.. We are close to the final solution but you say r1= 22r and.. It must supply 1A of secondary current plus the magnetizing current at 10us on time that depend of transformer inductance... So.. Maybe 1.5A. 12v/1.5A=8R...

[oldway]

I think you are not trying to trigger 1000A scr's....
This value of 1A peak is for high power scr's....

[aburastas]

Ok. Ok... I'm just begining with scrs... Searching for scrs I have found this reasonably priced "IXYS CS45-16IO1 Tiristor 1600V 100mA 48A 75A 4_32euro" and seem that with 100ma has a 10us delay until 300ma that has a flat line at 2us and Vgt=1.6. So assuming that R2 must be 1.6/0.3= about 5R?

The other thing that worry me is that in the scr search I've readed a semikron document that says that while the scr is blocking voltage you can't put gate voltage in order to avoid "a non permissible increase in off-state power losses and the formation of  hot spots in the thyristor chip"so perhaps the double secondary may be wrong and will be better to do the thing with two transformers and only send gate voltage to one scr in each half cycle.... What do you think about?

[oldway]

For low price scr's, look on ebay.de, there are a lot of them.
I like the "littlefuse", previous Teccor, TO218X isolated package S6065J (600V) or S8065J (800V) scr's but they are expensive.
I think that terminals of TO247 package are a little weak for high current.

For triggering, be aware to not exceeed the max Pgav (average gate power dissipation).
With CS45-16IO1, you need an Ig greater than 0.3Apeak to correctly trigger this scr.

For triggering circuit, better choose R1= 12R
R2 = R3 = 3R3
BC337 is perhaps too weak, use BD139, eventualy in darlington with BC337

you can't put gate voltage in order to avoid "a non permissible increase in off-state power losses

It is not recommended with high reverse voltage, but not forbidden...

Think a little ...
With 2 scr's connected head to tail is it possible that a scr has a gate signal and a high reverse voltage at the same time ?
NO because the other scr is conducting and reverse voltage is only about 2V ....

[aburastas]

Wow... Scrs seem to be more dificult that seemed in the begining... But expensive it is not my favourite word. As your experience and knowledge are priceless. We can do it from begining to avoid missmatchs. My welding machine has has a primary max curent of 26A and an "I1eff=8.4A" that I don't know it's mean and and primary Voltage of 230V. So. In order to search the cheapest correct scr what is the minimum Irms and Vblocking of the scr you recomend? I think that my safety margin is too much...

[oldway]

You have to respect certain principles into a project with scr's.

For example, always trigger the thyristor with sufficient gate current.

I worked in a large industrial electronics company where I developed a line of three-phase industrial battery chargers with full bridge of thyristors.
These chargers were manufactured up to 330V 600A.

My colleague was developing a single phase thyristorised battery charger for telecomunications.
I took a look at his project (he did not like that!) And I got him noticed that the trigger current was far too low.
He became angry, saying that it was working fine like this and I had to take care of it looking at me.
He did not modify his project and the charger was manufactured as is.

Result: The customer complained that the charger was not working in cold weather .... no wonder the scr sensitivity decreases with decreasing temperature ...

As regards the choice of the SCR's, there are many possible solutions between a small thyristor with a large heatsink or a large thyristor with a small heatsink.
This choice is also influenced by the possibility of protection by ultra fast fuse (ferraz example).

The i²dt of the UF fuse must be at least 30% higher than that of the SCR it must protect.

We often first choose the heatsink and the mounting type (1 scr/heatsink or all the scr's on the same heatsink).

Then we select a scr with higher rms current  than nominal rms current.
We calculates the junction temperature in the worst conditions and check whether it is compatible or not with a safe operation, including in case of overload.

Otherwise, we select a scr with a higher rms current or with less thermal resistance case.

For 230V, a 400V scr is enough with resistive load.
But with highly inductive load, a 600V or 800V is better choice.

[aburastas]

Well.... I'll assure you that where I live the scr need a lot of gate current in winter... Following your advices I arrived to the following conclusions.... Realy I hope I'm wrong...

To improve thermal dissipation I'd changed to a to220. LITTELFUSE  SK055R  Tiristor 1 kV 40mA 35A 55A TO-220 3,85euro

It has Vgt=1,5; Igt=40ma; Igm=4A; Tjmax=125C; Pgav=0,8; Rthj-c=0,5; Rthc-h=1,4 (screw, no grease, no mica)

It must pass 26A RMS but only half wave is 18,2A so it's dissipation will be 15W
To give a safe margin... We can put Tj=120C and Ta=35;

All that give us Rthh-a=(120 -35 -0,5*15 -1,4*15)/15= 3,77K/W

I find a 3.3k/w by 2.45euro...plus 3,85 of the scr is a reasonable value for each scr

Another conclusion is... Being D duty cycle. The pulse you purpose is 10us on 20us off so duty cycle is D=1/3. In that short on pulse must be D*Pg<=Pav

So Pg<=Pav/D Pg<=2,4W

Pg=Vgt*Ig; Igmax=2,4/1,5= 1,6A

I'm right for now???