Basics - AC Load swtiching. Thyristor or TRIAC

[anishkgt]

I Have a project that uses TRIAC to switch on an Inductive load, here a MOT (Microwave Oven Transformer). I switch it ON using arduino after detecting the zero crosssing. Since the load is an inductive load, the current lags behind the voltage and using an Arduino i have programmed it to switch of at the next zero cross but the since the load is inductive it does not switch off at the zero cross but a little latter. Since the voltage can only be controlled by the principle of a switch. The current is not controllable.

Is there anything like the TRIAC that can switch OFF an inductive load as soon as the gate current is stopped. Would a Thyristor help ?

[David Hess]

A TRIAC *is* a thryristor.  So is an SCR, DIAC, PUT, and other 4 layer devices.

What you want is some kind of transistor in a circuit that allows it to be used as an AC switch.  A power MOSFET or insulated gate bipolar transistor would be suitable and are commonly available in the form of SSRs (solid state relays) for exactly this type of application.

https://www.eevblog.com/forum/projects/mosfet-ac-ssr/

Watch out for the inductive kickback from the transformer when shutting it off.

[oldway]

One of the weak points of commun triacs is their low dV/dt. (2 to 5V/µs)
If dV/dt applied is greater then this max dV/dt, the triac switch on again without gate pulses and it goes to full conduction, loosing control.
An SCR has a max dV/dt of 500 or even 1000V/µs.
For this reason, we always prefered to use 2 SCR's in anti-parallel for controlling highly inductive load.
Gate drive was make with a pulse transformer with two secondaries.
But there is a new generation of triacs named "snubberless" or 3 quadrants triacs which has high dV/dt (500V/µs)
I recomend to use this kind of triacs to control the primary side of transformers.

[Cliff Matthews]

But there is a new generation of triacs named "snubberless" or 3 quadrants triacs which has high dV/dt (500V/µs)
I recomend to use this kind of triacs to control the primary side of transformers.

Thanks! I found a good read on that here: www.st.com/resource/en/application_note/cd00003865.pdf
and here! http://educypedia.karadimov.info/library/AN_3Q_TRIACS.pdf

[tautech]

Also check out Alternistors, a type of Triac sometimes used in the older transformer type Mig welders.
One I fixed used an Alternistor on a TO-3 package with push-on terminals.

An IC like U2008B (DIP-8) makes Triac control quite simple.
https://www.digchip.com/datasheets/parts/datasheet/000/U2008B-A-pdf.php


Also attached is a reverse engineered schematic of a single phase welder I used to have that used 2 SCR's in anti parallel like oldway recommends. To build it you must have a scope and appropriate channel isolation techniques in order to "tune" the Gate timings for the SCR's to share the load. The Unijunction transistors 2N2646 are getting hard to find though.
This welder had a 1/8" copper wire primary and 3/8 x 1/8 copper strap secondary, NO fan and you could weld with it all day @ full power. 
Twas the only 180A traditionally constructed welder I've been able to lift with one hand.

[anishkgt]

For this reason, we always prefered to use 2 SCR's in anti-parallel for controlling highly inductive load.
Gate drive was make with a pulse transformer with two secondaries.
But there is a new generation of triacs named "snubberless" or 3 quadrants triacs which has high dV/dt (500V/µs)
I recomend to use this kind of triacs to control the primary side of transformers.

I used TRIAC https://www.digikey.com/product-search/en?keywords=497-4172-5-ND. It is an Alternistor and Snubberless. Checked to find that the TRIAC by nature does not switch of exactly zero cross but after few milliseconds show here https://georgehobby.files.wordpress.com/2016/11/scopeview.png?w=580&h=348. None the less it worked well. So now thought i'd do an update on the circuit and implement a Thyristor with this https://www.digikey.com/product-search/en/discrete-semiconductor-products/thyristors-scrs/1376387?k=&pkeyword=&pv842=238&FV=1140050%2C1f140000%2Cfff40015%2Cfff80083%2C142c0092&mnonly=0&newproducts=0&ColumnSort=0&page=1&quantity=0&ptm=0&fid=0&pageSize=500 in Anto-Parallel just like The oldway.

i've attached the schematic with the the Anti-parallel, would this be ok ?

[oldway]

BTA26-800BWRG is a very good choice as good as 2 SCR's in anti-parallel.
It has a dV/dt of 1000V/µs.
I did not understood why you use anti-parallel thyristors in your schematic.
One of the 2 SCR's has no gate drive.
This schematic is ok with a triac, not with 2 scr's.

[anishkgt]

yes, the said TRIACs were used earlier. To actually move to the THYRISTOR was because the current being lagged behind at switch off. Would i be wrong in considering the two THYRISTOR idea ? since THYRISTORS switch off at the negative half cycle, i was thinking this would be ideal.

The Actually missed the second opto driver. So if i need to add a second opto driver should it parallel to the first opto or parallel to pin 4 ? would it ok to just connect the Gate of T2 parallel to T1 ?

The idea of two thyristors came to me from this circuit

[anishkgt]

I had the circuit done wrong, here is the corrected. I was thinking of using this THYRISTOR http://www.digikey.com/product-detail/en/stmicroelectronics/TN1610H-6T/497-15951-5-ND/5400217

[Zero999]

I had the circuit done wrong, here is the corrected. I was thinking of using this THYRISTOR http://www.digikey.com/product-detail/en/stmicroelectronics/TN1610H-6T/497-15951-5-ND/5400217

Hang on, that's not right either. R13 is now in series with the anode & cathodes of the SCR and load, so it will smoke when it's switched on.

[anishkgt]

Yup, your rite. Checked the datasheet of MOC3041 and i've corrected it now. How do i know if the MOC3023 and MOC3041 would be the same. Not talking about the datasheet comparison. Both are TRIAC drivers so would it be ok ?

[anishkgt]

ok figured it out. a quick look at both datasheets showed that MOC3041 has an inter Zero Crossing Circuit and MOC3023 is a random-Phase TRIAC Driver.

[oldway]

We are in the beginners section and therefore, it is normal that you make mistakes.

Let us put a little order in the ideas:

There are 2 ways to control the primary of a transformer:
- or by phase control
- or by applying the full waves and controlling the time where the voltage is applied to the transformer

You should first select the right option.

The phase control is of general use and allows to vary the effective voltage applied to the transformer from 0 to 100%

There are some conditions to respect:

- The control symmetry must be perfect so as not to create a continuous voltage that could saturate the transformer.
- Precautions must be taken to ensure that the triac or thyristors (pulse train or wide pulse) will conduct and stay conducting. (they may be turn off if current is to low and that gate signal is not applied)
The triac or scr's must have a large dV / dt to avoid re-conducting after zeroing the current and avoid loss of control
I recommend choosing either 2 scr's in anti-parallel (complicated solution) or a triac "snubberless" and provide a rc even on the "snubberless" triacs.
Do not forget that opto-triac is a weak point at dV / dt point of vew and that it needs a very efficient RC.

- the control by "train of sine waves" of variable length.
This is only used in electric heating applications where the load has a high inertia
What there is to know:
- the initial firing of the triac must not take place at the moment of the 0V passage of the voltage but at the moment of the passage of current by zero, which is between 45 and 80 degrees lagging on the passage of the voltage by 0.
This phase shift must be adjusted for the minimum inrush current during turn on.
If possible, there must be a logical order between the polarity of the turn on and turn off of the triac, the polarity of the next turn on must be  the reverse polarity of the last turn off.

I see you want use this controle for a spot welder....With or without current controle ?
You can apply full voltage to transformer and only controle time or controle applied voltage and time...What did you choosed ?

[anishkgt]

Thanks oldway i owe you one 

Yea my intention is to control the voltage and time. For that the TRIACs don't seem to be any good when it comes to cutting of at zero cross after tunred on at the peak of the Sine wave. Two SCRs in a back-to-back configuration solves the cut off time as SCRs by design turn off when there is a polarity change. Hence will switch off at the negative cycle.

If i could control the current too, then that would be advantage and something new for me to learn. Just a doubt would the SCRs i mentioned earlier be suitable for my setup ?

Regards,
Anish

[bktemp]

Yea my intention is to control the voltage and time. For that the TRIACs don't seem to be any good when it comes to cutting of at zero cross after tunred on at the peak of the Sine wave. Two SCRs in a back-to-back configuration solves the cut off time as SCRs by design turn off when there is a polarity change. Hence will switch off at the negative cycle.

No.
SCR and TRIACs behave identical, except a TRIAC conducts in both directions. So a TRIAC is basically two SCRs connectec in antiparallel.

SCRs and TRIACs stop conducting when the current is zero, so for AC at every zero crossing.
For resistive loads, both voltage and current zero crossing are at the same time.
For an inductive load, the current lags by 90°. A real transformer is somewhere between 0° (large resisitve load) and 90° (zero load), but you don't know unless you constantly measure the current and compare it to the voltage (or at least measure the voltage drop across the SCR/TRIAC to determine if it is turned on).

If i could control the current too, then that would be advantage and something new for me to learn. Just a doubt would the SCRs i mentioned earlier be suitable for my setup ?

You can't control the current using SCRs, you can only switch on for the remaining halfcycle and so adjust the avarage current. If you want an adjustable current you need to adjust the load resistance or your voltage.
In a welding transformer this is typically done by inserting a magnetic shunt into the transformer (changing its leakage inductance).
I don't say it is completely impossible, but adjusting the current of a transformer electronically is difficult.

[anishkgt]

This link here http://www.avdweb.nl/arduino/hardware-interfacing/spot-welder-controller.html got me to believe that TRIACs would not be the best way to go and actually on a scope there is about a few ms before the triac stopped conducting after zero crossing. So i was locked on to go for the 2 Thyristor configuration and what he says does make sense. The Thyristor stops conducting on every half cycle (meaning every time the polarity changed) that way i could be sure that it would stop conducting at every cycle, which is an advantage here. Secondly since TRIACs conduct both ways there could be a chance that it could conduct the next cycle as well.

Just clearing my doubts between the comments here and the article in the link above, correct me if am wrong. I already had this project completed with the TRIAC and that was when i read more about the article and did some reading on the web about THYRISOR. But nowhere was it mentioned about the turn off problems with TRIACs compared to THYRISTORs.

Attached the TRIAC configuration just in case somebody needs it latter.

Regards,
Anish.

[bktemp]

This link here http://www.avdweb.nl/arduino/hardware-interfacing/spot-welder-controller.html got me to believe that TRIACs would not be the best way to go and actually on a scope there is about a few ms before the triac stopped conducting after zero crossing. So i was locked on to go for the 2 Thyristor configuration and what he says does make sense. The Thyristor stops conducting on every half cycle (meaning every time the polarity changed) that way i could be sure that it would stop conducting at every cycle, which is an advantage here. Secondly since TRIACs conduct both ways there could be a chance that it could conduct the next cycle as well.

SCRs work better in this application, but 3 quadrant TRIACs (Alternistors or snubberless TRIACs) should also work.
SCRs simply make your life a bit easier, because they are more tolerant to voltage transients, but if you screw up the circuit design completely, you will get the same problems even when using SCRs.
If you design the circuit correctly, a TRIAC will turn off at every zero crossing, the same way a SCR does.

The description in the link doesn't tell the whole story, therefore it is a bit misleading.
It says solid state relays do not work with inductive loads. That is correct, but the actual reason is, most solid state relays have a zero crossing detection circuit buildin.
You can get solid state relays without the zero crossing circuit. They work fine even with inductive loads.

Here is an application note on how to use TRIACs for driving inductive loads (microwave oven transformers):
http://www.st.com/content/ccc/resource/technical/document/application_note/2f/f7/75/f5/9d/3d/44/04/CD00003867.pdf/files/CD00003867.pdf/jcr:content/translations/en.CD00003867.pdf
It shows that it can and has been done.

[bktemp]

You circuit looks ok, only some minor improvements:
According to the MOC302x datasheet R9 and C8 should be larger (100nF, 1k), because the TRIAC used in the optocoupler is much more sensitive than the main TRIAC, therefore it needs better filtering (by the way, that is a drawback if you use two SCRs instead of a TRIAC: It is more difficult to add a filter circuit for the optotriac).
D10 is useless at this place. A optocoupler shouldn't generate any voltage spike, but if you want to keep it, move it to the other side of R6.
If you don't do any filtering of the sync signal in software, a bit of low pass filtering and a stronger pullup than the internal one can help to avoid spikes triggering the zero crossing detection (and there are spikes because of the high currents involved). Add a small capacitor (1nF) between R2 and the AVR.

[anishkgt]

I've changed/added pars according to your suggestion. Now how did you come up with those values ? .

[anishkgt]

Forgot to attach the file.

Now with those changes, would there by any changes in the performance.

[bktemp]

The snubber values for the optotriac are taken from the datasheet (page 4, Figure 7):
http://www.ti.com/lit/ds/symlink/moc3022.pdf
It suggests even larger values (1.2k + 200nF), but using too large values generates other problems (it introduces a time delay and makes retriggering right after the zero crossing difficult for resistive loads).
The 1nF capacitor at the zero crossing detection circuit must be small enough to not introduce any significant time delay, but must be large enough to reject noise and spikes. I have taken the value from the original schematic. You often see 100pF-10nF being used on digital inputs in industrial applications to reject noise.

[anishkgt]

Thanks that was something new, i followed this https://www.fairchildsemi.com/datasheets/MO/MOC3023M.pdf from Faichild. From Page 8 Figure 13. I guess there is nothing like " which is true and which is not " it is just out of trial and error. But which value of which component do i change ?

[bktemp]

It is difficult, because your load can be anything between purely inductive (open load) and purely resistive (output loaded).
Therefore the snubber must be good enough to handle the transients of the inductive load, but must also be able to turn on resistive loads near the zero crossing. So the resistor must not be too large, but the snubber must reduce the slewrate at turn off enough for the optotriac to not trigger due to high dV/dt.
Because we need to switch on resistive loads, we can't increase the resistor further, so we need to increase the capacitor. But this will increase the leakage current of the circuit.

Often there isn't one perfect solution, instead you have to find the right balance.
In your case the best solution is probably somewhere between Figure 12 and  Figure 13. That's why I've suggestes 100nF for the capacitor. Using values between both circuits is a good starting point. If you experience problems, you can optimize the values. Depending on the gate current requirements, maybe 1k is too high and you need to reduce it but increase the capacitor.

[anishkgt]

Using values between both circuits is a good starting point. If you experience problems, you can optimize the values. Depending on the gate current requirements, maybe 1k is too high and you need to reduce it but increase the capacitor.

hmmm sounds a little hard as these capacitors are not available locally, have to order them through ebay or digikey. 1K resistors are ok to get. But i have a few of them caps ranging from 0.01uf~4.7uf. Is there any application or a tool to calculate the numbers on these Metallized Film Capacitors.

It is difficult, because your load can be anything between purely inductive (open load) and purely resistive (output loaded).

My load as earlier mentioned is an MOT (Microwave Oven Transformer) which is an Inductive load.

[anishkgt]

About the 2 thyristor setup are there anything more that i should be looking at or adding into this like earlier mentioned a cap or a resistor. was thinking of ordering two of these Thyristors https://media.digikey.com/pdf/Data%20Sheets/ON%20Semiconductor%20PDFs/MCR16NG.pdf and the other components in bulk about 100 each. The latest is attached. In the datasheet for 2 Thyristor config there is a resistor and a diode connected on Pin 6. i can understand the use of the resistor, current limiting for the opto but what does the diode do here ?