The care and feeding of Triacs.

PhilHornby

Senior Member
In the next phase of my project, I want to trigger two Triacs to control two 1KW heater elements. The elements are physically inter-connected and the centre point is currently used as the Neutral. The elements will be switched on either together, or separately for several minutes at a time.

I've never fully understood Triacs, and there seem to be different ways to connect them to the load. The following diagram shows the methods that seem the most common :-

TriacLayouts.jpg

Given that I have two Triacs, and two loads connected via their Neutrals, "Fig A" would seem to be the natural choice of configuration. However, since the opto-isolator will be triggered continuously for several minutes at a time, I am worried about the wattage rating of the resistor. (I can see that in "Fig C", when the Triac conducts it takes away the current flowing in the resistor - but I don't fully understand what happens in "Fig A" and "Fig B".)

If necessary, I can rewire the heaters so their common point is Live, rather than Neutral, but I'd prefer not to.
 

techElder

Well-known member
Phil, draw your triacs as two "back to back" SCRs. That may simplify it enough to make it more clear. You might even go so far as to draw the SCRs as a stack of diodes for clarity.

Only look at what happens during a single phase of the AC.
 

Reloadron

Senior Member
I see you are in the UK so I will assume 220 Volts mains voltage powering your two 1.0 KW Heaters, Not to discourage you building your own TRIAC switching circuits but you can likely just buy off the shelf a pair of 250 / 600 Volt SSRs rated at 10 Amps and using a 3 to 32 Volt control voltage, all in a single neat package for about maybe $15 to $20 USD each. Names like CRYDOM, OPTO22. NTE, OMRON, and others come to mind globally marketed, Everything in a potted turn key solution. With the 3 to 32 Volt control voltage range they work great with micro controllers or other 3.3 or 5 volt logic. Since the loads are 1 KW heater elements this makes for easier on/off control if On/Off control is all you need. Heater elements will be a resistive load. Anyway, considering time and cost as well as easy replacement I would consider using off the shelf SSRs (Solid State Relay). 240 VAC 10 Amp SSR or using a dew TRIACS you can roll your own.

Here is an Amazon example at less than $10 USD each with heat sink rated for 25 Amps.

Ron
 

Reloadron

Senior Member
+1 for ready built SSRs ( as long as they are from a decent supplier ! )
Oh yeah, you want to stick with a reputable supplier and brand name like the names I mentioned. Some not so good SSRs have been coming off the boat. :)

Ron
 

hippy

Technical Support
Staff member
However, since the opto-isolator will be triggered continuously for several minutes at a time, I am worried about the wattage rating of the resistor.
I don't know a lot about triacs but it may be that you don't have to be worried, though you are right to consider the issue.

Although the opto may be on for most of the time does not mean that the current through it or the resistor will be or, even if it is, that it will be large enough to cause a problem.

I agree; you would need to know the resistance and current which flows and that doesn't help towards that. The triac datasheet may indicate what the trigger current is, you could measure it, and perhaps look at similar circuits to see what they say about it. From the circuit I have for a commercial switching unit -

Code:
         .-------------------.---O==O-- L
     .---^--.                |   Fuse
 ====| OPTO |                |                
     `---.--'    ____    .---^---.
         `------|____|---| TRIAC |
                 150R    `---.---'
                             |
                            ( ) Load
                             |
                             `--------- N
No notes I could see about wattage for that 150R. That seems to be the same for a later circuit though that adds two back-to-back 2V7 Zeners in series with the 150R, though only presumably to keep from prematurely triggering the Triac.

Looking at Google images of "triac circuit board"; most don't seem to be using substantially large wattage resistors.

You can parallel two 300R to create a higher wattage 150R.
 

PhilHornby

Senior Member
It's just occurred to me, that I can test these configurations using low-voltage DC (manually reversed), rather than risking life and limb with the AC mains. I'll swap the opto-triac in each configuration for a simple switch, and measure the DC current that flows.

I'll report back...
 

Reloadron

Senior Member
If you are talking about the resistors labeled R1, R2 and R3 in figures 1, 2 and 3 their purpose is to limit the triac gate current. Really matters not the On Time of the triac be it a few min or a few days powering the load. The actual task of these resistor examples is that of limiting the peak instantaneous switch-on gate current of the triac to a safe value; its resistance (combined with that of the load) must be greater than the peak supply voltage (roughly 350V in a 240V AC circuit, 175V in a 120V circuit) divided by the triac's peak gate current rating (which is usually given in the triac manufacturer's extended data sheets). You may want to give this article about triacs a read.

Ron
 

fernando_g

Senior Member
All three circuits are functionally equivalent.
The only difference between A) and C) versus B), is that in the former once that the load turns on, the voltage across the resistor/optocoupler combination drops to very close to zero, significantly reducing their power dissipation.
 

PhilHornby

Senior Member
If you are talking about the resistors labeled R1, R2 and R3 in figures 1, 2 and 3 their purpose is to limit the triac gate current. ... The actual task of these resistor examples is that of limiting the peak instantaneous switch-on gate current of the triac to a safe value; its resistance (combined with that of the load) must be greater than the peak supply voltage (roughly 350V in a 240V AC circuit,...) divided by the triac's peak gate current rating ... You may want to give this article about triacs a read.

Ron
I'd looked at that article before, but hadn't noticed that there's a "Page 2" - which is where it starts to get more interesting. Yours is the first attempt I've seen at actually working out a value for the 'gate' resistor - however, it is different to that used on the page you referred me to. On Page 2, Figure 2 is the same as my "Fig 3" above. "Nuts & Volts" say of the resistor value:

Nuts&Volts said:
Note in Figure 2 that R1 is used to limit the peak switch-on current of the optocoupled triac (and thus the peak gate current of Q1) at IC1's absolute maximum VIH value minus 2V, i.e., typically at 23V; with the R1 value shown, the peak switch-on current is limited to 280mA.
In other words, they have used only 23Volts, rather than 350Volts (quite a difference :confused: ). They seem as concerned with protecting the Opto-coupler as the Triac. I need to study that in more detail and look at some datasheets.

In the meantime, I re-jigged my "Fig 3" circuit and took some measurements:-

Triac Layout 3.jpg Fig 3 Voltages.jpg

"The more I find out, the less I know" ;)
 

Reloadron

Senior Member
I'd looked at that article before, but hadn't noticed that there's a "Page 2" - which is where it starts to get more interesting. Yours is the first attempt I've seen at actually working out a value for the 'gate' resistor - however, it is different to that used on the page you referred me to. On Page 2, Figure 2 is the same as my "Fig 3" above. "Nuts & Volts" say of the resistor value:
You want to take note that the gate resistor value depends on the voltage we are switching. You will see a big difference between for example a common US circuit of 120 volts and a UK circuit of 220 volts. You are only looking to turn the heaters (load) ON / Off correct?

Ron
 

PhilHornby

Senior Member
You want to take note that the gate resistor value depends on the voltage we are switching.
That's a bit at odds, with the regard to the measurements I have taken so far though ...

My previous measurement, using a 6.3VAC source:

Fig 3 Voltages.jpg and the same circuit with a 37VAC source Fig 3 Voltages Big Transformer.jpg

The scale of the gate resistor voltage trace is the same in both images; the scale of load voltage trace is different). Unfortunately, I used a different timebase for each test. The only changes to the circuit, apart from the input voltage, is that the load resistor was increased from 100R to 1K and the LED dropping resistor was upped to 1K from 10K.

The voltage across the gate resistor is clearly not sinusoidal (or even symmetrical); I'm going to struggle to apply Ohms' law to that!
 

hippy

Technical Support
Staff member
The voltage across the gate resistor is clearly not sinusoidal (or even symmetrical); I'm going to struggle to apply Ohms' law to that!
Just take its peak voltage, then you have the worst case it could ever be.
 

PhilHornby

Senior Member
Varying supply voltage

Just take its peak voltage, then you have the worst case it could ever be.
That's probably sensible - once I have satisfied myself that it's a 'fixed' Triac-related parameter, rather than supply-related.

I repeated my measurements, with different supply voltages - but managed to keep the scale and time-base the same this time. The Rise-time of the gate resistor voltage seems to change marginally, but the maximum value is the same (from 18Vrms -> 55Vrms). There's an interesting change in timing, if the load is insufficient to allow the Triac to latch:-

Triac voltages with different supplies.jpg

(PS - I note the forum software is better behaved these days, wrt to uploading images!)
 

hippy

Technical Support
Staff member
I note that you are getting a gate pulse at the end of each cycle, presumably because you wait for a zero crossover, set the gate via the opto and the gate conducts until the trigger point is reached when the gate stops conducting. At the end of the cycle, it falls below the trigger point and, because the opto is still on, the gate starts conducting again, until actually at the crossover..

It may not be absolutely necessary but one trick to avoid that is to trigger, wait half a cycle, then turn the opto off, only turn it on again at the next crossover. Two benefits there are it makes it even more clearer where the actual triggering is, and also reduces the average current flowing.

Thinking about those back-to-back Zeners mentioned earlier; adding those will also reduce average current because there won't be any until the trigger voltage gets closer to the triac conducting. You will also end that risidual current on the gate which continues through the cycle if you choose a zener higher than that voltage.

If you have one or two Zeners it might be worth playing with those.
 

PhilHornby

Senior Member
I note that you are getting a gate pulse at the end of each cycle, presumably because you wait for a zero crossover...
No, these waveforms are generated with the opto-coupler out of the circuit, using the simple circuit shown here:

Triac Layout 3.jpg(The gate resistor voltage being measured across TP1-TP2 and the load voltage TP1-TP3)

I did have a play with triggering a MOC3041 (zero-crossing detect opto), using a single 500µS pulse just after zero-crossing, but I don't really want to tie up another Picaxe just doing that job. It also struck me, that there was absolutely no point using a MOC3041 - a non-zero crossing MOC3021 gave exactly the same results.
 

hippy

Technical Support
Staff member
True; in the test set-up you have you can't so easily gate the triggering.

I would not agree that there is no point using a MOC3041, a non-zero crossing MOC3021 giving exactly the same results. That's only true in the case where it has been turned on and remains turned on.

What the zero-crossover switching avoids is switching the triac on mid-cycle, causing in-rush current, spikes, noise and all sorts of related problems.

You may however be able to achieve that with software, ensuring it never happens, only switches at the zero crossover.
 

PhilHornby

Senior Member
I would not agree that there is no point using a MOC3041, a non-zero crossing MOC3021 giving exactly the same results.
I mean that there's little point using software to drive a MOC3021 to give the same functionality that's already present in the MOC3041, which is effectively what I'd ended up doing. For heater control, I do want zero-crossing turn ON/OFF.

I had a look at the gate voltage (& therefore current), in my original Fig A configuration, reproduced here in simplified, switch controlled form:-

Triac Layout 1.jpg This is representative of the configuration I would ideally like to use.

Again, the voltage across the resistor does not appear to change with supply voltage. It is the same (mysterious) 2 and a bit volts, across 100R - I wonder what sets it at that particular level?.

I've included the resistor voltage traces for 6.3,18,37 and 55Vac - but you can't really tell them apart.

Fig A voltages.jpg

It's worth noting, that I cannot get close to 2V across the gate resistor, when using DC. I get about 0.4V or 0.38V, depending on the polarity, but regardless of supply voltage. That seems odd!

UPDATE

With hindsight, that's because it triggers immediately and the gate current drops - taking the resistor voltage with it. Effectively, this is the region after the initial peak, in the above graphs
 
Last edited:

rq3

Senior Member
I mean that there's little point using software to drive a MOC3021 to give the same functionality that's already present in the MOC3041, which is effectively what I'd ended up doing. For heater control, I do want zero-crossing turn ON/OFF.

I had a look at the gate voltage (& therefore current), in my original Fig A configuration, reproduced here in simplified, switch controlled form:-

View attachment 21543 This is representative of the configuration I would ideally like to use.

Again, the voltage across the resistor does not appear to change with supply voltage. It is the same (mysterious) 2 and a bit volts, across 100R - I wonder what sets it at that particular level?.

I've included the resistor voltage traces for 6.3,18,37 and 55Vac - but you can't really tell them apart.

View attachment 21544

It's worth noting, that I cannot get close to 2V across the gate resistor, when using DC. I get about 0.4V or 0.38V, depending on the polarity, but regardless of supply voltage. That seems odd!
The triac is basically two back-to-back silicon controlled rectifiers (SCR). It will always measure roughly two diode drops to any terminal under load. Think of placing a single diode directly across the AC mains, with enough series resistance to keep the diode from becoming destroyed. The voltage measured across the diode will always be the diode drop, at most.
 

PhilHornby

Senior Member
The triac is basically two back-to-back silicon controlled rectifiers (SCR). It will always measure roughly two diode drops to any terminal under load...The voltage measured across the diode will always be the diode drop, at most.
Don't forget, the voltage I'm measuring, is that across the Gate resistor (because it is Gate current and therefore Resistance Value and Power I am trying to ascertain). Using the 'scope, with AC signals, it is a very non-linear thing, with a maximum value of about 2V (a bit over).

There is a paragraph in the document hippy linked to a few posts back that seems pertinent:

The gate is essentially at the same potential as the anode when an SCR is conducting. When the SCR is non-conducting, the gate potential is not related to anode potential within the normal operating range.
I'm guessing the voltage trace is the shape it is, because once the Triac fires, it effectively shorts the resistor out. Maybe the '2 and a bit volts' is time-related - as in, that's the potential the Mains has happened to reach when the Triac fires? This could/would? explain the different answer I get using DC - because the Triac fires a lot sooner?

Addition: This also might explain why the voltage across the resistor climbs again at the end of the AC cycle - as the Triac begins to 'unlatch'?
 

PhilHornby

Senior Member
I think I've got this sussed ;-)

Back to my original Fig (A) circuit, complete with MOC3041, with its 'Emitter' connected to 5V, via a 150R resistor (and a switch).

Triac with MOC3041.jpg

The voltages across the MOC3041 and the load resistor are as follows (18Vac supply) :-

MOC3041 Voltages.jpg

I believe that when the MOC3041 is activated, current flows through the Gate resistor until the (main) Triac fires. The effective 'shorting' of the Triac Anode and Gate then stops the current flowing through the MOC3041 and the Triac within it drops out. This effectively transforms my previous 'odd' gate resistor voltage waveform, into a pulse (800µS in the above example).

Guesswork/measurement/S.O.T. seems to be the only way to select the Gate resistor - the Triac's firing point has to be reached before the MOC3041 reaches its inhibit point (VIH), and decides that it is no longer in a 'Zero-crossing' period. For the example I tested, VIH is approx. 12V. The datasheet says 20V max.
 

PhilHornby

Senior Member
... Maybe the '2 and a bit volts' is time-related - as in, that's the potential the Mains has happened to reach when the Triac fires?
Well sort of ... this particular Triac appears to need approx. 25mA to 'fire'...

Since I was using a 100Ω resistor, V=0.025 x 100 = 2.5V (ish). Using a 330Ω; resistor, I measured 8.25V instead (and it obviously triggered slightly later). Even using only 6.8Ω; with a 55Vrms supply, neither MOC3041 nor Triac went bang :) (although my 1K load resistor rated at only ¼W started to smoke somewhat :eek:) - I measured 170mV peak across the resistor.

I reckon (after all this!), stick to about 100Ω ish - whatever the supply voltage and it'll be fine. Probably... :p
 
Last edited:
Top