SCR gate control

[chipwitch]

No matter how much theory I read, when I finally get to the point of experimenting and designing a simple circuit, the delicate balancing act between I, V, R and P never fails to challenge.  I first decided to tackle the seemingly simple task of resistor selection to limit current to the gate.  The SCRs I have have a limit of 4 amps, so I didn't expect much problem.  Oh yeah, I forgot power!  4 amps isn't the limiting factor... the resistor is.  4A at 120V is 480W.  Not an efficient choice even if I could find a resistor that large.  I probably couldn't afford it anyway.  So, I settled on 10k.  I have some 2W resistors, so that would work.  Problem is, I only get 12mA of current to the gate, which is okay because it only needs 2mA, but it also means an increase in the firing angle (is that the correct term?) and a decrease in output.  The decrease in output is probably reasonable, but I was wondering... is there a practical way to get the firing angle closer to 0 degrees while not consuming high power as the angle approaches 90 degrees?

I thought about using a small transformer to step down the voltage. Would that cause a lag in current?  If so, would that make timing the trigger at 0 deg more difficult?  Is adding a capacitor an option to adjust the PF back to unity?

I have some 36V, 5W zeners.  I was even trying to incorporate them into the design.  Any suggestions.

<edit> Attached a basic representative circuit.

[Kremmen]

Back in the day when i used to work with heavy DC drives the motors were always driven using a "6 pulse" thyristor rectifier bridge, in various cofigurations depending on the application. These were the big puck type thyristors - i cannot recall any more what the gate triggering power requirements were, but they were considerable. Just passing some phase voltage from the AC side to the gate was a non-starter so what was always used was a dedicated firing circuit. The principle was to discharge a capacitor into the gate at the instant of triggering. There were several evolutions of the firing circuit, some using transistors, others with smaller SCRs but all had the common features of separate capacitor charging, timing control and firing circuit.
Remember that a thyristor does not need constant gate current, just a good kick to get it conducting and you are good for the rest of the cycle.

[SeanB]

Simplest is to use a small transformer to make a low DC voltage and drive the gate with that. Not quite recommended as the SCR will be biased on with reverse voltage but most are very happy with that. Otherwise use a TRIAC which will happily work with both polarities. That way you control an AC load with only a low current though non isolated switch

Just remember that with a SCR you basically have a half wave rectifier, as it will not conduct the one half of the mains cycle. Thus power to the load will be under half of the rated power. TRIAC gives full wave control.

[TerminalJack505]

There was another thread not too long ago and the poster had a similar issue.  (They were using a TRIAC to drive a table saw or something.)  I can't remember all of the suggestions but mine was to use a capacitive power supply to drive the TRIAC's gate.

Beware that capacitive power supplies aren't isolated from mains, though, so be careful.

[Kremmen]

OK, now that there is the schematic...
So you are not attempting to control the power, just switch it on & off? Is there a particular reason to do it using a thyristor - as SeanB pointed out already, if nothing else the thyristor will half wave rectify the AC. If you want both half waves using thyristors, you will need at least a half-controlled bridge, i.e. 2 diodes and 2 thyristors but just for switching that would be kinda pointless.
Do consider a triac - much easier for full wave results. An easy alternative - gasp - would be a relay. Any particular reason why not?

[SeanB]

I have a similar issue replacing a SCR controller, looks like i will have to butcher something together until the stuff I ordered off fleabay arrives nect month or so.PO is slow this month, must be strike season again.

[chipwitch]

Thanks for all the replies!  I apologize for not mentioning my target output... 30V.  If I'm thinking right, that should address the choice of SCR over Triac.  I have both.... just thought the SCR was a more elegant way to buck that much voltage.  My reason for choosing a thyristor over other PS methods was half need of a PS and half simply to learn about them.  I'm not particularly opposed to Triacs, but my thinking was that learning the basic element of the triac (the scr) would be the best starting point. 

The capacitive PS was one thought I had... rather just a capacitor and a series resistor, but I don't have much in the way of non-polarized caps AND I thought the pf would throw off the timing.  Since I wasn't sure how to calculate the effect on timing, I thought the simpler solution might be to consider other options like the zener diode.

@Sean, you're saying just leave the gate triggered all the time?  If it was half wave rectified that would lag the current(?) so I'm guessing you mean full wave rectification?  Also, because I have so little experience, SCRs to me (incorrectly, I know), imply a variable output when used with AC.  Like variable speed drills (Triacs... I know).  So, my immediate concern is NOT for variable output, rather I'd like a circuit for which I'm able to achieve a better understanding of SCRs through being able to actually design and calculate all the values necessary for all the components.  Right now, I'm a little intimidated by reactance and that's the first thing I think about when I think of transformers.

I guess I'm just going to have to buck up (pun intended) and refresh myself on reactance.  It isn't that I can't do it, but it just adds another level of complexity.  The simpler I can keep it, the better I'll be able to understand using SCRs.

Is any of that poor logic?

[TerminalJack505]

It sounds like you are trying to make a dimmer-style circuit.  Do a Google image search for "dimmer circuit" for some of the ways they are implemented.

You should also grab the free simulators, LTspice and Tina-TI and use them to learn about these sorts of circuits.  Or any circuit, for that matter.

[xDR1TeK]

Simplest is to use a small transformer to make a low DC voltage and drive the gate with that. Not quite recommended as the SCR will be biased on with reverse voltage but most are very happy with that. Otherwise use a TRIAC which will happily work with both polarities. That way you control an AC load with only a low current though non isolated switch

Just remember that with a SCR you basically have a half wave rectifier, as it will not conduct the one half of the mains cycle. Thus power to the load will be under half of the rated power. TRIAC gives full wave control.

Agree with SeanB, SCR is used in high energy applications like the starter for neon lighting or laser applications that require voltages in the 1kV range.  Triac is better suited for dimmer style applications as    TerminalJack505 said.

chipwitch:
The angle of firing? You mean delay in opening the gate w.r.t. phase angle? You only need a positive pulse to activate the SCR.
Negative pulse to the gate turns it off or when the SCR is in blocking mode.
In your case, since you are interested in making the angle between off-trigger and phase equal to 0, you can use zener to wait around zero volts with a voltage divider circuit. In any case, your comparative duty cycle is 50% or less.

[T3sl4co1l]

If nothing else, you need a few volts to begin with, otherwise the SCR doesn't have anything to work with (plus whatever your load is dropping at whatever voltage/amperage), so out of 30V, the conduction angle has to be at least 3.4 degrees shy of "full circle".

There are a number of ways to provide gate triggering; one way would be the same as BJTs (indeed, an SCR below holding current behaves as a BJT), where you supply some DC voltage, with enough current to keep hFE happy (or in the SCR case, to satisfy Ig(max)).  You can also use a constantly-pulsating transformer, which saves on circuitry -- no need for isolated supplies, just a transformer.

A somewhat more naive approach might send a single pulse through a transformer -- of course you can't just pass DC -- but this may not trigger the SCR if it's still reverse-biased, or if the load is below holding current.

You could also optoisolate it, using the load to supply gate bias; high voltage phototransistor type isolators are available (not that you need much voltage in this case).  The downside is slow reaction time (optos are generally slow, to the tune of 10s of microseconds -- well, it's slow to a circuit anyway) and higher "dropout" voltage (since the transistor needs some voltage to do its job, plus you need a protection diode in series, then the normal gate-cathode voltage).

Tim

[chipwitch]

It sounds like you are trying to make a dimmer-style circuit.  Do a Google image search for "dimmer circuit" for some of the ways they are implemented.

Thank you, but I'm looking for a single SCR solution, if one exists.

[chipwitch]

chipwitch:
The angle of firing? You mean delay in opening the gate w.r.t. phase angle? You only need a positive pulse to activate the SCR.
Negative pulse to the gate turns it off or when the SCR is in blocking mode.
In your case, since you are interested in making the angle between off-trigger and phase equal to 0, you can use zener to wait around zero volts with a voltage divider circuit. In any case, your comparative duty cycle is 50% or less.

If it isn't called "firing angle," then I don't know what else it's called.  I see that term frequently applied to SCRs in this context.  It could be known as "delay," I just haven't read anything that put it in that term.  But, I think I understand you and we are talking about the same thing. 

I'm not asking the right question, I guess.  Apparently, I do that a lot on this forum.  Sorry.  I thought more people would be able to fill in the blanks because I'm still not well versed in electronics vernacular.  I am quite capable of building a trigger circuit to fire the SCR into conduction.  I can even design the circuit to do it close to zero.  Using a capacitor, I could probably shift the firing angle to come after the positive peak (between 90 and 180 degrees phase angle).  What I'm finding challenging is simply finding a simple solution to fire the SCR from as close to zero degrees as possible up to 90 degrees.  The simple way that I'm able to design the trigger circuit would simply use a current limiting resistor between the mains and the gate.  By adjusting the value of that resistor, I can easily calculate the correct resistance to trigger the gate at any voltage I desire.  But, the closer to 0 degrees one gets, the lower the resistance required.  That's all well and good until the phase reaches 90 degrees (120V) and suddenly that 1k resistor is drawing 125 mA (consuming 15W).  That's a big resistor.  So ideally, I'd like to clip the top of the wave, clamp down the voltage, shunt the current or by some means divert the current away from the resistor.  Maybe it can't be done without a huge resistor.  Thus, the reason for my question (OP).

Ideally, I'd like to be able to create a simple module that could be used with a variety of circuits (projects) where I could keep a few on the shelf and by simply tweaking a trimmer, adjust the output to whatever I need between a range.  0 to 60VDC would be awesome.  Again, maybe that isn't possible with an SCR.  Before I simply discard the SCR, I'd like to understand more about it, how it works, why it will or won't work.  The goal isn't to create a circuit, rather to create a circuit with an SCR.  It's a good lab project.  Or at least, so I think.

Simply suggesting I abandon the idea of using an SCR and use a Triac does little in the way of learning.  Half (if not more) of the circuits people are working on right now could be swept aside in lieu of Arduino or PIC.    In many cases it would be faster and easier.  But, I doubt it would do much in the way of helping anyone's understanding of electronic circuits.

If I'm asking an SCR to do something for which they are incapable, then fine.  I'm okay with that.  Even through defeat, we learn.

In any case, I sincerely appreciate the guidance (and patience) many of you have demonstrated and accept the blame for not being clear.  In fairness, I hope you'll all agree that the OP asks how one triggers an SCR while meeting certain criteria.  It doesn't ask how one goes about producing a particular output.

[chipwitch]

If nothing else, you need a few volts to begin with, otherwise the SCR doesn't have anything to work with (plus whatever your load is dropping at whatever voltage/amperage), so out of 30V, the conduction angle has to be at least 3.4 degrees shy of "full circle".

There are a number of ways to provide gate triggering; one way would be the same as BJTs (indeed, an SCR below holding current behaves as a BJT), where you supply some DC voltage, with enough current to keep hFE happy (or in the SCR case, to satisfy Ig(max)).  You can also use a constantly-pulsating transformer, which saves on circuitry -- no need for isolated supplies, just a transformer.

A somewhat more naive approach might send a single pulse through a transformer -- of course you can't just pass DC -- but this may not trigger the SCR if it's still reverse-biased, or if the load is below holding current.

You could also optoisolate it, using the load to supply gate bias; high voltage phototransistor type isolators are available (not that you need much voltage in this case).  The downside is slow reaction time (optos are generally slow, to the tune of 10s of microseconds -- well, it's slow to a circuit anyway) and higher "dropout" voltage (since the transistor needs some voltage to do its job, plus you need a protection diode in series, then the normal gate-cathode voltage).

Tim

Thank you.  Exactly the kind of thing I was looking for.  I can google search on some of the terms you provided for more detail.  I don't understand the "3.4 degrees shy of full circle" remark.  That's an awful specific number.  Full circle?  Aren't we talking about half a circle with SCRs?  If I can fire the gate at 7 volts (3.4 degrees?), I'd consider that close enough to zero and a success, especially if an alternate power source wasn't required to fire the gate.

Using a BJT... wouldn't that still be the same problem?  Whatever current limiting resistor I use to trigger at low voltage when the phase reaches 90 degrees, the power and current become the same issue I'm experiencing with the SCR?

10s of uSeconds is plenty fast enough for a 60Hz input.  I don't need any better resolution than that.  BTW... I don't understand the last parenthetical statement.  Does it contain grammatical errors?

[KerryW]

If it was ME, I would replace Rg with a JFET in constant current mode (Drain to switch, Gate and Source connected to SCR gate) and omit Rgk.

You would need a JFET with a minimum IDSS of 2mA, and a max Vds of at least 200V.

The SCR would turn on when the AC reached ~1.2V.  If you replace the diode with a Schottky, ~.9V.

[TerminalJack505]

You might play around with this circuit.  (See attached.)  Taken from here.

The output waveforms are for various parametrized values of R2.

Edit: Added png showing a minimum turn-on voltage of about 2V.

[chipwitch]

If it was ME, I would replace Rg with a JFET in constant current mode (Drain to switch, Gate and Source connected to SCR gate) and omit Rgk.

You would need a JFET with a minimum IDSS of 2mA, and a max Vds of at least 200V.

The SCR would turn on when the AC reached ~1.2V.  If you replace the diode with a Schottky, ~.9V.

Thanks for the suggestion.  I'll look into it.  It sounds good to me, but what do I know?

[chipwitch]

TerminalJack, that is one of the sites I reference often.  Tons of stuff for beginners.  Am I missing something?  It won't get to say, 59V?  It would be fine for only 30V, I suppose (which admittedly is my immediate concern).  I'm still not ready to give up on the notion a single circuit could handle the full range from ~0 - ~60V.  Maybe a two circuit solution is in order, 1 for 0-30 and 1 for 30-60V.

JFETs may be a solution?

Thanks BTW.  The output graph was a nice enhancement of the circuit.  When I first saw the circuit, the load dependency bothered me.  Maybe having a variable load wouldn't be much of a problem, but it just didn't feel right to me.  Other examples I saw tended to tap the line ahead of the load.  The cleverness of the shift in phase by the capacitor was NOT lost on me though.

[TerminalJack505]

I guess I don't know what you're trying to accomplish.  As you can see from the simulation that circuit can trigger before, at, or even beyond the peak voltage (which I assume you are referring to as 60V.)

[chipwitch]

I guess I don't know what you're trying to accomplish.  As you can see from the simulation that circuit can trigger before, at, or even beyond the peak voltage (which I assume you are referring to as 60V.)

No, you DO see.  I'm the one having trouble.  I see it now.  When I first read the article, I mistakenly thought it was only conducting only from 90 degrees to 180.  That would be a maximum of 30V, no?  Your graph doesn't include a superimposed graph of the input voltage.  That was throwing me off.  Now I see it.  What I don't see is how the current is being limited to 6mA?  Is it because once the SCR is triggered it's shunting the load current??  If so, what if you have a high impedance load (higher than the variable resistor).

Take into consideration, one thing I may be wrongly assuming is that the gate/cathode is low impedance, essentially grounding the trigger signal.  There is no (little) impedance within the SCR, right?

[KerryW]

If you move the load to be from the cathode to ground, gate impedance becomes essentially infinite.  You could tie the diode between the anode and the gate with no resistor needed.

[Kremmen]

I guess I don't know what you're trying to accomplish.  As you can see from the simulation that circuit can trigger before, at, or even beyond the peak voltage (which I assume you are referring to as 60V.)

No, you DO see.  I'm the one having trouble.  I see it now.  When I first read the article, I mistakenly thought it was only conducting only from 90 degrees to 180.  That would be a maximum of 30V, no?  Your graph doesn't include a superimposed graph of the input voltage.  That was throwing me off.  Now I see it.  What I don't see is how the current is being limited to 6mA?  Is it because once the SCR is triggered it's shunting the load current??  If so, what if you have a high impedance load (higher than the variable resistor).

Take into consideration, one thing I may be wrongly assuming is that the gate/cathode is low impedance, essentially grounding the trigger signal.  There is no (little) impedance within the SCR, right?

Thyristor as a component does not see any phase angles - those are properties of the circuit where the thyristor finds itself. For a thyristor there is only voltage across anode-cathode and current through the same terminals. Plus gate signals of course. But any angles in the cyclical circuit voltages just translate to instantaneous terminal voltages and those define the behavior of the thyristor.
A thyristor is a 4 layer device, schematically PNPN. You can think of them as a combination on PNP and NPN transistors (see pic below).
Assume there is a positive voltage across A and K (V_AK >0). Initially both transistors are in a nonconducting state. If you now input a positive voltage to the gate terminal the virtual NPN transistor starts to conduct. This pulls base current from the PNP transistor turning that on as well. There is positive feedback between the base currents of the transistors, quickly driving both to saturation and you don't need the original gate current any more to maintain conduction. The only way to turn off a normal thyristor is to cause the anode current to decrease below holding current, by removing or reversing the anode-cathode voltage. The V/I curve is in the second picture.
The detailed behavior of a thyristor is relatively complex and depends heavily on the fabrication details of the component - doping and such things. All of them share common properties such as the minimum holding current required to maintain conduction, as well as the general progression of turn-on and turn-off.
At turn-on there is an initial delay defined by the charge transit times of the 2 virtual transistor base regions. Once the gate cherge has had time to propagate to the bases the anode current rises in proportion to the charge build-up in the base regions. The values again depend on the details of doping and geometry and may be available in component datasheets.
Turn-off requires that the minority and majority carriers must be depleted from the junctions, either by imposing an external reverse electric field, or by recombination. Sometimes gold or platinum acceptor doping is used to speed up recombination. That however is a manufacturing decision which you cannot influence once the component has been chosen. On the other hans, turn-off can be speeded up by applying negative gate voltage to assist the recombination. This does not turn an ordinary thyristor into a GTO one so you will still need to decrease the anode current below I_th, but it does make the turn-off faster.

[TerminalJack505]

Here are a couple more screenshots from the same circuit.  The first set was showing the currents.  These show the voltages.  (See the last attached image to see which voltages are being measured.)

The first one shows the load voltage in relation to the input voltage for various values of R2.  (This is simulating different settings of a potentiometer or trimmer.)

The second one shows the voltage across R2 in relation to the input voltage for the same various values of R2.

The third one shows the input voltage, the voltage across the load, and the voltage across R2 for an R2 value of 100k.

Note that the input voltage is 120VAC.  That's an RMS voltage which is why the peaks are nearly 170V and -170V.

The fourth image shows what happens to the voltage across the SCR when it turns on.

You can see from the waveforms that the reason the current through R2 is limited is due to the voltage across the SCR dropping (to just 1.6V), as you surmised.

The circuit is designed to work with low impedance loads.  By putting the load in series with R2 it is basically working off of the assumption that most of the voltage is going to drop across R2 when the load is off.

[T3sl4co1l]

Thank you.  Exactly the kind of thing I was looking for.  I can google search on some of the terms you provided for more detail.  I don't understand the "3.4 degrees shy of full circle" remark.  That's an awful specific number.  Full circle?  Aren't we talking about half a circle with SCRs?  If I can fire the gate at 7 volts (3.4 degrees?), I'd consider that close enough to zero and a success, especially if an alternate power source wasn't required to fire the gate.

"Full circle" assuming you use a pair to fill in the full cycle.  I guess... you haven't indicated exactly what it is you're switching, and how you want to switch it (half wave, full wave, partial..)?

I was guessing about 1V for turn-on, which means it can't tun on until 1/30th, or 1.9 degrees.  Actually.. now I wonder if I did that wrong but got a roughly correct answer anyway...    Maybe there was a sqrt(2) in there, too.  Since it's not clear if you're speaking of 30V RMS AC, or 30V peak, or even something arbitrary.

Likewise, it turns off below a similar voltage, doubling it to 3.8 degrees.  Actually, that should be "half circle", although if you're only talking about one SCR, there's no other half to the circle to speak of, so I guess that's still true in a manner of speaking.  A complementary pair would miss a total of 7.6 degrees though.

But anyway yeah, if it's self driven, it has to have some dropout somewhere.

Using a BJT... wouldn't that still be the same problem?

No, I meant it behaves like one.  You'd have to provide external drive to keep it going under that condition (I_load < I_[h).  Which could be an external gate driver, isolated power, transformer coupled, you name it.

10s of uSeconds is plenty fast enough for a 60Hz input.  I don't need any better resolution than that.  BTW... I don't understand the last parenthetical statement.  Does it contain grammatical errors?

No, I mean:
Power source -- load -- SCR (A to C) -- GND

Opto is a HV phototransistor type, collector to SCR anode, emitter to SCR gate.

Except, if voltage reverses, so does the transistor, which is bad.  So it needs a series diode.

So your minimum dropout is Vce(sat) + Vf + Vg(on).

An optoisolator acts roughly as a constant current source, so this should turn on the SCR fairly easily, independent of operating voltage.  The transistor can potentially dissipate a lot of peak power (like 10mA * 100s of V -- in general SCR applications, not for your case), though the pulse width should be small (the SCR should turn on in a few microseconds), and has relatively high capacitance (a G-C resistor and C-G diode would help protect against transients and leakage).  For general SCR applications anyway, you'd probably need a darlington (photo- or otherwise), to deliver enough gate current to get it running, since bigger SCRs aren't as sensitive (>50mA Ig(max)?).

Tim

[T3sl4co1l]

You might play around with this circuit.  (See attached.)  Taken from here.

The output waveforms are for various parametrized values of R2.

Edit: Added png showing a minimum turn-on voltage of about 2V.

This circuit will not work, for reasons already explored: the ensure required gate current at any voltage, the resistance has to be exceptionally small.

Your subsequent simulation showing a 100k resistor is just that, a simulation.  I have no idea what kind of sim model you have there, but SCRs are notoriously difficult to model anyway, and it would hardly be surprising that it might trigger on microamperes.

Tim

[T3sl4co1l]

If it was ME, I would replace Rg with a JFET in constant current mode (Drain to switch, Gate and Source connected to SCR gate) and omit Rgk.

You would need a JFET with a minimum IDSS of 2mA, and a max Vds of at least 200V.

The SCR would turn on when the AC reached ~1.2V.  If you replace the diode with a Schottky, ~.9V.

Thanks for the suggestion.  I'll look into it.  It sounds good to me, but what do I know?

Just one question...

Where the fuck are you going to find one?  And, I want some.

Tim