How to drive SCR using Pulse transformer

[chinuhark]

I have been searching for this everywhere including the books I have but I just can't find the design procedure for the circuit used to drive an SCR using a Pulse transformer.
It is a pair of antiparallel SCRs being driven by a 1:1:1 Pulse transformer.
The SCR is TYN612 with - a) Igm(Peak gate current) = 4A b) Igt = 0.2mA to 15mA c) Vgt(max) = 1.3V d) Latching current = 60mA e)Holding current = 30mA
Assume a simple resistive load for now.

I have narrowed it down to the attached circuit but cannot understand how to calculate R1 and R2.
I am finding it very difficult to imagine how exactly the pulse transformer works. As N1=N2(turns ratio), is it as simple as I1 = I2 ? Some circuits don't have R2 at all. Wouldn't that short circuit the secondary? What is the max pulse width that one can use without causing saturation? Please help!
Note that the Pulse transformer was bought locally and has no name or model no, let alone a datasheet.

 

[chinuhark]

54 views and no replies!
Does no one use SCRs in the Power Electronics forum? I thought this must be a standard question with a straight forward answer...
Please help!!!

 

[mtwieg]

Assuming an ideal transformer, the current into the gate will be roughly (3.3-Vce-VD-Vgt)/(R1+R2), where Vce is the saturation voltage of the transistor, VD is the diode drop, and Vgt is the gate threshold voltage of the SCR. I'd choose R1+R2 to give around 100mA into the gate. Also since your supply voltage is only 3.3V you may want to use a MOSFET instead of a darlington for the driver.

 

[D.A.(Tony)Stewart]

SCR's and Triac's are simple in concept but difficult to master. I recall making a 3 channel Light Organ with Triacs and home-made pulse transformers in 1968. It was simple and worked. But you should read **broken link removed**before hand from other SCR Mfg who had experts in SCR applications unlike Hao Pin who merely copied the process.

Also when you assume anti-parallel of any SCR for any design, you will be wondering where are the negative quadrant specs, such as they have in 2, 3 or 4 Quadrant Triacs. However it is common to use only Q2 & Q3 quadrants and protect the gate from high positive EMF voltages.


For inductive loads choose a 3 quadrant Triac so that turn-off current does not create a false trigger, otherwise known as Snubberless Triacs or you must add a Snubber to suppress.

Understanding how to saturate and unsaturate power NPN and PNP may have differences will help to understand why Quadrant IV is missing in 3 quadrant Triac devices but this is an advantage for faster commutation. Also why ultralow Vce transistors work better for current sensitivity. You may recall most transistor switches are rated at Ic/Ib=10 and an ultra-low Vce transistor will be rated for Ic/Ib=50. Keep this in mind when you attempt to understand what surge current your load will have, be it a stopped motor or cold tungsten lamp to estimate what gate current you need. Although SCR's and Triacs latch and regenerate the required base current, they have to saturate first before they do this and the hFE drops rapidly to this value of 50 or 10 ( not shown in SCR specs) depending on the device characteristics.

Don't be misled by the best case "ultrasensitive" gate current when driving a non-linear resistive load such as a bunch of 300W Halogen lamps taking 10A when hot, but up to 100A when starting. Since Ig max is 4A and yet Imax peak is >>100A, you can see when operating in this mode high gate current pulses are required.

Also note from your data sheet, that pulse width has an effect on the thermal resistance.


If your load dI/dt is in the range of 20A/us from a resistive load, then the gate current risetime is equally important and pulse drivers have a big advantage here, as the drive current is independent of phase voltage and resulting gate current from pull-up type gate current drivers.


If you intend to run it cold or below room temp. then double the current.

 

[schmitt trigger]

In addition to what S.S.Guy mentioned, is that the gate transformer, must be able to support the required primary volt-second product without saturation.

For powerline gate drive, usually you will require as much volt-seconds as possible, since the Triacs take some time to latch.

For instance, Coilcraft's SD250-3L is capable of withstanding 375 V-usec, which translates to about 125 microseconds at 3 volt of applied primary voltage (I'm considering 0.3 volt parasitic voltage drop at the primary).
If your microcontroller can reliably ensure that this pulse width value is not exceeded, then the transformer will not saturate and the current limiting resistor R1 is not required.

If the transformer is too large, there is an alternative: use a smaller device with lower volt-seconds, and use picket-fence triggering. Which is a fancy way of saying a closely spaced burst of narrower pulses.
I've used this last method to trigger Triacs succesfully since the days of unijunction transistors (back in the mid 1970s).

 

[D.A.(Tony)Stewart]

I recommend you look here for pulse transformers. http://www.pulseelectronics.com/

 

[chinuhark]

Firstly thanks to all for the excellent replies and links which were very helpful.
After reading all these, I am thinking along the following lines:
1) Use unregulated 8.4V (directly from o/p of rectifier) with a nice little 470uF cap and decoupling caps because 3.3V just cannot cut it as the various on state voltages and forward drops are just too high (TIP122 - 2V, 1N4148 - 1V, Vgt - 1.3V : total 4.3Volts)
2) If I am not wrong, the total resistance is important and so R2 can be removed and R1 can be made larger for limiting the current.
3) As for the current, the guide said that it should be atleast twice Igt. As Igt = 15mA, I think 50mA is a safe value to ensure triggering.

So is it OK to take R1 = (V-Vce-Vf-Vgt)/50mA ?
Which gives R1 = (8.4 - 4.3)/50mA ~ 80 Ohms

The BIG ISSUES:
1) Are the calculations and more importantly logic used above correct?
2) I am certainly in favor of train of pulses method but how to decide the individual pulse widths and no of such pulses which will be required for successful latching? (Assume a simple incandescent lamp load for now)

The only thing written on the pulse transformer is Rp type 4503.
Is there a way to measure the Volt time product using a scope and test signals. If so how? And after getting this value, how to exactly use it for getting pulse width and no of pulses in the pulse train.

 

[FvM]

Driving two SCRs didn't yet appear in your calculation. It means that the primary current must be double the intended individual gate current. R2 could have a purpose to gurantee equal current sharing between both SCRs, despite of possible gate voltage differences.

- - - Updated - - -

Basically a single short pulse can trigger the SCR. Burst triggering may be appropriate if you're unsure about the exact zero-crossing time, e.g. due to reactive load currents. Still current and duration of each individual pulse should fulfill the datasheet trigger conditions.