Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

How to fire an SCR?

Status
Not open for further replies.

jpanhalt

Well-Known Member
Most Helpful Member
In a sense this question is related to an earlier thread about spot welders, but it is sufficiently different that I thought a new thread was justified. In brief, I now have an operating system for charging the capacitor(s) for a capacitor discharge spot welder. I am now looking on the firing circuit for the SCR. I am planning to use a Semikron SKKT92 with a 2000 A pulse rating. The main reason to chose this device was its big screw connections, and it was cheap ($2 USD each for three) on ebay.

I searched on SCR firing, triggering, and related key words and didn’t get much in the way of explanation, just lots of examples. The datasheet gives what’s required, but not how to get it. Something in my gut tells me that just switching a 12V power supply to the gate manually and momentarily is not the best way to go. Some of the Linear Technology circuits for switching regulators reminded me of circuits used for CDI, so I went down that path and now have two versions that work on the bench (see attachments).

Version A is the simplest and is pretty close to manual, except the turn off is controlled by the capacitor, not my poor reflexes. Version B is a little more complex and most similar to the CDIs with which I am familiar. When I monitor gate potential on my scope, Ver. A gives a good pulse but a long and gradual decay. Ver. B gives a very sharp pulse and fast decay, except one also sees negative spikes (ringing?). In both circuits, I added a resistor (R8 and R3, respectively) on the premiss that it would bleed off unwanted accumulation of charge and possibly prevent false firing.

Bench tests were done with only a 2200 ufd capacitor across the anode and cathode of the SCR at 12V. I suspect that when this is scaled up to the welder size (0.4 to 1.0 F) differences that weren’t apparent on the bench may become VERY apparent. :) I want to avoid or at least reduce the likelihood of something becoming too apparent for a brief millisecond.

My specific question are:
1) Is there anything wrong and what are the problems, if any, with simply firing the SCR with a relay?

2) Is the shorter pulse of Ver. B an advantage over the longer decay of the pulse from Ver. B?

3) Are negative spikes bad, or do they help close the gate?

4) What steps can be taken to reduce the ringing, while still retaining the negative pulse?

5) The datasheet shows (attached) a best trigger zone (BSZ) that seems to show that at lower voltages (0.5 to 1.3V), lower currents should be used, and one should not exceed 20V through a 20 ohm resistor. My experiments with firing at 9V vs. 12V showed that the former would not reliably trigger; whereas, the latter never failed to trigger. These tests were done using Ver B. Is there a recommended practice for choosing the trigger voltage for these devices, perhaps related to the anode voltage?

6) Will the bleed resistor (R8 and R3, see above) help or is it foolish?

This project is for getting clean tack welds on SS hypodermic tubing, so I can then use brazing for cluster joints in a large model airplane.

Thank you for any help and advice. Unfortunately, this post suffers from wordiness, but maybe that offsets “please help me” post with no other information. John
 

Attachments

  • skkt092_09.jpg
    skkt092_09.jpg
    61.9 KB · Views: 1,004
  • SCR FIRING.jpg
    SCR FIRING.jpg
    149.2 KB · Views: 1,547
(1)Why not just use a relay then? My understanding is that unless you need a really low power control circuit or fast switching, a relay would work. Using a relay to trigger the SCR defeats both these, doesn't it?

(3)A SCR, once triggered, will continue to conduct regardless of what happens at the gate. THe only way it will turn off is when the primary current reaches zero. If the negative spike you are talking about is in the primary voltage and somehow causes the effect described, that is how the SCR will shut off. Negative spikes might also cause latching (maybe)- I don't understand it completely, but something can happen with MOSFETs I believe. I do not know if it carries over to SCRs.

(4)Ringing by definition reduces the negative pulses. WHy do you want them, but at the same time reduce them.
 
Last edited:
dknguyen said:
(1)Why not just use a relay then? My understanding is that unless you need a really low power control circuit or fast switching, a relay would work. Using a relay to trigger the SCR defeats both these, doesn't it?

That is the question. After reading gobs of ANs from Linear and others, I did not see that question answered per se. Of course, a relay works on the bench at low power. My intended use in the welder will have high currents. The various references indicate that reliable turn-on in such cases is facilitated by high gate currents. I was going to use a simple wall wart to power the gate. Its max current is on the order of 1 A, so I thought a capacitor discharge would help supply a higher peak current for triggering. From the ONSemi Handbook (ONSemi Handbook (https://www.onsemi.com/pub/Collateral/HBD855-D.PDF):

"Turn−on of the SCR is thus accomplished with a short, high current pulse. Commonly used trigger devices are programmable unijunction transistors, silicon bilateral switches, SIDACs, optically coupled thyristors, and power control integrated circuits.

"Generally, a design starting point is selection of a capacitance value which will reliably trigger the thyristor when the capacitor is discharged."

An additional question I forgot to ask initially is whether the higher gate current is an advantage in causing faster spread of the turned on region of the gate. Again, that might be an advantage in the welder circuit.

dknguyen said:
(3)A SCR, once triggered, will continue to conduct regardless of what happens at the gate. THe only way it will turn off is when the primary current reaches zero. If the negative spike you are talking about is in the primary voltage and somehow causes the effect described, that is how the SCR will shut off. Negative spikes might also cause latching (maybe)- I don't understand it completely, but something can happen with MOSFETs I believe. I do not know if it carries over to SCRs.

Understood, but a small negative pulse at the gate will raise the minimum latch-on current and can help close the gate. See: ONSemi Handbook (https://www.electro-tech-online.com/custompdfs/2008/01/HBD855-DPDF.pdf)

"Reverse voltage on the gate of an SCR markedly increases the latch and hold levels.

" Negative gate bias will decrease the turn−off time.

"However, it [negative gate bias] does have a marked effect at low anode current levels; it can be put to advantage by using it to modify certain device parameters. Specifically, turn−off time may be reduced and hold current may be increased. Reduction of turn−off time and increase of hold current are useful in such circuits as inverters or in full−wave phase control circuits in which inductance is present."


The particular SCR I have seems to turn off quite easily. The minimum holding current listed in the datasheet is 250 mA. The decay of the pulse from Ver. A is quite long compared to Ver. B, which raised in my inexperienced mind a question of whether the length of the pulse could be a too long. (Too short a pulse can also be a problem; my concern here, though, is whether it can be too long and too slowly decaying.)

dknguyen said:
(4)Ringing by definition reduces the negative pulses. WHy do you want them, but at the same time reduce them.

I don't necessarily. It was simply an observation. Again, if there is no problem with Ver. A or even a direct manual switch (relay) that is probably the way I will go. The question is basically whether Ver. B, which causes a brief, sharp rising pulse has any advantages over a mechanical switch for something that will be triggered at less than 1 Hz (the large capacitor will take several seconds to recharge between welds).
John
 
CORRECTION/REVISION

Left the circuit for awhile and on returning with some fresh ideas found that the "ringing" or negative pulses metioned above for Ver. B occur on re-making of the trigger switch. That is, there is a nice trigger pulse when the switch opens. If held open, the SCR recovers as normal. Then when the trigger switch re-makes, the SCR is triggered again with much more noise (probe between the capacitor and SCR gate resistor). Then added a 1K resistor in the GND line from the NC contact. In other words, it is not a dead short to ground, but rather 1K to ground. That stopped the re-triggering on re-make and the trigger looks almost like Ver. A.

There are still parts of my question for which I need advice:

1) Is there any disadvantage to mechanically switching the SCR gate?
2) Is the rationale for the capacitor trigger (i.e., greater current) valid?
3) Does anyone have better trigger circuits to recommend?

Thanks. John
 
FOLLOW-UP

After more study, it seems di/dt for the trigger should be maximized (within limits) as that tends to raise the allowable critical value for the main di/dt of the SCR. For my device (SKKT 92), the recommended minimum di/dt for the gate is at least 1 A/uS.

I looked at the amount of switch noise from the foot switch and was surprised by the duration. It appears that producing multiple transitions that fail to trigger or even a single short transition that triggers, but is less than the recommended period (100 uS), would not be good for the gate and would reduce performance. Please comment, if that assumption is wrong. Thus, I decided to scrap the earlier designs, de-bounce the switch, and use a single pulse generator based on the 555. The de-bounce circuit is derived from an MIT circuit and is shown below. Components C8, R9, and R10 act to prevent re-triggering, even if the foot switch is held closed indefinitely. Unfortunately, the 555 produces a positive pulse, and its 200 mA is on the low side of what the SCR needs.

Questions:
1) Any advice on a simpler way to get more current?
2) I ruled out using a high-side N-channel mosfet, because I wanted to avoid the complexity of its gate drive. Any other suggestions?

We all like pictures, so I have included a few with this post. Remember, this is done on a breadboard, there are lots of stray inductances, and my scope leads are a mess.

The first two are measurements of the SCR gate current and voltage taken across R14 (which was reduced to 10 ohm) with a sweep of 250 uS/div (TDS210 scope).

The third picture is the SCR current across the same R14 with a sweep of 100 nS/div. The cursors are set at the half height of the current increase, which if my math is correct gives a 10 A/uS di/dt.

I greatly appreciate any help and suggestions:D . My naive opinion is that the drive looks good enough to go to the next step, namely, to hook it up to the big capacitor and welding electrodes and see what happens. Fortunately, I have 3 SCRs, but I would like to avoid making a mess.

John
 

Attachments

  • SCR TRIGGER 555.PNG
    SCR TRIGGER 555.PNG
    13.3 KB · Views: 1,589
  • 180108_250 uS_SCR gate voltage2.png
    180108_250 uS_SCR gate voltage2.png
    215.9 KB · Views: 630
  • 180108_250 uS_SCR gate current across R14_2.png
    180108_250 uS_SCR gate current across R14_2.png
    262.9 KB · Views: 529
  • 180108_100nS_SCR current2.png
    180108_100nS_SCR current2.png
    257.7 KB · Views: 533
If you are only using DC, then you could use IGBT's instead of SCR's, you could also potentionally use them for AC as well.

I have used them for simple high power DC switching, and according to Wikipedia,

"Toyota's second generation hybrid Prius has a 50 kW IGBT inverter controlling two AC motor/generators connected to the DC battery pack."
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top