Thyristor latching current problem NXT bt169d ...

[killingtime]

Hello,

I've received a batch of bt169d (nxt semi) thyristors from farnells. The datasheet specifies 4mA latching current (anode to cathode), and yet I cannot get the device to latch below 80mA. Above 80mA it behaves like a thyristor would (remove th gate current and the current still flows between a->c).

Circuit is simple:

cathode tied to ground.
anode tied to 1K ohm, tied to +12v
gate tied to central wiper of a 10K pot. Pot legs tied to 0v and 12v.

I've had 10mA into the gate (way above the gate threshold), and it still doesn't latch.

I have to add additional resistors in paralle with the 1k to get the current above 80mA.

Either NXP have sent out a bad batch of thyristors, or I'm missing something.....

Anyone help?

Just put 'bt169d nxp' into google. 1st non sponsored link back takes you to the datasheet.

Thanks.

***EDIT***

OK,

On further investigation, it would appear that the device will latch below 80mA, but only if the gate goes to high impedance after switch on (instead of to 0V). If you send the gate to 0V below 80mA, the thyristor turns off again.

As a workaround, I can place a diode in series with the gate ...... but can someone please confirm that this is normal behaviour for a thyristor?

I thought you could ground the gate ater exceeding the latching current without interrupting the anode - cathode current flow?

Thanks.

 

[(*steve*)]

Firstly, let me say that I've never observed the behaviour you're talking about.

However, I'm not completely surprised.

If you compare a gate-turn-off thyristor with a normal one, you'll see that the basic structure is exactly the same. Indeed if you were to make a device from 2 transistors connected together you almost certainly could turn the device off from the gate.

The reason they don't turn off is that the gate exerts a relatively small influence on the behaviour of the device. It relies on feedback (essentially) to turn fully on. However at lower and lower currents the amount of that feedback is reduced and the gate current becomes significant. At very small currents (below the latching current) the gate drive is required to maintain conduction. It is not hard to imagine that there is a transition region between the gate drive being required, and the gate completely losing control. I am somewhat surprised that region is so large.

If you look at the datasheet (sorry for the Philips device) you will notice that the latching current is specified for a Rgk of 1k. Presumably if this varies, so will the latching current requirement.

A series resistor may also fix your problem.

edit: I *have* observed the behaviour at currents lower than the latching current, I'm not a complete newb

 

[killingtime]

Hello Steve,

After your post, I tried my circuit set-up with a 1K Ohm resistor in series with the gate - and it latched at 10mA with the gate eventually grounded via the resistor.

I took another look at the NXP datasheet, and the only reference to any gate resistor is as part of a foot note to a dv\dt graph......

Well, I've learnt something new, but still find it a little odd that in order to meet the publisheed specs of a device, you have to add additional components that are buried in the small print. I hope this trend does not catch on, otherwise it will make comparing devices very difficult.

Thanks for your help.

 

[(*steve*)]

Specifications never stand alone, they are always predicated on some set of assumptions.

If you look at the Philips datasheet (link fixed above), the specification for Rgk is not "hidden". It can often be instructive to look at the specifications for parts from other manufacturers as they may be clearer about certain aspects than others.

Another point well illustrated here is that you can't just stick in a component and expect it to work right to the limits of its specification.

A fairly obvious example of "needing something more" is a 2N3055 transistor which has a specification telling you how much power it can dissipate. Well, it won't get anywhere near that without a *huge* heatsink. And there are other issues (SOA for example) which may well limit the dissipation.

Pretty much every specification requires you to either make assumptions about the environment the device is in, or add components to (help) ensure they are maintained.