Maker Pro
Maker Pro

TRIAC dV/dT and fluoroescent lamps

A

Adam. Seychell

Jan 1, 1970
0
I'm building an electronic fluorescent lamp circuit with the basic lamp
circuitry shown below. The input frequency is typically 100kHz during
normal operation. The starting sequence begins by turning on the TRIAC ,
and setting the frequency so about 200mA flows through the filaments.
Once heated the circuit is powered off for about 1ms, and the TRIAC is
then allowed to turn off. Power is turned back on at the LC resonant
frequency thereby producing high voltage across the lamp to start the
arc. Once started the lamp reduces impedance and the circuit behaves
like conventional inductor ballast.


o---2.0mH--. ,-----.--------.
| | | |
|---| | |
| | | |
| | | |
180V | | | |
square wave | | | 1.0nF |
input | | --- |
| | --- |
| | | T2|
| | | ----- TRIAC
| | | /\ \/
| | | -----
|---| | T1|\___G
| | | |
o----------' '-----o--------`
|
GND


I'm concerned about reaching maximum off state dV/dT ratings of the
TRIAC and causing it false trigger during lamp operation or starting.
The device I have tested is a BT134 with a rated off state dV/dT of
250V/uS (type). Strangely the TRIAC does not seem to trigger when
exceeding the rated dV/dT. For example, if I temporary remove the lamp
so the circuit effectively becomes an unloaded series LC then the
voltage across the TRIAC was measured to be 700Vpk at 140kHz. This
corresponds to a dV/dT of over 600V/uS across the TRIAC, yet still did
not turn on. However, the TRIAC will sometimes trigger when the circuit
is powered instantly, but if the 140kHz voltage is ramped to 700Vpk over
a few milliseconds, then the TRIAC never triggers. Is there more to the
story than just an off state dV/dT rating ?
 
F

Fritz Schlunder

Jan 1, 1970
0
I'm concerned about reaching maximum off state dV/dT ratings of the
TRIAC and causing it false trigger during lamp operation or starting.
The device I have tested is a BT134 with a rated off state dV/dT of
250V/uS (type). Strangely the TRIAC does not seem to trigger when
exceeding the rated dV/dT. For example, if I temporary remove the lamp
so the circuit effectively becomes an unloaded series LC then the
voltage across the TRIAC was measured to be 700Vpk at 140kHz. This
corresponds to a dV/dT of over 600V/uS across the TRIAC, yet still did
not turn on. However, the TRIAC will sometimes trigger when the circuit
is powered instantly, but if the 140kHz voltage is ramped to 700Vpk over
a few milliseconds, then the TRIAC never triggers. Is there more to the
story than just an off state dV/dT rating ?


Unfortunately the critical dV/dT required for turn on isn't a very well
defined parameter. It varies significantly depending upon junction
temperature, gate drive impedance, and also surely to some extent from
device to device and the waveform/duration of the high dV/dT event. Using a
high impedance gate drive will significantly reduce the critical dV/dT, or
inversely, if you short the gate to MT1 with a low inductance/resistance
path during off state you will get the best dV/dT ability.

Given the ambiguity I don't think I'd risk it for a commercial or industrial
design, I think I would find a different approach.
 
L

Larry Brasfield

Jan 1, 1970
0
Adam. Seychell said:
I'm building an electronic fluorescent lamp circuit with the basic lamp circuitry shown below. The input frequency is typically
100kHz during normal operation. The starting sequence begins by turning on the TRIAC , and setting the frequency so about 200mA
flows through the filaments. Once heated the circuit is powered off for about 1ms, and the TRIAC is then allowed to turn off.
Power is turned back on at the LC resonant frequency thereby producing high voltage across the lamp to start the arc. Once started
the lamp reduces impedance and the circuit behaves like conventional inductor ballast.


o---2.0mH--. ,-----.--------.
| | | |
|---| | |
| | | |
| | | |
180V | | | |
square wave | | | 1.0nF |
input | | --- |
| | --- |
| | | T2|
| | | ----- TRIAC
| | | /\ \/
| | | -----
|---| | T1|\___G
| | | |
o----------' '-----o--------`
|
GND


I'm concerned about reaching maximum off state dV/dT ratings of the TRIAC and causing it false trigger during lamp operation or
starting. The device I have tested is a BT134 with a rated off state dV/dT of 250V/uS (type). Strangely the TRIAC does not seem to
trigger when exceeding the rated dV/dT.

Between the rating being conservative, device variation,
and your choice of temperature at which to do this test,
there is no need for surprise as this result.
For example, if I temporary remove the lamp so the circuit effectively becomes an unloaded series LC then the voltage across the
TRIAC was measured to be 700Vpk at 140kHz. This corresponds to a dV/dT of over 600V/uS across the TRIAC, yet still did not turn
on. However, the TRIAC will sometimes trigger when the circuit is powered instantly, but if the 140kHz voltage is ramped to 700Vpk
over a few milliseconds, then the TRIAC never triggers. Is there more to the story than just an off state dV/dT rating ?

Yes, there certainly is. That rating likely applies
during "commutation", when the carrier distribution
differs from what occurs during your testing.

By the way, (if I were you), I would be sure to take
a look at the dI/dt of your circuit. You might find
you are abusing the poor triacs.

And be sure to look at several datasheets unless
it's a hobby project or your employer craves to
get into vanishing-sole-source pickles. Several
manufactures make such a part and they are not
likely to have the same specs or margins.
 
A

Adam. Seychell

Jan 1, 1970
0
Larry said:
Yes, there certainly is. That rating likely applies
during "commutation", when the carrier distribution
differs from what occurs during your testing.



By the way, (if I were you), I would be sure to take
a look at the dI/dt of your circuit. You might find
you are abusing the poor triacs.


During filament heating period the on current is no more than a 300mApk
triangle wave at 100kHz, or 0.17A/uS which is about 300 times below the
rated dI/dT of most similar sized TRIACs.

And be sure to look at several datasheets unless
it's a hobby project or your employer craves to
get into vanishing-sole-source pickles. Several
manufactures make such a part and they are not
likely to have the same specs or margins.

Thanks for the tips, but what is a good safety margin to work with. For
example, I may find that a bunch of TRIACs typically turn on when
exceeding say 1000Vpk @ 120kHz sine wave across the terminals. Would it
be safe to build the product given that the nominal lamp starting
waveform is only 220Vpk 120kHz, then reduces to 140 Vpk during normal
operation.
 
M

Mark

Jan 1, 1970
0
would a cap on the gate to ground help?

I have always thought of the dv/dt problem as caused by stray C from
the anode to the gate coupling enough signal to reach turn on.

Mark
 
L

Larry Brasfield

Jan 1, 1970
0
Mark said:
would a cap on the gate to ground help?

Probably, as would other low impedances.
(From the OP's diagram, they gate might
already be kept at low impedance, or not.)
I have always thought of the dv/dt problem as caused by stray C from
the anode to the gate coupling enough signal to reach turn on.

That's not a bad mental model, but the capacitance
in question is not reachable from the gate without
some impairment by lateral resistance. This limits
what you can achieve with external shunting.
 
L

Larry Brasfield

Jan 1, 1970
0
Adam. Seychell said:
Larry Brasfield wrote: [snip]
By the way, (if I were you), I would be sure to take
a look at the dI/dt of your circuit. You might find
you are abusing the poor triacs.

During filament heating period the on current is no more than a 300mApk triangle wave at 100kHz, or 0.17A/uS which is about 300
times below the rated dI/dT of most similar sized TRIACs.

The dI/dt you need to worry about is not the current
that occurs while the device has been conducting for
awhile. The issue is dI/dt during the transition from
the off state to the fully conducting state. It takes
time for the conduction process to spread from the
metalized portions of the gate region across the whole
junction set. During that time, current density can be
much higher than after conduction is fully established.
I look at the cap you have across the triac and see a
potential for high dI/dt during switching. Maybe, if
you turn on at triac voltage zero-cross, that is not a
problem. But you should be ensuring that.
Thanks for the tips, but what is a good safety margin to work with.

Sorry, but that's too grand a topic for me just now.
For example, I may find that a bunch of TRIACs typically turn on when exceeding say 1000Vpk @ 120kHz sine wave across the
terminals. Would it be safe to build the product given that the nominal lamp starting waveform is only 220Vpk 120kHz, then reduces
to 140 Vpk during normal operation.

Let's hope some of the other hotshots here can go
thru the whole question and answer series with you
that would be necessary to address your concerns
in a responsible fashion. I have used only a handful
of 4-layer devices, and cannot make the judgment
you appear to want off the cuff.
 
Top