Maker Pro
Maker Pro

cascode switching

K

kell

Jan 1, 1970
0
I tried firing a car ignition coil by connecting
a 400 volt transistor in cascode, like this:

+12 v
|
|
+--+-------+
| |
| |
10R )
| ignition)
| coil )
| )
| _______)
| c/
+--| MJE13009
e\
|
|
points
|
|
gnd

but it blew out the transistor after a few cycles.
I have fired ignition coils using igbt's and mosfets
(although not in cascode)
with similar voltage ratings to the MJE13009 and nothing
ever went wrong.
Recently I also tried the MJE5852,
which is a 400 volt pnp transistor, in the following configuration,
and didn't have any problem

+12 v
|
|
)
ignition)
coil )
)
)
|
|
e/
+----| MJE5852
| c\
| |
points |
| |
| |
+---+--+
|
|
gnd

Did I get a bad MJE13009 or is there something wrong with
the arrangement in the top diagram?
 
kell said:
I tried firing a car ignition coil by connecting
a 400 volt transistor in cascode, like this:

+12 v
|
|
+--+-------+
| |
| |
10R )
| ignition)
| coil )
| )
| _______)
| c/
+--| MJE13009
e\
|
|
points
|
|
gnd

but it blew out the transistor after a few cycles.
I have fired ignition coils using igbt's and mosfets
(although not in cascode)
with similar voltage ratings to the MJE13009 and nothing
ever went wrong.
Recently I also tried the MJE5852,
which is a 400 volt pnp transistor, in the following configuration,
and didn't have any problem

+12 v
|
|
)
ignition)
coil )
)
)
|
|
e/
+----| MJE5852
| c\
| |
points |
| |
| |
+---+--+
|
|
gnd

Did I get a bad MJE13009 or is there something wrong with
the arrangement in the top diagram?

Yes, the top diagram is completely wrong.
 
J

John Popelish

Jan 1, 1970
0
kell said:
I tried firing a car ignition coil by connecting
a 400 volt transistor in cascode, like this:

+12 v
|
|
+--+-------+
| |
| |
10R )
| ignition)
| coil )
| )
| _______)
| c/
+--| MJE13009
e\
|
|
points
|
|
gnd

but it blew out the transistor after a few cycles.
I have fired ignition coils using igbt's and mosfets
(although not in cascode)
with similar voltage ratings to the MJE13009 and nothing
ever went wrong.
Recently I also tried the MJE5852,
which is a 400 volt pnp transistor, in the following configuration,
and didn't have any problem

+12 v
|
|
)
ignition)
coil )
)
)
|
|
e/
+----| MJE5852
| c\
| |
points |
| |
| |
+---+--+
|
|
gnd

Did I get a bad MJE13009 or is there something wrong with
the arrangement in the top diagram?

You don't show any ignition condenser in your schematic. Did you
remove the condenser from the points (you should)? I think the
stresses on the transistor would be reduced quite a bit if you
reconnected the condenser to the collector.
 
kell said:
I tried firing a car ignition coil by connecting
a 400 volt transistor in cascode, like this:

+12 v
|
|
+--+-------+
| |
| |
10R )
| ignition)
| coil )
| )
| _______)
| c/
+--| MJE13009
e\
|
|
points
|
|
gnd

but it blew out the transistor after a few cycles.

There are 2 problems with this circuit. The first is it doesnt make
sense. The tr does nothing of any use. Lose the tr, connect coil
direct, and it'll work fine.
The 2nd is the b/e jn will see massive ac voltages as the coil LC
rings, so no wonder it died.

I have fired ignition coils using igbt's and mosfets
(although not in cascode)
with similar voltage ratings to the MJE13009 and nothing
ever went wrong.
Recently I also tried the MJE5852,
which is a 400 volt pnp transistor, in the following configuration,
and didn't have any problem

+12 v
|
|
)
ignition)
coil )
)
)
|
|
e/
+----| MJE5852
| c\
| |
points |
| |
| |
+---+--+
|
|
gnd

Thats hard to imagine, connecting c to b is generally regarded as a Bad
Idea.

Did I get a bad MJE13009 or is there something wrong with
the arrangement in the top diagram?

No, yes, and in both I think.


NT
 
J

John Popelish

Jan 1, 1970
0
kell wrote:




There are 2 problems with this circuit. The first is it doesnt make
sense. The tr does nothing of any use. Lose the tr, connect coil
direct, and it'll work fine.
The 2nd is the b/e jn will see massive ac voltages as the coil LC
rings, so no wonder it died.

The schematic is not all that bad. The transistor has a current gain
less than 1, in that the points have to carry both the coil current
and the base current, but when the points open, the points have to
interrupt way less voltage than when they are connected directly to
the coil. So the transistor provides voltage gain.

At that moment the points open, all the coil current is detoured
through the base as reverse current, turning the transistor off very
quickly. This is great for fast switch off, but may be so fast that
the voltage will peak extremely high, because not all the coil primary
inductance is coupled to the secondary, so the collector voltage will
be proportional to the secondary voltage, but will have an additional
leakage inductance component. Slowing the rise of the coil voltage
with some capacitance (possibly a little less than the original point
condenser) will help the transistor get fully switched off before full
voltage is reached (lowering its switching losses), and lowering the
peak voltage the collector-base junction has to withstand.
 
K

kell

Jan 1, 1970
0
You don't show any ignition condenser in your schematic. Did you
remove the condenser from the points (you should)? I think the
stresses on the transistor would be reduced quite a bit if you
reconnected the condenser to the collector.

I removed the condenser from the points of course. I didn't put a cap
across the transistor because I never had to do that on any of the
other ignition circuits I built. I wonder if there's something about
cascode that escapes me. First time using it.
 
J

Jim Thompson

Jan 1, 1970
0
I removed the condenser from the points of course. I didn't put a cap
across the transistor because I never had to do that on any of the
other ignition circuits I built. I wonder if there's something about
cascode that escapes me. First time using it.

Put the cap across the COIL.

...Jim Thompson
 
K

kell

Jan 1, 1970
0
Jim said:
Put the cap across the COIL.
Thanks for reminding me I could do that.
If anybody's interested, I went with this circuit because it has the
lowest voltage drop of several circuits I tried. Coil driver IGBT's
are robust but the ones I've experimented with seem to have a much
higher voltage drop than bipolar transistors. The difference as much
as a whole volt. And I didn't want to drop any voltage because this is
going on a six volt vehicle. A volt is a terrible thing to waste!
 
K

kell

Jan 1, 1970
0
Jim said:
Put the cap across the COIL.
Thanks for reminding me I could do that.
If anybody's interested, I went with this circuit because it has the
lowest voltage drop of several circuits I tried. Coil driver IGBT's
are robust but the ones I've experimented with seem to have a much
higher voltage drop than bipolar transistors. The difference as much
as a whole volt. And I didn't want to drop any voltage because this is
going on a six volt vehicle. A volt is a terrible thing to waste!
 
B

Bobscar

Jan 1, 1970
0
I remember Maplin Electronics publishing a book for about ten pounds.
The circuit consisted of a BU205 tv line output transistor driven by
BC461. The BC461 is in turn driven from the breaker points/ trigger
head via a 555 timer wired as a scmitt trigger. Unfortunately this book
is no longer in print so try see if local library has it if in UK. The
circuit also contains RC/ varistor filtering to protect the driving
components, notably the 555 (max +Vcc = 18v).
 
K

kell

Jan 1, 1970
0
Bobscar said:
I remember Maplin Electronics publishing a book for about ten pounds.
The circuit consisted of a BU205 tv line output transistor driven by
BC461. The BC461 is in turn driven from the breaker points/ trigger
head via a 555 timer wired as a scmitt trigger. Unfortunately this book
is no longer in print so try see if local library has it if in UK. The
circuit also contains RC/ varistor filtering to protect the driving
components, notably the 555 (max +Vcc = 18v).

They're a bunch of nervous nellies if they think they have to use a
Scmitt trigger for something like this.
 
J

Jasen Betts

Jan 1, 1970
0
They're a bunch of nervous nellies if they think they have to use a
Scmitt trigger for something like this.

debounce might be handy though.
 
P

Pooh Bear

Jan 1, 1970
0
kell said:
I'm using the same
switching topology, with mechanical points where they have a mosfet in
the pictured circuit.

No you're not ! Look at what the base is connected to.

Graham
 
K

kelvin_cool_ohm

Jan 1, 1970
0
The schematic is not all that bad. The transistor has a current gain
less than 1, in that the points have to carry both the coil current
and the base current, but when the points open, the points have to
interrupt way less voltage than when they are connected directly to
the coil. So the transistor provides voltage gain.

At that moment the points open, all the coil current is detoured
through the base as reverse current, turning the transistor off very
quickly. This is great for fast switch off, but may be so fast that
the voltage will peak extremely high, because not all the coil primary
inductance is coupled to the secondary, so the collector voltage will
be proportional to the secondary voltage, but will have an additional
leakage inductance component. Slowing the rise of the coil voltage
with some capacitance (possibly a little less than the original point
condenser) will help the transistor get fully switched off before full
voltage is reached (lowering its switching losses), and lowering the
peak voltage the collector-base junction has to withstand.

Carrying your logic one step further, do we not come across a second
mode that might have destroyed the transistor:

Once the voltage produced by the inductive current collapse peaks, it
starts back down through zero on the negative swing of the ringing
wave form. At the point where the voltage at the collector drops 0.7
volts negative (referenced to its other end at B+), the collector-base
junction of the transistor becomes forward biased through the 10 Ohm
resistor. The collector-base junction becomes a 0.7 volt voltage clamp
for all the stored energy with only a 10 Ohm resistor for a load.
It would explain why this particular topology failed where the others
didn't.

I'd be interested to hear your response.

Rick
 
J

John Popelish

Jan 1, 1970
0
kelvin_cool_ohm said:
Carrying your logic one step further, do we not come across a second
mode that might have destroyed the transistor:

Once the voltage produced by the inductive current collapse peaks, it
starts back down through zero on the negative swing of the ringing
wave form. At the point where the voltage at the collector drops 0.7
volts negative (referenced to its other end at B+), the collector-base
junction of the transistor becomes forward biased through the 10 Ohm
resistor. The collector-base junction becomes a 0.7 volt voltage clamp
for all the stored energy with only a 10 Ohm resistor for a load.
It would explain why this particular topology failed where the others
didn't.

I'd be interested to hear your response.

I understand the situation you describe, so let's think about the
magnitude of the current pushed through the base by this mechanism.
The maximum forward biased emitter to base junction current was (6
volts - ~0.7 volts)/10 ohms = 530 mA. But the collector current peak
is (6 volts - Vcesat)/ primary resistance, which is almost certainly
quite a bit higher than 530 mA. The current peak on the first
rebound will be lowered from the original peak, by the energy that
went into the spark, so it is a bit of a guess for me to know how much
current we are talking about.

Hooking a scope probe to the base lead would be instructive.

I still think the worst thing the circuit does to the transistor is to
try to force it off while the primary is applying maximum voltage to
the collector. The load line shaping provided by a bit of parallel
capacitance really unloads this power dump into the collector each
pulse by getting the transistor almost completely turned off while
there is just a modest voltage on the collector, and then limiting the
peak voltage as the energy in the leakage inductance charges the cap,
instead of being consumed as heat by breaking down the transistor with
over voltage.
 
F

Frithiof Andreas Jensen

Jan 1, 1970
0
other ignition circuits I built. I wonder if there's something about
cascode that escapes me. First time using it.

2 things:

That the points are likely to bounce and wiring i.e. inductance.

A cascode is FAST, when the emitter opens the collector current is directed
through the collector-base junction until the collector-base storage has
been swept out. This takes a couple of usec. After that, the transistor will
switch of in a few nsec causing the collector voltage to shoot up.

Antisaturation diodes helps with the stored charge and a decoupling
capacitor on the base-ground will let the charge be dumped cleanly without
"exciting" long wires with usec current pulses of many A and fast risetimes.

If he switch bounces, the transistor will switch a high voltage few dosen
times every cycle - it might not like that.

An oscilloscope with decent probes would be useful.
 
Top