Maker Pro
Maker Pro

pnp transistor for level shift circuit.

D

David Grant

Jan 1, 1970
0
I need a pnp transistor for a basic level shift circuit that brings a 0-5V
swing 400kHz PWM signal down 50V. (supply voltages are +5V, -50V). Shifted
signal is going into a IR2011 gate driver.

The problem I have is that most transistors that have decent enough
switching times (we're aiming for < 40ns rise time and fall time and
something with similar turn on and turn off delays so as to not distort the
signal too much) aren't rated for the Vce or Vbe that we need (most are only
15V). Lots of spec sheets don't even quote the timing specs, or the input
capacitance (which I understand is what determines the switching times).
I've spent several hours looking through spec sheets, and haven't even come
up with something close.

I'm at a loss, does anyone know where to look, or another way to approach
this? I'm pretty inexperienced with searching for parts based on specs
needed. Let me know if more info will help.

Thanks,

Dave
 
L

Larry Brasfield

Jan 1, 1970
0
David Grant said:
I need a pnp transistor for a basic level shift circuit that brings a 0-5V
swing 400kHz PWM signal down 50V. (supply voltages are +5V, -50V). Shifted
signal is going into a IR2011 gate driver.

The problem I have is that most transistors that have decent enough
switching times (we're aiming for < 40ns rise time and fall time and
something with similar turn on and turn off delays so as to not distort the
signal too much) aren't rated for the Vce or Vbe that we need (most are only
15V). Lots of spec sheets don't even quote the timing specs, or the input
capacitance (which I understand is what determines the switching times).
I've spent several hours looking through spec sheets, and haven't even come
up with something close.

A single PNP level shifter is unlikely to yield the
switching speed you need, IMHO. The problem
is that either the load resistance will be too large
to switch fast, or the dissipation will be enough
that a big and slow transistor will be necessary.
I'm at a loss, does anyone know where to look, or another way to approach
this? I'm pretty inexperienced with searching for parts based on specs
needed. Let me know if more info will help.

I would look into AC coupling the pulse with
some kind of DC restore circuit appropriate
for the expected range of PWM values. You
don't need to rapidly change the -50V level,
do you?
 
T

Tim Wescott

Jan 1, 1970
0
Larry said:
A single PNP level shifter is unlikely to yield the
switching speed you need, IMHO. The problem
is that either the load resistance will be too large
to switch fast, or the dissipation will be enough
that a big and slow transistor will be necessary.




I would look into AC coupling the pulse with
some kind of DC restore circuit appropriate
for the expected range of PWM values. You
don't need to rapidly change the -50V level,
do you?
Or couple it with a transformer if the 50V level is expected to flop around.
 
S

Spehro Pefhany

Jan 1, 1970
0
I need a pnp transistor for a basic level shift circuit that brings a 0-5V
swing 400kHz PWM signal down 50V. (supply voltages are +5V, -50V). Shifted
signal is going into a IR2011 gate driver.

The problem I have is that most transistors that have decent enough
switching times (we're aiming for < 40ns rise time and fall time and
something with similar turn on and turn off delays so as to not distort the
signal too much) aren't rated for the Vce or Vbe that we need (most are only
15V). Lots of spec sheets don't even quote the timing specs, or the input
capacitance (which I understand is what determines the switching times).
I've spent several hours looking through spec sheets, and haven't even come
up with something close.

I'm at a loss, does anyone know where to look, or another way to approach
this? I'm pretty inexperienced with searching for parts based on specs
needed. Let me know if more info will help.

Thanks,

Dave

A 10-15Mbps optoisolator might be just the ticket. Guaranteed
30-40nsec response times (if you can get ahold of any). eg NEC PS9121


Best regards,
Spehro Pefhany
 
D

David Grant

Jan 1, 1970
0
I just had an idea... The PWM is generated with a comparator at the input to
the level shifter at present.. Does it make more sense to generate the PWM
using a comparator that's biased at -50V, and instead level shift the audio
signal input to the comparator, hence generating (essentially) the same PWM
signal, but at -50V?

Thanks,

Dave
 
D

David Grant

Jan 1, 1970
0
Larry Brasfield said:
A single PNP level shifter is unlikely to yield the
switching speed you need, IMHO. The problem
is that either the load resistance will be too large
to switch fast, or the dissipation will be enough
that a big and slow transistor will be necessary.

The reference design for the Class D power audio amp at
http://www.irf.com/technical-info/refdesigns/iraudamp1.pdf (page 14) uses an
MMBT5401 (http://www.fairchildsemi.com/ds/MM/MMBT5401.pdf) to accomplish
this very thing. Two problems: the datasheet doesn't specify the switching
characteristics (it gives an imput capacitance vs. reverse bias voltage
graph - but i'm not sure how/if timing can be derived from this), plus I
need a device with leads, not surface mount.

Any clarifactions/suggestions appreciated as usual....

Dave
 
R

Ross Herbert

Jan 1, 1970
0
The reference design for the Class D power audio amp at
http://www.irf.com/technical-info/refdesigns/iraudamp1.pdf (page 14) uses an
MMBT5401 (http://www.fairchildsemi.com/ds/MM/MMBT5401.pdf) to accomplish
this very thing. Two problems: the datasheet doesn't specify the switching
characteristics (it gives an imput capacitance vs. reverse bias voltage
graph - but i'm not sure how/if timing can be derived from this), plus I
need a device with leads, not surface mount.

Any clarifactions/suggestions appreciated as usual....

Dave


So, the obvious answer is to use a 2n5401 TO92 device
http://www.fairchildsemi.com/ds/2N/2N5401.pdf
 
K

Ken Smith

Jan 1, 1970
0
I need a pnp transistor for a basic level shift circuit that brings a 0-5V
swing 400kHz PWM signal down 50V. (supply voltages are +5V, -50V). Shifted
signal is going into a IR2011 gate driver.

The problem I have is that most transistors that have decent enough
switching times (we're aiming for < 40ns rise time and fall time and
something with similar turn on and turn off delays so as to not distort the
signal too much) aren't rated for the Vce or Vbe that we need (most are only
15V). Lots of spec sheets don't even quote the timing specs, or the input
capacitance (which I understand is what determines the switching times).
I've spent several hours looking through spec sheets, and haven't even come
up with something close.


How about using a P channel MOSFET? They have fast and near equal
switching times. Take a look at Supertex.

You could also move the entire PWM system down to the -50V rail by making
a -50+5V to run the circuit on and level shifting the input.
 
L

Larry Brasfield

Jan 1, 1970
0
David Grant said:
The reference design for the Class D power audio amp at
http://www.irf.com/technical-info/refdesigns/iraudamp1.pdf (page 14) uses an
MMBT5401 (http://www.fairchildsemi.com/ds/MM/MMBT5401.pdf) to accomplish
this very thing. Two problems: the datasheet doesn't specify the switching
characteristics (it gives an imput capacitance vs. reverse bias voltage
graph - but i'm not sure how/if timing can be derived from this), plus I
need a device with leads, not surface mount.

Any clarifactions/suggestions appreciated as usual....

That transistor has 6 pF Cob at 10V. This may go down to
3 pF at your 50V bias. Add that to, say, 5 pF of input C
on the gate driver, and your collector circuit already has a
54 nS time constant. Not very promising to getting to your
40 nS, semi-symmetric rise and fall time goal. The drive
circuit from that schematic will suffer from Miller effect for
the falling output transition, add more dissymmetry.

Note that I did not claim a simple PNP level shifter could
not be built. I claimed it would not deliver the switching
performance you expressed a desire for.

I still believe that an AC coupled PWM logic signal with
a simple DC restore diode would be simpler and give you
the performance you hoped for.

Another approach would be to level shift your low
frequency error signal down to the -50V rail. You can
arrange that its offset error is inside the feedback loop
so that it contributes no error in a class 1 or higher servo.
Then convert it to PWM at the rail where you need that.
 
D

David Grant

Jan 1, 1970
0
Larry Brasfield said:
That transistor has 6 pF Cob at 10V. This may go down to
3 pF at your 50V bias. Add that to, say, 5 pF of input C
on the gate driver, and your collector circuit already has a
54 nS time constant. Not very promising to getting to your
40 nS, semi-symmetric rise and fall time goal. The drive
circuit from that schematic will suffer from Miller effect for
the falling output transition, add more dissymmetry.

Thanks for pointing this out.
Note that I did not claim a simple PNP level shifter could
not be built. I claimed it would not deliver the switching
performance you expressed a desire for.

I still believe that an AC coupled PWM logic signal with
a simple DC restore diode would be simpler and give you
the performance you hoped for.

Not sure what you mean by this (sorry - fresh out of university still)

Are you referring to something like:


in --C--+-- out
|
R
|
-50Vwith a diode in somewhere?Dave
 
G

Genome

Jan 1, 1970
0
David Grant said:
I need a pnp transistor for a basic level shift circuit that brings a 0-
5V swing 400kHz PWM signal down 50V. (supply voltages are +5V, -50V).
Shifted signal is going into a IR2011 gate driver.

The problem I have is that most transistors that have decent enough
switching times (we're aiming for < 40ns rise time and fall time and
something with similar turn on and turn off delays so as to not distort
the signal too much) aren't rated for the Vce or Vbe that we need (most
are only 15V). Lots of spec sheets don't even quote the timing specs, or
the input capacitance (which I understand is what determines the
switching times).I've spent several hours looking through spec sheets,
and haven't even come up with something close.

I'm at a loss, does anyone know where to look, or another way to approach
this? I'm pretty inexperienced with searching for parts based on specs
needed. Let me know if more info will help.

Thanks,

Dave

Drive it common base

1K
,---/\/\/\------,
| |
| e \|
(VIN) |-------(GND)
| c /|
| |
(GND) |---------(OUT)
|
/
\ 1K
/
\
|
|
(-50V)

DNA
 
P

Pooh Bear

Jan 1, 1970
0
David said:
I need a pnp transistor for a basic level shift circuit that brings a 0-5V
swing 400kHz PWM signal down 50V. (supply voltages are +5V, -50V). Shifted
signal is going into a IR2011 gate driver.

Simple.

Don't use the IR2011. That's for one gnd referenced output plus a high side
drive. As in SMPSs.

HP ( or whatever they call themselves these days ) make a lovely opto-isolated
super fast gate driver. 2A peak gate current IIRC.

Sorry don't have the device number to hand but sounds the ticket to me.


Graham
 
F

Fred Bloggs

Jan 1, 1970
0
I just had an idea... The PWM is generated with a comparator at the input to
the level shifter at present.. Does it make more sense to generate the PWM
using a comparator that's biased at -50V, and instead level shift the audio
signal input to the comparator, hence generating (essentially) the same PWM
signal, but at -50V?

It's not a PWM like you think- it is a self-oscillating sigma-delta
modulator- the level shifter is driven by the integrator, the comparator
is already referenced to the -50V supply. Use the circuit as shown or it
will not work. And when you pose a misleading question to NG, all you
end up with are irrelevant replies.
 
L

Larry Brasfield

Jan 1, 1970
0
David Grant said:
Not sure what you mean by this (sorry - fresh out of university still)

Are you referring to something like:


in --C--+-- out
|
R
|
-50Vwith a diode in somewhere?Dave

Yes. You would have to ensure that the transient effect
of bringing up the -50V supply did not turn your FETs
on for too long. The diode basically provides a current
path such that the voltage across the output resistor can
have a DC average value. In this case, the anode would
be connected to the -50V rail.
 
F

Fred Bartoli

Jan 1, 1970
0
Genome said:
Drive it common base

1K
,---/\/\/\------,
| |
| e \|
(VIN) |-------(GND)
| c /|
| |
(GND) |---------(OUT)
|
/
\ 1K
/
\
|
|
(-50V)

Yep, simple and efficient. In case you want really low falling time, just
add a small cap.
Depending on the parasitics loading, 100p should be about a good value.
 
D

David Grant

Jan 1, 1970
0
Fred Bloggs said:
It's not a PWM like you think- it is a self-oscillating sigma-delta
modulator- the level shifter is driven by the integrator, the comparator
is already referenced to the -50V supply. Use the circuit as shown or it
will not work. And when you pose a misleading question to NG, all you
end up with are irrelevant replies.

Thanks for the tip. I realize the IR reference circuit is
self-oscillating... I'm modifying their design to use a comparator instead.
When I said "at present" I meant currently in *my* design. Sorry for the
confusion.
 
F

Fred Bloggs

Jan 1, 1970
0
David said:
input to



Thanks for the tip. I realize the IR reference circuit is
self-oscillating... I'm modifying their design to use a comparator instead.
When I said "at present" I meant currently in *my* design. Sorry for the
confusion.

You're not going to "modify" it without mucking it up- you have already
run into the requirement for a fast level shift which the IRF design
does not require. The IRF design looks streamlined to the minimum- you
will not improve it.
 
R

Ross Herbert

Jan 1, 1970
0
That'd do it, thanks!

Dave

It should do... after all, many smt devices are based on their
forebear leaded devices and have the same generic number identifier.
ie. MMBT5401 = 2N5401. There will usually be only slight differences
in characteristics and in many cases the leaded device will have
better specs.
 
Top