PWM MOSFET driver question

[Richard Rasker]

Hi all,

I'm currently designing a full-bridge PWM controlled DC motor driver with a
24V/10A supply voltage. For driving each MOSFET half-bridge, I selected the
LM5106 driver IC.

As the '5106 is only available in a 20mil lead spacing MSOP, I can't do any
bench tests without designing a PCB right away, so I'd like to check up on
things to prevent costly mistakes (this is my first PWM motor driver).
Also, the datasheet (http://www.national.com/pf/LM/LM5106.html) is rather
concise.

Apparently, the high side driver derives its positive gate drive voltage via
the bootstrap capacitor. This doesn't mean that the PWM signal *must* have
a duty-cycle of less than 100% (full positive DC) for the driver to
function properly, or does it? Or does the chip contain some sort of
internal oscillator-plus-charge-pump, with the bootstrap capacitor only
necessary for storing the energy?

This is important, because I'm using a rather simple way of generating a PWM
signal with a triangle wave and an LM339 comparator(*). I don't need
precise speed control; it's only important that the motors don't start or
stop abruptly. This also means that for most of the time, the "PWM" signal
is actually a logic-level DC voltage.

Any information on this (including examples) is very welcome.

*: I'd rather use a microcrontroller-generated PWM signal -- but I'd need
six PWM outputs at least, two (both directions) for each motor. The
PIC16Fxx controllers I can program don't have this many PWM outputs, and I
calculated that it's impossible to generate even a 4-bit resolution 20kHz
PWM output in software: the program would have to check/change outputs at
20x16=320kHz; with a PIC running at 20MHz (5MHz internal clock), this would
require an interrupt rate of once per 16 clock cycles, which is not
possible.
The PIC30F2010 does have enough PWM outputs -- but my development
environmnent (both the assembler and programmer) can't handle this device.

Thanks in advance, best regards,

Richard Rasker

 

[Steve]

Hi all,

I'm currently designing a full-bridge PWM controlled DC motor driver with
a
24V/10A supply voltage. For driving each MOSFET half-bridge, I selected
the
LM5106 driver IC.

As the '5106 is only available in a 20mil lead spacing MSOP, I can't do
any
bench tests without designing a PCB right away, so I'd like to check up on
things to prevent costly mistakes (this is my first PWM motor driver).
Also, the datasheet (http://www.national.com/pf/LM/LM5106.html) is rather
concise.

Apparently, the high side driver derives its positive gate drive voltage
via
the bootstrap capacitor. This doesn't mean that the PWM signal *must* have
a duty-cycle of less than 100% (full positive DC) for the driver to
function properly, or does it? Or does the chip contain some sort of
internal oscillator-plus-charge-pump, with the bootstrap capacitor only
necessary for storing the energy?

This is important, because I'm using a rather simple way of generating a
PWM
signal with a triangle wave and an LM339 comparator(*). I don't need
precise speed control; it's only important that the motors don't start or
stop abruptly. This also means that for most of the time, the "PWM" signal
is actually a logic-level DC voltage.

Any information on this (including examples) is very welcome.

*: I'd rather use a microcrontroller-generated PWM signal -- but I'd need
six PWM outputs at least, two (both directions) for each motor. The
PIC16Fxx controllers I can program don't have this many PWM outputs, and I
calculated that it's impossible to generate even a 4-bit resolution 20kHz
PWM output in software: the program would have to check/change outputs at
20x16=320kHz; with a PIC running at 20MHz (5MHz internal clock), this
would
require an interrupt rate of once per 16 clock cycles, which is not
possible.
The PIC30F2010 does have enough PWM outputs -- but my development
environmnent (both the assembler and programmer) can't handle this device.

Thanks in advance, best regards,

Richard Rasker

Bootstrapping designs will require PWM to maintain the charge. They cannot
be used for a DC signal to the load. In the case of the 5106, the HB UVLO
function will kill the upper driver if the bootstrap voltage falls. This
could happen on very high duty cycle (~100%), and it could also happen if
the gate capacitance discharges the bootstrap cap too far. Make sure the BS
cap is big enough for your application if you stick with this part.

Having HB UVLO disable the upper driver may not be fatal in your
application. Its OK to supply a DC input to the chip if your oscillator
stops. But if you need a DC output from the driver in those conditions, you
are in trouble with this part.

BTW - for prototyping, check Digikey for "Surfboard" products that let you
mount a single SMD part on a small FR4 board which can then be attached to a
perfboard and hand-wired. They sell for about a buck and are great for
sample evaluations. All kinds of SMD patterns are available.

Steve

 

[Richard Rasker]

Hi all,

I'm currently designing a full-bridge PWM controlled DC motor driver with
a 24V/10A supply voltage. For driving each MOSFET half-bridge, I selected
the LM5106 driver IC. ....
Apparently, the high side driver derives its positive gate drive voltage
via the bootstrap capacitor. This doesn't mean that the PWM signal *must*
have a duty-cycle of less than 100% (full positive DC) for the driver to
function properly, or does it?

[snip]

Bootstrapping designs will require PWM to maintain the charge. They cannot
be used for a DC signal to the load.

Hm, that's a bit of a bummer. I was hoping that the device would feature an
internal charge pump or the likes. Then again, the way the capacitor is
connected does indeed suggest a mandatory non-100% duty cycle (which leads
to complications in the analog PWM generating circuit ...).

In the case of the 5106, the HB UVLO function will kill the upper driver
if the bootstrap voltage falls. This could happen on very high duty cycle
(~100%), and it could also happen if the gate capacitance discharges the
bootstrap cap too far. Make sure the BS cap is big enough for your
application if you stick with this part.

Having HB UVLO disable the upper driver may not be fatal in your
application. Its OK to supply a DC input to the chip if your oscillator
stops. But if you need a DC output from the driver in those conditions,
you are in trouble with this part.

I've been searching quite thoroughly, but virtually all combined
low+high-side drivers use a bootstrap capacitor.
But perhaps it's an idea to generate a voltage of +12 volts above VDD, and
hook that up to HB? Or could this cause trouble with the HS line? The
datasheet isn't clear at all on this point.

BTW - for prototyping, check Digikey for "Surfboard" products that let you
mount a single SMD part on a small FR4 board which can then be attached to
a perfboard and hand-wired. They sell for about a buck and are great for
sample evaluations. All kinds of SMD patterns are available.

I know, I have created several "standard" mini-PCB's myself for easily
hooking up individual SOT23-5, SSOP and other SMD devices. I just don't
have an MSOP-10 board -- and the only one from DigiKey I could find
(551012876-001-ND) is a bit on the pricey side, at $600 ;-)
So I guess I'll have to fire up the 'ole etch tank, and create something
myself.

Anyway, thank you for your reply, it helped me avoid at least one mistake.

Best regards,

Richard Rasker

 

[Tim Wescott]

Hi all,

I'm currently designing a full-bridge PWM controlled DC motor driver with
a 24V/10A supply voltage. For driving each MOSFET half-bridge, I selected
the LM5106 driver IC. ...
Apparently, the high side driver derives its positive gate drive voltage
via the bootstrap capacitor. This doesn't mean that the PWM signal *must*
have a duty-cycle of less than 100% (full positive DC) for the driver to
function properly, or does it?

[snip]

Bootstrapping designs will require PWM to maintain the charge. They cannot
be used for a DC signal to the load.

Hm, that's a bit of a bummer. I was hoping that the device would feature an
internal charge pump or the likes. Then again, the way the capacitor is
connected does indeed suggest a mandatory non-100% duty cycle (which leads
to complications in the analog PWM generating circuit ...).

In the case of the 5106, the HB UVLO function will kill the upper driver
if the bootstrap voltage falls. This could happen on very high duty cycle
(~100%), and it could also happen if the gate capacitance discharges the
bootstrap cap too far. Make sure the BS cap is big enough for your
application if you stick with this part.

Having HB UVLO disable the upper driver may not be fatal in your
application. Its OK to supply a DC input to the chip if your oscillator
stops. But if you need a DC output from the driver in those conditions,
you are in trouble with this part.

I've been searching quite thoroughly, but virtually all combined
low+high-side drivers use a bootstrap capacitor.
But perhaps it's an idea to generate a voltage of +12 volts above VDD, and
hook that up to HB? Or could this cause trouble with the HS line? The
datasheet isn't clear at all on this point.

BTW - for prototyping, check Digikey for "Surfboard" products that let you
mount a single SMD part on a small FR4 board which can then be attached to
a perfboard and hand-wired. They sell for about a buck and are great for
sample evaluations. All kinds of SMD patterns are available.

I know, I have created several "standard" mini-PCB's myself for easily
hooking up individual SOT23-5, SSOP and other SMD devices. I just don't
have an MSOP-10 board -- and the only one from DigiKey I could find
(551012876-001-ND) is a bit on the pricey side, at $600 ;-)
So I guess I'll have to fire up the 'ole etch tank, and create something
myself.

Anyway, thank you for your reply, it helped me avoid at least one mistake.

Best regards,

Richard Rasker

If you must have 100% duty cycle to the motor, then make a bootstrap
supply for yourself -- you could either do this with a little flyback
circuit, or a charge pump.

You can arrange for a guaranteed maximum duty cycle of less than 100%
using analog components if you really want to.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" gives you just what it says.
See details at http://www.wescottdesign.com/actfes/actfes.html

 

[Wimpie]

Hi all,

I'm currently designing a full-bridge PWM controlled DC motor driver with a
24V/10A supply voltage. For driving each MOSFET half-bridge, I selected the
LM5106 driver IC.

Hi Richard,

This doesn't answers your question directly, but can you use a
HIP4080A/4081A? This one has a built-in charge pump and is a full
bridge driver.

Best regards,

Wim

 

[Richard Rasker]

Hi Richard,

This doesn't answers your question directly, but can you use a
HIP4080A/4081A? This one has a built-in charge pump and is a full
bridge driver.

Hm, somehow I completely overlooked these in my search ... it seems that
the '4080 is ideal for my application, as it even features a comparator for
direct use with a triangle wave.

Thanks for the tip-off, I'll order a handful of these first thing tomorrow
morning!

Richard Rasker