H-Bridge can't move motor if already moving?

[[email protected]]

Hi,
I have a 15:1 gear ratio in my permanent magnet DC motor/gear assembly.
When i turn the motor manually i can generate up to 40V of regenerative
voltage.

My H-bridge circuit is one of those high voltage 3-phase international
recifier modules, where i'm only using 2 of the phases to make an
h-bridge. I have bootstrap capacitors, which limits my duty cycle to
around 95%; anything higher may cause the bootstraps to not properly
charge. All is fine here.

My problem happens though when i'm generating some voltage by manually
moving the motor in the SAME direction where my H-bridge wants to move
the motor. I suspect that this voltage is causing my bootstraps
capacitors to not charge, causing me to go to "undervoltage lockout".
That's what tech support told me.

How do I get around this? I want to be able to start the motor moving
even if someone is already pushing the thing manually in that
direction.

 

[Winfield Hill]

[email protected] wrote...

I have a 15:1 gear ratio in my permanent magnet DC motor/gear assembly.
When i turn the motor manually i can generate up to 40V of regenerative
voltage.

My H-bridge circuit is one of those high voltage 3-phase international
recifier modules, where i'm only using 2 of the phases to make an
h-bridge. I have bootstrap capacitors, which limits my duty cycle to
around 95%; anything higher may cause the bootstraps to not properly
charge. All is fine here.

My problem happens though when i'm generating some voltage by manually
moving the motor in the SAME direction where my H-bridge wants to move
the motor. I suspect that this voltage is causing my bootstraps
capacitors to not charge, causing me to go to "undervoltage lockout".
That's what tech support told me.

How do I get around this? I want to be able to start the motor moving
even if someone is already pushing the thing manually in that
direction.

Is your H-bridge on (but at zero PWM) when someone starts pushing the
motor? Does it act as a brake to the manual pushing? When you turn
on the bridge, spending some time with both switches in the LO state
should get your caps charged. One thought: you could add a pair of
floating gate supplies (9V batteries?) to test the theory, or to get
the caps charged to begin with, or to keep them charged during long
off periods...

 

[[email protected]]

Thanks for the insight Winfield.

Is your H-bridge on (but at zero PWM) when someone starts pushing the
motor?

Yup, i have the H-bridge "armed" in the direction i want to move it.
That is, the bottom switch (IGBT) is constantly on, allowing the
bootstrap to charge up, while the top switch is at 0 duty cycle (thus
off).

Does it act as a brake to the manual pushing?

In the direction that I armed the h-bridge in, it does not act as a
brake. As a feature of the product, it is supposed to be easy to move.
When i sense this manual movement (via encoder changes), i want to
start PWM'ing. Or of course, when someone presses an actuator button,
i'll start moving. If i set the h-bridge to be armed in the opposite
direction (with the opposite bottom switch turned on), this will act as
a brake, but then this is contrary to the intended purpose of wanting
an "easy push" feature of the motor.


When you turn on the bridge, spending some time with both switches in
the LO state
should get your caps charged.

When you say "LO", you mean both of them turned on? I thought that
spending some time with the bottom switch turned on while the top is
turned off is what charged the bootstrap. Thus too long of an on duty
cycle ( >98% e.g.) would starve the bootstrap from charging up in time
for the next PWM cycle.

Okay, so I have the cap charged up while keeping the bottom switch low.
But like I said, any regenerative voltage manually generated will screw
up the role of the bootstrap. I think it takes something like 10V above
the motor rail voltage to properly charge the bootstrap cap to have
enough power to turn on the top switch. Since the bootstrap is charged
to 15V (my power supply to the drive), any manually generated voltage
above say, 5V (15 - 5 <= 10) will not properly turn on the top switch.
(I think this is what is happening)


One thought: you could add a pair of
floating gate supplies (9V batteries?) to test the theory, or to get
the caps charged to begin with, or to keep them charged during long
off periods...

Like i said above, the caps are fully charged, even for long periods.
In fact, it only takes a few microseconds to charge them. Any manually
generated voltage does screw it all up while the motor is stationary.
Once i get the motor going though, any manually generated speedup in
the same direction keeps the motor going though, so i have no problem
there.

I take it this is a drawback to using an H-bridge for a PMDC motor? Or
perhaps i'm missing something.

Thanks for the battery suggestion. I guess what your're getting at is
to eliminate the bootstraps altogether and to put a constant >10V
source in series with the motor rail voltage.

 

[legg]

Okay, so I have the cap charged up while keeping the bottom switch low.
But like I said, any regenerative voltage manually generated will screw
up the role of the bootstrap. I think it takes something like 10V above
the motor rail voltage to properly charge the bootstrap cap to have
enough power to turn on the top switch. Since the bootstrap is charged
to 15V (my power supply to the drive), any manually generated voltage
above say, 5V (15 - 5 <= 10) will not properly turn on the top switch.
(I think this is what is happening)

The bootstrap should charge independently, as the motor node is
externally forced, if the controler's supply is live. The bootstrap
circuit is simply a diode and cap reacting to peak output power node
voltages. These externally applied voltages cannot exceed the attached
driver's rails without being clamped to those rails.

If the externally applied voltage is out of sync with the controller's
drive, you might be triggering protective features in the controller,
or confusing it's logic to a point where it locks out.

RL

 

[Winfield Hill]

[email protected] wrote...

Thanks for the insight Winfield.

Is your H-bridge on (but at zero PWM) when someone starts pushing the
motor?

Yup, i have the H-bridge "armed" in the direction i want to move it.
That is, the bottom switch (IGBT) is constantly on, allowing the
bootstrap to charge up, while the top switch is at 0 duty cycle (thus
off).

Does it act as a brake to the manual pushing?

In the direction that I armed the h-bridge in, it does not act as a
brake. As a feature of the product, it is supposed to be easy to move.
When i sense this manual movement (via encoder changes), i want to
start PWM'ing. Or of course, when someone presses an actuator button,
i'll start moving. If i set the h-bridge to be armed in the opposite
direction (with the opposite bottom switch turned on), this will act as
a brake, but then this is contrary to the intended purpose of wanting
an "easy push" feature of the motor.


When you turn on the bridge, spending some time with both switches in
the LO state
should get your caps charged.

When you say "LO", you mean both of them turned on? I thought that
spending some time with the bottom switch turned on while the top is
turned off is what charged the bootstrap. Thus too long of an on duty
cycle ( >98% e.g.) would starve the bootstrap from charging up in time
for the next PWM cycle.

Okay, so I have the cap charged up while keeping the bottom switch low.
But like I said, any regenerative voltage manually generated will screw
up the role of the bootstrap. I think it takes something like 10V above
the motor rail voltage to properly charge the bootstrap cap to have
enough power to turn on the top switch. Since the bootstrap is charged
to 15V (my power supply to the drive), any manually generated voltage
above say, 5V (15 - 5 <= 10) will not properly turn on the top switch.
(I think this is what is happening)


One thought: you could add a pair of
floating gate supplies (9V batteries?) to test the theory, or to get
the caps charged to begin with, or to keep them charged during long
off periods...

Like i said above, the caps are fully charged, even for long periods.
In fact, it only takes a few microseconds to charge them. Any manually
generated voltage does screw it all up while the motor is stationary.
Once i get the motor going though, any manually generated speedup in
the same direction keeps the motor going though, so i have no problem
there.

This makes little sense to me. If both lower MOSFETs are turned on,
which means both H-bridge outputs are LO (that's what I was referring
to by LO), then the caps have to be charged. And you should experience
motor braking. Then when you detect a little externally-forced motion
you can release the brakes, and start PWM, which should also maintain
the cap charge. Hmm, as an aside, how would you hold on a slope?

I take it this is a drawback to using an H-bridge for a PMDC motor? Or
perhaps i'm missing something.

Thanks for the battery suggestion. I guess what your're getting at is
to eliminate the bootstraps altogether and to put a constant >10V
source in series with the motor rail voltage.

As a test, yes.

 

[Rich Grise]

Hi,
I have a 15:1 gear ratio in my permanent magnet DC motor/gear assembly.
When i turn the motor manually i can generate up to 40V of regenerative
voltage.

My H-bridge circuit is one of those high voltage 3-phase international
recifier modules, where i'm only using 2 of the phases to make an
h-bridge. I have bootstrap capacitors, which limits my duty cycle to
around 95%; anything higher may cause the bootstraps to not properly
charge. All is fine here.

My problem happens though when i'm generating some voltage by manually
moving the motor in the SAME direction where my H-bridge wants to move
the motor. I suspect that this voltage is causing my bootstraps
capacitors to not charge, causing me to go to "undervoltage lockout".
That's what tech support told me.

How do I get around this? I want to be able to start the motor moving
even if someone is already pushing the thing manually in that
direction.

What are the symptoms that you're having right now? What's the
supply voltage? What's the rating of this motor that it generates
40V?

What happens when somebody's pushing it, and somebody presses "start?"

Thanks,
Rich

 

[[email protected]]

What are the symptoms that you're having right now? What's the
supply voltage? What's the rating of this motor that it generates
40V?

It's a permanent magnet DC motor rated at something like 120VDC. Supply
voltage comes from a tap off the primary in my autotransformer, which
is 100VAC rectified, thus it's at 100 x 1.414 = 141VDC

What happens when somebody's pushing it, and somebody presses "start?"

I've got a PIC microcontroller optocoupled from all four IGBTs on the
drive module. The application is a pedestrian sliding door like you see
in a shopping center, hotel, airport, etc.

Okay, so when i press a button to "start" it, the motor is initially
stationary, my PWM duty cycle starts off slow, accelerates at a rate so
that the operation doesn't look "jerky". This part works okay.

When I am manually pushing it while the motor is still off, then i
press the 'start' button, i see my PWMs running on my oscilloscope
coming out of the microcontroller, but the motor drive module still
isn't kicking in. Once I release my hand, a split second later, the
motor will jump as if the drive module finally kicked in. I was told by
international rectifier's tech support that i am being inhibited by an
undervoltage lockout of the module.

Since the bootstrap/charge pump consists of a cap (and a diode in the
module), if the manually generated voltage is negative, then the diode
won't conduct.

 

[Winfield Hill]

[email protected] wrote...

When I am manually pushing it while the motor is still off, then i
press the 'start' button, i see my PWMs running on my oscilloscope
coming out of the microcontroller, but the motor drive module still
isn't kicking in. Once I release my hand, a split second later, the
motor will jump as if the drive module finally kicked in. I was told
by international rectifier's tech support that i am being inhibited
by an undervoltage lockout of the module.

What? The undervoltage lockout must refer to your H-bridge's
chip supply voltage, is that sagging for some reason? What's
the part number of your IR module?

 

[[email protected]]

It's the IRAMS10UP60B, supplied at 15V.

 

[Chopper]

SNIP

It's a permanent magnet DC motor rated at something like 120VDC. Supply
voltage comes from a tap off the primary in my autotransformer, which
is 100VAC rectified, thus it's at 100 x 1.414 = 141VDC

SNIP

Hi

Just a small point, noted you are using an autotransformer. Do you have an
isolating transformer in front of that? If not be VERY VERY VERY CAREFUL. If
do, just be VERY CAREFUL

Do you have a circuit for the H Bridge?

Regards

Chopper

<Remove 'Chopper' in Email address>

 

[[email protected]]

Do you have a circuit for the H Bridge?

The reference circuit is here:
http://www.iit.edu/~valemic1/hbridge.jpg

which is just a snapshot of the IRAMS10UP60B data manual.

By the way, i only use two of the phases and make it into an H-bridge,
while what is shown is a 3phase circuit.

In the circuit, the motor's return is also the 15V ground. When the
motor is manually turned, then i have a hunch it causes the diodes
connected to the bootstrap capacitors (CB1, CB2 in the diagram) to get
reverse-biased, and not conduct. Thus the bootstrap capacitors never
get enough juice to fire the top switch.

So i'm PWM'ing from my micro, but the drive's output is not responding
until this generated voltage subsides. It's gotta be that the
bootstrap capacitors are not charging up because the diode connected to
it gets reverse-biased.

Perhaps it's not an undervoltage lockout after all. UVLO is said to
occur at 11.1 volts for this drive module. I don't think this is the
case.

So not being too device-specific, has anyone in general ever had a
problem in starting up a DC motor that is already moving from some
external force?

 

[Winfield Hill]

[email protected] wrote...

The reference circuit is here: http://www.iit.edu/~valemic1/hbridge.jpg
which is just a snapshot of the IRAMS10UP60B data manual.

By the way, i only use two of the phases and make it into an H-bridge,
while what is shown is a 3phase circuit.

In the circuit, the motor's return is also the 15V ground. When the
motor is manually turned, then i have a hunch it causes the diodes
connected to the bootstrap capacitors (CB1, CB2 in the diagram) to
get reverse-biased, and not conduct. Thus the bootstrap capacitors
never get enough juice to fire the top switch.

I've examined the datasheets and I still don't buy your argument.
http://www.irf.com/product-info/datasheets/data/irams10up60b.pdf
BTW, the interior IC is their IR21363, similar to the ir2131, see
http://www.irf.com/product-info/datasheets/data/ir2136.pdf

You've got to show how the diodes get reverse biased, which implies
the lower IGBTs somehow fail despite their gates being driven on.
These IGBTs are capable of sinking or clamping more than 15A with
no more than 3.5V Vce drop (e.g. see irg4bc20k datasheet, fig 1,
http://www.irf.com/product-info/datasheets/data/irg4bc20k.pdf ).

Furthermore, you should be able to observe the actual situation with
your scope. For example, you can monitor the bridge current across
module pins 12 and 22. The module has an internal 33-milliohm sense
resistor, which means iTrip = 0.5/0.033 = 15A. You can monitor the
sense voltage with a pair of probes in differential fashion.

 

[Winfield Hill]

Winfield Hill wrote...

[email protected] wrote...

I've examined the datasheets and I still don't buy your argument.
http://www.irf.com/product-info/datasheets/data/irams10up60b.pdf
BTW, the interior IC is their IR21363, similar to the ir2131, see
http://www.irf.com/product-info/datasheets/data/ir2136.pdf

You've got to show how the diodes get reverse biased, which implies
the lower IGBTs somehow fail despite their gates being driven on.
These IGBTs are capable of sinking or clamping more than 15A with
no more than 3.5V Vce drop (e.g. see irg4bc20k datasheet, fig 1,
http://www.irf.com/product-info/datasheets/data/irg4bc20k.pdf ).

Sorry, that's figure 2.

 

[[email protected]]

I've been told recently by tech support to try adding diodes between
VS1 and VSS, as well as VS2 and VSS, to clamp any negative voltages.

I'll report what happens.

Thanks for reading into it Winfield. I'm more of an embedded software
guy than a motor guru. Good to have others looking into it too.

 

[[email protected]]

I just realized that the diodes i was told to add are merely redundant
to the body diodes already in the bottom IGBT. oh well.

 

[[email protected]]

CONCLUSION:

The bootstraps are definitely getting discharged when I push the motor
manually. I measured the voltage across the bootstrap capacitor with a
scope, and found that it gets fully discharged.

One kludge would be to put a diode and relay in parallel with each
other in series with the motor. If the relay is open, then the diode
will block the discharge path of the bootstrap capacitor. However, the
motor will now not move in the opposite direction unless the relay is
closed. So if anything, this is just a kludge, and in fact, the
mechanical relay will eventually fail.

That said, the method of turning on the bottom FET/IGBT to charge the
bootstrap capacitor in an h-bridge circuit is fundamentally flawed.

The advised method would be to alternately clock each diagonal pair in
the h-bridge at all times. So i guess at 50/50 duty cycle, you then
have two opposing equal forces.

 

[Winfield Hill]

[email protected] wrote...

CONCLUSION:

The bootstraps are definitely getting discharged when I push the motor
manually. I measured the voltage across the bootstrap capacitor with a
scope, and found that it gets fully discharged.

One kludge would be to put a diode and relay in parallel with each
other in series with the motor. If the relay is open, then the diode
will block the discharge path of the bootstrap capacitor. However, the
motor will now not move in the opposite direction unless the relay is
closed. So if anything, this is just a kludge, and in fact, the
mechanical relay will eventually fail.

That said, the method of turning on the bottom FET/IGBT to charge the
bootstrap capacitor in an h-bridge circuit is fundamentally flawed.

The advised method would be to alternately clock each diagonal pair in
the h-bridge at all times. So i guess at 50/50 duty cycle, you then
have two opposing equal forces.

BTW, we assume that you've got a so called "freewheeling" diode across
each of your IGBTs, as shown in the irams10up60b datasheet, right?

Not to be too stern, but unless you explain some aspect we are missing,
or make the relevant measurements, we have to discount your conclusion,
and your dramatic "fundamentally-flawed" assertion. If instead you want
to assert there's something fundamentally flawed within the irams10up60b
module, that may be.

You're saying the bottom IGBT is turned on, the driver-chip's Vdd power
supply is present, yet under this circumstance the associated high-side
driver cap can't become charged or if charged becomes discharged? As I
pointed out** - for the IGBTs in your machine, it shouldn't be possible
for the motor, acting as a generator, to overwhelm the turned-on IGBT.

Please tell us what happens when your push the motor:
(1) Does the driver IC's Vdd / Vcc power supply remain AOK?
(2) Is the bottom IGBT's gate voltage high, and the IGBT turned on?
(3) Does the IGBT's collector voltage stay near ground (+/- 1 volt)?
(4) Yet somehow the high-side driver capacitor tied to this same
IGBT's collector/drain becomes discharged? Nah, can't be.

Perhaps you can also tell is what's going on with the other IGBTs
in your bridge. We're keeping in mind it's your uP program that
decides which of the IGBT gates in the module to activate.


** copy of the post, the details of which you didn't address:

Author: Winfield Hill
Date: 30 May 2005 05:30:04 -0700
Message-ID: <[email protected]>

[email protected] wrote...

The reference circuit is here: http://www.iit.edu/~valemic1/hbridge.jpg
which is just a snapshot of the IRAMS10UP60B data manual.

By the way, i only use two of the phases and make it into an H-bridge,
while what is shown is a 3phase circuit.

In the circuit, the motor's return is also the 15V ground. When the
motor is manually turned, then i have a hunch it causes the diodes
connected to the bootstrap capacitors (CB1, CB2 in the diagram) to
get reverse-biased, and not conduct. Thus the bootstrap capacitors
never get enough juice to fire the top switch.

I've examined the datasheets and I still don't buy your argument.
http://www.irf.com/product-info/datasheets/data/irams10up60b.pdf
BTW, the interior IC is their IR21363, similar to the ir2131, see
http://www.irf.com/product-info/datasheets/data/ir2136.pdf

You've got to show how the diodes get reverse biased, which implies
the lower IGBTs somehow fail despite their gates being driven on.
These IGBTs are capable of sinking or clamping more than 15A with
no more than 3.5V Vce drop (e.g. see irg4bc20k datasheet, fig 2,
http://www.irf.com/product-info/datasheets/data/irg4bc20k.pdf ).

Furthermore, you should be able to observe the actual situation with
your scope. For example, you can monitor the bridge current across
module pins 12 and 22. The module has an internal 33-milliohm sense
resistor, which means iTrip = 0.5/0.033 = 15A. You can monitor the
sense voltage with a pair of probes in differential fashion.