Mosfets for my stepper driver

[Iman Habib]

Hi guys.


I have been reading, experimenting with steppers (only unipolar steppers)
and building stepper driver for a couple of days now. I must say it ha been quite fun.

Now to my problem. Well actually its not really a problem.
I already have a working driver for my stepper but I'm thinking about
building another more simple construction using mosfets.

Not being much familiar with fets I thought I would ask you guys.

I checked the specs for the mosfet: BUZ10 from STMicroelectronics
( http://www.elfa.se/pdf/71/07100142.pdf ) and a couple of other
high effect mosfets, from other manufacturers, and they all seem to be quite
good at handling inductive kicks and capable of absorbing a lot of the
unwanted energy (of course some cooling would probably be needed).

So my question is whether or not its possible to simply drive some mosfets
with a hex buffer/driver (like the 7407) and let the mosfets directly drive
a unipolar stepper without any protection for the mosfets. Such as external
resistors, diodes or zeners. This would in effect reduce the cost
(the BUZ10 costs around a 1$) for building my stepper drivers and also
reduce pcb size a bit.

I guess I wont really know until I set this up later today but I thought I would
ask you guys and se if any of you have some objections or thoughts in this matter.


thank you for reading my post

//iman

--
So remember, kids! Every time you download a song off the Internet, you kill a baby panda!

 

[Iman Habib]

"Paul B" <pb@osiris.notthisbit.co.uk> wrote in message
news:a3bmnvc22a2psnfbsb1asheda6j5hpqfv9@4ax.com...

On Wed, 1 Oct 2003 12:53:17 +0200, "Iman Habib"
imanhabibREMOVETHIS@eml.cc> wrote:

Hi guys.


I have been reading, experimenting with steppers (only unipolar steppers)
and building stepper driver for a couple of days now. I must say it ha been quite fun.

Now to my problem. Well actually its not really a problem.
I already have a working driver for my stepper but I'm thinking about
building another more simple construction using mosfets.

Not being much familiar with fets I thought I would ask you guys.

I checked the specs for the mosfet: BUZ10 from STMicroelectronics
( http://www.elfa.se/pdf/71/07100142.pdf ) and a couple of other
high effect mosfets, from other manufacturers, and they all seem to be quite
good at handling inductive kicks and capable of absorbing a lot of the
unwanted energy (of course some cooling would probably be needed).

So my question is whether or not its possible to simply drive some mosfets
with a hex buffer/driver (like the 7407) and let the mosfets directly drive
a unipolar stepper without any protection for the mosfets. Such as external
resistors, diodes or zeners. This would in effect reduce the cost
(the BUZ10 costs around a 1$) for building my stepper drivers and also
reduce pcb size a bit.

I guess I wont really know until I set this up later today but I thought I would
ask you guys and se if any of you have some objections or thoughts in this matter.

Really difficult to say without knowing the power requirement of the
motors in question...

Oh i forgot that.

They are steppers that i yanked out from a couple of old HP printers.
PM55L series from minebea.

http://www.minebea-ele.com/en/product/motor/F_1000/pdf/motor-PM55L048.pdf

The specs say that the current/step is about 800mA but is more like
around 750mA last time i mesured. The drive voltage i use is around 12 to 24v.

cheers
//iman

 

[Scott Stephens]

Iman Habib wrote:

So my question is whether or not its possible to simply drive some mosfets
with a hex buffer/driver (like the 7407) and let the mosfets directly drive
a unipolar stepper without any protection for the mosfets. Such as external
resistors, diodes or zeners. This would in effect reduce the cost
(the BUZ10 costs around a 1$) for building my stepper drivers and also
reduce pcb size a bit.

Why are you so cheap? You better use the zener's. If its a 12V stepper,
use a 24 volt zener and feel them for warmth. If they got too hot to
touch, increase the power (or you can calculate the power from the
inductance and pulse width and rate - P = (1/2 L I^2)*pulse width*rate;
of course its more complicated than that, but that will give you an idea
of how much power you must dissipate if your motor doesn't convert it
into kinetic energy or heat )

Checkout my web site and the references to the app notes. Crappy little
chips like the 7407 can't handle much power, and if your motor doesn't
turn (fast enough), you will hammer your chips to death.

--
Scott

**********************************

DIY Piezo-Gyro, PCB Drill Bot & More Soon!

http://home.comcast.net/~scottxs/

**********************************

 

[Paul B]

On Wed, 1 Oct 2003 12:53:17 +0200, "Iman Habib"
<imanhabibREMOVETHIS@eml.cc> wrote:

Hi guys.


I have been reading, experimenting with steppers (only unipolar steppers)
and building stepper driver for a couple of days now. I must say it ha been quite fun.

Now to my problem. Well actually its not really a problem.
I already have a working driver for my stepper but I'm thinking about
building another more simple construction using mosfets.

Not being much familiar with fets I thought I would ask you guys.

I checked the specs for the mosfet: BUZ10 from STMicroelectronics
( http://www.elfa.se/pdf/71/07100142.pdf ) and a couple of other
high effect mosfets, from other manufacturers, and they all seem to be quite
good at handling inductive kicks and capable of absorbing a lot of the
unwanted energy (of course some cooling would probably be needed).

So my question is whether or not its possible to simply drive some mosfets
with a hex buffer/driver (like the 7407) and let the mosfets directly drive
a unipolar stepper without any protection for the mosfets. Such as external
resistors, diodes or zeners. This would in effect reduce the cost
(the BUZ10 costs around a 1$) for building my stepper drivers and also
reduce pcb size a bit.

I guess I wont really know until I set this up later today but I thought I would
ask you guys and se if any of you have some objections or thoughts in this matter.

Really difficult to say without knowing the power requirement of the
motors in question...
--

"I expect history will be kind to me, since I intend to write it."
- Winston Churchill

 

[John Jardine]

Iman Habib <imanhabibREMOVETHIS@eml.cc> wrote in message
news:blebmm$bae72$1@ID-168056.news.uni-berlin.de...

Hi guys.


I have been reading, experimenting with steppers (only unipolar steppers)
and building stepper driver for a couple of days now. I must say it ha
been quite fun.

Now to my problem. Well actually its not really a problem.
I already have a working driver for my stepper but I'm thinking about
building another more simple construction using mosfets.

Not being much familiar with fets I thought I would ask you guys.

I checked the specs for the mosfet: BUZ10 from STMicroelectronics
( http://www.elfa.se/pdf/71/07100142.pdf ) and a couple of other
high effect mosfets, from other manufacturers, and they all seem to be
quite
good at handling inductive kicks and capable of absorbing a lot of the
unwanted energy (of course some cooling would probably be needed).

So my question is whether or not its possible to simply drive some mosfets
with a hex buffer/driver (like the 7407) and let the mosfets directly
drive
a unipolar stepper without any protection for the mosfets. Such as
external
resistors, diodes or zeners. This would in effect reduce the cost
(the BUZ10 costs around a 1$) for building my stepper drivers and also
reduce pcb size a bit.

I guess I wont really know until I set this up later today but I thought I
would
ask you guys and se if any of you have some objections or thoughts in this
matter.


thank you for reading my post

file://iman

--
So remember, kids! Every time you download a song off the Internet, you
kill a baby panda!


The BUZ10 is no good for this job. Even with 5v on it's gate it's only just

starting to switch on. A TTL 7407 is no good as a driver either, as it's
logic high output is in the 4V area.
You need a 'logic level' FET driven straight from a 'HC' type 7407
(74HC07?).
Also you won't get away with running the FET barefoot. At switch off the
coils will generate 1000's of volts and take the FETS out.
A diode across the coil is no good either, as auto-transformer action with
coils wired as unipolar, causes the unswitched half of the winding to rise
to twice the supply voltage.
Therefore instead of a diode you need to use a Zener with a voltage of
greater than 2x the motor voltage. It won't get in the way of the normal
switching action but will short the switch off transients.
regards
john

 

[Dave M.]

John Jardine <john@jjdesigns.fsnet.co.uk> wrote in message
news:blfr0a$th$1@news8.svr.pol.co.uk...

Iman Habib <imanhabibREMOVETHIS@eml.cc> wrote in message
news:blebmm$bae72$1@ID-168056.news.uni-berlin.de...
Hi guys.


I have been reading, experimenting with steppers (only unipolar
steppers)
and building stepper driver for a couple of days now. I must say it ha
been quite fun.

Now to my problem. Well actually its not really a problem.
I already have a working driver for my stepper but I'm thinking about
building another more simple construction using mosfets.

Not being much familiar with fets I thought I would ask you guys.

I checked the specs for the mosfet: BUZ10 from STMicroelectronics
( http://www.elfa.se/pdf/71/07100142.pdf ) and a couple of other
high effect mosfets, from other manufacturers, and they all seem to be
quite
good at handling inductive kicks and capable of absorbing a lot of the
unwanted energy (of course some cooling would probably be needed).

So my question is whether or not its possible to simply drive some
mosfets
with a hex buffer/driver (like the 7407) and let the mosfets directly
drive
a unipolar stepper without any protection for the mosfets. Such as
external
resistors, diodes or zeners. This would in effect reduce the cost
(the BUZ10 costs around a 1$) for building my stepper drivers and also
reduce pcb size a bit.

I guess I wont really know until I set this up later today but I thought
I
would
ask you guys and se if any of you have some objections or thoughts in
this
matter.


thank you for reading my post

file://iman

--
So remember, kids! Every time you download a song off the Internet, you
kill a baby panda!


The BUZ10 is no good for this job. Even with 5v on it's gate it's only
just
starting to switch on. A TTL 7407 is no good as a driver either, as it's
logic high output is in the 4V area.
You need a 'logic level' FET driven straight from a 'HC' type 7407
(74HC07?).
Also you won't get away with running the FET barefoot. At switch off the
coils will generate 1000's of volts and take the FETS out.
A diode across the coil is no good either, as auto-transformer action with
coils wired as unipolar, causes the unswitched half of the winding to rise
to twice the supply voltage.
Therefore instead of a diode you need to use a Zener with a voltage of
greater than 2x the motor voltage. It won't get in the way of the normal
switching action but will short the switch off transients.
regards
john

I've stayed away from zeners for high speed suppression, because they are
very slow. I guess you can't beleive everything you read about these
devices.

Dave M.

 

[Fritz Schlunder]

I've stayed away from zeners for high speed suppression, because they are
very slow. I guess you can't beleive everything you read about these
devices.

Dave M.

Actually zener diodes will clamp overvoltages extremely fast. They normally
will respond to overvoltage conditions in sub nanosecond time spans, so for
pretty much all practical intents and purposes they respond instantly.

I'm not sure about how fast zener diodes undergo reverse recovery, they
aren't well characterized for this type of operation. There was a thread
about a year and a half ago on S.E.D. titled "Reverse Recovery of Zener
Diodes". Some people believed (but indecisively) they were slow while at
least one person (R. Legg) seemed to be confident they had very fast reverse
recovery time as well. Assuming he was right, or that you won't be
operating your zener diode in the forward conducting region then you
shouldn't have any problems with zener speed.

A diode that doesn't conduct forward current won't be subject to a reverse
recovery time/current when it becomes reverse biased. I should think in a
stepper motor application there is no reason to forward bias the device.

 

[Garrett Mace]

The BUZ10 is no good for this job. Even with 5v on it's gate it's only
just
starting to switch on. A TTL 7407 is no good as a driver either, as it's
logic high output is in the 4V area.
You need a 'logic level' FET driven straight from a 'HC' type 7407
(74HC07?).
Also you won't get away with running the FET barefoot. At switch off the
coils will generate 1000's of volts and take the FETS out.
A diode across the coil is no good either, as auto-transformer action with
coils wired as unipolar, causes the unswitched half of the winding to rise
to twice the supply voltage.
Therefore instead of a diode you need to use a Zener with a voltage of
greater than 2x the motor voltage. It won't get in the way of the normal
switching action but will short the switch off transients.
regards
john

What about running a diode from ground to coil, then another from coil to
power supply? For bipolar drivers you do this at both ends of the coil.

Actually ends up being a full-wave bridge across the coil. I should look
into integrated full-wave bridges to see what's available in terms of
switching speeds; might be a nice way to reduce part count and cost. I'm
just getting into the power and current control areas of a microstepping
bipolar driver board I'm designing, I already have the four-axis step
translator and microstep sequencer/sine table analog reference generators
implemented in a CPLD.

 

[Fritz Schlunder]

So remember, kids! Every time you download a song off the Internet, you
kill a baby panda!

Haha... Very funny. Perhaps it rings a little too true though for what
they teach kids these days.

The BUZ10 is no good for this job. Even with 5v on it's gate it's only
just
starting to switch on. A TTL 7407 is no good as a driver either, as it's
logic high output is in the 4V area.

Actually the 7407 isn't a regular 7400 series TTL type part. It is a hex
buffer with high voltage open collector outputs capable of sinking up to
40mA each and blocking up to 30 volts. The output doesn't have a pull up
driver. You normally use a pull up resistor to some higher voltage supply.
In the OP's case he should probably use some reasonable pull up resistance
like 1k Ohms up to a 12V supply. Used in this fashion it will work great
with the BUZ10.

You need a 'logic level' FET driven straight from a 'HC' type 7407
(74HC07?).
Also you won't get away with running the FET barefoot. At switch off the
coils will generate 1000's of volts and take the FETS out.

Actually if you look at the BUZ10 datasheet graciously provided by the OP,
it has a section titled "avalanche characteristics." Although the device
is rated to block 50V, you can in fact apply more than this if you so
desire, and if you externally limit the current through the device (as well
as total dissipation). At some voltage higher than 50V, the device will
begin to breakdown just as a zener diode does. For the BUZ10 it will likely
start really clamping maybe somewhere around 65 volts for instance. The
device will not be destroyed by an avalanche event so long as the datasheet
avalanche characteristics are not exceeded. For the BUZ10, that means you
must never let the avalanche current exceed a maximum of 10A peak. Beyond
this value and the device may somehow latch up and no longer be gate
controllable. As a result the device can very easily be destroyed at that
point. The other parameter that must never be exceeded while operating a
MOSFET in avanlanche mode is the die temperature should never exceed the
rated operating temperature. It is difficult to know precisely how the die
temperature will increase when you start dumping avalanche energy into it
since it depends upon the current waveform and duration. To make things
easier for designers the manufacturers often provide a maximum repetitive
and non repetetive avalanche energy rating under certain conditions of
inductive avalanche operation. Sometimes they also provide a graph
depicting maximum single pulse avalanche energy versus starting die
temperature to give you a better idea so you don't exceed the maximum die
temperature rating.

Some orthodox engineers don't like the idea of using a MOSFET (or any device
besides a zener) in avalanche mode. I say if it will simplify your circuit
while still remaining fully functional and reliable you should do it. One
drawback of using a MOSFET in avalanche is you don't get to choose the
avalanche voltage very precisely like a zener. You only guarantee the
minimum avalanche voltage, but the real voltage will likely be somewhat
higher. How much higher you don't get to know for certain but usually
several percent at least. Nevertheless using a MOSFET as a zener has some
other advantages such as the choice of package. You can easily get a cheap
TO220 power MOSFET that can claim to handle some beefy power rating like 100
Watts. Meanwhile 100 Watt zener diodes are practically nonexistant, and
they don't necessarily come in a nice easy to use heatsinkable package
either.

That said the stepper motor coils will not generate kVs of juice blasting
the BUZ10. When the MOSFETs switch off they avalanche non destructively at
some voltage probably around 65V or so for this 50V rated device. Suppose
the coils have 50mH of inductance. 0.5LI^2 energy storage at 800mA is 16mJ.
If the supply voltage is 24V to the coil, that will appear in series with
the induced coil voltage when the MOSFET switches off. Since the MOSFET
will avalanche around 65V, that means the coil will produce an extra
65-24=41V as seen outside of the coil. Since the coil has substantial
resistance some of the stored inductive energy will get absorbed in the
coil's own resistance. In addition to the inductive stored energy some
additional energy will be drawn from the supply since it is in series with
the coil. The energy from the supply, and the energy of the inductor not
dissipated in the coil's own resistance will be dissipated by the MOSFET in
avalanche. Without doing the actual exact calculations that energy is most
likey something in the vicinity of 16mJ for a 50mH coil. Since that is well
under the BUZ10's maximum rating of 150mJ, and we won't be repeating this
too frequently, the BUZ10 will probably take it without twitching an eye.
Some people will argue it will cause thermal heating and cooling cycle
stress on the die/die-package bond which may decrease reliability. To what
extent I know not, but from my personal experience with MOSFET avalanching I
would say they are very reliable and I wouldn't worry too much about it
unless maybe I was selling millions of them. A zener diode must also
undergo the same type of themal cycling stress, yet they seem to be quite
reliable when you don't seriously abuse them. Also I should imagine a
physically relatively very large power MOSFET die would naturally be more
rugged than a much smaller zener diode die.

A diode across the coil is no good either, as auto-transformer action with
coils wired as unipolar, causes the unswitched half of the winding to rise
to twice the supply voltage.
Therefore instead of a diode you need to use a Zener with a voltage of
greater than 2x the motor voltage. It won't get in the way of the normal
switching action but will short the switch off transients.
regards
john

 

[Paul Burridge]

On Thu, 2 Oct 2003 00:22:16 -0700, "Fritz Schlunder" <me@privacy.net>
wrote:

Actually the 7407 isn't a regular 7400 series TTL type part. It is a hex
buffer with high voltage open collector outputs capable of sinking up to
40mA each and blocking up to 30 volts. The output doesn't have a pull up
driver. You normally use a pull up resistor to some higher voltage supply.
In the OP's case he should probably use some reasonable pull up resistance
like 1k Ohms up to a 12V supply. Used in this fashion it will work great
with the BUZ10.

[...]

Very interesting. You certainly seem to know what you're talking
about, Fritz!
Just curious, but is your time-stamp set correctly? You appear to be
posting from Germany but your stamp's set for somewhere in the
mid-western USA.
--

"Windows [n.], A thirty-two bit extension and GUI shell to a sixteen bit patch
to an eight bit operating system originally coded for a four bit
microprocessor and produced by a two bit company."

 

[Iman Habib]

"Fritz Schlunder" <me@privacy.net> wrote in message
news:blgjh2$bj3ds$1@ID-203926.news.uni-berlin.de...

So remember, kids! Every time you download a song off the Internet, you
kill a baby panda!


Haha... Very funny. Perhaps it rings a little too true though for what
they teach kids these days.




The BUZ10 is no good for this job. Even with 5v on it's gate it's only
just
starting to switch on. A TTL 7407 is no good as a driver either, as it's
logic high output is in the 4V area.


Actually the 7407 isn't a regular 7400 series TTL type part. It is a hex
buffer with high voltage open collector outputs capable of sinking up to
40mA each and blocking up to 30 volts. The output doesn't have a pull up
driver. You normally use a pull up resistor to some higher voltage supply.
In the OP's case he should probably use some reasonable pull up resistance
like 1k Ohms up to a 12V supply. Used in this fashion it will work great
with the BUZ10.


You need a 'logic level' FET driven straight from a 'HC' type 7407
(74HC07?).
Also you won't get away with running the FET barefoot. At switch off the
coils will generate 1000's of volts and take the FETS out.


Actually if you look at the BUZ10 datasheet graciously provided by the OP,
it has a section titled "avalanche characteristics." Although the device
is rated to block 50V, you can in fact apply more than this if you so
desire, and if you externally limit the current through the device (as well
as total dissipation). At some voltage higher than 50V, the device will
begin to breakdown just as a zener diode does. For the BUZ10 it will likely
start really clamping maybe somewhere around 65 volts for instance. The
device will not be destroyed by an avalanche event so long as the datasheet
avalanche characteristics are not exceeded. For the BUZ10, that means you
must never let the avalanche current exceed a maximum of 10A peak. Beyond
this value and the device may somehow latch up and no longer be gate
controllable. As a result the device can very easily be destroyed at that
point. The other parameter that must never be exceeded while operating a
MOSFET in avanlanche mode is the die temperature should never exceed the
rated operating temperature. It is difficult to know precisely how the die
temperature will increase when you start dumping avalanche energy into it
since it depends upon the current waveform and duration. To make things
easier for designers the manufacturers often provide a maximum repetitive
and non repetetive avalanche energy rating under certain conditions of
inductive avalanche operation. Sometimes they also provide a graph
depicting maximum single pulse avalanche energy versus starting die
temperature to give you a better idea so you don't exceed the maximum die
temperature rating.

Some orthodox engineers don't like the idea of using a MOSFET (or any device
besides a zener) in avalanche mode. I say if it will simplify your circuit
while still remaining fully functional and reliable you should do it. One
drawback of using a MOSFET in avalanche is you don't get to choose the
avalanche voltage very precisely like a zener. You only guarantee the
minimum avalanche voltage, but the real voltage will likely be somewhat
higher. How much higher you don't get to know for certain but usually
several percent at least. Nevertheless using a MOSFET as a zener has some
other advantages such as the choice of package. You can easily get a cheap
TO220 power MOSFET that can claim to handle some beefy power rating like 100
Watts. Meanwhile 100 Watt zener diodes are practically nonexistant, and
they don't necessarily come in a nice easy to use heatsinkable package
either.

That said the stepper motor coils will not generate kVs of juice blasting
the BUZ10. When the MOSFETs switch off they avalanche non destructively at
some voltage probably around 65V or so for this 50V rated device. Suppose
the coils have 50mH of inductance. 0.5LI^2 energy storage at 800mA is 16mJ.
If the supply voltage is 24V to the coil, that will appear in series with
the induced coil voltage when the MOSFET switches off. Since the MOSFET
will avalanche around 65V, that means the coil will produce an extra
65-24=41V as seen outside of the coil. Since the coil has substantial
resistance some of the stored inductive energy will get absorbed in the
coil's own resistance. In addition to the inductive stored energy some
additional energy will be drawn from the supply since it is in series with
the coil. The energy from the supply, and the energy of the inductor not
dissipated in the coil's own resistance will be dissipated by the MOSFET in
avalanche. Without doing the actual exact calculations that energy is most
likey something in the vicinity of 16mJ for a 50mH coil. Since that is well
under the BUZ10's maximum rating of 150mJ, and we won't be repeating this
too frequently, the BUZ10 will probably take it without twitching an eye.
Some people will argue it will cause thermal heating and cooling cycle
stress on the die/die-package bond which may decrease reliability. To what
extent I know not, but from my personal experience with MOSFET avalanching I
would say they are very reliable and I wouldn't worry too much about it
unless maybe I was selling millions of them. A zener diode must also
undergo the same type of themal cycling stress, yet they seem to be quite
reliable when you don't seriously abuse them. Also I should imagine a
physically relatively very large power MOSFET die would naturally be more
rugged than a much smaller zener diode die.

A diode across the coil is no good either, as auto-transformer action with
coils wired as unipolar, causes the unswitched half of the winding to rise
to twice the supply voltage.
Therefore instead of a diode you need to use a Zener with a voltage of
greater than 2x the motor voltage. It won't get in the way of the normal
switching action but will short the switch off transients.
regards
john

Thank you very much Fritz.
You helped me out there a lot.
Just recently set up my mosfet based
stepper driver and it all seem to work fine. =)

cheers
//iman

 

[Alan]

"Fritz Schlunder" <me@privacy.net> wrote in message news:<blgjh2$bj3ds$1@ID-203926.news.uni-berlin.de>...

The BUZ10 is no good for this job. Even with 5v on it's gate it's only
just
starting to switch on. A TTL 7407 is no good as a driver either, as it's
logic high output is in the 4V area.


Actually the 7407 isn't a regular 7400 series TTL type part. It is a hex
buffer with high voltage open collector outputs capable of sinking up to
40mA each and blocking up to 30 volts. The output doesn't have a pull up
driver. You normally use a pull up resistor to some higher voltage supply.
In the OP's case he should probably use some reasonable pull up resistance
like 1k Ohms up to a 12V supply. Used in this fashion it will work great
with the BUZ10.

Yes I actually do this with a 7406 or 7 and my BUZ104's, works well
enough. One thing to watch out for if you make an all N channel
H-bridge though, turn on the high side first to reference the coil to
+V. If you turn on the low side first, the other side's mosfet source
is referenced to ground through the coil. You apply say 14V to it's
gate to turn it on, and that source pulls up to +V through the FET.
Well there is source-gate capacitance, as the source pulls up it can
push the gate over 20V and blow the gate. I didn't care about turn on
time, and was using 180K gate pull ups to lightly load the charged
pumped gate voltage. May not overshoot with a lower gate resistance,
but still something to watch out for, made for the occasional blown
gate if things were powered on with the running pattern going and the
motors attached. 10K resistors from each motor terminal to the
motor's +V and turning on high sides first keeps everything referenced
to the +V and are good ideas so this can't happen.

Alan

 

[Jim Thompson]

On Thu, 2 Oct 2003 11:42:53 -0700, "Fritz Schlunder" <me@privacy.net>
wrote:

"Paul Burridge" <pb@osiris.notthisbit.co.uk> wrote in message
news:sl4onv4tashcbe3vnvesbeg61p79qei8pg@4ax.com...

Very interesting. You certainly seem to know what you're talking
about, Fritz!
Just curious, but is your time-stamp set correctly? You appear to be
posting from Germany but your stamp's set for somewhere in the
mid-western USA.
--


Thanks. Yeah my time stamp actually is set correctly. I reside in Phoenix,
Arizona of the United States. I post using a free news server
(news.individual.net) which actually provides quite good service but is
located somewhere in Europe. My local ISP's news server (which is actually
newsfeeds.com) provides terrible service. Half the time the service simply
doesn't work, when it does work allot of messages don't appear at all, and
anytime I post (sometimes it eats my posts without posting them) using them
it automatically puts spam (advertising newsfeeds) on my messages. Not a
very desirable combination.

Aha! A fellow desert rat. Where are you in Phoenix? I'm in the
Ahwatukee Foothills.

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[Fritz Schlunder]

"Paul Burridge" <pb@osiris.notthisbit.co.uk> wrote in message
news:sl4onv4tashcbe3vnvesbeg61p79qei8pg@4ax.com...

Very interesting. You certainly seem to know what you're talking
about, Fritz!
Just curious, but is your time-stamp set correctly? You appear to be
posting from Germany but your stamp's set for somewhere in the
mid-western USA.
--

Thanks. Yeah my time stamp actually is set correctly. I reside in Phoenix,
Arizona of the United States. I post using a free news server
(news.individual.net) which actually provides quite good service but is
located somewhere in Europe. My local ISP's news server (which is actually
newsfeeds.com) provides terrible service. Half the time the service simply
doesn't work, when it does work allot of messages don't appear at all, and
anytime I post (sometimes it eats my posts without posting them) using them
it automatically puts spam (advertising newsfeeds) on my messages. Not a
very desirable combination.

 

[Fritz Schlunder]

"Jim Thompson" <invalid@invalid.invalid> wrote in message
news:t8sonvso7im5o4k4f40cueabukcrpm9tkt@4ax.com...

Aha! A fellow desert rat. Where are you in Phoenix? I'm in the
Ahwatukee Foothills.

...Jim Thompson

Hehe, yeah I'm a desert rat too. Ahwatukee Foothills eh'. They have a
funny name. I'm in west Mesa. I often read your posts bragging about how
hot it is or otherwise how our weather is going. I find them amusing
because I always think to myself, "Aww, yeah big deal. My weather is just
as extreme as yours. If you want to impress me with your weather you need
to boast about how much hotter it is than my own weather."

 

[Jim Thompson]

On Thu, 2 Oct 2003 12:49:39 -0700, "Fritz Schlunder" <me@privacy.net>
wrote:

"Jim Thompson" <invalid@invalid.invalid> wrote in message
news:t8sonvso7im5o4k4f40cueabukcrpm9tkt@4ax.com...


Aha! A fellow desert rat. Where are you in Phoenix? I'm in the
Ahwatukee Foothills.

...Jim Thompson


Hehe, yeah I'm a desert rat too. Ahwatukee Foothills eh'. They have a
funny name. I'm in west Mesa. I often read your posts bragging about how
hot it is or otherwise how our weather is going. I find them amusing
because I always think to myself, "Aww, yeah big deal. My weather is just
as extreme as yours. If you want to impress me with your weather you need
to boast about how much hotter it is than my own weather."

"West" Mesa? Near Tempe?

I tout the heat to keep people from moving here and crowding up the
place. Most Easterners *do* think 100°F is hot ;-)

...Jim Thompson
--
| James E.Thompson, P.E. | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona Voice480)460-2350 | |
| E-mail Address at Website Fax480)460-2142 | Brass Rat |
| http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.

 

[John Jardine]

Fritz Schlunder <me@privacy.net> wrote in message
news:blgjh2$bj3ds$1@ID-203926.news.uni-berlin.de...

So remember, kids! Every time you download a song off the Internet,
you
kill a baby panda!


Haha... Very funny. Perhaps it rings a little too true though for what
they teach kids these days.




The BUZ10 is no good for this job. Even with 5v on it's gate it's only
just
starting to switch on. A TTL 7407 is no good as a driver either, as it's
logic high output is in the 4V area.


Actually the 7407 isn't a regular 7400 series TTL type part. It is a hex
buffer with high voltage open collector outputs capable of sinking up to
40mA each and blocking up to 30 volts. The output doesn't have a pull up
driver. You normally use a pull up resistor to some higher voltage
supply.
In the OP's case he should probably use some reasonable pull up resistance
like 1k Ohms up to a 12V supply. Used in this fashion it will work great
with the BUZ10.

This will teach me not to trust my memory and to try be more attentive in
future!. Yes, although switching speed will be low, the 7407 will run the
BUZ10 OK.

You need a 'logic level' FET driven straight from a 'HC' type 7407
(74HC07?).
Also you won't get away with running the FET barefoot. At switch off the
coils will generate 1000's of volts and take the FETS out.


Actually if you look at the BUZ10 datasheet graciously provided by the OP,
it has a section titled "avalanche characteristics." Although the device
is rated to block 50V, you can in fact apply more than this if you so
desire, and if you externally limit the current through the device (as
well
as total dissipation). At some voltage higher than 50V, the device will
begin to breakdown just as a zener diode does. For the BUZ10 it will
likely
start really clamping maybe somewhere around 65 volts for instance. The
device will not be destroyed by an avalanche event so long as the
datasheet
avalanche characteristics are not exceeded. For the BUZ10, that means you
must never let the avalanche current exceed a maximum of 10A peak. Beyond
this value and the device may somehow latch up and no longer be gate
controllable. As a result the device can very easily be destroyed at that
point. The other parameter that must never be exceeded while operating a
MOSFET in avanlanche mode is the die temperature should never exceed the
rated operating temperature. It is difficult to know precisely how the
die
temperature will increase when you start dumping avalanche energy into it
since it depends upon the current waveform and duration. To make things
easier for designers the manufacturers often provide a maximum repetitive
and non repetetive avalanche energy rating under certain conditions of
inductive avalanche operation. Sometimes they also provide a graph
depicting maximum single pulse avalanche energy versus starting die
temperature to give you a better idea so you don't exceed the maximum die
temperature rating.

Some orthodox engineers don't like the idea of using a MOSFET (or any
device
besides a zener) in avalanche mode. I say if it will simplify your
circuit
while still remaining fully functional and reliable you should do it. One
drawback of using a MOSFET in avalanche is you don't get to choose the
avalanche voltage very precisely like a zener. You only guarantee the
minimum avalanche voltage, but the real voltage will likely be somewhat
higher. How much higher you don't get to know for certain but usually
several percent at least. Nevertheless using a MOSFET as a zener has some
other advantages such as the choice of package. You can easily get a
cheap
TO220 power MOSFET that can claim to handle some beefy power rating like
100
Watts. Meanwhile 100 Watt zener diodes are practically nonexistant, and
they don't necessarily come in a nice easy to use heatsinkable package
either.

That said the stepper motor coils will not generate kVs of juice blasting
the BUZ10. When the MOSFETs switch off they avalanche non destructively
at
some voltage probably around 65V or so for this 50V rated device. Suppose
the coils have 50mH of inductance. 0.5LI^2 energy storage at 800mA is
16mJ.
If the supply voltage is 24V to the coil, that will appear in series with
the induced coil voltage when the MOSFET switches off. Since the MOSFET
will avalanche around 65V, that means the coil will produce an extra
65-24=41V as seen outside of the coil. Since the coil has substantial
resistance some of the stored inductive energy will get absorbed in the
coil's own resistance. In addition to the inductive stored energy some
additional energy will be drawn from the supply since it is in series with
the coil. The energy from the supply, and the energy of the inductor not
dissipated in the coil's own resistance will be dissipated by the MOSFET
in
avalanche. Without doing the actual exact calculations that energy is
most
likey something in the vicinity of 16mJ for a 50mH coil. Since that is
well
under the BUZ10's maximum rating of 150mJ, and we won't be repeating this
too frequently, the BUZ10 will probably take it without twitching an eye.
Some people will argue it will cause thermal heating and cooling cycle
stress on the die/die-package bond which may decrease reliability. To
what
extent I know not, but from my personal experience with MOSFET avalanching
I
would say they are very reliable and I wouldn't worry too much about it
unless maybe I was selling millions of them. A zener diode must also
undergo the same type of themal cycling stress, yet they seem to be quite
reliable when you don't seriously abuse them. Also I should imagine a
physically relatively very large power MOSFET die would naturally be more
rugged than a much smaller zener diode die.

A diode across the coil is no good either, as auto-transformer action
with
coils wired as unipolar, causes the unswitched half of the winding to
rise
to twice the supply voltage.
Therefore instead of a diode you need to use a Zener with a voltage of
greater than 2x the motor voltage. It won't get in the way of the normal
switching action but will short the switch off transients.
regards
john

On the no-extra-protection-needed, I've parked firmly in the orthodox camp.
As the OP requested and also for my own interest, I read the BUZ10 data
sheet wrt it's avalanche performance. I saw no data there that would allow
freedom from the usual tyranny of extra protection components.
They make mention of 'single pulse avalanche energy <=150mJ' a low duty
cycle and that's about it!. As you mention, it would be usual to supply
extra graphs to back up the safe working area of the repetitive impulsed
energy so that the circuit can be designed with these limits in mind.
There's nothing there, hence as an orthodox designer and from normal data
sheet experience, I know not to make -any- assumptions on this aspect.
Yes ... 150mJ is (on the face of it) a fair amount of pulsed energy that
can be absorbed but take the example of a big size 34 'Slo Syn' stepper
I've just measured at 20mH (3.5Amps), that's 122mJ of energy being dumped
into the FET at say 250 impulses (1000 steps) a second. This is a lot of
watts of avalanche dissapation and up near to the FET's limit even on a
single impulse.
Yes, if the repetitive impulse v. temp graphs are supplied and if the motors
are small (low L, I, hence Joules) running without external protection is
worth a try. Otherwise, from that teacher called experience, it's a case of
65V ... 72V ... 1000's of volts
regards
john

 

[John Jardine]

Garrett Mace <g.ryan@macetech.com> wrote in message
news:yQPeb.4472$pg7.3385@twister.rdc-kc.rr.com...
[clip]

What about running a diode from ground to coil, then another from coil to
power supply? For bipolar drivers you do this at both ends of the coil.

Actually ends up being a full-wave bridge across the coil. I should look
into integrated full-wave bridges to see what's available in terms of
switching speeds; might be a nice way to reduce part count and cost. I'm
just getting into the power and current control areas of a microstepping
bipolar driver board I'm designing, I already have the four-axis step
translator and microstep sequencer/sine table analog reference generators
implemented in a CPLD.


Sounds a fascinating project!. I keep eyeing up affordable CPLD kit. One

I've been doing on and off these past two years (6Amps and only to 1/8th
step) is run by a cheap PIC outputting sine data to a 'R' ladder plus a
couple of 'hold' caps. It uses FET's though, so the 'free' body diodes can
be put to good use.
regards
john

 

[Garrett Mace]

Sounds a fascinating project!. I keep eyeing up affordable CPLD kit. One
I've been doing on and off these past two years (6Amps and only to 1/8th
step) is run by a cheap PIC outputting sine data to a 'R' ladder plus a
couple of 'hold' caps. It uses FET's though, so the 'free' body diodes can
be put to good use.
regards
john

Well, I decided to go with a CPLD because I figured it would actually be
cheaper than four separate microcontrollers and D/A modules and whatnot. So
far it fits into a 256 macrocell Xilinx Coolrunner, and I'm paring it down
pretty far so it might even make it into the 128 macrocell version. You can
buy a 256 macrocell XPLA3 from Xilinx's online store for $13. I'd say that's
pretty decent for handling all the necessary logic, plus being completely
reconfigurable.

I don't have a CPLD dev board, though Xilinx sells a Coolrunner dev board
for $50. I'm doing the initial testing in my Spartan IIe FPGA dev board.

Using R+2R ladders and smoothing caps is still a really nice approach, as it
can scale to much higher speeds. But I decided to use PWM for my analog
reference because I can clock an 8-bit PWM over 50 MHz, which keeps the
final PWM frequency well above my expected step rate and the current chopper
rate. Plus it uses much fewer output lines. With PWM I have a range of 256
values for each coil current, and can approximate a sine quite well. I tried
PWMs up to 12 bits, but did not notice much of an improvement over 8.

The power sections are detachable, and the logic output is configurable to
allow for different types of drive chips and bridges. So theoretically you
can start with an L298 bridges, then build ten-amp FET boards if you need to
later.