Searching \ for 're [EE] Design Challenge - A better H Bridge MK2' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=design+challenge
Search entire site for: 'Design Challenge - A better H Bridge MK2'.

Truncated match.
PICList Thread
're [EE] Design Challenge - A better H Bridge MK2'
2005\08\16@202537 by Russell McMahon

face
flavicon
face
Criticism sought.
:-)

A month or so ago I was talking about H bridges.
Object was to drive a motor from 4 x C cells at about 150 mA - 300 mA run
current.
Motor ideally needed 4v+, less still worked within reason. Motor was unhappy
with more than say 6 volts and supply CAN rise to 8v or even 10v when a
mains adaptor is added.. Existing bridge dropped 2v plus across bridge. Also
had no voltage limiting.
Low component cost was an object. Operation on as flat as possible batteries
important. Off current should be under say 10 uA.
Commercial H bridge ICs tended to suffer from higher cost, poor saturation
voltage, too high minimum operating voltage.

It's worth noting that at the currents concerned (about C/10 for Alkalines)
a battery pack with open circuit or low current voltage of over 5 volts may
drop to 4 volts or less on load - ie internal impedance at this level of
load can be very poor.

____________

This is what I came up with.
It's more complex than may be expected but not as bad as it feels at 1st
glance.
Original had 6 transistors. This has 9. Plus 3 more in optional brake which
wasn't part of original.
It has been suggested that the voltage regulator may more elegantly be able
to be incorporated in the bridge proper. I agree, but it proved hard to do
while also maintaining essentially zero off current.

It works superbly.

Resistor values may not be final values.

Circuit (14 kB gif) is at

       http://russell.servepics.com/temp/r100b.gif

Motor Brake at top right is an optional extra. It works very well. R42 is
pivotal for proper operation. 100k values in brake could be lower but easily
met timing needs.

H bridge proper is at bottom.
Its job is to provide two saturated transistors for the motor drive when
driven on.
It does.

The "voltage limiter" is in fact a voltage regulator. It's job is to prevent
motor voltage from rising too high when supply is above spec. It does NOT
limit motor current except in its role as a voltage limiter. Rather than
grounding Q33 and Q32 emitter they are taken to effective ground by D31 or
D32 when the bridge is driven on. When the bridge is off the emitters float
and this reduces idle current to zero. The regulated voltage need not be
precise - just not overly high ever.

The motor brake has two MOSFETS in opposing series. When the bridge is off
the FETs are driven on by R41. Q43 is off. When the bridge is driven on Q43
is on and the FETs are off. The result is that the motor is shunted by R45
when the bridge is off and is not shunted at all when the bridge is on. R45
can be selected to suit required brake severity.

Q12, Q13, Q22, Q23, Q31 are something like BC337 / 2N2222.
Q31 dissipation needs to be watched.
Q41, Q42 need to be logic FETs to allow low battery voltages. The FETs are
the most expensive components.

Overall quiescent current is about zero. (Sub uA).

Q12/13/21/22 are NOT provided with base pulldowns as they have no voltage
applied when the bridge is off.

R33 is essential for correct voltage regulator action.
R35 is a good idea.

R42 is needed for good braking in both directions. Without it the floating
load makes FET clamping suspect in one motor direction.

It would be nice to think that an IC would do this job more easily and
cheaply. It would have to provide low Vsat (under 0.5 v total at say 300
mA), < 3mA input drive, very low off current (< 10 uA, motor voltage
limiting. Brake is an option which would be nice in an IC. `


Criticism invited.
How would I do it better while meeting the requirements ?



       Russell McMahon

2005\08\16@204711 by Richard Prosser

picon face
By changing the values of R33 & R31 to make them a voltage divider you
may be able to loose Z31 - at the expense of some temperature
variation.

What do the Anodes of D31, D32 connect to externally? It seems they
form a near short circuit through Q11 Q21, collectors (if turned on) &
Q13, Q23  bases. If limited by external resistor then OK, otherwise
you may sink excess current into this node. Or is it just a test
point?

RP

On 17/08/05, Russell McMahon <spam_OUTapptechTakeThisOuTspamparadise.net.nz> wrote:
{Quote hidden}

> -

2005\08\16@214901 by Russell McMahon

face
flavicon
face
> By changing the values of R33 & R31 to make them a voltage divider
> you
> may be able to loose Z31 - at the expense of some temperature
> variation.

It would work, but I have found in the past that (as you say) the
temperature dependence is very much higher. Also the regulation with
input voltage is poorer. Overall the zener seems a good compromise for
the price.

> What do the Anodes of D31, D32 connect to externally?

Nothing.

> It seems they
> form a near short circuit through Q11 Q21, collectors (if turned on)
> &
> Q13, Q23  bases. If limited by external resistor then OK, otherwise
> you may sink excess current into this node. Or is it just a test
> point?

In fact the circuit is drawn wrong !!!! - thanks for bringing that to
my attention! BUT it should work as intended as it was shown. The
cathodes SHOULD connect to Q12 and Q22 collectors (middle of bridge /
motor connections). In both cases the results would be about the same.
When the bridge is on one of these two diodes is pulled to nearly
ground and Q32 & Q33 have an earrth path. The "voltage regulator" uses
this as its ground reference. When the bridge is off this point floats
and the voltage regulator has no ground path and draws no current. I
could have used a seperate transistor as per Q43 (and did during
development) but the arrangement shown works well and saves a
transistor and a few Rs. It has an added bonus. As the bridge
saturation voltage rises with current the "ground" for the voltage
regulator also rises and the regulated voltage rises
slightly to compensate. This is a second order effect but welcome.


   Corrected circuit at http://russell.servepics.com.temp/r100b.gif


       Russell McMahon

2005\08\16@235047 by Vasile Surducan

face picon face
On 8/17/05, Russell McMahon <.....apptechKILLspamspam@spam@paradise.net.nz> wrote:
> Criticism sought.
>
>         http://russell.servepics.com/temp/r100b.gif
>
> Criticism invited.

Russel, wrong or right drawing this design can't be more cheaper that
a standard integrated H bridge. You forgot from the total cost of this
design the time spent with PCB design and component soldering. In fact
I think is more expensive than most of the standard H bridges adapted
for your requirements. Except maybe if you plan to produce it in China
with chinese transistors.

2 eurocent criticism done,

Vasile

2005\08\17@002501 by Russell McMahon

face
flavicon
face
>>         http://russell.servepics.com/temp/r100b.gif
>>
>> Criticism invited.


> Russel, wrong or right drawing this design can't be more cheaper
> that
> a standard integrated H bridge. You forgot from the total cost of
> this
> design the time spent with PCB design and component soldering. In
> fact
> I think is more expensive than most of the standard H bridges
> adapted
> for your requirements. Except maybe if you plan to produce it in
> China
> with chinese transistors.

A number of people have said this. But nobody so far has suggested a
"Standard H Bridge" that will meet my specs or be cheaper, let alone
both.

Initial production is in Taiwan. If it goes to decent volume
production will indeed be in China.

Design time is not an issue for several reasons. Soldering time is
relevant but as it will be autoinserted/placed this is not a vast
factor. PCB area matters but as it is part of a PCB of fixed existing
dimensions this also is less of a factor than otherwise.

Most H bridges will not run down to say 3v supply. 5v supply is OK for
the logic portion as long as the bridge proper is happy with less.

Most H bridges (not all) have terrible saturation voltages - 1.5v or
worse. I require say 0.5v total in most cases.

Many H bridges do not provide the dynamic braking function that my FET
circuit provides - but let's leave this out of cost and consideration
unless there is a bridge which does this.

Many bridges have substantial quiescent currents.

Most H bridges do not limit the maximum motor voltage when supply is
too high.

I looked at all the parts that people suggested previously. One came
closer than most. I couldn't find a source of supply but I imagine it
would have been available. But it still didn't meet all specs.

Even if I use a standard H bridge that I'm aware of I may have to eg
add a voltage regulator or a power shut down circuit to limit
quiescent current.

I'd be very happy to know of an H bridge that met all these specs
(except perhaps motor braking) and sold for say under $US0.40 in 1000
quantity.
By all means please do suggest ones that do. I'd be most grateful.




       Russell McMahon

2005\08\17@065736 by Gerhard Fiedler

picon face
Russell McMahon wrote:

>         http://russell.servepics.com/temp/r100b.gif

It seems that russell.servepics.com doesn't run an http server.

Gerhard

2005\08\17@073024 by Russell McMahon

face
flavicon
face

>>         http://russell.servepics.com/temp/r100b.gif

> It seems that russell.servepics.com doesn't run an http server.

It does, but my ISP has been down for about an hour with "network
problems" - just came back up. The server isn't much good without it's
ISP ;-).

           RM

2005\08\18@064832 by Gerhard Fiedler

picon face
Russell McMahon wrote:

>>>         http://russell.servepics.com/temp/r100b.gif
>
>> It seems that russell.servepics.com doesn't run an http server.
>
> It does, but my ISP has been down for about an hour with "network
> problems" - just came back up.

Seems it's down again. Maybe they go down regularly during my daytime hours
:)  Haven't had much luck trying to look at your schematics so far...

Gerhard

2005\08\18@073442 by Russell McMahon

face
flavicon
face
>>>>         http://russell.servepics.com/temp/r100b.gif

>>> It seems that russell.servepics.com doesn't run an http server.

>> It does, but my ISP has been down for about an hour with "network
>> problems" - just came back up.
>
> Seems it's down again. Maybe they go down regularly during my
> daytime hours
> :)  Haven't had much luck trying to look at your schematics so
> far...


My fault probably !!!!!!!!!!!
The file WAS named R100B.GIF
NOW it is named r100b.gif as above.

It seems that some attempts aren't case sensitive and some are.
There have been over 80 downloads of the file in the last few days.

I'll try not to do that again :-(


       Russell


2005\08\18@155355 by Dwayne Reid

flavicon
face
At 06:20 PM 8/16/2005, Russell McMahon wrote:
>Criticism sought.
>
>Motor Brake at top right is an optional extra. It works very well. R42 is
>pivotal for proper operation. 100k values in brake could be lower but easily
>met timing needs.

Just a silly thought - could you implement the motor brake by adding only 2
extra PNP transistors?  Collectors grounded on both, bases go to nodes D1 &
D2 on the schematic, emitters to matching motor leads (D1:M1, D2:M2).  Both
transistors are ON when both inputs are LOW, the transistor connected to a
motor terminal is turned OFF when that terminal is brought to Vbatt.

Downsides:

1) generates lots of heat and wastes much battery power while motor is
running if/when when Vbatt exceeds Vprocessor by more than 1 Vbe drop - you
would want to add resistors in series with the emitters to limit current in
that case.

2) The brake transistors don't saturate - reduces the effective brake action.

But possibly worth a try.

dwayne

--
Dwayne Reid   <dwaynerspamKILLspamplanet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(780) 489-3199 voice          (780) 487-6397 fax

Celebrating 21 years of Engineering Innovation (1984 - 2005)
 .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-
    `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'
Do NOT send unsolicited commercial email to this email address.
This message neither grants consent to receive unsolicited
commercial email nor is intended to solicit commercial email.

2005\08\19@002728 by Russell McMahon

face
flavicon
face
>>Motor Brake ...

> Just a silly thought - could you implement the motor brake by adding
> only 2 extra PNP transistors?  Collectors grounded on both, bases go
> to nodes D1 & D2 on the schematic, emitters to matching motor leads
> (D1:M1, D2:M2).  Both transistors are ON when both inputs are LOW,
> the transistor connected to a motor terminal is turned OFF when that
> terminal is brought to Vbatt.

It's a starter for 10 points, but the detail looks like it will make
it end up no better than the existing scheme.

- The motor needs a current path on both sides so a PNP on one side
alone isn't enough. Diodes appear ... .

- In the brake mode the transistor is turned on leading to the drear
quiescent current - and quite significant due to the current to be
shunted initially, were it not for the fact that ...

- The bridge is depowered by the top transistors being off so once
motor energy vanishes the PNPs turn off.  This may be OK but mat not -
depends on slowing motor characteristics at low voltage. Some brakes
have allowed the motor to slow rapidly and then coast slowly.

The currently used FETs have the advantage of being actively turned
hard on for as long as needed and drawing essentially no power in
standby mode, even though still on.



       RM

2005\08\19@082147 by Gerhard Fiedler

picon face
Russell McMahon wrote:

> My fault probably !!!!!!!!!!!
> The file WAS named R100B.GIF
> NOW it is named r100b.gif as above.

I don't think that this was the problem. If it had been, this would have
given me a successful server connection resulting in a 404 "file not found"
error or so. But the error was that my browser could not connect to an http
server at your address.

> It seems that some attempts aren't case sensitive and some are.
> There have been over 80 downloads of the file in the last few days.

I don't think case sensitivity is/was an issue. http is case sensitive, but
the file access from Apache on Windows probably isn't. Apache will ask the
OS for r100b.gif (as instructed by the http request), and the OS will be
happy to serve it R100B.GIF instead -- and Apache will never know this.
Probably.

Anyway, today I'm in.

Gerhard

2005\08\19@090415 by Russell McMahon

face
flavicon
face
> Anyway, today I'm in.

Do you like the brake circuit ? :-)

   http://russell.servepics.com/temp/r100b.gif

Too many parts, but it works well.
C41 delays the brake turn on slightly for reasons related to
implementation.

To get good switching in all cases theory and practice diverged a
bit - my theory anyway. The FET gates need to be taken well above the
V+ rail to get good switching. Even when R42 is changed to 1k (helps
pull the floating motor low) it needs the high level gate drive.

A single reed relay with a normally-closed contact and diodes D41 &
D42 to drive it would do as well.



       RM



2005\08\19@105100 by Michael Rigby-Jones

picon face
part 1 1365 bytes content-type:text/plain; (decoded quoted-printable)

Russell,

I had a look at the design, and figured if you are using MOSFETs, why not put them in the H bridge?

I haven't simulated this design, my worry would be making sure the MOSFETs turn off before the PNP switches on ( which I guess may not be possible without a few more components ).  However, it does solve the motor braking problem without any additional components, and without a high quiescent current.  Using two zeners does feel a bit clunky, I guess you could use steering diodes, and one zener and turn on Q5 and Q6 at the same time with no problems.

Anyway, that's my coffee break gone, back to work!

Cheers

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

part 2 8082 bytes content-type:image/gif; (decode)


part 3 35 bytes content-type:text/plain; charset="us-ascii"
(decoded 7bit)

2005\08\19@105559 by Michael Rigby-Jones

picon face
part 1 935 bytes content-type:text/plain; (decoded quoted-printable)

Dammit, literaly as my finger descended onto the mouse button to click 'send' I noticed the rather fatal flaw with that circuit!

Corrected (but probably still useless) version attached.

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

part 2 6069 bytes content-type:image/gif; (decode)


part 3 35 bytes content-type:text/plain; charset="us-ascii"
(decoded 7bit)

2005\08\19@120343 by Gerhard Fiedler

picon face
Russell McMahon wrote:

> Do you like the brake circuit ? :-)
>
>     http://russell.servepics.com/temp/r100b.gif
>
> Too many parts, but it works well.

I'm sure it does -- if you say so :)

I can't figure out why it should, though. How is the basis of Q44 ever 700
mV more negative than its emitter?

Gerhard

2005\08\19@122151 by Michael Rigby-Jones

picon face


>-----Original Message-----
>From: .....piclist-bouncesKILLspamspam.....mit.edu [EraseMEpiclist-bouncesspam_OUTspamTakeThisOuTmit.edu]
>Sent: 19 August 2005 17:04
>To: Microcontroller discussion list - Public.
>Subject: Re: re [EE] Design Challenge - A better H Bridge MK2
>
>
>Russell McMahon wrote:
>
>> Do you like the brake circuit ? :-)
>>
>>     http://russell.servepics.com/temp/r100b.gif
>>
>> Too many parts, but it works well.
>
>I'm sure it does -- if you say so :)
>
>I can't figure out why it should, though. How is the basis of
>Q44 ever 700 mV more negative than its emitter?

When Q43 is conducting, C42 is charged via D43.  When Q43 is switched off, R41 pulls the negative side of C42 up to the positive supply rail, so the positive side of C42 (and also the emitter of Q44) is at a potential of around double the supply voltage.

It's a fairly classic voltage multiplier circuit, though used in a "one shot" mode here.

Regards

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\19@122626 by Spehro Pefhany

picon face
At 01:03 PM 8/19/2005 -0300, you wrote:
>Russell McMahon wrote:
>
> > Do you like the brake circuit ? :-)
> >
> >     http://russell.servepics.com/temp/r100b.gif
> >
> > Too many parts, but it works well.
>
>I'm sure it does -- if you say so :)
>
>I can't figure out why it should, though. How is the basis of Q44 ever 700
>mV more negative than its emitter?

I've not tried to analyze this whole circuit, but...

C42/Q43/R41 forms a voltage doubler connected to the emitter of Q44.
Turn Q43 on for a while, then off, and the emitter of Q44 shoots up to
about 2*Vbat and decreases exponentially with time constant tau = R41*C42.

Best regards,

Spehro Pefhany --"it's the network..."            "The Journey is the reward"
speffspamspam_OUTinterlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com
->> Inexpensive test equipment & parts http://search.ebay.com/_W0QQsassZspeff


2005\08\19@140657 by Dwayne Reid

flavicon
face
At 10:24 PM 8/18/2005, Russell McMahon wrote:
>>>Motor Brake ...
>
>>Just a silly thought - could you implement the motor brake by adding only
>>2 extra PNP transistors?  Collectors grounded on both, bases go to nodes
>>D1 & D2 on the schematic, emitters to matching motor leads (D1:M1,
>>D2:M2).  Both transistors are ON when both inputs are LOW, the transistor
>>connected to a motor terminal is turned OFF when that terminal is brought
>>to Vbatt.
>
>It's a starter for 10 points, but the detail looks like it will make it
>end up no better than the existing scheme.
>
>- The motor needs a current path on both sides so a PNP on one side alone
>isn't enough. Diodes appear ... .

Nope - try it.  The reverse-biased PNP doesn't work all that well but it
does work.  But - a pair of diodes is pretty darned cheap.

>- In the brake mode the transistor is turned on leading to the drear
>quiescent current - and quite significant due to the current to be shunted
>initially, were it not for the fact that ...

Nope again - when both inputs are LOW, the quiescent current is no higher
than your original circuit.  While quiescent, both upper transistors are
turned OFF.  That means that both motor terminals are floating.  Only while
the motor is coasting and acting as a generator do the PNP brake
transistors conduct.

>- The bridge is depowered by the top transistors being off so once motor
>energy vanishes the PNPs turn off.  This may be OK but mat not - depends
>on slowing motor characteristics at low voltage. Some brakes have allowed
>the motor to slow rapidly and then coast slowly.

That's pretty much how this one would work.  It's intent is to allow
quicker brake action than just that single low-value resistor in parallel
with the motor.

>The currently used FETs have the advantage of being actively turned hard
>on for as long as needed and drawing essentially no power in standby mode,
>even though still on.

Exactly.  No question that it is a better brake circuit.  Question is: does
the extra cost justify the extra performance?

The real downside to what I suggested is that the brake transistors start
to turn on when Vmotor exceeds Vprocessor by more than 1 Vbe drop.

dwayne

--
Dwayne Reid   <@spam@dwaynerKILLspamspamplanet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(780) 489-3199 voice          (780) 487-6397 fax

Celebrating 21 years of Engineering Innovation (1984 - 2005)
 .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-
    `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'
Do NOT send unsolicited commercial email to this email address.
This message neither grants consent to receive unsolicited
commercial email nor is intended to solicit commercial email.

2005\08\19@190652 by Russell McMahon

face
flavicon
face
>>> Do you like the brake circuit ? :-)
>>>     http://russell.servepics.com/temp/r100b.gif
>>> Too many parts, but it works well.

>>I can't figure out why it should, though. How is the basis of
>>Q44 ever 700 mV more negative than its emitter?

> When Q43 is conducting, C42 is charged via D43.  When Q43 is
> switched off, R41 pulls the negative side of C42 up to the positive
> supply rail, so the positive side of C42 (and also the emitter of
> Q44) is at a potential of around double the supply voltage.
> It's a fairly classic voltage multiplier circuit, though used in a
> "one shot" mode here.

__

The voltage step up is standard enough - but by itself it didn't do
the job. The problem was that without Q44, diode D43 raises the gate
to V+ while the motor is being driven and the FETs would turn on to a
limited extent. I needed something which would completely remove the
gate drive until it was boosted to ~~~2 x V+. Q44 does this. Its base
is held at V+ so as long as C42 is less than V+ then Q44 is off. When
the right hand end of C42 and therefore Q44 emitter is raised to ~~ 2
x V+, Q44 is turned on and it's collector drives the FET gates high.
They stay high unto EITHER a new input cycle starts (when C42 is
pulled low and removes drive) or until C42 discharges via Q33 base and
R46. This sets the maximum brake fully on time and also allows it to
be instantly removed if a new cycle starts. R47 is optionally provided
to give the gates a defined low value relative to their sources when
Q44 is turned off but in practice  is hardly needed. I've never seen
this combination of voltage booster and emitter switched output stage
in use before but it's so useful that I'm sure it must have been
"invented" many times before. Odds are it's a standard FET high side
driver which I've failed to notice :-). I suspect it's switching speed
may be limited by practical aspects, but it works well here. Unlike
most voltage doublers it is not allowed to have an output "reservoir"
capacitor as the voltage must be removed as soon as a new cycle
begins.



           RM

2005\08\20@082157 by Gerhard Fiedler

picon face
Russell McMahon wrote:

> Do you like the brake circuit ? :-)

Now that I understood it... :)  (Thanks to Spehro and Michael.)

I don't remember who it was who suggested this before, but I liked the idea
of using the brake FETs in the bridge and getting the brake effect "for
free". Did you check out that circuit?

Gerhard

2005\08\20@091838 by Russell McMahon

face
flavicon
face
> I don't remember who it was who suggested this before, but I liked
> the idea
> of using the brake FETs in the bridge and getting the brake effect
> "for
> free". Did you check out that circuit?

I've looked at it but haven't yet breadboarded it. Because of the way
the motor behaves its VERY advisable to try ideas in real world
situations.Simulations and theory can let you down badly here.

You may recall the discussion a few weeks ago where various in bridge
ideas were tried. My first design applied regulation to the bridge
proper. The ideal design would integrate bridge, voltage regulator and
brake. In many designs the need for a near zero quiescent current
makes life harder.

One issue with having the FET's in the bridge proper is the need to
turn them on only AFTER the drive is removed and while the top half of
the bridge is powered off. This isn't hard but adds to circuit
complexity. Something like what I have done to turn on the FETs now
would probably work. Having the motor provide drive to its brake
transistors (bipolar or FET) is an obvious idea as it's potentially
self controlling. I tried back to back transistors which 'reached
across the motor' for their base drive. All such schemes so far are
either less than ideal brakes and/or require extra control circuitry.
TANSTAAFL seems to apply thoroughly here.

A good thing about the FET circuit that I have so far is that it works
well and is essentially independent of other interactions. I can
control its braking effect, attack time and on time easily and
quiescent current is essentially zero.

I'll happily use a simpler / cheaper design if it's as effective (or
even good enough). I'll probably be playing with some of the ideas
suggested in the next few days.


       RM

2005\08\21@094838 by Gerhard Fiedler

picon face
Russell McMahon wrote:

>> I don't remember who it was who suggested this before, but I liked the
>> idea of using the brake FETs in the bridge and getting the brake effect
>> "for free". Did you check out that circuit?

> You may recall the discussion a few weeks ago where various in bridge
> ideas were tried. My first design applied regulation to the bridge
> proper. The ideal design would integrate bridge, voltage regulator and
> brake. In many designs the need for a near zero quiescent current
> makes life harder.

Yes, and on second thought, the 3.5 V minimum supply voltage doesn't make
the use of FETs as bridge switchers exactly easy.

With the types of products I mostly work on, I rarely get to dive so deep
into how to employ single transistors :)  It's quite interesting to follow
this development. Thanks for sharing -- and please keep the updates coming.

(How's that for a model of open source development? In a sense, that's
pretty much what is going on here...)

Gerhard

2005\08\21@235645 by Russell McMahon

face
flavicon
face
> I had a look at the design, and figured if you are using MOSFETs,
> why not put them in the H bridge?

I looked at the design quite carefully at the time, I thought, but
missed it's key point. It's brilliant. I don't know if it's workable,
but brilliant things don't have to work. I looked at it just now, saw
it didn't do what you said (I thought) started to explain why, and
then realised I'd misunderstood it. The FETs being turned hard on in
the both-inputs-off state is of course the key.

**** NOTE ****     Anyone intetrested DO look at Michael's circuit.
The key is that BOTH FETS are on as of right when the bridge is at
rest. This is acheieved with no quiescent current (other than FET gate
leakage of typically << 1 uA). This provides an "automatic" motor
brake. The voltage regulator may work :-). Q5, Q6 will need base
pulldowns (based on my experiences with this sort of regulator).

__________

I'm sure it will have problems as is (life's like that) BUT it has the
makings of an excellent solution.

I'll go and simulate it in the only way that works when you are using
things like motors (hard silicon) and see what happens. Expect a
report and quite possibly a slightly modified real world design
"soon".

> I haven't simulated this design, my worry would be making sure
> the MOSFETs turn off before the PNP switches on
> ( which I guess may not be possible without a few more components ).

A small amount of "shoot though" may be tolerable in this case,
although I'll try to avoid it.

> However, it does solve the motor braking problem without any
> additional components,

It does !!!!!!!

> and without a high quiescent current.

Yes.

> Using two zeners does feel a bit clunky, I guess you
> could use steering diodes, and one zener and turn on
> Q5 and Q6 at the same time with no problems.

Probably. but you will probably need to retain both R8 and R9 to allow
'spreading' of  Q5 & Q6.

___

I've just seen a potential problem, similar to what I had when I used
a similar arrangement of the bridge to limit current rather than
voltage. As you start to switch off eg Q2 to limit current you'll also
start to turn on M1. This is a way to limit motor voltage but at the
expense of dissipation in M1. Fortunately the Vth of the FET helps us
here (for a change). As long as M1 gate is below pinchoff voltage it
wont shunt current from the motor. Suitable design of R7/R4 can
arrange that Vth is never reached during normal voltage regulating
operation.

Also - with my design the brake and FETs are an add on option. With
yours the FETs (and brake) are integral to the bridge operation. This
is not terrible but is worth noting.

> Anyway, that's my coffee break gone, back to work!

Brilliant. Have a virtual coffee on me at this stage pending
investigation.
At this stage the gains look moderately modest over my working design
but very worthwhile if they can be made to work in the real world. I
have 2 x FET and  10 bipolars. You have 2 FETS and 6 bipolars. Glue to
be decided.

The latest console turned up by Courier from Taiwan today and has been
taken away to courier's not-at-home-must-sign-see-you-tomorrow-land.
So I have a day to get this right :-).


               Russell McMahon




Cheers

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly
prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================


--------------------------------------------------------------------------------


> --

2005\08\23@113835 by Vasile Surducan

face picon face
On 8/19/05, Michael Rigby-Jones <KILLspamMichael.Rigby-JonesKILLspamspambookham.com> wrote:
> Dammit, literaly as my finger descended onto the mouse button to click
> 'send' I noticed the rather fatal flaw with that circuit!
>
> Corrected (but probably still useless) version attached.

 Maybe is brilliant, but those two FET's used as turn OFF the motor
with inputs both on zero, may be replaced with four diodes (BTW which
are lost from the initial design) mounted from the motor  ends to Vcc
and GND. In this way they suppress the self induction when motor is
turned ON and assure a curent path when the motor is turned OFF.

Vasile

2005\08\24@045311 by Michael Rigby-Jones

picon face


{Quote hidden}

Are you looking at the wrong circuit?  If you replaced the two FET's in my design with four diodes, the circuit would not longer work.

Mike

=======================================================================
This e-mail is intended for the person it is addressed to only. The
information contained in it may be confidential and/or protected by
law. If you are not the intended recipient of this message, you must
not make any use of this information, or copy or show it to any
person. Please contact us immediately to tell us that you have
received this e-mail, and return the original to us. Any use,
forwarding, printing or copying of this message is strictly prohibited.
No part of this message can be considered a request for goods or
services.
=======================================================================

2005\08\24@071316 by Russell McMahon

face
flavicon
face
>>  Maybe is brilliant, but those two FET's used as turn OFF the
>>motor with inputs both on zero, may be replaced with four
>>diodes (BTW which are lost from the initial design) mounted
>>from the motor  ends to Vcc and GND. In this way they suppress
>>the self induction when motor is
>>turned ON and assure a curent path when the motor is turned OFF.

> Are you looking at the wrong circuit?  If you replaced the two FET's
> in my design with four diodes, the circuit would not longer work.

I'm fairly sure that he means that if you take the original H bridge
circuit that you started from, or indeed almost any std bipolar
transistor H bridge, that by if reverse polarity diodes are placed
across the 4 bridge transistors they will conduct when the motor runs
as a generator and the circulating currents will brake the motor.
Which is true. To a point. This will work quite well down to around
1.5 volts (2 diode droops under substantial current) but will fade
away completely at the end. Depending on drive voltage, motor load
(which in this case varies with direction) the consequent slowing may
be enough. Or may not. The motor is geared down and drives a
heavilyish spring loaded cable which controls an eddy current brake.
To meet regulatory standards the applied load needs to be positioned
to within certain accuracy limits.

If I was in overall control of the project I would use simple software
to ensure accurate positioning, by stopping a little before position
and then 'nudging' the motor appropriately. This would be entirely
acceptable in this application. BUT I have at best advisory or less
role in that area and NIMBY is very strong.


       RM


2005\08\24@203306 by Peter

picon face

On Wed, 24 Aug 2005, Russell McMahon wrote:

> heavilyish spring loaded cable which controls an eddy current brake. To meet
> regulatory standards the applied load needs to be positioned to within
> certain accuracy limits.

Imho in this case a passive brake is not enough. The strong tension on
the gearing will cause stopping to be faster in one direction than in
the other. Especially if a worm gear mechanism is involved.

Peter

2005\08\24@213745 by Russell McMahon

face
flavicon
face
> ... and NIMBY is very strong.

That should have been NIH, but there's NIMBY too :-)

   http://en.wikipedia.org/wiki/Not_Invented_Here
   http://en.wikipedia.org/wiki/NIMBY




       RM

2005\08\25@205236 by Vasile Surducan

face picon face
On 8/24/05, Russell McMahon <RemoveMEapptechspamTakeThisOuTparadise.net.nz> wrote:
{Quote hidden}

Russel, I love to talk with you!
If will be so easy for the all others to understood my ideeas, my life
could be much simple.

To lower the voltage thresold of the OFF voltage I was thinking to
your original design, where the supply voltage can be shutted down to
0.2V or so when both commands are low. A 3.5V motor running at 1.5V
could means a stopped motor. Or not, as you say. Indeed with a gear
you coud have some problems (I've guess there are minor).

cheers,
Vasile


The motor is geared down and drives a
{Quote hidden}

> --


're [EE] Design Challenge - A better H Bridge MK2'
2005\09\07@095220 by Russell McMahon
face
flavicon
face
part 1 1898 bytes content-type:text/plain; (decoded 7bit)

Michael said -

> I had a look at the design, and figured if you are using
> MOSFETs, why not put them in the H bridge?

and provided a circuit.

Brilliant!

As is to be expected when one tangles with Murphy, it had problems in
practice. But the idea of the two low side FETs which are *ON* when
the bridge is off thereby acting as an active brake, proved extremely
successful. The FET / bipolar gate/base drive circuit was undesignable
as originally shown due to the supply voltage range - something which
is not obvious until you do some calculations.

The integral voltage regulator proved too hard to make work well. I've
taken it outside the bridge at the cost of only one more transistor.
It can be omitted if not needed. While such a simple regulator *looks*
trivial to design it can take some hair tearing to make it work well.

I'll explain the circuit operation further in due course if there's
any interest, but here's an outline:

All turned-on transistors saturate nicely until the voltage gets above
maximum when the voltage regulator limits available bridge voltage.

D31, D32 reduce quiescent current from typically 30+uA to well under 1
uA.

D41, D42 turn on  the regulator only when there is drive.

R41, R46 turn both FETs on when the bridge is off.

Component values suit the current design and may need to be altered
for other loads.
Reverse polarity diodes on the bridge transistors are not used and
largely not needed due to the FET clamping action.

This design works superbly in practice. I haven't quite finalised it.
Said reverse protection diodes may yet appear, but probably not.
Oscilloscope will tell.  Circuit is drawn "strangely" to meet certain
drawing constraints caused by a hobbled antiquated PCB package. After
next week my circuits may never look like this ever again :-).



       RM





part 2 68455 bytes content-type:image/jpeg; (decode)


part 3 35 bytes content-type:text/plain; charset="us-ascii"
(decoded 7bit)

2005\09\07@141449 by Dwayne Reid

flavicon
face
At 07:37 AM 9/7/2005, Russell McMahon wrote:

>This design works superbly in practice. I haven't quite finalised
>it. Said reverse protection diodes may yet appear, but probably not.
>Oscilloscope will tell.  Circuit is drawn "strangely" to meet
>certain drawing constraints caused by a hobbled antiquated PCB
>package. After next week my circuits may never look like this ever again :-).

I have some minor suggestions that might reduce component count slightly:

Approach #1: Eliminate D31, D32.  Move lower ends of R41, R46 to
bases of Q12, Q22 and reduce to 10K or lower.  Disconnect lower end
of R34 from D41, D42 and move it to incoming Vbatt.  Move D41, D42:
anodes are tied to emitters of Q32, Q33 and R33.  Connect cathodes to
collectors of Q11, Q21.  Idea is that the voltage regulator is
enabled by Q11 or Q21.

Advantage of this is there is no leakage path to gnd - quiescent
current should approach 0.

Approach #2: Your circuit as posted: use a resistive sum to turn on
the voltage regulator instead of D41, D42.  In other words, replace
D41 & D42 with two resistors, R34 is now a short.  This depends on
the micro having enough sink capability to ensure the side that is
off remains off.

dwayne

--
Dwayne Reid   <dwaynerEraseMEspam.....planet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(780) 489-3199 voice          (780) 487-6397 fax

Celebrating 21 years of Engineering Innovation (1984 - 2005)
 .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-.   .-
    `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'   `-'
Do NOT send unsolicited commercial email to this email address.
This message neither grants consent to receive unsolicited
commercial email nor is intended to solicit commercial email.

2005\09\07@193249 by Russell McMahon

face
flavicon
face
Ken said (offlist)

> I like it (but feel am somewhat miffed as this is largely along the
> lines
> that I suggested some time ago  - namely that the brake should be
> implemented using the H-bridge itself).

All input always much appreciated, but, to be fair to all, I'm  afraid
I can only allow a tip of the hat in your direction on this occasion
:-).
The aim has always been to provide braking and voltage regulation
using only 4 transistors, less if possible :-).

ie yes, you did suggest both braking and voltage regulation within the
bridge and I had of course wanted to achieve both of these as well
both before and after you said it.

Others also suggested FETs and of course I wondered about FETs.
Despite their great advantages I sought to avoid them due to the
potential problems with getting the people doing the work to acquire
them. Not an inconsiderable problem potentially.

BUT what made me take the leap was the concept of having BOTH bottom
FETs always on in the inactive state with no drive current required.
Measuring current on a 200 uA range there is no reading at rest!
Removing the two gate drive diodes adds about 30+ uA quiescent when
the top bipolars are reverse biased when turned off.

Even in retrospect using the two lower on-when-off FETs is not quite a
wholly obvious idea. Just an obviously excellent one. It may well be
standard practice elsewhere but I have never encountered it before.
One more thing to add to my armoury.

The low side FET drive state relative to input drive is similar to the
high side case when bipolar emitter followers are used, as was the
case in the original. In the original the high side drivers are turned
on when there is no drive and the unused high side driver is turned
off when drive is applied leaving the other to remain biased on. This
is what contributes to the original very poor voltage drop as the high
side driver must not only drop 1 Vbe (as it is an emitter follower)
but also the extra voltage required across its base drive resistor,
which increases until it is high enough to provide 1/beta-th of the
load current. The result is about 1.5 volts drop across the high side
driver, which is unacceptable with flattish batteries.

The relevant low side FET is also "passively" turned on under drive
but the full battery supply is available to bias it into full
saturation (enhancement) and there is no gate drive current required,
allowing a high value resistor to be used.

"Logic" FETs are required due to the need to operate down to under 4
volts supply.


       Russell McMahon

More... (looser matching)
- Last day of these posts
- In 2005 , 2006 only
- Today
- New search...