Searching \ for '[EE] SN754410 H-Bridge Driver' 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/io/motors.htm?key=h%2Dbridge
Search entire site for: 'SN754410 H-Bridge Driver'.

Exact match. Not showing close matches.
PICList Thread
'[EE] SN754410 H-Bridge Driver'
2011\02\02@195528 by Nathan House

picon face
I'm going to use a SN754410 IC to drive two small motors and I'm
wondering if I need to put flyback diodes in my circuit? From the
reading I've done, there doesn't really seem to be any consensus as to
whether this is necessary or not (everybody says something
different..).

Appreciate your help

2011\02\02@201704 by IVP

face picon face
> I'm going to use a SN754410 IC to drive two small motors and I'm
> wondering if I need to put flyback diodes in my circuit?

Nathan,

the datasheet says you should

I use ultrafast diodes on the outputs of the L298, otherwise it would
be a junker almost immediately, even with the smallest of steppers

For DC motors I always include something like an MUR1560 and a
Schottky plus at least one filter cap right at the motor terminals

Jo

2011\02\02@224329 by Nathan House

picon face
Thanks for your response.

Where does it say that in the datasheet? I believe you, I just can't
find it in there.. I bought a bunch of 1A schottky diodes on ebay,
will those work as well as the MUR1560 you mentioned? Do you put the
filter cap (.1uF?) on the motor itself or on the pcb where the motor
wires connect?

Also, do you think it would be a bad idea to power the motors and
microcontroller from the same power supply (batteries)?

Thanks for your help

2011\02\02@234720 by RussellMc

face picon face
> > I'm going to use a SN754410 IC to drive two small motors and I'm
> > wondering if I need to put flyback diodes in my circuit?

> the datasheet says you should
>
> I use ultrafast diodes on the outputs of the L298, otherwise it would
> be a junker almost immediately, even with the smallest of steppers
>
> For DC motors I always include something like an MUR1560 and a
> Schottky plus at least one filter cap right at the motor terminals

Joe's answer will usually be correct.
Sometimes it will be wrong.
Because:

The question is not specific enough.
*IF* you have uncontrolled energy stored in an inductor such that  it
will "ring" when turned off, then it WILL be dissipated in some
manner. If your turned off H bridge switch is the only or lowest
breakdown path for that energy then it WILL get turned on - either
temporarily or permanently, or rendered hors de combat thereafter. If
there is enough capacitance that i/1LI^2 can transfer to 1/2CV^2 such
that V is below breakdown voltage then the switch will survive.

SO generally, as Joe says, you have to do something with the inductive
energy either with a diode or some other way of directing the stored
energy.

BUT

in some motors (and you have not specific topology or motor type, the
inductive energy may NOT be uncontrolled, and clamping it with a diode
may be a very bad idea [tm].

A very real world example is a stepper motor with a centre tapped
inductor where you apply DC to the centre tap and then alternately
apply ground to one or other ends of the inductor. In this case, if
centre tap is at v+ the "open" end will rise to 2V+ by transformer
action. if you diode clamp to V+ (as seems normal and tempting)
"thinks will not work as expected" [tm]. Smoke may be green or orange
and/or stepper may walk funny. Solution is to either not clamp the
drive points or to diode clamp to at least 2V+ or via clamping diodes
to eg a capacitor + resistor combo designed to suit.



         Russel

2011\02\03@001850 by IVP

face picon face
> Where does it say that in the datasheet?

Hi, Nathan. Take a look at Figure 3. Also, have a look at the SGS
L293, which the SN754410 is compared with

http://www.st.com/stonline/books/pdf/docs/1328.pdf

and the L298

http://www.st.com/stonline/books/pdf/docs/1773.pdf

> I bought a bunch of 1A schottky diodes on ebay, will those work
> as well as the MUR1560 you mentioned?

Better would be the UF400x series. Same packages as 1N, but fast

The MUR1560 is a fairly hefty fast diode, 600V, 15A. I mentioned
it as one type of drive I use is PWM with a large FET and a 1/2HP
12VDC motor. It generates a lot of back-EMF and hash noise, which
the diode takes care of, leaving the FET largely unaffected. Schottky
diodes, ISTR, have no recovery time. However, the application notes
of the ICs above do not use Schottky, but ultrafast. FETs can be
more sensitive than bipolars in some instances and need over-voltage
protection so I tend to use both low Vf and fast diodes. No failures
yet, touch wood. It's cheap insurance anyway

> Do you put the filter cap (.1uF?) on the motor itself or on the pcb
> where the motor wires connect?

The best place to get rid of noise is at the source. Doing it there will
help prevent noise being radiated from wiring. I would suggest a cap,
say 100nF ceramic, across the motor terminals, and also one from each
terminal to the motor case. Another option is to add a pair of back-to-
back zener diodes across the terminals. The value will depend on what
voltage the motor is driven with and what noise voltage you want to
suppress. For a 12V motor, perhaps 15V

terminal - cathode - anode - anode - cathode - terminal

> Also, do you think it would be a bad idea to power the motors and
> microcontroller from the same power supply (batteries)?

With adequate filtering it should be OK. The only issue might be
when the motor turns on or if it stalls. There could be big dip in the
supply. Generally the micro will cope if it's above the brown-out
voltage, although watch out for noise on MCLR if you have it enabled
or any noise on inputs that could affect operation. A dip in power at
start-up could probably be avoided by ramping up the power to the
motor with PWM (ie soft start)

If possible, you could isolate the SN754410 supply from the micro
with a couple of diodes (Schottky would be lower Vf than a silicon
at higher amperage) and give the SN754410 and motor their own
reservoir cap

micro - cathode - anode +V+ anode - cathode - IC/motor

Jo

2011\02\03@005120 by IVP

face picon face
> Schottky would be lower Vf than a silicon at higher amperage

A little more on that. Generally Schottky would be lower, but you
should check the I-V curve for a particular device. For example a
BAT46 has a much lower Vf than a 1N4148 over the spec range,
but the curves for comparable larger diodes may be closer at higher
amperages. eg at 1A, 1N5401 (0.75V) vs 1N5822 (0.52V)

2011\02\03@013629 by RussellMc

face picon face
If you care about every fraction of a Volt (as you sometimes do), Vf
on Schottly diode tends to be noticeably lower for lower Vmax ratings.
If this matters use the lowest voltage rating that the circuit allows.

eg I selectively use 1N5817 & 1N5819 diodes in 2 locations the same
design where voltages vary and diode drop matters (slightly)  in each
case). eg 0.1V drop driving a 3.3V LED from a smps is 3% less energy
or 3% more "free" battery life. For a say 300 minute battery life
that's 10 more free minutes at full brightness. That's enough energy
to put a long diminishing brightness tail on a portable light. Free is
good. Cost of either diode is usually the same. If you can convince
somebody to sell the lower voltage part at lower cost you get more for
less :-).

For flyback clamping this is usually not overly relevant until you get
to serious dissipation levels.
For very low voltage rectification, such as from low voltage PV
(solar) panels or low voltage inverter outputs, the difference can
make a  few percent difference to the efficiency.

Note also that Schottky may have VERY high reverse leakage currents at
high temperature. So much so that reverse power dissipations may
become serious at high voltage and temperature for high current (say
10A plus) devices.

In some circuits high temperature Schottky leakage may be so high as
to render them unusable. Look at spec sheets, determine the worst case
operating device temperatures 9ambient + dissipation induced) and
check the data sheet. Note that for reverse biased dissipation this
may be an iterative procedure as increased leakage leads to increased
temperature leads to .. . "Over relaxation" allows shortening this
process - or use eg SPICE as long as your model properly models
reverse leakage with \temperature.
Not a concern in most applications.


   Russell McMahon






























Applied_Technolog

2011\02\03@083616 by Olin Lathrop

face picon face
Nathan House wrote:
> I'm going to use a SN754410 IC to drive two small motors and I'm
> wondering if I need to put flyback diodes in my circuit? From the
> reading I've done, there doesn't really seem to be any consensus as to
> whether this is necessary or not (everybody says something
> different..).

You didn't say what a 754410 is or how you are using it, but any time you
switch a inductive load you have to keep in mind that the instantaneous
current thru the inductor won't change.  That current will go somplace.  If
there isn't a nice path for it, it will make one.  That means the voltage
goes up until something arcs, smokes, punches thru, or otherwise gives up
and allows the current to flow.

If you are always driving the motors with a voltage in only one direction,
you can put a catch diode right accross each motor.  It's a good idea to
give the inductive kickback current the shortest possible path.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\02\03@085659 by RussellMc

face picon face
> If
> there isn't a nice path for it, it will make one.  That means the voltage
> goes up until something arcs, smokes, punches thru, or otherwise gives up
> and allows the current to flow.

Consistent with that, (ie Olin and I are in agreement :-) ) do
understand the "special case" that I mentioned where  you  drive one
or other  end of a centre tapped stepper motor winding and the other
end of the winding rises to 2 x Vdrive. You may still need to deal
with the stored inductive energy * but you must not clamp it to less
than 2 x Vdrive or interesting things will happen (Ask me how I know
:-) ). This is not what happens with most motor drive situations, but
this sort of stepper motor is common enough that some people will
encounter this sooner or later.
* Nominally the open winding is hard coupled to the driven winding,
but Occam says that leakage inductance and Murphy mean that some sort
of clamp makes sense.


        Russell

2011\02\03@105218 by William \Chops\ Westfield

face picon face

>> I'm going to use a SN754410 IC to drive two small motors and I'm
>> wondering if I need to put flyback diodes in my circuit?

THe SN754410 is a dual H-bridge (quad half-h bridge, actually), pin  compatible with the L293D and with somewhat higher current  capability.  The complication is that the internal schematic shows  diodes in the proper places, but the suggested external schematic also  shows additional diodes.  Rumor has it (from Adafruit, IIRC) that the  internal diodes are primarily for ESD-like protection and external  diodes should be used for flyback purposes.  But like the original  poster says, it would be nice to see a definitive written statement to  that effect...

BillW

2011\02\03@120121 by Wouter van Ooijen

face picon face
< Rumor has it (from Adafruit, IIRC) that the
> internal diodes are primarily for ESD-like protection and external
> diodes should be used for flyback purposes.  But like the original
> poster says, it would be nice to see a definitive written statement to
> that effect...

(teacher hat on)
Now, a question to the class: where do you go for answers? 'the piclist'! OK, that's a good second, but actually the first choice should be 'the datasheet'!

Where to get the datasheet? Use http://www.findchips.com to find a supplier, most (at least Digikey and Mouser) have a link to the datasheet on their product page. You can of course just google, but your result will be polluted by alldatasheets an other services that want to sell you info that is otherwise freely available.

Basic assumption when reading a datasheet: everything is forbidden, except what is shown in the "normal (or recommended) operating conditions" section. If you have a pair of scissors, cut out the "absolute maximum ratings" section now and burn it.

So for this chip we get http://focus.ti.com/lit/ds/symlink/sn754410.pdf , which shows on page 4 the "VOKH High-level output clamp voltage" at -0.5A and +1.0A. There is something I don't get, IMHO those two currents should have the same sign, but it is clear that a clamp current is allowed, at least up to 0.5 (or maybe 1.0) A. The clamp voltage is a lousy 2.5V at 1A, so this is not a very good diode, but this is within the "normal operating condition" for this chip. Of course, if you think 2.5V is way too much, or you fear you will exceed the maximum power dissipation of the chip, or you are afraid these diodes will be too slow (I see no switching time info), external (better!) clamping diodes might be a good idea (as shown on page 6), but basically this chip has internal clamping diodes.

Note about L293: Watch out, the L293D version has such diodes, the L293 has not!!

--
Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

2011\02\03@122522 by Michael Watterson

face picon face
On 03/02/2011 17:00, Wouter van Ooijen wrote:
> lousy 2.5V at 1A, so this is not a very good diode, but this is within
> the "normal operating condition" for this chip. Of course, if you think
> 2.5V is way too much, or you fear you will exceed the maximum power
> dissipation of the chip, or you are afraid these diodes will be too slow
> (I see no switching time info), external (better!) clamping diodes might
> be a good idea (as shown on page 6), but basically this chip has
> internal clamping diodes.

All what he says and:

For clamping, Fast and current rating may be more important than Schottky. Schottky is more important for series diodes in normal current path, especially in low voltage high current PSUs.

Also you may want Snubber Network at Motor connections to reduce rise time and initial peak current.

A reasonable overview
http://en.wikipedia.org/wiki/Snubber

On a Bridge driven motor you might put four diodes (ground to out1, out1 to Vs, ground to out2, out2 to Vs) on the PCB, but have transorb (back to back zener rated more than supply volts) and/or R & C snubber network to reduce RFI and current in cable to the on-board diodes. You might have transorb at 1.5 to x3 the supply volts. Power rating of C and R may be important! Many years ago a I saw C explode when large motor 1st turned off.

A stepper or any multiple winding motor is a more complex issue as explained earlier

2011\02\03@124136 by Charles Craft

picon face
On 2/3/2011 12:25 PM, Michael Watterson wrote:
{Quote hidden}

Vishay application notes for Transzorb

2011\02\03@155304 by Nathan House

picon face
Thanks for your help, everyone.

>You didn't say what a 754410 is

Well, the topic title *is* "SN754410 H-Bridge Driver" :)

>If you are always driving the motors with a voltage in only one direction,
>you can put a catch diode right accross each motor.

The motors need to run in either direction.


Here is the datasheet for anyone interested:
http://focus.ti.com/lit/ds/symlink/sn754410.pdf

I'm wanting to use the sn754410 to drive the two small BLDC motors in
this product: http://www.sparkfun.com/products/319 (a gearbox/motor
combination)

The diodes I purchased are 1N5819 Schottky diodes:
http://www.vishay.com/docs/94614/1n5819.pdf. The datasheet says "High
frequency operation," but they might not be fast enough

2011\02\03@163034 by IVP

face picon face
> The diodes I purchased are 1N5819 Schottky diodes:
> http://www.vishay.com/docs/94614/1n5819.pdf. The datasheet says
> "High frequency operation," but they might not be fast enough?

"fast" in this context applies to the reverse recovery time, rather than
a frequency

http://en.wikipedia.org/wiki/Flyback_diode

Past this point I'm no great expert, but I would guess that a 1N5819
is suitable if its temperature is limited. As Russell said, as a Schottky
heats up it becomes more conductive in the reverse direction and so
becomes useless as a diode, and in fact could be a liability

Filtering that suppresses heat-producing noise will help the diode
stay cool and reactive to fast transient

2011\02\03@170448 by Nicola Perotto

picon face


On 03/02/2011 20.53, Nathan House wrote:
> Well, the topic title *is* "SN754410 H-Bridge Driver" :)
>
Haven't read forum rules??? Olin is dispensed from reading the subject ;-)

>
> I'm wanting to use the sn754410 to drive the two small BLDC motors in
> this product: http://www.sparkfun.com/products/319 (a gearbox/motor
> combination)
>
This is not a BLDC but a brushed DC motor.
I made a small controllor for this gearbox with the SN754410 without diodes..
The current is around 0.5A so it's better to heatsink the chip.

2011\02\03@175147 by Michael Watterson

face picon face
On 03/02/2011 20:53, Nathan House wrote:
> The diodes I purchased are 1N5819 Schottky diodes:
> http://www.vishay.com/docs/94614/1n5819.pdf. The datasheet says "High
> frequency operation," but they might not be fast enough?
they are for SMPSU use rather than Motor snubbers.
Actually as long as you don't use something slow like 1N4007!
I've tested the 1N4007 successfully as a varicap (reverse bias) and RF current controlled attenuator/Switch as it's so slow due to maybe the large inverse working voltage rating. It's probably a poor choice.

Most Schottky are fast. It's just you need a Fast rectifier of suitable power rating and PIV rating rather than extra cost of Schottky, which tend to have lower forward voltage (good for low voltage SMPSU such as 3V or 5V) and tend to have poor reverse leakage and low PIV ratings.

If the 1N5819 will take the current (depends on Motor inductance and on Current). Instantaneously when you turn the bridge off, whatever current there was will flow in a pair of the diodes in series. When the current was the other direction, it flows in the other two diodes when bridge off.

Turning entire bridge high impedance for a suitable time before trying to reverse the volts may be a good plan, so nasty currents flow in external diodes and not wrong direction in output switches. It's about 12 years since I did this stuff so I'm hazy on some details

2011\02\03@175553 by Michael Watterson

face picon face
On 03/02/2011 21:30, IVP wrote:
> ast this point I'm no great expert, but I would guess that a 1N5819
> is suitable if its temperature is limited. As Russell said, as a Schottky
> heats up it becomes more conductive in the reverse direction and so
> becomes useless as a diode, and in fact could be a liability

Yes. I'd have chosen a non-Schottky, fast recovery diode. But the current and operating volts decide if the 1N5819 is OK.
For PIV you don't want to over specify as higher PIV is often a slower diode

2011\02\05@181431 by Nathan House

picon face
>Yes. I'd have chosen a non-Schottky, fast recovery diode.

I'm kind of confused; Isn't a Schottky diode a fast recovery diode?
Could you give an example as to what diodes you would have used for
this?

>This is not a BLDC but a brushed DC motor.
>I made a small controllor for this gearbox with the SN754410 without diodes.
>The current is around 0.5A so it's better to heatsink the chip.

My mistake, I thought they were brushless. Do you think a ground plane
will provide a suitable heatsink for the IC?

I'm not sure if I connected the flyback diodes correctly in my
circuit. Here's the schematic:
www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r8.png.
Is that the right way to do it?

If you have any other advice for the circuit, I would appreciate
hearing that as well!

Thanks,

Natha

2011\02\05@183915 by Dwayne Reid

flavicon
face
At 04:14 PM 2/5/2011, Nathan House wrote:

>I'm not sure if I connected the flyback diodes correctly in my
>circuit. Here's the schematic:
>www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r8.png.
>Is that the right way to do it?

I see one oddity: the motor driver chip is powered from Vdd but the positive side of the clamp diodes goes to Vcc.  The positive side of the clamp diodes should go to the same power supply line as what powers the driver chip.

Question: why do you have 10k resistors in series with all of the control inputs to the driver chip?  If the driver chip has internal pull-up or pull-down resistors, you will have problems.  Even if the driver doesn't have pull-up or pull-down resistors, I'd be tempted to reduce those series resistors to 1k or so.

dwayne

-- Dwayne Reid   <spam_OUTdwaynerTakeThisOuTspamplanet.eon.net>
Trinity Electronics Systems Ltd    Edmonton, AB, CANADA
(780) 489-3199 voice          (780) 487-6397 fax
http://www.trinity-electronics.com
Custom Electronics Design and Manufacturing

2011\02\05@191336 by Nathan House

picon face
> I see one oddity: the motor driver chip is powered from Vdd but the
> positive side of the clamp diodes goes to Vcc.  The positive side of
> the clamp diodes should go to the same power supply line as what
> powers the driver chip.

Sorry, that's not very clear. Vdd is the 5V supply and is connected to
VCC1 (that's the name of that pin in in the IC's datasheet). VCC is
connected to pin VCC2 (again, the pin name in the datasheet) and the
diodes. So the motors will be powered by the VCC supply, while Vdd is
the logic supply.


> Question: why do you have 10k resistors in series with all of the
> control inputs to the driver chip?

I put the resistors there because I read on a website that it was a
good idea, as if the motor drive blew up or something, the resistors
would keep the microcontroller from getting fried. Maybe they're not
necessary, though.

2011\02\05@195802 by RussellMc

face picon face
Is there a / tell us about balancing robots.
Inquiring minds want to know ...

> I see one oddity: the motor driver chip is powered from Vdd but the
> positive side of the clamp diodes goes to Vcc.  The positive side of
> the clamp diodes should go to the same power supply line as what
> powers the driver chip.

Sort of.
If returned to a fixed supply clamp diodes should be returned to a
supply which is greater than or equal to the  supply being used to
drive the outputs.

1. If Vclamp is less than vdrive then the driver will be shorted to
Vclamp via the relevant diode when driving.

2. Here if Vcc >= Vdd then all is well.
Vdd is connected to +5V at top middle.
If Vcc >= 5 V it will work OK as long as the drivers are open
collector / open drain and rated for whatever Vcc is.

3. As noted several times  - in the special case of driving a centre
tapped mtor - usually a stepper but not always, you must satisfy:

 Vcc >= 2 x Vdd "or else".

4. Vcc could be a resistor to ground, or to a power supply, or a zener
to ground of Vz >= Vdd (or >= 2 x Vdd for centre tapped motor) or
similar energy dissipation or storage means.

If,  Vcc < Vdd then "there will be problems" [tm].

Russell

2011\02\05@220020 by IVP

face picon face
> >Yes. I'd have chosen a non-Schottky, fast recovery diode.
>
> I'm kind of confused; Isn't a Schottky diode a fast recovery diode?

Yes

> Could you give an example as to what diodes you would have used
> for this ?

Common types would be UF400x, UF540x. I also use the BYx ultra-
fast in small SMPS. There are many varieties of eg BYV, BYW

For reference

1N540x, Standard Recovery > 500ns

http://parts.digikey.com/1/parts/412455-diode-standard-3a-100v-do-201ad-1n5401-e3-54.html

UF400x, Reverse recovery time 50ns/75ns

http://www.fairchildsemi.com/ds/UF%2FUF4007.pdf

UF540x, Reverse recovery time 50ns/75ns

http://www.ygchina.com/sdygpdf/zxejg/uf/UF5400THRUUF5408.pdf

BYV28, Reverse recovery time 50ns max

http://www.vishay.com/docs/86043/86043.pdf

Which you choose depends on specific requirements especially re
amperage. A Schottky is perfectly acceptable, but I suspect that it
is more applicable to protect FETs rather than BJTs, due to its low
Vf rather than speed

As I mentioned though, IANA diode expert

Joe

2011\02\05@223831 by IVP

face picon face
> I'm kind of confused; Isn't a Schottky diode a fast recovery diode?

PS I meant to include a link to the 1N58xx

http://www.st.com/internet/analog/product/62111.jsp

You'll note that Schottkys are much more sensitive to temperature
than other fast diodes, and that they will run away thermally

Compare for example the leakage currents. 1N58xx (2000uA @
100C) vs BYV28 (150uA @ 150C)

Jo

2011\02\05@225907 by IVP

face picon face
> Compare for example the leakage currents. 1N58xx (2000uA @
> 100C) vs BYV28 (150uA @ 150C)

Correction - 20,000uA @ 100C vs 150uA @ 150C

2000uA @ 25C vs 5uA @ 25C

Jo

2011\02\06@060205 by Michael Watterson

face picon face
On 05/02/2011 23:14, Nathan House wrote:
> I'm not sure if I connected the flyback diodes correctly in my
> circuit. Here's the schematic:
> www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r8.png.
> Is that the right way to do it?
>
Are there two motors?

if it's one motor with two windings then the induced emf in 2nd winding might or might not be a problem for that circuit.

If it's two motors, the circuit looks OK.

Have you an Resistor and Capacitor (snubber rated suitably) at the motor contacts too

2011\02\06@060505 by Michael Watterson

face picon face
On 06/02/2011 00:13, Nathan House wrote:
> I put the resistors there because I read on a website that it was a
> good idea, as if the motor drive blew up or something, the resistors
> would keep the microcontroller from getting fried. Maybe they're not
> necessary, though.
It's a good idea, though value may be far too high. 100 Ohms to 1kOhm depending on IC spec.

If there is an interconnect cable, then putting 1/2 the value at driving end and 1/2 at receiving end and suitably very small capacitors will also reduce EMI out and in

2011\02\06@060912 by Michael Watterson

face picon face
On 06/02/2011 02:50, IVP wrote:
{Quote hidden}

yes, All Schottky are "fast". But many have high reverese leakage or low PIV rating.
But not all Fast Recovery diodes are Schottky type.

low voltage SMPSU, series protection etc will be Schottky.

Higher voltage SMPSU ( > 25V?) and parallel protection might not be Schottky.

2011\02\06@083110 by Olin Lathrop

face picon face
Nathan House wrote:
> I'm kind of confused; Isn't a Schottky diode a fast recovery diode?
> Could you give an example as to what diodes you would have used for
> this?

Yes, Schottky diodes have very fast recovery times, usually so fast as to be
ignorable.  The only drawback to Schottkys in this case is that they may
have excessive reverse leakage.  That is usually a function of their voltage
rating and your operating temperature.  Check the datasheet, and if the
leakage is acceptable then go with them.

As for which silicon diodes I would use, I would start with Schottky and use
silicon if the Schottkys weren't good enough.  Recovery time is a serious
issue and slow recovery can really beat the crap out of both the diode and
the switch.  There are various silicon diodes labeled as "fast" and
"ultrafast".  However, that's often a load of marketing.  You have to check
the datasheets carefully.  I've seen cases of a ultra fast diode being
slower than a fast one.  All the labels tell you is that recovery time was
probably at least some consideration in the design of the diode.

> My mistake, I thought they were brushless.

Wait a minute!  How can you make a driving circuit without knowing what kind
of motor you are driving.  The two are driven quite differently.

> I'm not sure if I connected the flyback diodes correctly in my
> circuit. Here's the schematic:
>
www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r8.png.
> Is that the right way to do it?

That looks right.  Next time don't be so lazy and rotate the part
designators properly before asking anyone else to look at the schematic.

I see some other potential issues, but I will wait to comment until your
schematic shows you are taking this project seriously.  If you don't care,
why should I?


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\02\06@113330 by Nathan House

picon face
>Is there a / tell us about balancing robots.
>Inquiring minds want to know ...

I'm trying to build a small balancing robot. The board I'm making here
has a PIC18F2550 as the "brain," and an accelerometer and gyroscope
will be used to detect the tilt and rate of rotation (the op amp /
line follower is there because the sensors have a higher resistance
than the ADC allows for inputs).

I'm hoping a PID loop will be enough to keep it standing up, although
I'm sure there will be lots of jitter. I know the proper way to do
this is to use a Kalman filter, but I don't have the mathematical
knowledge (yet) to implement one, and I don't know if a PIC18 could
even handle a Kalman filter?


>Are there two motors?

Yes, there are two motors. One motor is connected to 1Y & 2Y, the
other to 3Y & 4Y.


>Have you an Resistor and Capacitor (snubber rated suitably) at the motor
>contacts too?

Someone else mentioned that I should put .1uF capacitors from each
motor terminal to the casing, and one across the terminals. Where
would the resistor go?


>> My mistake, I thought they were brushless.

>Wait a minute!  How can you make a driving circuit without knowing what kind
>of motor you are driving.

Ignorance.


>That looks right.  Next time don't be so lazy and rotate the part
>designators properly before asking anyone else to look at the schematic.

This will probably drive you nuts, but what's a part designator and
which way should it be rotated?


Thanks for your help

2011\02\06@121107 by Olin Lathrop

face picon face
Nathan House wrote:
> This will probably drive you nuts, but what's a part designator

The name of a part, like "R1", "C27", "IC5", etc.  However the same applies
to part values or any other text.

> and which way should it be rotated?

So it can be comfortably read, of course.

The point is not to try to get away with living up to the letter of some
law, but to make the schematic clearly and easily readable.  This applies to
all parts of the schematic and any other documentation you generate, whether
someone explicitly asks for it or not.  Text from two different places
overlapping, for example, is also bad.  If I remember right, you had that in
your PIC symbol.  I was actually going to use that to judge whether you
really cared or not, and therefore how much I was going to help.  Oh well,
can't use that anymore.

Again, clarity is something you should be striving for on your own, whether
I or anyone else bugs you about it or not.  Lack of attention to detail and
sloppiness in general is going to cause others to take you and your work
less seriously.  Most of the time they won't bother to tell you.  They'll
just walk away to various degrees and you'll never know you didn't get what
you could have from them or why.

Sometimes it may not even be concious or deliberate.  For example, a sloppy
looking homework assignment may get a worse grade than a neat one with the
same content when part of the grade is subjective, like a essay.  Sometimes
this may be the grader thinking "what a jerk dumping this slop on me", but
other times the grader may not even be aware of the bias.

What looks like crap is likely to be perceived as crap.  The totally amazing
thing is how many people don't seem to get this obvious and very true
concept.  Duh.

A homework assignment is a rather benign example.  There are far more
serious examples.  Be extra extra careful with your resume, for example.
This may be obvious, but you'd likely be surprised at some of the resumes
I've seen.  For example, in one case I offered to do someone I knew a favor
in passing his resume to a contact I had at a place he was interested in.  I
asked him to forward me his resume so I could forward it to the manager with
a nice introduction and recommendation attached.  What he sent me was plain
text crap formatted to oblivion.  Now I had four choices:

 1 - Send it on as is.

 2 - Fix the formatting myself and pass it on as intended.

 3 - Tell my friend his resume was a mess and he needs to send me a
     cleaned up copy before I'll forward it.

 4 - Silently do nothing.

I chose #4.  #1 would have made me look bad because I would have been
endorsing the mess.  No thanks.  #2 would have been considerable work, and I
didn't feel it was my job.  #3 could have created bad blood "what do you
mean my resume sucks?".  #4 was just easier, and I felt a lot less obligated
to help when I saw the mess that got dumped on me.  I never told the guy.
He probably has no idea why he didn't get a call from that company.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\02\06@135041 by Nicola Perotto

picon face

On 06/02/2011 16.33, Nathan House wrote:
> That looks right. Next time don't be so lazy and rotate the part
>> designators properly before asking anyone else to look at the schematic.
> This will probably drive you nuts, but what's a part designator and
> which way should it be rotated?
>
>
Vcc1 is logic supply and is better to the left side with logic pins.
Vcc2 is motor supply and is better to the right with output pins.
Having swapped can drive you to a mistake...

2011\02\06@135115 by Nicola Perotto

picon face


On 05/02/2011 23.14, Nathan House wrote:
> Do you think a ground plane will provide a suitable heatsink for the IC?

If I remember correctly these motors need 3-6V and less then 0.5A (but minimum for the chip is 4.5V) so you must dissipate the heat because the absolute max power dissipation is only 2W, i think that a ground plane is not enough.

2011\02\06@151135 by Olin Lathrop

face picon face
Nicola Perotto wrote:
>> Do you think a ground plane will provide a suitable heatsink for the
>> IC?
>
> If I remember correctly these motors need 3-6V and less then 0.5A
> (but minimum for the chip is 4.5V) so you must dissipate the heat
> because the absolute max power dissipation is only 2W, i think that a
> ground plane is not enough.

That makes no sense.

You say the maximum motor power is 6V x 500mA = 3W, which should be
considerably more than what the driver chip dissipates.  I don't know if the
driver chip uses bipolar or FET switches, but either way dissipation should
be low.

The datasheet must say somewhere, but let's say the switches are bipolar
with 300mV saturation each.  There are two in the current path so that would
be 600mV x 500mA = 300mW.  If they are FETs, lets say 100mOhm.  Again there
are two, so 200mOhm x (500mA)**2 = 25mW.  Of course check the datasheet and
do the calculation yourself, but unless they are doing something strange or
this is a inappropriate chip (darlington drivers, for example), the above
should be pessimistic guesses.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\02\06@173446 by IVP

face picon face
> 0.1uF capacitor ... across the terminals. Where would the resistor go?

In series with that capacitor. As the page suggests, 100 ohms + 0.1uF
is common for small motors

http://en.wikipedia.org/wiki/Snubbe

2011\02\06@175920 by RussellMc

face picon face
> That looks right.  Next time don't be so lazy and rotate the part
> designators properly before asking anyone else to look at the schematic.

Also: brush your teeth, wear a clean shirt, carry a spare handkerchief
perchance a lady may wish to borrow one from you, smile nicely for the
camera and remember to say please and thank you on all appropriate and
some inappropriate occasions.

Note also that some very few people have problems with spatial
orientation of labels and may need or want everything arranged just-so
to suit them, or have personal preferences which they demand must be
considered as if they were international standards which MUST be
adhered to, or need to find some other "errant" detail to put you in
your place and cause you to appreciate  your inherent lack of worth
and rank in the order of things.

_____

Misc thoughts: Neatness counts, being polite costs little, improved
readability is always a good idea BUT needs to be balanced against
other factors.

The "correct" orientation of component labels and values on a diagram is moot.
This is very very very much a matter of personal style and preference,
and one person's "laziness" is another's "improved readability.
But also, a given individuals' assessment of maximum readability is a
personal choice and may seem strange or just plain wrong to others. As
is, apparently, the case here. Olin's assessment that you should
"rotate the part designators properly" seems bizarre to me. I can see
a few designators/labels/whatever that I would have rotated
differently BUT in no case would i say that my choice was "right". I
could explain why I would do it differently but would not expect to
get universal acceptance of my reasoning. Olin is, apparently, made of
sterner (or something) stuff and is the (apparently) final judge and
arbiter of good rotation.

None of your choices is confusing (to me), hard to read (for me) or so
obviously totally wrong (for me) that I (unlike Olin) would demand
that it be done my way. Maybe Olin has problems with spatial
orientation which I (or perhaps also he) am not aware of?

Having the axis of  the labels aligned with the direction of the
associated circuit traces as you have done for eg D1-D8 has the very
great (IMHO) advantage of increasing the reliability of correctly
associating the component with its label /designator when component
density is tight and there are many similar components in close
proximity. I would have aligned the D1-D8 labels the same way you did.

I would have aligned the Vcc & Vdd labels horizontally. I think it
looks prettier and is more "euphonic" (or drawn equivalent) but your
way is fine. I MIGHT have rotated the pin numbering on IC3 90 degrees
clockwise for pins 14, 8, 19 IF the drawing package allowed that (or
assigned them that way when creating the footprint IF that was a
choice) but your way is fine.

Your C2 and c4 labels follow different rules BUT what you have done is
how i would have done it.

I would have rotated output connector labels 1Y 2Y 3Y 4Y 90 degrees
clockwise AND I think that is better than what you did BUT your method
is fully readable and causes no problems in clarity.

FWIW - my wife could quite possibly read this diagram "upside down"
every bit as well as right side up. I couldn't. I could handle it
inverted but it would definitely slow me down. That's just a matter of
brain wiring and experience and ... .  Whereas I can understand what
the circuit does when viewed from any orientation, while my wife would
have no idea whatsoever regardless of its orientation. Which is
obviously irrelevant to how it is drawn.

it may well be that Olin's expectations are based not only on a desire
for neatness and order but on personal "brain wiring" which he is not
aware of. We all have limitations and preferences based on how we rare
put together. That in itself is not bad, but attempting to impose our
preferences on others when opposing choices can readily be argued by
others is liable to help one to have an interesting life. Olin does
seem to have an interesting life :-).

> I see some other potential issues,

Vcc < Vdd, Vcc = Vdd, Vcc > Vdd ???

:-)



> but I will wait to comment until  ...  why should I?

:-)


                Russell

2011\02\06@180906 by RussellMc

face picon face
> For example, in one case I offered to do someone I knew a favor
> in passing his resume to a contact I had at a place he was interested in.  I
> asked him to forward me his resume so I could forward it to the manager with
> a nice introduction and recommendation attached.  What he sent me was plain
> text crap formatted to oblivion.  Now I had four choices:
>
>  1 - Send it on as is.
>  2 - Fix the formatting myself and pass it on as intended.
>  3 - Tell my friend his resume was a mess and he needs to send me a
>      cleaned up copy before I'll forward it.
>  4 - Silently do nothing.

> I chose #4.

> #1 would have made me look bad because I would have been
> endorsing the mess.  No thanks.  #2 would have been considerable work, and I
> didn't feel it was my job.  #3 could have created bad blood "what do you
> mean my resume sucks?".  #4 was just easier, and I felt a lot less obligated
> to help when I saw the mess that got dumped on me.  I never told the guy.

That's an immensely interesting and possibly useful offering.

You have supplied in very compact form a real world personal "case
study", 4 perceived options, what you perceived the 'costs' and
benefits to be in each case and what you chose and why.

It MAY be useful to people as a guide to how to interact with you, or
not, as the case may be.
That' NOT a comment on the merit of your choice (I'll carefully
refrain from that :-) ).


 Russell

2011\02\06@192623 by Nathan House

picon face
>Having the axis of  the labels aligned with the direction of the
>associated circuit traces as you have done for eg D1-D8 has the very
>great (IMHO) advantage of increasing the reliability of correctly
>associating the component with its label /designator when component
>density is tight and there are many similar components in close
>proximity. I would have aligned the D1-D8 labels the same way you did.

I already went ahead made the changes Olin suggested earlier today,
but I kind of liked the diode labels the other way too. Whatever, it
doesn't matter to me..

First schematic:
www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r8.png
New schematic: http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r9.png

Is that better

2011\02\06@194501 by RussellMc

face picon face
> I already went ahead made the changes Olin suggested earlier today,
> but I kind of liked the diode labels the other way too. Whatever, it
> doesn't matter to me..

> First schematic:
> www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r8.png
> New schematic: http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r9.png
>
> Is that better?

No.
Certainly not.
It's different.
An it has the important aspect of Olin having been able to arbitraily
impose his will on someone and get them to change something from
"entirely OK as was if not actually 'perfect'" to "probably OK as is
but shows Olin is important".

D1-D* are fine as they are now shown - and were fine as they were
previously shown.

I dislike the way the resistors are now labelled and I would probably
never do it that way myself. BUT that way is fine, certainly nothing
I'd consider complaining about and may be useful in some cases. I do
not consider that either change contributed in any significant way to
the readability of the drwaing. And, that has NOTHING to do with the
fact that the change was made to meet Olin's perceptions. They were
just unnecessary make work which are a matter of personal preference
(and maybe of olin's mind wiring).

Noting that Vcc must not be less than Vdd may be useful. It usually
wouldn't be, but it could happen.
If Vcc < Vdd then D1-D8 need to be returned to Vcc or someother
suitable point / circuit.



Russell.


Russell

2011\02\06@201229 by IVP

face picon face
> Do you think a ground plane will provide a suitable heatsink
> for the IC?

Adding heatsinking would be no bad thing

There are clip-on and solderable IC heatsinks available. You might
find one on a monitor/CRT board. Or make your own with a bit of
'tin', eg an actual tin can or screening metal found in many appliances.
Easy to work with, solderable, fre

2011\02\06@205051 by Oli Glaser

flavicon
face
On 07/02/2011 00:26, Nathan House wrote:
> First schematic:
> www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r8.png
> New schematic:http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r9.png
>
> Is that better?

Both look fine - I wouldn't worry too much about what random, extreme obsessions with neatness etc, people may have as long as it's "sensible", and legible to the majority.
While this stuff matters to some extent, most will not write you off simply because you failed turn a designator round 90 degrees, and if they do, you're probably better off not working for/with them anyway.
One thing - IMHO there's no harm in making the names more descriptive, e.g. +5V and Vmotor for the two supplies (then you don't need the explanatory text) With more complex designs which may have many nets/pages this can make things easier.

2011\02\06@205531 by Richard Prosser

picon face
On 7 February 2011 13:44, RussellMc <.....apptechnzKILLspamspam@spam@gmail.com> wrote:
{Quote hidden}

My only comment would be a pet hate of D1-D* being connected at the
crossover point. I far prefer to see no more than 3 connections at a
node on the circuit as the way it is presented here only needs a small
dot to be added to make it indistinguishable from wires crossing but
not joining.

Another thing I might add is a note to minimise the loop caused by the
crystal capacitors to the ground pin of the IC (Pin8) and that the
decoupling capacitors should be mounted as close as practical to the
appropriate ICs. It largely depends how much confidence you have in
the person doing the layout.

RP


>

2011\02\06@212540 by Nathan House

picon face
>It largely depends how much confidence you have in
>the person doing the layout.

Not much confidence at all, then. I'm doing the layout ;-) This is
just a hobby project, FYI.

This is what I have so far (for the pcb):
http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-pcb-r9-2.png

I'd appreciate advice anyone might have. This is the first time I've
used a ground plane, so I'm hoping I did it correctly

2011\02\06@232907 by Mark Rages

face picon face
On Sun, Feb 6, 2011 at 8:25 PM, Nathan House <nathanpiclistspamKILLspamgmail.com> wrote:
>>It largely depends how much confidence you have in
>>the person doing the layout.
>
> Not much confidence at all, then. I'm doing the layout ;-) This is
> just a hobby project, FYI.
>
> This is what I have so far (for the pcb):
> http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-pcb-r9-2.png
>
> I'd appreciate advice anyone might have. This is the first time I've
> used a ground plane, so I'm hoping I did it correctly.

You will want "thermals" between ground-connected pins or pads and the
ground plane.  Otherwise, the ground plane will act as a heat sink and
make those pins or pads hard to solder.

Regards,
Mark
markrages@gmail
-- Mark Rages, Engineer
Midwest Telecine LLC
.....markragesKILLspamspam.....midwesttelecine.co

2011\02\07@074711 by Olin Lathrop

face picon face
Nathan House wrote:
> New schematic:
>
http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r9.png

That's definitely better, despite what Russell says (Russell feels it's his
mission to apologize for anyone I catch being lazy or irresponsible).

In general that looks pretty good.  The thing I was going to mention was the
1K resistors in series with the driver chip inputs.  If these are regular
digital inputs of IC2 and IC2 and the PIC are on the same board, then there
is no need for them.  They will only slow down the edges.

In some cases you do want to put resistors in series with digital signals,
but that's when they are transmission lines.  In any case 1KOhm is way too
much.  If these inputs go directly to the driver FET gates you might want a
little resistance in series, but more like 10 Ohms.  I would loose them
altogether.

A few other nits:

Why is Vcc on the input side and Vdd on the output side of IC2?  That seems
backwards from your description of those supplies at the top of the
schematic.  Maybe there is a reason that makes sense for that chip.  I
didn't look at the datasheet.  However, if Vdd powers the digital side and
Vcc the power bridges, then that is misleading.

There is no capacitance on Vcc I can see anywhere.  There should be some
right at the input connector, and at least a bypass cap at IC2.

What's with the "buffer amplifier" in the upper left corner?  It doesn't
have a component designator and doesn't look anything like a amplifier.  I
didn't look up a LM358, but from your usage it looks like a regular opamp.
By not using the common symbol, you are going out of your way to say
something about that amplifier, but I have no idea what that is.  This is
adding confusion to your schematic, if for no other reason than people have
to wonder what special thing you mean.  It also takes more time to visually
"decode" what you have versus a normal opamp symbol.  That's why we have
standard symbols.

I just looked at that amp more closely, and it looks like (from the pin
names only) that it's a dual opamp with both amps being used as voltage
followers.  However, both inputs are floating and there is no bypass cap.
Both of those are bad ideas.  If it had been drawn properly, this would have
been immediately obvious.

Surely you can think of better values for some of the test points than U$n,
which is what Eagle sets it to by default in some cases.  It will be very
useful to have the pads labeled on the board, and this label should be shown
on the schematic too.  This is usually done by setting the part value.  Such
values for connectors or pads become silkscreen text, at least in my Eagle
libraries.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\02\07@083600 by RussellMc

face picon face
> > New schematic:
> >
> http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r9.png
>
> That's definitely better, despite what Russell says (Russell feels it's his
> mission to apologize for anyone I catch being lazy or irresponsible).

Tut tut.
Your failure to learn is most troubling :-).
That's demonstrably untrue, even in this instance, as I'm certainly
not being an apologist for you on either count in this instance. 'Tis
also a shame that you continue to hide behind that pretense so long
after it lacks any consistent credibility. Laziness and
irresponsibility are easily applied labels. using them discerningly
and appropriately takes a bit more work.

eg now that Nathan has followed you suggestions, ask him if he
considers that he was being lazy or irresponsible initially. (Not
that, based on most other prior 'challenges', I hold out much hope for
you doing that).

Are you indeed suggesting that your way is THE only way, that no other
way has merit and that your judgement on subjective matters is
necessarily correct?

> In general that looks pretty good.  The thing I was going to mention was the
> 1K resistors in series with the driver chip inputs.  If these are regular
> digital inputs of IC2 and IC2 and the PIC are on the same board, then there
> is no need for them.  They will only slow down the edges.

For all values of "no" not involving magic motor juice leaking through
the driver IC into the PIC. He noted that these were to add a degree
of PIC protection if/when gang aglae. Others have commented on
appropriate values.

I largely agree with the following, which I won't quote in full.
I often agree with Olin once he gets into the technical explanation
mode. My disagreements often relate to snap judgements of people with
accompanying rudeness when inexperience, English as a second language
or unexpected but reasonable other factors may be involved. Knock
about 10% off the sharp edge and you'd have an extremely good
"product" indeed.

> Why is Vcc on the input side and Vdd on the output side of IC2?  Tha
....
> Both of those are bad ideas.  If it had been drawn properly, this would have
> been immediately obvious.




           Russell

2011\02\09@192011 by Nathan House

picon face
I changed the schematic to use the "standard" op-amp symbol. Is it
more readable now?

www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r11.png
http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-pcb-r11.png

I put a 10uF capacitor between the power connector and the sn754410
(which is the only place VCC is used). Do you think that is enough, or
should I put a few 10uF caps in parallel? I figured it would be
pointless to put a .1uF bypass cap by the VCC input on the sn754410
since the motors draw up to an amp, but I could put one of those on
there, too, if you think it would be a good idea.

Here are the Eagle files if anyone feels like taking a closer look..
http://roboticsguy.com/files/balancing_robot_2550.zip

Appreciate all of your advice

2011\02\10@082456 by Olin Lathrop

face picon face
Nathan House wrote:
> I changed the schematic to use the "standard" op-amp symbol. Is it
> more readable now?
>
>
http://www.roboticsguy.com/images/misc/balancing_robot/balancing-robot-board-schematic-r11.png

That looks pretty good.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\02\10@122911 by Adam Field

flavicon
face
On Wed, Feb 9, 2011 at 7:20 PM, Nathan House <EraseMEnathanpiclistspam_OUTspamTakeThisOuTgmail.com> wrote:
> I changed the schematic to use the "standard" op-amp symbol. Is it
> more readable now?
>

One thing I noticed, which has nothing to do with readability (which
is quite good) is that the USB VCC pin goes no where. When you have a
self powered PIC, you should monitor the status of the USB VCC pin so
you don't have the pullups enabled on the D+ and D- lines when there
is no USB VCC present. This can happen if the host computer goes to
sleep or otherwise shuts down the USB controller. The microchip usb
stack supports this feature, I think I remember it being defined in
hardwareprofile.h.

A typical circuit to do this is the put a 100K resistor from USB VCC
to ground and then another 4.7K resistor from USB VCC to a pin
configured as an input. The 100K resistor keeps the input pin from
floating, and won't draw too much from the USB when it's on

More... (looser matching)
- Last day of these posts
- In 2011 , 2012 only
- Today
- New search...