On Fri, Sep 16, 2016, at 12:31 PM, 'Luke Steele' via Machinekit wrote:
> Hi John,
> On Thursday, September 15, 2016 at 3:30:31 PM UTC+1, John Kasunich wrote:
> >
> >
> > You mentioned the LMD18200, and I looked it up.  That is a chip, not a 
> > complete driver.  It takes a significant amount of electronics knowledge 
> > to make a reliable system, even starting with a chip like that which has 
> > a lot of built-in functionality.  Power distribution, ground bounce, 
> > common 
> > mode voltages, choice of switching frequency, electrical noise, thermal 
> > management, circuit board layout, etc, etc, are all non-trivial issues. 
> >
> > Because you mentioned a naked chip, I assumed that you were 
> > comfortable with all those issues. 
> >
> > Fair enough - I have used that chip a few times successfully, but perhaps 
> the applications were simple enough to avoid addressing the more complex 
> issues you mention.

Many of those issues scale with power level.  I work at 10's of kW
up  to over 1MW, and they are critically important.  I maybe be
overestimating risk level at 10's of watts, but if you get into 100+ 
watts some care is definitely needed.

> I haven't actually heard of a hysteresis controller before. Is it a 
> bang-bang controller with a deadband?

Yes.  The motor load is primarily an inductance, with a resistance
and counter-emf term.  An analog comparator with deadband 
has delay measured in nano-seconds.  If the current is below the
lower limit, turn on FETs and apply full voltage, current starts
ramping up.  When current reaches upper limit of deadband,
turn them back off to apply zero volts and current starts ramping
back down until it hits the lower threshold again.

Actually with a full H-bridge there are three possible states,
full positive, zero, and full negative, and the control is slightly
more complex.  Not much.  Or you can use only full positive
and full negative, at the expense of a bit more motor heating
due to higher ripple current.

Switching frequency depends on deadband, inductance, and
counter-emf, and is not constant.  But that is usually OK - a
suitable deadband can keep it in a safe range.

Advantages are that it is simple to implement in hardware, needs
no real tuning (except setting deadband), is unconditionally stable,
and no other control method can be faster.

Disadvantage is that switching frequency varies (often sounds
like a hiss instead of a whine), and current can wander around
inside the deadband.  How important that is depends on accuracy
requirements and how big of a deadband you need to keep 
switching frequency reasonable.

  John Kasunich

website: http://www.machinekit.io blog: http://blog.machinekit.io github: 
You received this message because you are subscribed to the Google Groups 
"Machinekit" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to machinekit+unsubscr...@googlegroups.com.
Visit this group at https://groups.google.com/group/machinekit.
For more options, visit https://groups.google.com/d/optout.

Reply via email to