The best stepper driver??? was Re: Stepper questions

>In CCED, "Mariss Freimanis" <mariss92705@...> wrote:
>The two step motor phase windings don't share a common flux path
>and thus don't interact. Proof: try using a 4-wire motor as a
>transformer. Put a <snip of technical details, reposted below>

Hello Mariss, CNC'ers,

All true, and good to know.

The laboratory I'm interested in is the one where we make parts; the
shop. The shop/work/hobby area of myself and our customers and
potential customers.

And in that shop, over many years I've used: unipolar linear,
unipolar L/R, unipolar Bi-level, unipolar chopper, bipolar L/R, and
bipolar chopper drives. From MANY mfrs.

Nearly ALL of them have worked reliably, and ALL of them have made
parts. Each HAS been worth the money spent.

All have NOT made parts at the same rate, OR with the same ease to
put into use. Some, we still use because "we've got them", but won't
be replacing with the same drive style in the future. Others we will
ikely buy again over time.

So to replace long-term generalities with present specifics:

We've got Gecko, Xylotex, PMDX, THS, EAS, HCNC, GeoM, and many
others at work. (FWIW, I run each of these mfrs. drive nearly every
day doing real work in a real shop. And others not listed here as
well.)

Gecko's are clearly the best, and their price refects that. They are
also overkill for MANY projects.
I use them and appreciate their performance and features where the
added price--and in my case the larger difficulty to wire up--since
my GeoM drives are plug-n-play.

Xylotex, THS, and PMDX all use the same chip; a bipolar chopper
rated by its MFR. at 35VDC and 2-1/2A. (PMDX says their drive will
handle 3A reliably and I trust Steve's ability, so I'll not question
his spec. I've also been running his drive with a 3A motor for a
year without fail.)

IMO/E all these drives are within a hairbreadth of each other
performance wise. Note that Jeff at xylotex also makes/made a 7062
based board; but it does not seem to be his focus. I suspect that is
because the surface mount A3977 based board is less "expensive" for
him to produce in many ways. Or he just feels it is a better choice
for his customers.

IMO the early xylotex were not as robust in actual shop use, as the
other two. There was also a noise problem in the step/dir inputs
which Jeff to my knowledge has fully corrected in present boards.
Saying they were not as robust may be an unfair assessment, since
their popularity meant a larger sample and response was presented in
these groups where much of my NON-personal experience is gotten.

But I still read more often about xylotex boards going bad than
HobbyCNC boards, and I've blown one myself. (It WAS my fault, and I
DID do something stupid. Reverse polarity on the 5v supply. But I've
done stupid things with other boards too, and they've survived.)
Jeff pointed out in a discussion we had on another list that might
be due to the higher voltage diodes in the unipolar chopper used in
the HobbyCNC boards. And he's likely right.

BUT...

At the end of the day in MY "laboratory"; I don't really care what
the "technical" reason is, the results are what I'm after. And in
side-by-side testing, I simply got better results with the HobbyCNC
board than I did with the Xylotex and other A3977 based boards.

Our target--for this project--did not warrant higher costs and
capability, so Gecko, Rutex, and others were left out of
consideration.

So when I set out to design a FULLY plug-and-play system for folks
who DON'T want to need to know all about stepper drives and how and
why they fail, and how one is more efficient, etc....

I went with the chip used in the HobbyCNC board; a Sanken 7062M.
A unipolar chopper, rated for 3amps at 44VDC; it has higher initial
specs than the "competing" chips like the A3977, L298, L6208, etc.

It is less expensive than the chips with higher ratings like the
LMD18245 and L6203--and I only need one per axis, so my board can be
the same size as the xylotex 3 axis drive. It is less expensive in
our quantities than the discrete components used to make a Gecko.

The 7062 costs a bit more than the A3977, but based on all the
above, I designed a 3 axis board and a 1 axis board using it, and
put those into our geoM plug-n-play MCU. I designed the board to use
EITHER DIN 5pin OR screw terminal outputs; so those who wish to use
specific cables to their motors are able to do so.

The HobbyCNC board is very large, and its layout and IO
handling/pinout is IMO non-standard and incomplete. Mounting
inexpensive effective heatsinks is difficult because the screw
terminals get in the way. (you can use large tall expensive
heatsinks to get past this, again increasing overall size.) I don't
think electronics in 2006 need to be any bigger than they need to
be; I like economy and elegance.
I DO understand that Dave is going after a kit market, so having a
larger board makes things easier for HIS customers. The HCNC board
is noisier than the A3977 based boards. You have to put it together
yourself.

Against this, the board works. People use it, and like the RESULTS
they get with it.

Now, MY goal for MY customers, is that they DON'T have to know
ANYTHING about CNC electronics to USE their CNC machines. (Except
not to unplug things under power!<G>)

With THAT goal in mind, I created a small, integrated, expandable 3
axis board using a unipolar chopper style drive. I've tested it with
the "typical" motors folks will want to use and with the inexpensive
and widely available cables so they don't have to do any wiring.
Like the HobbyCNC board which was its predecessor, it works.

I gave it an expansion connector to provide for updates and
configuration changes that seem likely both as a person grows in
their CNC experience, and also as the market itself changes over the
next few years.

Then I created a box with expansion panels, and created a list of
likely expansion products to fit those panels, including probes, IO,
and power control. Also potential 2nd paralell, serial, USB and
Ethernet input/ output. (these expansion products will be added to
GeoM offerings as they are tested and available.

I gave this box a power supply which has been tested with the
expected motors and uses and again, it works. I spent a LONG time
trying to get the box size "right". It is far smaller than anything
having remotely the capabilities (that I'm aware of).

And then I did my best to price it all fairly.

So while we can debate the relative merits of one style drive over
another--and I DO enjoy such debates as longtime members here WELL
know<G>--it is MY opinion that for MY customers (those who just want
to have a reliable and EASILY completed CNC machine) this new GeoM
controller box will get them where they want to be.

Ballendo

P.S. And I've hedged my bet by designing our GeoM box in such a way
that nearly any of those other drives/boards may be used, if that
makes sense for YOUR needs...


--- In CAD_CAM_EDM_DRO@yahoogroups.com, "Mariss Freimanis"
<mariss92705@...> wrote:

>
> The two step motor phase windings don't share a common flux path

and

> thus don't interact. Proof: try using a 4-wire motor as a
> transformer. Put a function generator on one winding

(the "primary"),

> a scope on the other winding (the "secondary"). You will see

nothing

> on the scope.
>
> ------------------------------
>
> Unipolar drives have some severe drawbacks in high performance
> designs. It is because a single center-tapped winding DOES act as

a

> transformer and the consequences are all bad:
>
> 1) The center-tap goes to +V supply. When one end-wire is taken to
> GND by a transistor (the "on" one), the other end-wire goes to

twice

> the +V supply. This voltage is seen by the "off" transistor.
>
> Transistors in an 80V bipolar drive see 80V stresses, transistors

in

> an 80V unipolar drive see at least 160V stresses.
>
> 2) Inductive energy recovery. The "off" center-tap to end-wire
> winding carries (L*I^2)/2 energy when turning off. There is no

easy

> means of returning it to the supply, so it must be dissipated
> (wasted). Usually zener clamps do this dirty work.
>
> A bipolar drive easily returns this energy back to the power

supply

> for re-use.
>
> 3) Circuit topology. A high-performance (switching type) unipolar
> drive requires 3 transistors and 3 power diodes per winding to
> efficiently handle energy return. However the transistors still
> see "twice supply voltage" stresses.
>
> At this point the unipolar drive requires 12 power devices vs. a
> bipolar drive's 8, the unipolar drive transistors see twice the
> voltage of those in a bipolar drive and 6 motor wire connections

are

> required vs. 4 for a bipolar drive. All that to get the same
> performance as a bipolar drive.
>
> 4) Failure modes. A failed transistor in a high performance

unipolar

> drive cooks the motor. A failed transistor in a bipolar drive

shorts

> the supply, blows a fuse and spares the motor.
>
> Most designers ask "All that for what?" Then they discard the
> unipolar drive topology and go for the bipolar one.
>
> --------------------
>
> If anyone is interested, I can try to dig up a high performance
> unipolar drive power section design and post it. Something I

designed

> and we sold about 30 years ago. I may still have it around on some
> old 10MB hard-drive.:-)
>
> Mariss