Re: PWM stepper driver with external bridge

Hi, it's my first post here, hope I don't offend
anyone but I want to dispute a couple of points.
:o)

> 1. You may be reinventing the wheel, as there are several good

similar designs

> already on the market.
>
> Dan Mauch's design, for example, uses an F84 for logic, will run up

to 40

> volts and gives 1/4 step resolution. While rated for 5 amps, it

could

> likely run close to 10 with just a wee bit of tweaking.

I believe Dan's design is 1/2 step, and i'm not
sure if it is current compensated half step
(which reduces resonance) or traditional half
step. Also, there is a BIG difference between
5A and 10A, like 4 times the power dissipation
and exceeding the package dissipation specs of
many smaller components like TO220 mosfets etc.
I must say I haven't seen Dan's design, but
having designed power control hardware for many
years I can tell you 5A to 10A drivers are as
different as 2A to 5A drivers are...

> 2. You cannot expect that the "inbetween" steps in a microstepping

controller

> to be at all accurate and as the number of intermediate steps goes

up, so does

> the error. ( Ericcson/NJR has a useful app note describing the

problem )

> If your aim is higher resolution, you would do better to use 1/2 or

1/4 step

> and different pulleys to gain the increase.

Wrong! Yes microstepping CAN be "at all" accurate,
the question is "how accurate"?
Positional accuracy is *excellent* in 1/4 and 1/8th
stepping, and still quite good in 1/16th stepping.
PROVIDING you "tune" the motor currents at each
microstep ratio to suit that model motor. I have
spent the last couple of weeks doing just this. :o)
Remember for 8th stepping there are only 3 current
levels that need tuning, and the full step and
half step standards. Not hard to tune 3 ratios.
Only 1/32 or smaller starts to get inaccurate
due to the ratio of step size to static friction.

> 3. The a3955 IS interesting in that it mixes current decay modes

automatically,

> but that's about all the chip has going for it as a conventional

controller.

>
> It is ABSOLUTELY UNUSEABLE for your application, as it is designed

to drive a

> motor ( and hence has two outputs ) whereas a bridge driver has

FOUR outputs.

The a3955 will do as Lawrence requires, it will
drive ONE PHASE perfectly as it was designed.
Obviously you need two for both phases. It will
drive a h-bridge fine with a few resistors.
Remember a stepper motor requires TWO h-bridges
for bipolar operation.

> 4. Since you're already planning on using a PIC, use it for ALL the

functions.

> ( I've designed such a circuit, so I know it's possible ).
>
> PIC -> Bipolars -> FETs OR
>
> PIC -> IR drivers ( as mentioned previously ) -> FETs
>
> Add a dual comparator and a resistor ladder of some sort and you're

done.

This sounds somewhat similar to the commercial
project I am currently working on. :o)

>
> Here's the advantages of such a system. You now CAN go to 1/8 or

1/16 step (or

> even further) with accurate step sizes by tailoring the reference

voltage

> for each step IN BOTH DIRECTIONS of each motor using a table inside

the PIC.

Sounds like you are talking static friction again,
this is not really an issue for 1/4 and 1/8th
steps, they will have excellent accuracy without
having to compensate for directional effects of
static friction.

> You would start with the classic sin(a)/cos(b) current relationship,

but then

> measure the ACTUAL step size and adjust the currents accordingly.

These new

> values are then programmed into the PIC and you have a controller

customized

> for your particular motor.

Yes, that's how you do it.

> The PIC is certainly able to perform all the on/off logic and since

it knows

> about the current levels ( having output VRef itself ), the PIC can

also

> decide upon a current decay mode strategy.

I am curious how you generate vref with the PIC,
by hardwired resistor ladder? Or pwm??

> The design of the bridge should present no particular problem, if it

is

> understood that a relatively high current capacity totem pole must

be used to

> drive each base. ( You must charge and discharge that GATE cap

quickly or

> your maximum step frequency suffers and the FETs get hot from living

too long

> in the linear region ) There ARE totem pole driver chips out there

to ease the

> design complexity, but they start to drive the price of the project

up.


There is not a lot of complexity in a h-bridge
if it is designed well. Gate capacitance should
not be huge problem.

Hope you don't think I am having a go at you,
just a few things you said were quite wrong and
may put people off microstepping which would be
a shame. 8th stepping is an excellent way to
increase positioning accuracy by four times over
half stepping. I do agree that VERY small
steps like 1/32 and 1/64 etc may have nothing to
offer in accuracy but 1/8th and even 1/16th do.
:o)
-Roman