Dear all,

hi,
my name is george (yioryos in greek) dimitriadis, and this is my first
post.
I would be grateful if everyone is interested to reply to my post and please
excuse me for any
non-conformities according to the newsgroup policy.... I will do my best to
be as typical as possible.

I am trying to build a stepper microstepping driver for my small cnc mill. I
am not new to microstepping and
I am aware of chips such as lmd18245 or the popular a3977 from Allegro. I am
also aware of the L6207 with dual bridge
and built-in monostables and comparators. The thing is that since I am a
hobbyist and passionate with designing my
own projects, I like building from scratch in order to both learn and have a
tailor-made system according to my needs.

I have experimented with L6205, a 7555 monostable triggered by a pic
comparator at slow decay mode that worked fine.
When I chose to put the comparator output directly, (without the
monostable), to one of the 2 bridge logic inputs for slow decay again,
or at enable input for fast decay, it also worked well but the currents
should be kept low otherwise due to the relatively high Ron
and the resulting high frequency losses, the chip's temperature would rise
significantly.

So, I decided to build my own h bridge with low Ron mosfets, being able to
achieve high frequencies and currents, utilising fast or slow decay,
in order to have fine chopper control of coil current, necessary for
microstepping. The circuits attached, are not intended as
the stepper motor final drivers, they are just for the development and
debugging of the chopper function.

I found that there was a very handy chip, HIP4081A, that would drive
properly an n-channel fet h-bridge using charge pumps
and bootstrap capacitors. Furthermore, deadband delays are implemented and
input logic has been taken care of, by means that
the bridge can be fully controlled by the lower side inputs.(ie if the lower
inputs are high, the higher inputs are "not care", thus
by having the higher inputs tied at logic high, the full bridge is
controlled via the lower inputs).

But, there was a problem.... Although I think I was careful in setting up
the circuit; proper bootstrap capacitors according to
mosfet input capacitance,shotky diodes at bootstrap circuit, 1n5822 shotky
diodes at mosfets, (I was amazed that both the hip4081a
application note and datasheet does not mention anything about diode
protection), decoupling caps in all chips and using the same
succesful strategy as with L6205, the results were not as expected.

In L6205, the Vref I was setting for the current control,(fed in the
non-inverting input of the pic comparator),was the upper magnitude limit of
the chopper frequency measured at Rsense with my oscilloscope. The target
current was always met with L6205.It was also dynamically linked with Vref,
ie when I was changing Vref by turning the preset, current limiting was
following the theory. With the HIP4081A though,
the waveform at Rsense was very ugly: Instead of the usual chopper
trapezoidal ripple with a magnitude=Vref voltage, there was a spiky waveform
with no
trapezoidal form at all, and 2 important drawbacks: I had a magnitude much
higher than the one set by Vref, there was no whatsoever dynamic
link between changing Vref, (it was locked at that magnitude although I was
changing Vref), and the spike was extending in the negative region
as well at approximatelly 1/3 of the positive spike magnitude.

One thing is for su I used the exact same strategy and working hardware
as the one that succesfully operated with L6205 but
I am missing something....

Although I hoped I would have an easy to read post for my first one, I ended
up having all these.I am sorry for this.

I would be grateful if anyone is interested to tell me his/her
opinion/advice considering the problem I am facing.
Please excuse both my english and drawing quality.

Hope to hear from anyone out there!

Thanks,

yioryos

"yioryos dimitriadis" tainaros2004(remove wrote in message
...
[...]
yioryos

Switching Fets ON creates a nasty edge at the front of the current sense
waveform. The drawing has been altered a little to show a simple filter.
Although it is actually easier to do the filtering in the PIC program.





--
Posted via a free Usenet account from http://www.teranews.com