开云体育

Hi All,

A warning to anyone using a LM2907/2917 Frequency to Voltage
converter for spindle control. The chip data sheet does not specify
an upper frequency limit, I have found that it is very non linear
above about 13KHz, I did not realize the problem with Mach1 because I
was using a D-FF as a pre-divider, however when I set up Mach2 for a
Kernel speed of 35KHz I found that at above half spindle speed I was
getting very strange results. The answer was to use the second D-FF
on the chip, so there is now a divide by 4 pre-scalar to the FtoV.
This modification should be good for M2 up to a Kernel frequency of
45KHz.

Apologies to Art for any time he wasted on another non M2 problem.

Bernard

On Thu, 07 Aug 2003 00:36:53 -0000, you wrote:

Hi All,

A warning to anyone using a LM2907/2917 Frequency to Voltage
converter for spindle control. The chip data sheet does not specify
an upper frequency limit, I have found that it is very non linear
above about 13KHz, I did not realize the problem with Mach1 because I
was using a D-FF as a pre-divider, however when I set up Mach2 for a
Kernel speed of 35KHz I found that at above half spindle speed I was
getting very strange results. The answer was to use the second D-FF
on the chip, so there is now a divide by 4 pre-scalar to the FtoV.
This modification should be good for M2 up to a Kernel frequency of
45KHz.

Bernard - Can you explain this a little more, a circuit diagram might
help? The National spec sheet shows it as pretty linear ie less than
0.4% across a 500kHz input range??
--
Steve Blackmore

Steve,

Linearity is only defined over the range 1kHz to 10khz, see note 5 in
the spec sheet. For practical purposes it extends from about 200Hz
to 12 - 13kHz, the bottom end depends on the filter capacitor. I am
retired now and without the necessary test equipment I can't do full
tests, but from empirical tests the response falls very rapidly, e.g.
@4.25kHz Vout= 2.5 @8.5kHz Vout=5, @12.75kHz Vout=7V & @17.5kHz Vout=
7.3V.

I will post a hand drawn circuit of my revised circuit later to-day.

Bernard

--- In mach1mach2cnc@..., Steve Blackmore <steve@p...>
wrote:

On Thu, 07 Aug 2003 00:36:53 -0000, you wrote:

Hi All,

A warning to anyone using a LM2907/2917 Frequency to Voltage
converter for spindle control. The chip data sheet does not

specify

an upper frequency limit, I have found that it is very non linear
above about 13KHz, I did not realize the problem with Mach1

because I

was using a D-FF as a pre-divider, however when I set up Mach2 for

a

Kernel speed of 35KHz I found that at above half spindle speed I

was

getting very strange results. The answer was to use the second D-

FF

on the chip, so there is now a divide by 4 pre-scalar to the

FtoV.

This modification should be good for M2 up to a Kernel frequency

of

45KHz.

Bernard - Can you explain this a little more, a circuit diagram

might

help? The National spec sheet shows it as pretty linear ie less than
0.4% across a 500kHz input range??
--
Steve Blackmore

On Thu, 07 Aug 2003 15:05:39 -0000, you wrote:

Linearity is only defined over the range 1kHz to 10khz, see note 5 in
the spec sheet. For practical purposes it extends from about 200Hz
to 12 - 13kHz, the bottom end depends on the filter capacitor. I am
retired now and without the necessary test equipment I can't do full
tests, but from empirical tests the response falls very rapidly, e.g.
@4.25kHz Vout= 2.5 @8.5kHz Vout=5, @12.75kHz Vout=7V & @17.5kHz Vout=
7.3V.

Have a look at the graph "Tachometer linearity Vs R1" on page 6

With an input voltage of 12V, Frequency of 1000Hz and a R1 value of
350K it's 0.2%

The input frequency is set by steps per unit and velocity, get your
min/max step rate between 1KHz and 10K and it's about as linear as it
gets.

I will post a hand drawn circuit of my revised circuit later to-day.

Are you putting it in Mach1Mach2 files section?

--
Steve Blackmore

--- In mach1mach2cnc@..., Steve Blackmore <steve@p...>
wrote:

On Thu, 07 Aug 2003 15:05:39 -0000, you wrote:

Linearity is only defined over the range 1kHz to 10khz, see note 5

in

the spec sheet. For practical purposes it extends from about

200Hz

to 12 - 13kHz, the bottom end depends on the filter capacitor. I

am

retired now and without the necessary test equipment I can't do

full

tests, but from empirical tests the response falls very rapidly,

e.g.

@4.25kHz Vout= 2.5 @8.5kHz Vout=5, @12.75kHz Vout=7V & @17.5kHz

Vout=

7.3V.

Have a look at the graph "Tachometer linearity Vs R1" on page 6

With an input voltage of 12V, Frequency of 1000Hz and a R1 value of
350K it's 0.2%

I'm missing something as I can't extrapolate a frequency response
from this graph. In your earlier post you mentioned a frequency
response of 500kHz, is this the graph you were referencing?

The input frequency is set by steps per unit and velocity, get your
min/max step rate between 1KHz and 10K and it's about as linear as

it

gets.

Maximum resolution is obtained by dividing the Kernel Frequency by
Motor RPSec, this gives steps per unit. The purpose of the warning
was to advise group members that the LM2907 has a frequency limit way
below what M1/M2 is capable of outputting. From direct email I know
that there are a lot of queries about this chip, and while it is
possible to maintain it within linear limits by manipulating the
tuning paramaters, the pre-scalar prevents one source of non
linearity and makes for very simple motor calculations.

I will post a hand drawn circuit of my revised circuit later to-

day.

Are you putting it in Mach1Mach2 files section?

Listed as FtoV with Pre-scalar.

Bernard

--
Steve Blackmore