Re: [Emc-users] Latest version

[Jon E. <el...@pi...>]

David A. Frantz wrote:

> Hi John;
>
> While Jitter is certianly an issue the servo update loop speed is also 
> a factor.    Five years ago I was working on CNC lathes with sub us 
> update rates.    I actually had the good fortune to talk to the 
> engineer responsible for the DSP card that drove the system, he 
> insisted that the update rate was need to do what they where doing.   
> Since at the time I had considerable experience with various diamond 
> turning technologies, I had to believe him as his equipment had rather 
> impressive abilities to interpolate radius and maintain surface finish.

What?  Less than a microsecond?  There has to be a mistake here.
Even the fastest DSP processors only execute 100 instructions or
so in 1 uS.  But, if the servo motor is driven by a PWM driver,
any updates to the PWM duty cycle are not going to have any
effect if it is faster than the whole PWM cycle.  Were the servo
motor drivers running PWM faster than 1 MHZ?  That is possible,
I suppose, but it still doesn't make sense.  Can the motors/leadscrews/
carriage/tools really respond with x hundred KHZ bandwidth?  That
is totally preposterous.

Ahh, diamond turning.  Well, the whole problem must be a resolution
problem, and they were using tricks with the DSP to meet the needs for
resolution when making optical mirrors.  That is a far cry from the usual
machine tools, though.

>
> Now I fully realize that a gantry type machine is a different animal 
> than a diamond turining lathe cutting optical parts, but I would 
> suspect that at some point the update rate will impact the abiltiy to 
> further increase a systems performance.    If his system can actually 
> hit 400 ipm,  a 20us update rate will probally do, but I do wonder how 
> much faster he would be able to to go and still keep the resultant 
> quality high.   On the type of machine being discussed I would think 
> mechanical issues would be huge before controls issue becmae a 
> problem, but then agian I've not worked on that type of equipment.
>
> Also I'm not convinced that a one milli second update rate would be 
> suitable for any CNC machine doing any sort of interpolation unless a 
> rather loose end result is suitable.    I geuss this depends somewhat 
> I what one wants to accomplish with the system.

HUH?  Why do you say that?  For **ANY** CNC machine?  Do you know what 
the basic
mechanical accuracy of a Sherline, Taig or Harbor Freight machine is? 
 Do people
who cut wood signs need NANOMETER precision?

Do you know any home shop types who are running their Sherlines at 400 IPM?
Do you know how much power it takes to make a sharp corner at 400 IPM on
a Bridgeport-sized machine?

I think one must match the servo control system to the rest of the machine.
If you have a 1/8 Hp servo motor moving a 400 Lb table, it is TOTAL 
nonsense to
talk about microsecond servo loops, when typically cutting at under 10 IPM
speeds.  My machine is more accurate than I can measure, most of the time,
and tool deflection is the largest source of error, by a factor of 3 to 
5.  My
servos are updated at 1 KHz, although I must add that they are true velocity
servos, and so the amps actually have a velocity bandwidth greater than 
1 KHz.
This is all meaningless, however, because the low motor torque and high 
table
mass make it impossible to respond significantly above 40 Hz or so, in the
mechanical sense.

I urge you to consider that a lot of people making their own CNC machines
are NOT doing diamond turning lathes.

Jon