#emc-devel | Logs for 2008-07-23

[00:19:43] <skunkworks> jmkasunich: http://www.electronicsam.com/images/KandT/servostart/newschem4.png
[00:23:59] <skunkworks> another stupid question.. should I only hook logic ground to servo supply ground at the resister? (so technically - I have 2 separate grounds with the low end of the resister the only thing tying the logic ground to servo supply ground..
[02:58:33] <jepler> hmph. seems like given this 'wire p = q - 100; always @(posedge clk) if(p < 50) t <= 0; else t <= 1;' the software could infer just a comparator. unfortunately, it seems to infer a subtractor plus a comparator.. I guess I see now that it couldn't because the subtraction could "wrap around" to higher values ...
[03:42:11] <CIA-34> EMC: 03jmkasunich 07TRUNK * 10emc2/src/emc/usr_intf/axis/extensions/minigl.c: add support for a couple more primitives
[11:49:20] <cradek_> cradek_ is now known as cradek
[20:35:27] <Roguish_> hey, all you programmer guys. how do i determine the largest encoder count possible?
[20:35:45] <SWPadnos> start counting
[20:35:58] <Roguish_> i have a rotational axis with 7118 counts per degree.
[20:36:08] <cradek> ouch
[20:36:20] <Roguish_> my customer does not want to have to unwind.
[20:36:53] <Roguish_> i32? i64? floating point counter???????
[20:37:07] <SWPadnos> floats are effectively 24-bit
[20:37:15] <Roguish_> need to set the max and min limits really high.
[20:37:31] <SWPadnos> once you hit that limit, you can't add a single count to the number
[20:37:55] <SWPadnos> the encoder drivers should extend whatever the hardware can do to 32 bits
[20:37:56] <Roguish_> what kind of number is used ?
[20:38:17] <SWPadnos> it should be a 32 bit signed int, but it could be hardware dependent
[20:38:27] <Roguish_> m5i20
[20:38:29] <SWPadnos> (ie, some drivers may not extend the value to 32 bits)
[20:38:33] <SWPadnos> I think that one is 32 bits
[20:39:42] <SWPadnos> that's about 300k degrees, or ~838 revolutions
[20:39:49] <SWPadnos> (+/- that amount)
[20:40:22] <Roguish_> sounds good, hope it's high enough.
[20:41:15] <SWPadnos> unwinding is another problem entirely
[20:42:13] <Roguish_> ok, for a non programmer. in binary > 32 1's converted to decimal?
[20:42:31] <SWPadnos> 31, it's signed
[20:42:46] <Roguish_> at least i was close. thanks.
[20:42:54] <SWPadnos> 2^31 is 2147483648
[22:41:50] <cradek> woo, the lathe is in place