« first day (4228 days earlier)      last day (705 days later) » 

TQQQ
TQQQ
2:18 AM
Hey. How would i get an analog signal proportional to something in mechanics?

More specifically, i want to detect position according to the profile of an aluminum part. It can be machined to anything - can be different distance (within 1mm to 5mm) or thickness or anything. I am looking for an analog signal that could be digitized to 10 bit resolution with reasonable noise. Speed is not going to be an issue, can move slowly
 
 
10 hours later…
Tony Stewart EE75
Tony Stewart EE75
12:30 PM
1 mm resolution at 1m is about 10 bit resolution. but that does not define accuracy
 
TQQQ
TQQQ
yeah, no, it's inside the device, meaning almost in contact. anyway, i recalled the magnetic encoders, hopefully problem solved
 
Tony Stewart EE75
Tony Stewart EE75
Laser deflection is possible from a mirror or repeatable image
alum. is not magnetic when still
 
TQQQ
TQQQ
alum is not, magnet can be attached.
 
Tony Stewart EE75
Tony Stewart EE75
What's wrong with a depth gauge
You still never gave accuracy or range of 10 bits
10cm 10 m?
 
TQQQ
TQQQ
12:46 PM
10 bits would cover 1mm to 5mm. or 1 to 1.2mm. The point is to measure position, the shape can be defined accordingly
nothing wrong with a depth gauge. i am looking for cheap solutions, but this is a conversation that always ends badly
 
Tony Stewart EE75
Tony Stewart EE75
1:00 PM
That is a 1 micron resolution. Can you guarantee the reference point is that stable? Then I might use interferometry with a Bragg reflector or Fresnel plastic sheet and a laser reflector to count cycles relative to a fixed datum reflector
That is the wavelength of 1 cycle of Infrared
I know a friend who used that for measuring the sharpness of a blade sharpness and coplanarity using interferometry for a steel sheer blade
with adjustments to machine .The surface table was used and large so $$=?
@TQQQ FWIW look up laser interferometry
 
rdtsc
rdtsc
1:22 PM
If you can motorize the moving component, 5rev:1" linear screw + 200 pulse/revolution stepper = 0.001" per step.
 
Nick Alexeev
Nick Alexeev
Folks, I need to estimate losses in a BLDC motor (for thermal management purposes). I can calculate copper losses I²R. I can't calculate the rest of the losses.
I'll have to measure them, but I haven't yet got a setup to measure with.
Is there a rule of thumb which allows to estimate the rest of the losses?
Maybe something along the lines of "Copper losses are usually XX% to YY% of total losses".
 
W5VO
W5VO
If you didn't need "cheap", I'd say that Keyence (and probably others) has some laser rangefinders that work very well and give you micron-level accuracy with decent distance.
 
TQQQ
TQQQ
@TonyStewartEE75 yes, I know what that is. I was thinking more about a sine/cosine type of measurement for improved accuracy. anyway, a 12-bit encoder with a magnet will do the work and will probably cost within 10-15 dollars.
 
Tony Stewart EE75
Tony Stewart EE75
1:40 PM
@TQQQ if calibrated accurately, maybe as mu is very temperature dependent. I have used ferrite Eddy Current measurements to detect 1mm with 12 bit accuracy and > 40 dB SNR with sin/cos at 100k/200kHz
Best case 60dB SNR with PLL's
to measure 1um flaws inside Monel steel.
but for Alum purity affects coupling free of Fe so viscous force is more accurate
using motion
but my system would be far more accurate than a magnet based on my SmCo experience with HDD rotary motors
based on tests done with Fujitsu HDD division
that is why recalibration was done with temperature compensation for Rotary actuator Force constant due to Mu variation
every power up and every change of 5'C a recal was done as it affected overshoot in the servo feedback system
 
TQQQ
TQQQ
2:01 PM
well.. my system is going to cost 200 bucks for the end customer :) so accuracy is not my greatest concern
 
 
1 hour later…
rdtsc
rdtsc
3:16 PM
Nick, I think this largely depends on outside variables. A faster (and higher voltage?) driver could be much more efficient than a slow, lower-voltage one, etc. Maybe make an estimation from the reported efficiency of the driver if known?
 

« first day (4228 days earlier)      last day (705 days later) »