Gallery, Projects and General > Project Logs

Electronic Leadscrew for the New Lathe

<< < (21/32) > >>

RussellT:

--- Quote from: vtsteam on April 01, 2015, 04:26:11 PM ---For Integer inch pitch threads, there is no error in the ability to send steps on time in relation to the spindle position. which is what counts as far as the thread is concerned. There is no periodic error in those threads.

--- End quote ---

Yes - I wasn't clear but I'd also looked at other threads.

Russell

DMIOM:

--- Quote from: PekkaNF on April 01, 2015, 02:44:31 AM ---I like the direction you are heading. Keep it clean and simple.

Very long time ago I used EPROM as a look uptable - without using a uP. Really simple. Incremental encoder to address lines (all extras pulled), data lines were output. Eprom was mostly wasted but small partion of was this lookup table. You need reset and clock, but they could be generated from index and quadrature count pulses?
.........
--- End quote ---

OT(ish):

This reminded me of a project where I did something similar, only using volatile storage instead of EPROM.

Many years ago, I was involved in designing what was thought to be the first =digital= scanning electron microscope (SEM) in Britain, if not in Europe.

I was asked if I could use any digital techniques to improve the control and discrimination of the display; as previously the totally analog display channel typically only had the possibility of adjusting the gain and upper/lower thresholds (albeit often with multi-turn pots).

Computers at that time were far far too slow to process the image on-the-fly, so what I came up with was a hardware solution where each scan spot, instead of going through analog shaping, was digitised, modified and then converted back to analog for display. The essential central converter was a "register file" (what we would now recognise as a small block of memory) with 256 locations and stacked to 8 bits deep. The values from the ADC (Analog to Digital Converter) were used as the addresses into the register file, and the contents of that location were used to drive the DAC (Digital to Analog Converter) thence to the display. The register file was dual-ported, so you could load any pattern into the file - initially from thumbwheels and an 8048 micro, then later from a Commodore PET. Preparation of what we called "the profile", which is now more widely described as a lookup table (LUT) on the PET meant that you could add as as many knees, stretches etc. as you wanted; and the profiles could be stored and shared - between runs, between researchers and even between sites.

Although the input from the SEM was monochrome, by fanning-out from the DAC to a stack of three register files, we added false-colour ability; modification of the colour lookup tables on the PET allowed various features to be highlighted, and we added a facility similar to the zebra-stripes in professional video camera viewfinders that if the video settings meant that detail was being lost at either end, we would highlight that area on-screen - at the bottom, crushed blacks would be highlighted in blue and over-exposed/burnt-out areas would be highlighted in red.

Dave

PekkaNF:
I hope that VT will post here something, othervise we are running amok with OT :D

I remember using hardware FIFO and LIFO registers with hardware multiplier or something. The best thing with these assyncrous lookuptables and regfisters was that you could syncronize them with the "process". Eliminitating a whole lot of quanisizing error and timing problems. You could actually have PLL to lock the "clock" and inhibit "data" when the loop was not locked, that saved whole lot of housekeepping if you were logging data. Now I'm probably good just for blowing up electrolytes.

Pekka

vtsteam:
Pekka, thanks! I'm working on this in the evenings but not ready to write anything here yet because I have several options I want to explore, and I am waiting for parts I ordered to arrive by mail so I can build and test.  I should have everything here within the week.  :beer:

PekkaNF:
I love OT almost as much as pictures!

Two of my friends are working on one polar cordinate measurement system. They need several thousand pulpes per revolution and they choose to use quadrature pulse sensor on X4 mode. They found one chip that is normaly sold in 1000s, but ebay rescues. There is a Eage file, they are going to test it, probably even next week. They are pretty confident that they are going to all that 4x and +/- 1 non accumulating pulse error.

Pekka

Navigation

[0] Message Index

[#] Next page

[*] Previous page