Centroid CNC4

forhire wrote:

Here is my shopping list so far:
1 Power supply www.antekinc.com PS-15N70R5R12 $240
3 Stepper drives www.geckodrives.com G203V $147 each x 3 = $441
1 Universal Stepper Control (USC) pico-systems.com/univstep.html $250

I'm still working on the jog control portion. Does anyone have a recommendation? I'm not sure I want to use a touch screen or mouse.

I love a pendant with jog dial (MPG). I bought an HEDSS jog dial on eBay for about $52.
With a 3-axis system using the USC board, you have one more encoder input available to use for the
jog dial. (You can also use general digital inputs, these are sampled 1000 times a second, so even the
software encoder counter can keep up with your hand unless you really twirl the dial.
Here's my first try at a pendant: pico-systems.com/pendant.html
The HEDSS one is quite a bit smaller and fits in the hand.

I am still thinking about the stepper encoders. I believe the USC can read encoders. The USC can also counts pulses... not sure how that compares to encoder counts.

By default in an open-loop system, the USC counts the step pulses it issues to keep EMC in sync with the
stepper drivers. You can put encoders on the motors, this will allow you to detect stalls and missed steps.
You want fairly high resolution encoders, so there won't be positions on the motor that you can't get to because
there's no encoder count at that position.

I ordered new pulleys so I can get better resolution. Can't wait for them to arrive

While this WILL increase resolution, it may have an adverse effect on speed. Torque on stepper motors falls off rapidly at higher speeds, so if the motor has to turn twice as fast, you may actually deliver less torque at the same table speed. But, the Gecko drives are so MUCH better than hideous old full-step unipolar drives that you may still end up with more resolution plus more speed.

Jon

BigJohnT wrote:

You might want to do the math on the USC to see if you gain any performance over a simple breakout board like the C1G from cnc4pc. With a decent latency I "think" you can generate pulses faster than the stepper can take them... but Andy would know for sure

www.cnc4pc.com/Store/osc/product_info.ph...th=33&products_id=49

In my case my steppers are rated for 1800 rpm and larger steppers can be less that than. You might do some math to see what the highest rpm your machine would drive the steppers for a ballpark idea of the maximum step rate you will need.

John

A MAJOR benefit to a hardware step generator is the smoothness of pulse trains, due to the finer granularity of the hardware. For instance, the USC has a clock of 100 ns, about 200 times finer than most software step generators can handle. 1800 RPM with a Gecko 200-series step driver requires 60,000 steps/second.
At 25,000 steps/second, EMC2's software step generator has typically a 50% jump in step timing. The USC has a 0.3% jump.

Jon

jmelson wrote:

Here's my first try at a pendant: pico-systems.com/pendant.html

Thanks for the information on your pendant. I think that will work well. I found the wiring diagram and I think I understand how it works

By default in an open-loop system, the USC counts the step pulses it issues to keep EMC in sync with the
stepper drivers. You can put encoders on the motors, this will allow you to detect stalls and missed steps.
You want fairly high resolution encoders, so there won't be positions on the motor that you can't get to because
there's no encoder count at that position.

I have encoders already on the steppers. I'll likely have to take off the covers and get a part number off the encoder. What would be a recommended resolution for the encoder?

I ordered new pulleys so I can get better resolution. Can't wait for them to arrive

While this WILL increase resolution, it may have an adverse effect on speed. Torque on stepper motors falls off rapidly at higher speeds, so if the motor has to turn twice as fast, you may actually deliver less torque at the same table speed. But, the Gecko drives are so MUCH better than hideous old full-step unipolar drives that you may still end up with more resolution plus more speed.

This will be the real test. Obviously I won't know until I try it. I did talk to another CNC4 user who says this was a common and much needed upgraded, he did mention some speed decrease... but with a 3000 rpm spindle... speed isn't my issue. I finished boring and broaching the new pulleys today. Unfortunately the Y axis was 1.5 inches longer between centers than the X axis... they looked so similar... that will teach me for not measuring everything. Also, I was 10mm too long on the X axis belt... I should have centered the adjuster when I measured the center. I have to order new belts

jmelson wrote:

A MAJOR benefit to a hardware step generator is the smoothness of pulse trains, due to the finer granularity of the hardware. For instance, the USC has a clock of 100 ns, about 200 times finer than most software step generators can handle. 1800 RPM with a Gecko 200-series step driver requires 60,000 steps/second.
At 25,000 steps/second, EMC2's software step generator has typically a 50% jump in step timing. The USC has a 0.3% jump.

That re-confirms the reason I'm looking to use the hardware USC. I also wanted the IO and encoder capability.

Photo of new 2:1 pulley and belt that is too long:

forhire wrote:

I have encoders already on the steppers. I'll likely have to take off the covers and get a part number off the encoder. What would be a recommended resolution for the encoder?

Well, the encoders were probably selected to be a reasonable match to the motors. A "500 line"
encoder will give 2000 counts when counting all quadrature transitions. That will roughly match
the Gecko microstepping resolution, but due to slight errors in the motor components, its steps
will not line up perfectly with the encoder. So, you will likely have to set a deadband of about 1.5
encoder counts to eliminate hunting at some spots.

Jon

BigJohnT wrote:

Yea, I didn't even think of the I/O part of it... would the new Mesa cards be worth waiting for? As I understand they are going to be out soon and more cost effective in the I/O area.

John

My question is which card is easier to configure? I dont really need anything more than spindle control on top of my motors, so if I can save $200 I would prefer to get the C1G.

Also, how important are the encoders in this setup? I was thinking that I could just not use them when I upgrade if they aren't doing a whole lot.