[Mariss Freimanis]

Hi,

Short story is I decided to learn Verilog. Verilog is a language for programmable logic ICs. Learning it meant having a hardware target for the code being written while learning. I chose a DC servodrive as the target because it's relatively simple.

Learning it turned out to be much easier than I figured; what was supposed to be an exercise target has unexpectedly turned into a real drive. If you have used our G320 drives before, then I think the G380V will have significant improvements over the G320. Here are the differences:

1) G203V protection features. Short-circuit, over-voltage, under-voltage, over-temperature, reversed power supply polarity and internally fused. Also running a servomotor into a hard-stop won't hurt it. It should be unkillable.

2) All PID coefficients are now trimpot settable. The integral setting is included now. Top adjust trimots like on the G203V.

3) A true current-mode power amplifier section. The G320 has a quasi current-mode amplifier. This should help by adding phase margin stability.

4) Servo locking range is increased from +/-128 counts to +/-2,048 counts. That's +/- 1 full motor revolution with a 500-line encoder.

5) Encoder CH_A and CH_B inputs are now optoisolated. Multiple axis encoders can be driven with a single external 5VDC power supply now without any ground-loop problems. The encoders can also be read by the PC now if the CNC program allows it.

6) Separate FAULT and RESET terminals. NO more ERR/RES terminal. FAULT out is an optoisolated open collector, RESET in is an optoisolated LED input.

7) An optoisolated WARN open collector output. This output indicates when the motor is at full torque but developing an increasing following error. This gives warning the drive will FAULT if the error reaches +/-2,048 counts, giving ample warning in time to do something about it.

8) An IN_POSITION indicator showing the motor is within +/-2 encoder counts of the command position.

9) A 5-second 20A timer. Set the LIMIT trimpot to the motor's rated continuous current. The 5-second timer allows for max torque while accelerating or decelerating, then reverts to the LIMIT trimpot setting. This prevents the motor burning down when overloaded yet allows max torque for pulse loads.

10) Anti-dither (dead-band) enables 1/100 of a second after the last step pulse is sent. This keeps the motor quiet while stopped yet keeps servo stiffness while running. This feature is option-block selectable.

11) Four LED indicators. POWER, FAULT, WARN and IN_POSITION.

12) Built-in encoder divider. Settings are divide by 1, 2, 5, 10. Kind of like a G340. This feature is also option-block selectable.

13) The 7 optoisolator channels have a G203V interface specification, meaning it will work with 2.5V, 3.3V and 5V logic signals. All optoisolators are socketed. Maximum step pulse frequency is >500kHz.

14) It all fits in the same outline G320 package, price will be similar to a G203. All other specs (supply voltage, motor current, etc.) are the same as a G320. The prototype boards were let to the PC fabricators today. If everything goes as planned, look for it in the 1st quarter of 2008.

Mariss

[davo727]

Sounds awesome, I am working on the mechanicals of a mill conversion. Thinking about February, hopefully I can go with the 380 drives otherwise I may have to use 320s and upgrade them later when 380s are out.

[bclark6716]

Marris,
Is this the stepper/ servo that has been discussed in earlier post, or an upgrade to the 320?

Brian

[ger21]

Upgrade to the G320

[Xerxes]

Mariss, why not add option jumpers that allow supplying ENC_GND and ENC_5V from drive? Would be useful for most of the users.

[rokag3]

hey Mariss,

1) G203V protection features. Short-circuit, over-voltage, under-voltage, over-temperature, reversed power supply polarity and internally fused. Also running a servomotor into a hard-stop won't hurt it. It should be unkillable.

I am going to use my 3*G320 with the KL34-170-90 at 75 V with a French software (NINOS) which is in the beta version sometime unpredictable and so i may experiment hard stop
i have a schema circuit from gecko June 16 2002 to shunt to the ground current returned from the load. Is it preferable to implement it ?
by the way you said that this circuit will be implemented with a led indicator 51/2 years after have you done it
also i fried one Fairchild 2531 opt there is no way to find it in Greece how to get one (and even 3)

Except that i think that in your upgrade you should open the way for optic fiber (see in diy servo uhu)
http://www.cnczone.com/forums/showth...14217&page=122
#457
because it's very cheap now and you solve electric and interference problems. For sure someone is going to make a cheap quadrature encoder with toslink soon . but even now it's very easy to connect a cheap encoder with the existing circuit (3 pin:signal,+5V,0V)

For the 5v supply of the encoder i think that a small regulated 5v7805 based must be implemented close to the machine so you can avoid all electrical connections why spend money for optocoupler if you let 5v wire jeopardize your drive !!!

[davo727]

Hello Mr. Mariss Freimanis, Wondered how the G380 deal is going. is it too early to estimate a time on having units shipping out? Thanks, Dave

[CoAMarcus]

We are currently working on the prototypes of the G380, and will be fine tuning the firmware on it to get what we like. We are looking for 100% utilization of the CPLD, and will add most (hopefully all) of the features we are looking to get in there. I would say if all goes according to plan, you should look for it within the next three months. We are not ones to give definite dates to product releases if they are still in research, but here's crossing my fingers there are no roadblocks.

-Marcus Freimanis

[mlaws1172]

Sounds like you are going to make it harder for to blow one up. How about a setup for diff. encoders? If I read this right, the 50ma encoder limit went away?
mike

[Mariss Freimanis]

The prototype boards should be populated by next week and be ready for testing and debugging. I'm tending away from having the production G380Vs powering the encoders:

1) Someone will always try to run an encoder that draws too much current, causing the G380V to malfunction. Doing so can over-dissipate the internal voltage regulator. Numerous G320 failures trace back to chronic regulator overheating due to excessive encoder supply current loads.

2) The G380V encoder inputs are opto-isolated so the encoders can be powered by the PC or breakout board +5VDC. A lot of encoders have an index channel (CH-I) which becomes useless if the drive powers the encoder.

3) I'm loathe to use a different main connector. The normal 12-position connector is sufficient if the encoders are externally powered. Just parallel all encoder +5VDC and GND supply inputs to the breakout board or PC 5V supply and run encoder CH_A and CH_B to the G380V just as you would STEP and DIRECTION.

4) No encoder supply requirements means no need for an on-board 5VDC supply and attendant regulator. This reduces the G380V regulator count to 12VDC, 3.3VDC and 1.8VDC, making the drive less expensive.

5) External 5VDC encoder supply means differential encoders can be used single-ended. The G380V CPLD logic uses an integrating type digital noise filter and a quadrature decode algorithm that cannot be fooled by common-mode noise on the CH_A and CH_B inputs. This cancels the main advantage of a differential input.

6) Having a +5VDC "port" open exposes the G380V to potential damage the protection features of the drive cannot control and eliminate. If the G380V is to be truly bulletproof then this "port" has to be closed. In the long run, this slight inconvenience will be outweighed by the survivability its elimination imparts to the G380V.

----------------------------------

New stuff just keeps on happening department: I just closed out the board layout for a G213V, an until now unannounced new drive. This is a G203V with an integral pulse multiplier. The new pulse multiplier is a 4-quadrant PID servo-type summing node circuit design borrowed from the G380 which simply cannot be fooled by any constant-contouring application. That means the multiplier cannot make mistakes no matter how or when or at what speed axis direction reversals are made.

The new pulse multiplier is integrated into the new G213V meaning there is no separate plug-in multiplier board; it's part of the main board. This means the manufacturing costs are only a few percent more than a G203V and the price will reflect that.

Mariss

[mlaws1172]

You are right about not powering the encoders with the drive. Cant wait to try some
mike

[fseger]

any ETA on this new drive?

[CoAMarcus]

Not yet. Because we all have ADD at Geckodrive and can not choose just one project at a time, this one will be finished around the same time as the G250 and G540. I would expect around May at the latest, but don't quote me. This is a very rough ETA, not anything planned.

-Marcus

[fseger]

Thanks,
Looks like I'll be using the 320 again for next project.
Cheers
Fred

[bcromwell]

Note to self: send Marcus large quantities of Wellbutrin so he can get G380's complete

[nicad]

A little late in the thread, I know....
I didn't see a cost referenced anywhere.
Any idea on cost range for the G380? I assume more than a 320... but by how much?

[CoAMarcus]

It will probably be comparable in price to our G203V because it will be using the same CPLD based architecture. I would speculate that it will cost around $150. Just as a disclaimer, this is a ballpark and is not to be taken as anything but speculation and good guessing.

-Marcus

[pminmo]

Mariss why did you go with Verilog over VHDL?

[MrWild]

Once this is done, maybe a "cheap servo drive?" Lower power, lower amps, smaller servo motors. Fewer bells and whistlkes. Something that will run a 24/48volt locog motor. Now that'd be sweet.

[CoAMarcus]

pminmo,

Verilog was chosen over VHDL out of pure laziness. Verilog is easier to learn than VHDL, and it does just as good of a job. We learned Verilog in about three weeks, and we only use about 6 different commands in it.

MrWild,

We may consider that, but with all of the things on our plate it will probably be a while. We still have the G250, G251, G540, G380, G213V, G201 R12, and the stepper servo on tap before that. So the answer is probably, but it will be a while. ;-)

-Marcus Freimanis