Wishlist for the next drive (describe your dream drive!)

[Eson]

It looks good! Does the MOSFETs require a heatsink?

[guru_florida]

I like the discussion here and I have some related questions. I have posted a new topic so as not to hijack this thread! Can I request that you pcb pros read the following post:
http://www.cnczone.com/forums/showth...674#post380674


Also, I also agree that screw terminals or the PowerMod connectors would be a better idea for motor connection. Soldering 6 pins on the db25 is more of a job for me although my solder skills are very good. At 3am, I could easily end up soldering the wrong pin or bridging two pins by mistake and bammm! (its a common rookie mistake right.) Raise the price 5$ and sell the mating connector with the board perhaps. Just my 2c worth.

Excellent work though!

Thank you,
Colin

[Xerxes]

Eson: yes, top side of drive will be covered by black anodized alu plate.

Colin: most of the users don't need 6 pins in parallel if maximum amps are not required. Shorting pins also will only blink some leds thanks to short circuit protection. But indeed there could be an option for different connector.

I did some more research on short circuit protection. It appears that MOSFETs can take hundreds of amps in short peaks safely so I can still sleep my nights well. Also competing products have similar characteristics, or worse (I tried shorting couple of them) :-)

[CNCAddict]

I'm also beginning to wonder about soldering 18 wires for a motor connection. It would of course be possible to bridge the 6 pins with a solid copper bar, then just have to make 3 wire connections. That might not be a bad solution, but this "tip" might not be obvious to everyone. It's also a problem that if one uses a ready made DB25 cable, then it will still be a bad experience stripping/soldering all 18 wires at the motor end.

Right now I'm VERY partial to this sort of connector (see below). They have them in different current ratings and the prices aren't too high. The advantage is that they are pluggable AND screw terminal so you can have your cake and eat it too This is the same style connector used on Kollmorgens Servostar PD drive.

http://www.mouser.com/catalog/632/1349.PDF

[Xerxes]

CNCaddict, thank you for sharing your opinion! I was thinking that kind of connector as alternative. However, it lacks strain relief which is bad when drive is installed so that connectors reach outside world without intermediate wiring.

I'm planning to make also simple "motor breakout board" which converts between screw terminals and D-Subs. Just add some molded D25 and D15 cables between drive and breakout and you have robust shielded cabling and zero solder joints. Breakout should be mounted near motor.

[CNCAddict]

OK, the breakout board would be a fantastic solution! The thing I really hate about the BIG companies is that the cables cost about 1/3 the price of the drive and they're usually proprietary connectors. This is a dishonest way of hiding the true costs of a system until it is too late

Another suggestion would be a way to actually change servo settings and retrieve actual position from the drive inside Mach3. So it would be possible to do things like track (save to file) actual position vs. commanded position right from inside the CNC software; or do cool things like shut down the control loop and just use the servo drive as a DRO for guys who keep handwheels on their machine. All industrial CNC machines that I know of allow position feedback right at the control...so far none of the hobby systems allow this easily. It usually takes a servo drive AND some sort of encoder counter board. I think this is a very inelegant solution since the servo drive already has all this software/hardware built in. It would also be nice to view fault conditions and/or servo current from inside the control software. Your thoughts???

[Xerxes]

All that is possible by using SPI bus but no CNC software supports that at the moment. Serial bus could also used as step/dir replacement without its restrictions.

However, VSD-E makes it possible to use SPI and step/dir simultaneously so it would be possible to use separate program to monitor drive status/positon/following error etc while CNC software is using step/dir.

-----

I did some more testing today. Analog circuity seems to work fine including the new differential +/-10V inputs and analog temperature sensing. Analog temp sense makes it possible to add adaptive output current limits. Drive can automatically reduce current limits if it's running hot and finally shut down if temperature still rises.

Analog signal quality seems to be exceptionally good. ADC noise is well below 2 bits (I calculated 0.82 bits RMS from 1000 samples), this is as good as it gets. I think 4 layer board might have something to do with this :-) Attached picture (zoomed) shows captured ADC signal noise.

[Episs]

ADC noise is really low, in which layer did you rough thous ADC traces, and what was nearest layers ? was it something like
1.Ground
2.ADC
3.Ground
Or something else.

How about PC Power supply connectors They have a locking device with arms to keep them securely connected here is a picture. My fend use such type connectors on his self made driver for old Russian 6phase stepper motor.
connector ratings for LR-02-2V (form Elfa catalog)
Max current: 15 A / 10 A (2-pole / 9-pole)
Max voltage: 300 V
Contact resistance: 15 mΩ max
they are cheap, high power and secure.

[Xerxes]

Episs, the analog section layer order is:
1. Analog signal
2. AGND
3. AVCC
4. GND+other signals

AVCC has been filtered by LC filter.

Good point about connectors! Those are available for PCB mounting as well.

[jahonen]

Current measurement looks very good. Is that under powered conditions (the power stage operating with the load) or just the basic noise of the measuring amplifier + ADC? I recommend testing with powered (dynamic) conditions, since that might change the situation somewhat (not necessarily very much, though).

I have also found that proper arrangement of the sections and carefully thinking where the current flows, produces very good analog signal fidelity with multilayer boards (and usually low EMI). This is just because the current loops are so small, so there is almost no crosstalk (inductive coupling, which is by far the most important coupling mode at high frequencies) between them.

Regards,
Janne

[Xerxes]

Janne,
I noticed no difference when power stage is active & loaded and when it's off. That can be since ADC samples are taken at center of PWM cycles when there is 'radio silence' (no switching during ADC operation).

[hunserv]

Eariler I started developing my own Servo system, but got stuck with Assembly programming. Higher level languages are not meant for me.
I wanted to use a bit more advanced PID algorhythm than the classical one described for this product, but 32bit calculations for 8bit microcontrollers are hard, and then I gave up instead of switching to a 16bit one (my main profile is not programming)

So, this drive VSD-A almost does everything what I wanted, and I could check it out at one of my friends who ordered 3 pieces for his Sanyo-Denki servomotors.

I would be more than glad to see the (already mentioned by others) modular version of this item, because controlwise it is quite advanced, but I miss the 300V rating. Also a higher-current power-stage would be handy. In case of modular design this would be easily achievable.

For the digital part:
When I realized my PID control algorhythm I took ideas from this site:
http://bestune.50megs.com/typeABC.htm
http://bestune.50megs.com/piddesign.htm

I think the rest is just good as it is, some "cream" could be a few configureable analog outputs for torque/current/load monitoring...
As others mentioned using the servo as a DRO is a good idea, but as everyone uses different systems and have different needs it would be hard do find a "good for all" solution to output position coordinates.