[Arquibaldo]

Dear Russel & Michael:

Thanks a lot for your replies to my questions - now I know how to
handle the E543 & its encoder. I am waiting for the driver parameters
when your machine is in its new home.

Let me now start building the "Elm-Chan servo driver" and this leads
to a few other questions:

Russel: Would it be too much to ask for the schematics of the latest
versions of the two boards (the processor board and the H-bridge board)
you are now using (the ones pictured in post #353 on page 30 of this thread)?

The reason I ask is that there are so many slightly different versions of
the "Elm-Chan servo driver" that I am confused, and I would like to start
precisely with your latest version, which is known to work with the E543's.

I can program AVR's and draw PCB's, so I have analysed very carefully
you pictures, and I think I understand most of what you did, but not all:

- are you using IRF540 MOSFET's and BYV28 supression diodes, like most people?
some seem to prefer IRPF240's, which have better thermal characteristics
but cost twice as much.

- what is the DIP4 IC in the H-bridge board? an optocoupler?

- I see you have 2 DIP4 IC's in the processor board, probably PC817 optocouplers
for the step & direction signals from the PC, but what is the DIP8 IC in
this board?

I've ordered the IR2184's (the DIP8 version - SMD is too much for my 67-yo-
cataract-operated-eyes) and intend to base my PCB on the design on thread
56395, posts #26 & #28, which I like because it allows a very compact and
efficient mounting of the heatsink, parallel to the board. I think a standard
PC processor cooler, with heatsink & fan, would be ideal.

However, I intend to make a few modifications, which I describe for your
comments (may be this should be on a different thread, but...)

- the encoder cable will use RS422 TX & RX, to minimize noise problems:
there will be a small board, attached to the servo, with the RS422 TX.
I will use RJ45 connectors and a standard CAT5 ethernet cable between
the encoder & the driver board.

- I will use the MAX232 RS232 driver, which I trust more than the 74HC14
hack used in Chan's original implementation.

- like you, I will add optical insulation for the step and direction signals,
using either PC817's or the HCPL2530/31 which some people seem to prefer.

- I will remove the 48V, 12V & 5V rectifiers, filters and regulators from
the board. I think these belong on a separate power board, supplying
the 3 or 4 drivers. Also, this allows the implementations of a power
sequencing scheme, which is the object of my final ( for this post...)
question:

There are at least 5 different supplies that must be considered when
powering up or down the system:

- PC power, waiting for the execution of the control program (EMC2) and
the proṕer levels on the parport & breakout board
- 5V for the encoders
- 5V for the microcontroller (AVR)
- 12V for the IR2184's
- 48V for the servos

It seems obvious that this is the logical sequence when powering up the
system (which would obviously be reversed when powering down).
I would implement this on the PC itself, which, when EMC2 is ready, would
sequence a series of relays with the other supplies. But this is kind of
expensive, and may be overkill. Some simplifications might be acceptable.

What are your thoughts about the power sequencing problems? What are you
using in your systems?

I am also worried about excess heat: ideally, we should have
thermistors attached to the MOSFET's and servo motors, monitoring their
temperature and shutting down the system if certain limits are reached.
But this is getting too complex and expensive. Is anybody monitoring
the temperature in their systems?

Nelson

[epineh]

Hi Nelson, I only have the board layout, as I was using Eagle at the time I found it easier to just do the layout and not draw the schematic first. Also I am away from the PC with the layout, I can probably post it tomorrow.

The schematic is as per the original Elm Chan high power version, I am using IRFZ44 FET's (50Volt, 50 Amp) and IR2184 gate drivers, I could only get the SMD package, I would prefer the DIL package but that is all I can get here. The FET's are only warm to touch under full use and I don't use any external bypass diodes, there has been much discussion on this forum of this and some say to use them and some say to just use the internal parasitic diode of the MOSFET, which is what I did. My heatsink is just a 10mm chunk of aluminium bolted to the 4 FET's (with insulators) I think it is in a pic somewhere, if not let me know and I can take one.

The FET's I use are about 60c Australian each so they are quite cheap, I blew a few at the start before I knew what dv/dt induced turn on is, lets just say don't use a battery bank for your motor supply while testing

The 8 pin DIL on the H Bridge was going to be an opto (2630) on the PWM signals from the processor board, but I just bypassed it as I think it is overkill, after all I only want to protect the PC from damage, protecting a $1.70 micro with a $3.00 opto didn't make sense to me...and the opto on the step/dir has the PC covered.

You are right, the 4 pin DIP's are opto's for the enable and fault (pc817's), I put them on there but did not connect them through to the PC, I got lazy...

The 8 pin is a HCPL 2630 I used for the step/dir input to the drive, it is just faster than the 817's

A MAX 232 should work for the serial, I am using the 74hc14 only as I am cheap, it is a fraction of the cost of the MAX chip. I haven't had any problems with noise on the serial link. Another good idea I have seen others do is to put the MAX232 or 74HC14 on the connecting cable and use a header on the servo board, saves board real estate and parts.

For power, I bring in +5V from the PC as I am switching the p.port active low, as this has higher current capacity than active high. This is all contained in the corner of the PCB including the opto's so it is the only section the PC "sees".

I have a seperate logic power supply, with its own transformer that outputs 3.3, 5 and 12 volts, the 3.3 was for FPGA stuff just in case, I use the 5V for all drive logic, encoders, micro etc. The 12V is for the 2184's as you mention.

I don't like using switchmode power supplies for logic though I know a lot of people do, it is just so easy to rig a transformer, some caps, rectifier and a few 7805's 7812's ... whatever.

My startup sequence is to start the PC and EMC, then all my servo power supplies run off one switch, so I just flick the switch and the lot comes on, drives startup and lock the servo's. I like to keep it as simple as possible.

I used to startup almost exactly as you had mentioned, but found once everything is sorted there really is no need.

As for MOSFET heat, I check mine when I change tuning parameters and during/after heavy usage, I have a special piece of equipment for this...my finger

In my opinion, once the design is working and no more changes are happening things will stay much the same, so if the heatsinks are only just a little warm to touch at any stage then things are good. I think if you are building a high voltage/power version then things will be a lot more critical, but keeping voltages and current down is a lot more forgiving.

Besides my servo's have huge amounts of power compared to any steppers I have used and will go faster than the router and PC can handle with heaps of torque, gotta be happy with that ! No midband resonance, missed steps or loss of torque at speed, just sweet motion

Phew..hope this answers your questions.

Cheers.

Russell.

[Arquibaldo]

Russel:

" Phew..hope this answers your questions"

It certainly does... Thanks A LOT for your attention and kindness.

I will give you a few weeks break, while I build the electronics, before
bothering you with my next round of questions, which will be about the
mechanical aspects of the red machine.

I hope this break will replenish your amazing patience with me.

Cheers

Nelson

[epineh]

Hi Nelson, no probs, feel free to ask anytime, I don't get out much these days anyway

Things are a little busy right now but I will get that PCB layout and parameter settings for you soon. The layout is on about three different PC's with different versions, I have to double check to make sure I give you the current one...

Cheers.

Russell.

[epineh]

OK here we go, now first of all, I don't classify my design's as particularly good, these were done a while ago and looking over them I will do the layout different next time.

I am posting them for viewing only, if you choose to make this layout exactly good luck, you may need it

I don't label my boards all that well but you should be able to work out what I have done, I used Eagle vers 4.16r2 I don't know if the later version opens these files as I don't use Eagle any more.

Cheers.

Russell.

[epineh]

Hmmm... went to the shed today thinking I would just cut out a few small jobs that I wanted to do...looks like the controller PC HDD has crashed (chair)

I managed to boot a live CD and get the configuration files off the old HDD before the whole thing went completely south, so next is to replace it, do a new install and try the config file...fingers crossed

Russell.

[Khalid]

Russel i just visited here to admire your hardwork, indeginious problem solving approach, beautiful mind and hardship

[contactirfu]

And I have to thank you for showing me how to use the opto's with the 24V for limits, estops etc - though it is common knowledge - I only learned it from you.

My best to You Russell -

so whats new you are cutting ? sorry that the HDD gave up - its a 2 to 3 hr pain to set all up

do ya use a windows live cd?

RGDS
IRfan

[epineh]

Hi Irfan, thanks for the kind words

I am using EMC so I used a live install disk of Ubuntu to save my machine's config files from the old HDD, I will download the latest version of the "Brain Dead Install" of Ubuntu with EMC ready to go... so with a bit of luck all I have to do is re-install and transfer my old config settings across...then I can sleep

Cheers.

Russell.

[epineh]

Oops, missed your other question, I was doing some inlay work for a friend, nothing too complex.

Russell.

[epineh]

I managed to get Linux installed after many attempts, not an OS problem but I had to search around for a Hard Disk that didn't sound like it had been run over by a truck (chair)

I was a bit concerned that with the failure of the previous disk the config files I saved from it may have been corrupt but all is well, got the files where they needed to be (took longer due to my Linux noobishness) , started EMC, then fired up the drives...success !!!

The whole process took about 15 minutes, including Linux install, not bad for a total hard disk crash recovery

I guess the moral to the story is to backup your machine configuration, I only got away with it so easily as the hard disk let me access the config files, if it was a total crash the process would have taken longer.

I am probably like most people, think to backup AFTER a crash... at least this time I was lucky !!!

Russell.

[crocky]

Sounds like fun, not

Hope you are handling the all water well, that is one place I would not want to be during the wet.

[epineh]

Hi Bob, yeah not my idea of a good time, lol. Unfortunately when I re-installed I used the latest version of EMC and now I can't enter anything except 0.0 for my G54 offsets, where I could before, not the end of the world but I may have to actually read the manual...not one of my strongest points

As for the water, where we live is the pretty much the wettest town in Australia, so we have really good drainage ... we don't call it rain unless it records over 100mm in one shower

Russell.

[epineh]

I was playing around with the servo tuning tonight, and after I was finished I ran a small bit of code to test things out, without the spindle going...only problem was I hadn't zeroed the Z axis and the machine drove the brand new V cutter straight into the table, then proceeded to drag it across...(chair)

So one broken cutter and one large gouge in the table later I have learned to check these things before running any code, guess I won't forget that in a hurry, the cutter wasn't cheap and had only been used a couple of times... ah well

Russell.

[chich2]

Nooooooooooooooooooooooooooooooooooooooooooooooooo oo!

Chich

[BobF]

Arrrrrgh ;-(

[chicof]

Hi Russel,

Decided to make myself "visible" to the community today, here at your beautiful thread.

I am working on the surfboard scanner as I told you through PM and today the Digikey encoders arrived, a friend sent them to me from Tweed Heads.

Straight away I went to the bench and set everything up. Encoder, power supply, 5VDC, Paralel port cable and mach3 up and running, set the pinnage at ports and pins, input pins 10 for channel A and 11 for channel B, both at portA.

But... No Game.

The encoder1 DRO does shows action but it's very inconsistent, to the point where you can't tell if it's going CCW or CW. It's not a smooth count like the one that you a video at EMC2.


for cables I am using those HDD LED and power button cables as they have a right connectors for the encoders.

Later I also tried using a shielded 6 core, but no difference.


Any bit of help appreciated

chico

[epineh]

Hi Chico, welcome to the zone ! (officially )

The closed loop test I did was running EMC with a Pluto-P FPGA based card, I don't actually run Mach but there are lots of people here that do, and I have seen a couple looking at your post so I would imagine somebody will chip in with Mach related help...

I used the same cables for my encoders, I ran 4 core shielded cable up to the encoder then soldered and heatshrunk this to the PC headers, very handy (and practically free ) I don't get any noise issues with this setup even with unshielded motor cabling that generates enough noise to take out my shed radio...

I have not done a direct encoder connection to the parallel port but I will try help as best I can.

The parallel port is a very low current device, I would consider putting resistor's inline with the encoder output, say 1000 ohm for starters, and see how that goes. Also you say 5VDC power supply, where is this coming from ? Without a common ground of some sort to the PC the encoder output voltage will have little meaning to the parallel port input pin.

If you are using 5VDC from the PC itself then this shouldn't be a problem.

My machine isn't closed loop to the controller PC (yet) I have standard step/dir output from EMC and the servo drives take care of the rest, encoder reading, position capture, PID loop etc.

What are the mechanical's of what you have in mind, do you plan to manually move something and take readings from the DRO ? I need a little more info on what you are trying to achieve so we can get it sorted.

Cheers.

Russell.

[epineh]

I just looked again and see where you get the 5VDC from, though I would consider still grounding one of the pins 18-25 on the parallel port for the voltage reference.

Russell.

[epineh]

I also checked your original PM, you did mention using encoders to make a 3D input device for Mach, sorry for not remembering ... in my defence I get a few PM's and I have a very bad memory

Russell.