[tse]

Hello,
We are very enjoy to help you with our investigation off reverse engineer SX3 spindel control.

An other information:
If you push on the START buton, you push on the SPEED touch an obtain for exemple 1000 RPM. you touch the STOP buton, the bruch stop, after stoped, you push on the START buton again, you abtain 1010 RPM.
All push on STOP and START after, you abtain 1020, 1030 RPM etc...
Strange but this is the result...
Very good chance for your developement for spindel control for SX3.
A2P team in Belgium
Patrick Bayard - Patrick Bunet
http://www.a2p.be

[devers6]

I've been flipping back and forth all weekend over the proper approach to the modification of the daughterboard to accomplish 0-10v control over spindle speed. My first inclination was to replace the entire daughterboard with an older motor controller chip, but as I looked at it it seemed like such a waste - everything necessary for interfacing to the existing circuitry (optoisolators, smt resistors and capacitors, a known-working circuit board) all exist on the current daughterboard. All of it would have to be duplicated on a replacement daughterboard; and I hate dealing with surface-mount passive components (my eyes just don't focus on those damn things any more). As much as my skill set pushes me toward a different, pin-compatible daughterboard, I just can't shake the feeling that the only proper way of doing this - lowest cost, most minimal hardware intervention, ease of upgrade - all points to simply replacing the ATMEGA chip with one reprogrammed with firmware that adds the analog speed control without changing any of the rest of the daughterboard.

This pushes me way out of my level of experience, but it seems like the correct way to do it. I'm sure most here would agree with this conclusion; it seems to be historically the most discussed approach - the problem being no one seems to have the time to take it on. So I am going to concentrate my efforts on identifying relevant application notes that accomplish BLDC commutation with the addition of analog input for speed control. One that I have found seems to be the most applicable so far:

http://www.atmel.com/dyn/resources/p...ts/doc7827.pdf

The difference is that this application note addresses the ATTiny861. Could you Atmel experts please chime in on how this chip compares to the one used in the SX3 (it is labelled ATMEGA8-16 and is a 28 pin chip)? Is this application note suitable as a starting point for porting to the ATMEGA? I have done no microcontroller programming (I am a Windows programmer by trade), but I do know my way around a C program; and I ought to be able to modify code to match up to the pinout of the SX3 chip.

My thinking is to modify the code so the output uses the pins as-is on the 20 pin SIP header (interface to the power section), but re-purposes the lines coming from the front panel to allow for a 0-10v DC input from either a pot on the front panel, or the external speed control source. I would not propose to maintain the existing front panel matrix switches, instead going for just On/Off and Direction inputs from simple pushbutton and toggle switches. A plus would be a re-use of the existing tachometer display, but I would consider that disposable - if necessary, a cheap 2x16 LCD display could be put in it's place to give whatever feedback seems appropriate.

Can I get some help from some of you Atmel gurus out there? I would be learning as I go, and am willing to take on some of the heavy lifting as no one else seems to be able to invest the time. But I am determined to finally see this thing through - it seems so close, it just needs a final push to get this product up to its full potential.

[devers6]

Can anyone say Arduino?

From what I have been able to gather, the ATMEGA8 chip was used in early versions of the Arduino general-purpose controller. I think they have since moved on to newer chips, but a lot of the code should be somehow applicable. I found this Instructable that seems to have much useful information - I downloaded it as a PDF and have attached it here.

[devers6]

I took the code in the Instructable in the above post and stripped out what I consider to be the irrelevant bits (regenerative braking, debugging output, etc), and we're left with this:

/*
* BLDC_congroller 3.1.1
* by David Glaser
*
* The 3.x series of programs is for the ST L6234 3-Phase Motor Driver IC
*
* Runs a disk drive motor clockwise
* With regenerative braking
* Motor speed and braking is controlled by a single potentiometer
* Motor position is determined with three Hall-Effect sensors

* The Arduino receives outputs from 3 hall sensors (pins 2,3,4)
* and converts their combination to 6 different commutation steps
* PWM outputs on pins 9,10,11, at 32 kHz (corresponding to EN 1,2,3 respectively
* 3 DO on pins 5,6,7 (IN 1,2,3)
* Analog in 0 is connected to a potentiometer to change the PWM duty and change
* between motoring and regenerative braking.
*/

int HallState1; //Variables for the three hall sensors (3,2,1)
int HallState2;
int HallState3;
int HallVal = 1; //binary value of all 3 hall sensors

int mSpeed = 0; //speed level of the motor
int throttle = 0; //this variable is used with analog in to measure the position of the throttle potentiometer

void setup()
{
pinMode(2,INPUT); // Hall 1
pinMode(3,INPUT); // Hall 2
pinMode(4,INPUT); // Hall 3

// Outputs for the L6234 Motor Driver
pinMode(5,OUTPUT); // IN 1
pinMode(6,OUTPUT); // IN 2
pinMode(7,OUTPUT); // IN 3
pinMode(9,OUTPUT); // EN 1
pinMode(10,OUTPUT); // EN 2
pinMode(11,OUTPUT); // EN 3


/* Set PWM frequency on pins 9,10, and 11
// this bit of code comes from
Arduino Diecimila and the Atmega168: Changing PWM Frequency on the Arduino Diecimila and the Atmega168
*/
// Set PWM for pins 9,10 to 32 kHz
//First clear all three prescaler bits:
int prescalerVal = 0x07; //create a variable called prescalerVal and set it equal to the binary number "00000111" number "00000111" number "00000111"
TCCR1B &= ~prescalerVal; //AND the value in TCCR0B with binary number "11111000"

//Now set the appropriate prescaler bits:
int prescalerVal2 = 1; //set prescalerVal equal to binary number "00000001"
TCCR1B |= prescalerVal2; //OR the value in TCCR0B with binary number "00000001"

// Set PWM for pins 3,11 to 32 kHz (Only pin 11 is used in this program)
//First clear all three prescaler bits:
TCCR2B &= ~prescalerVal; //AND the value in TCCR0B with binary number "11111000"

//Now set the appropriate prescaler bits:

TCCR2B |= prescalerVal2; //OR the value in TCCR0B with binary number "00000001"//First clear all three prescaler bits:

}

//MAIN LOOP OF THE PRGROM
void loop()
{


throttle = analogRead(0); //value of the throttle potentiometer
mSpeed = map(throttle, 512, 1023, 0, 255); //motoring is mapped to the top half of potentiometer

HallState1 = digitalRead(2); // read input value from Hall 1
HallState2 = digitalRead(3); // read input value from Hall 2
HallState3 = digitalRead(4); // read input value from Hall 3

HallVal = (HallState1) + (2*HallState2) + (4*HallState3); //Computes the binary value of the 3 Hall sensors

// Commutation for Motoring
// Each binary number has a case that corresponds to different transistors being turned on
// Bit Math is used to change the values of the output
// For tutorial on bitmath with the Arduino: Arduino playground - BitMath
// PORTD contains the outputs for the IN pins on the L6234 driver
// that determine whether the upper or lower transistor of each phase is used
// The outputs for the EN pins are controlled by the Arduino command analogWrite, which
// sets the duty of the PWM (0 = OFF, 255 = ON or throttle value that is controlled by the potentiometer).

switch (HallVal)
{
case 3:
//PORTD = B011xxx00; // Desired Output for pins 0-7 xxx refers to the Hall inputs, which should not be changed
PORTD &= B00011111;
PORTD |= B01100000; //

analogWrite(9,mSpeed); // PWM on Phase A (High side transistor)
analogWrite(10,0); // Phase B off (duty = 0)
analogWrite(11,255); // Phase C on - duty = 100% (Low side transistor)
break;
case 1:
//PORTD = B001xxx00; // Desired Output for pins 0-7
PORTD &= B00011111; //
PORTD |= B00100000; //

analogWrite(9,mSpeed); // PWM on Phase A (High side transistor)
analogWrite(10,255); //Phase B on (Low side transistor)
analogWrite(11,0); //Phase B off (duty = 0)
break;
case 5:
//PORTD = B101xxx00; // Desired Output for pins 0-7
PORTD &= B00011111; //
PORTD |= B10100000;

analogWrite(9,0);
analogWrite(10,255);
analogWrite(11,mSpeed);
break;
case 4:
//PORTD = B100xxx00; // Desired Output for pins 0-7
PORTD &= B00011111;
PORTD |= B10000000; //

analogWrite(9,255);
analogWrite(10,0);
analogWrite(11,mSpeed);
break;
case 6:
//PORTD = B110xxx00; // Desired Output for pins 0-7
PORTD &= B00011111;
PORTD = B11000000; //

analogWrite(9,255);
analogWrite(10,mSpeed);
analogWrite(11,0);
break;
case 2:
//PORTD = B010xxx00; // Desired Output for pins 0-7
PORTD &= B00011111;
PORTD |= B01000000; //

analogWrite(9,0);
analogWrite(10,mSpeed);
analogWrite(11,255);
break;
}
}


Keep in mind that I haven't changed the port and pin variables, and I haven't modified the code to change the mapping of the analog input to eliminate the regenerative braking; so this code would not work as-is. It should be considered just pseudo-code for the time being. But the basis for commutation and speed control is here. First impression is that this is insanely easy; but I have worked in software long enough to know that things that seem insanely easy rarely are, so I would appreciate any input from those of you in the know about power and motor control as to what safeguards to include.

I don't own an Arduino, but I do own a Netduino (Netduino :: Netduino : Overview). It seems that I should be able to remove the ATMega chip, connect the appropriate ports from the Netduino to the ATMega socket, and be able to drive the SX3 motor. I don't have the balls to do that right off the bat - I will probably build up a small power stage to control a small BLDC motor and get the code working perfectly first. I have seen enough H-bridges smoke because the two sides of the bridge were enabled simultaneously, and I don't want to do that with the 300+ volt multi-amp driver circuits on the SX3 power board until I am absolutely sure that the software is right.

Once the code is tested, it should be possible to download it to an ATMEGA8-16 (which seem to be discontinued and perhaps hard to get), so the only thing necessary would be to remove the existing ATMEGA, plug in a new one, and wire up the replacement front panel and external circuitry using the existing front panel cable. That should make the upgrade about as painless as possible.

[devers6]

Well, I had high hopes for this weekend. I spent all week tracing out connections and creating a simplified schematic of the daughterboard; I bought an Arduino Uno and some blank ATMEGA8 chips so I could try to create a replacement chip for the daughterboard; I worked out the pinout so I could plug in the Arduino in place of the ATMEGA chip and work as an emulator while I revised the code; and I bought a little 12v BLDC motor that I could use as a proxy for the big motor while I tested the code out, hoping that I could run just the power section or the drive on 12 volts and minimize the risk of blowing something up.

Things got off to a very bad start. I dug out my little Owon digital LCD scope so I could check out some waveforms on the daughterboard, but I decided to take a look at the main DC supply. I set it for 50v per division and hooked up the probe across R505 and R504 on sheet 1 of the ATT00024 schematic posted previously, and turned on the power. I got a healthy snap and a lot of smoke poured out the the scope. Turns out this scope has a maximum input of 300V, and the main supply measures out at 340v. Guess I shouldn't have done that.

I pulled out my old, old Tektronix scope and was a lot more careful about what I looked at. What I was really looking for was how best to isolate the daughterboard and drive stage from the high voltage DC. I wanted to leave the driver electronics unpowered from the AC line, find an appropriate place where I could inject 12 volts, and try to drive my little Maxon BLDC motor using the daughterboard and my Arduino to control the speed. What stymied me is that the low voltage DC rails (+5V, +5VA, and VBB from Sheet 2) are driven by a separate small transformer, and I couldn't find a convenient way to power that circuit while leaving the main DC voltage unpowered.

In the process of doing this, I measured the VBB voltage at around 14 volts on my multimeter. It looked like it was just 1/2 wave rectified, so I wanted to see what the waveform looked like. I still had my multimeter probes connected, so I went to attach my scope probe ground to the probe I had attached to GNDF. Big spark, and the mill went dead. The onboard fuse had blown, so I replaced it and tried turning it on again. Fuse blew immediately. So now I have a dead driver board and who knows how much else; and my forward progress is now going backward.

At this point, I still don't really know what I did wrong. The last time I had something spark at me when I connected a scope ground, turned out to be a miswired outlet that had hot and ground reversed. Can't be the case here, because both the mill and the scope were powered by an extension cord with a single duplex outlet wired correctly as far as I can tell.

I guess I should be happy that I didn't short something out with a body part - I have enough respect for power electronics that I was very careful with everything I did; usually making my connections with power off and energizing the circuit standing well away. I even found that the 340V DC didn't bleed off when power was turned off - it would take 20 or 30 minutes for the capacitors to discharge.

But I am dead in the water now. I suspect that one or more of the IGBT's are shorted, or perhaps the IR2103 drivers feeding them. I can't rule out further damage on the daughterboard or the rest of the driver board as well. I'm taking a break from it right now to reflect on things and consider my options. Grizzly wants $250 for a new driver board - any of you guys who replaced your motor have one you want to sell?

[ddexd]

dever6,

The attached link talks about isolation transformers and their use with oscilloscopes.

<http://www.toddfun.com/2011/04/30/isolation_transformers/>

Also, if you have a dual input oscilloscope with invert and add capability, you could invert one of the inputs and add it to the first to get a differential reading. In this mode you do NOT connect the grounds.

[devers6]

Thanks for that. It explains exactly how I managed to blow up $500 worth of gear. I learned much of this 35 years ago, but have actively avoided it for the past 30 years by working strictly in digital electronics where ground is ground and voltages seldom exceed +/- 15 volts, and managed to avoid it entirely for the past 15 years by working exclusively in software. Sometimes you get reminded of what you don't know (or have forgotten) the very hard way.

Upon re-examining the schematics, what should have been my tip-off was the furious use of optoisolation in all areas of the digital to power interface. Over the years, I have become accustomed to the use of optoisolation as a way to protect digital circuitry from accidental exposure to high voltages and transients - what I forgot was that it is also used in the case of intentional connections. Power electronics is rife with points that are considered "common" because other points go positive and negative in relation to them, but "common" and "ground" are not necessarily the same thing.

So for any of you looking at these schematics, make sure you treat the signals labeled "GND F" and "GND PWR" as points not necessarily at ground potential.

With this in mind, I could use some help from those of you who know this stuff better than I do. What I would like to do is to devise a way to power the driver stages with low voltage DC (like 12 volts) so that I can safely drive a small motor through the same interface circuitry as the high voltage BLDC motor while I am working out the software. My initial idea was to apply the 12 volts across the two resistors R505 and R504 on sheet 1 of the schematic, and also to the anodes of diodes D207, D208, D209 and D210 to power +5V, +5VA, VBB and the Power Good signal going to pin 14 of C302 of the daughterboard. Does this seem reasonable? This way I can avoid all of the high voltage issues during the time of development when problems can be expected. By using a low current supply (like a wall wart) I can probably drive this little motor:

http://test.maxonmotor.com/docsx/Dow...f/05_170_e.pdf

which has all the hall effect sensors to replicate the big motor but wouldn't draw enough current to hurt anything. Seems like this should work, but there may be gotchas that I haven't considered; and I would appreciate any input you guys may have.

By the way, the driver board doesn't seem to have any obvious serious damage (no burnt traces, no smell). I found a pair of IGBT's shorted, but the other 4 seem OK. I have some replacements coming in from Mouser, and I have ordered some of the IR2103's from China in case one or more was blown. So the driver board may be repairable, I hope.

[ddexd]

I've always looked at my memory as having a 5 year half life. So whatever I have learned and haven't used actively, I will loose half of that knowledge in 5 years. In the next 5 years I would loose half of what wasn't lost in the first 5 years and so on.

So in 35 years I could only remember <1% of what I originally started with, if I hadn't actively used that knowledge in that time, as it were.

Hopefully you get your board repaired and up and running. I'm following your thread with high interest as I have the daughter board controller on my SX3. I would love to have a solution to getting the spindle under CNC control.

[devers6]

I got the IGBT's in last night. They were about $3.00 each from Mouser. I replaced a pair of them that measured shorted between all legs. Now the driver does not blow the fuse anymore, and the front panel beeps like it used to on startup. However, pressing the Start button on the front panel does not start the motor; instead the word "Err" shows on the tach display. So apparently there is more fixing to do on the board. I carefully examined all the surface mount circuitry around the shorted IGBT's, and I didn't see any evidence of burned components or traces; so the most likely things still wrong are 1 or more of the 1/2 bridge driver chips (IR2103). I have some on order, but they are coming from China so I'm going to have to wait a bit for them (most electronics stuff I get directly from China takes about 3-4 weeks). I would have liked to get the IR2103's from Mouser or DigiKey, but they both show them as non-stocked items with long lead times. I may look around a little more to see if there is a domestic supplier anywhere in the US that might get them to me a little sooner. They weren't expensive - $2.20 or so each - so I could afford to buy a couple to use while I wait for the China order.

I'm still looking for advice on my strategy of powering the driver board from low voltage DC during the development of the ATMEGA software. I guess my biggest unknown is if the power circuits will work at that low of a voltage (I am considering 12 volts) since they are optimized for a 300+VDC environment. The spec sheet on the IGBT's (G60N100) seems to indicate the device will work at low voltage, but I've never used them before and I don't know the key specs to consider. So any of you who deal with power electronics could really be of help here.

One other area I could use some help on is the CTRL1 and CTRL2 circuits on page 3 of the ATT00024.pdf schematics. It looks to me as if CTRL1 is a comparator circuit that compares an input voltage developed from the R414 shunt against a variable voltage set by R703, and then sends a signal to the ATMEGA to indicate overcurrent and perhaps invoke a shutdown. That should be easy enough to accomodate, but I'm not really sure what the CTRL2 signal is attempting to do. Could any of you speculate on what it might be for?

I'm quite anxious to get back in this, because I feel I am really close to being able to start experimenting with replacement code in the ATMEGA chip. I wish I hadn't blown it up, but at least it looks like repair is a possibility - it just puts me on hold for several days.

[ddexd]

The way U601A is configured it looks like an integrator. Could they be using this to create a time delay on power up?

[devers6]

Could very well be. I sort of thought that function was handled by the circuitry at the bottom of sheet 2, which presents a signal on pin 14 of the daughterboard when logic voltage is present and switch S1 is enabled (I don't exactly know what switch that is). But there is no kind of delay on that signal, and it might make sense that the processor would want to wait for a period of time before continuing. I guess it would be easy enough to probe the output signal and monitor it during startup to see what it does. It seems to be a steady state signal once it is asserted - its only input is the voltage-divided +5VF that I don't think ever would go away while the mill is powered.

Thanks for the feedback.

[devers6]

What I see on this pin is a short positive pulse a second or two after startup, that roughly corresponds with the beep you hear on the front panel. So perhaps it is a signal that says the various +5v signals are stable. Doesn't seem terribly important in the overall scheme of things, since motor commutation does not (and should not) start on powerup, but rather when commanded from the front panel. I believe that I will provide for both of these signals in the software, but ignore them for the time being until I can decide how best to integrate them with the rest of the motor control scheme.

[neilw20]

Circuits for SX3. Read carefully. A few smokey gotchas in there.:wee:
http://www.cnczone.com/forums/bencht...tml#post958939

[devers6]

I spent some time yesterday trying to see if I could power the driver board with low voltage DC - no luck. First of all, the connections to the low voltage supplies 5v, 5va, etc can't be made as I originally indicated (to the anode of the 4 diodes. It has to be made to the cathode side, or else the input voltage will be shorted to ground through the transformer). But applying +12v correctly still did not allow the daughter board to power up as it does when running off line voltage. I suspect it is Vbb that is critical to the startup sequence as it is used to derive +5vf which is used for the overcurrent and startup delay circuits on page 3. +12 volts doesn't seem to work - I have no idea what Vbb should be (remember it was my trying to measure Vbb that got me into trouble in the first place) and I can't quite get myself to stick a probe anywhere with full AC power applied. I think my next step is to get a variac and dial down the input voltage, and use NeilW's suggestion of putting a light in series to limit current.

In the meantime, I needed to get some milling work done; so I have put the AMC driver back in and tried to get the G540 speed control working. I now have basic speed control, although at reduced RPM's. I think this will work for me for now, until I can fully repair my driver board.

I'll try to put up some video of the speed control once I get everything buttoned up.

[cornbinder23]

Did you get the IR2103's?

JTCUSTOMS

[devers6]

Yes, I did get the IR2103's - unfortunately, swapping them out made no difference. They were about the last obvious things to change - everything else seems to be surface mount chips and passives; and before I look into changing them I need to probe around to see if I find something that doesn't look right. And I'm pretty gunshy about probing around these days.

I do appreciate you letting me "borrow" these chips to aid in the troubleshooting - I was really hoping they would be the final fix. My China chips are due in soon - I figured I would replace yours with the new chips on the off chance that something on my board might have damaged yours - I've got no real way of testing them, so I thought I would take the safe way out and replace yours with the brand-new ones.

Thanks again for the prompt help - too bad it didn't work out.

Dave

[neilw20]

Trial and error replacement only works if you are lucky enough to hit on the right component(s)
Initially when these things blow up they can take many other parts with them.
You need to do some detective work to find the bad bits.

Output devices blown?
Replace all six, or at least pairs of devices. If one went short, then the other one in the leg supplied the excessive current, and can't be trusted.
The zener diodes across the gates of the IGBTs often get destroyed. That becomes their protective function at smoke time. :banana:

U201 switchmode supply is the first section to get working properly, and it must be run from mains voltage.
This is a common failure after a blowup - usually caused by a rogue wire to chassis that started the whole episode.

Doing any of this testing without an isolating transformer is asking for trouble.
The exception to this is tacking on a few leds and resistors then switching on to see if they light.
The slightest slip with a meter probe can cause smokey (or more) problems due to the vast potential differences on adjacent areas of the PCB.
Really dumb design is well proven as evidence by the difficulty doing repairs.

Once you've socketed the drivers it is all a bit easier.
The drivers are fairly easy to test on a low voltage supply.

1. Does the relay come on. Then see if your leds work below.

If the EStop switch is used on a regular basis the relay contacts eventually fail. Every tiome you reset it the current surge kills them.
Only use in an emergency. Power off. Reset the EStop, then the relay will survive.

2. The low voltage section ,must be fully operational. Put a led and 470 ohm resistor hall sensor supply pins 1 and 2 of CN101.
If this does not light up then the switchmode supply needs attention.

3. The low voltage section ,must be fully operational. Put a led and 470 ohm resistor at the output of each regulator.sensor supply.
EACH of these supplies are isolated from one another.

4. Once the supply is working, verify that the hall sensor signals are OK. Did you check them?

5. Hall sensors OK.

6. Now look to get the output sections working.

[devers6]

Thanks for all these tips, Neil.

I like the idea of tacking on an LED and then powering up to see state - I will do that. I think I may also tack on a little wire loop in strategic places, attach my multimeter with the test clip, and then power up to take a measurement. It is just too scary to be probing around on that circuit board. I will have to round up an isolation transformer as well.

I had to replace one pair of IGBT's - both measured dead shorts across all terminals. With the new ones in, the drive no longer blows the fuse, so if the other IGBT's are bad, they are open. I have replacements for all 6 - I may just replace them all just to be sure.

The halls are OK because the motor works fine on the AMC drive, just runs 1/2 speed because it is running off rectified 120v instead of the voltage-doubled 120v on the Grizzly driver board.

[rwskinner]

Hows it working out for you? It did okay for me but I started having issues and the motor kept stalling. I bumped up the a 1 hp Baldor from Ebay. I haven't built the mounts yet or bored the pulley but I will soon.

Richard

rwskinner:

Thanks! I just ordered one.

bbox

[Com_VC]

Hi devers6,

Out of interest did you ever end up repairing your SX3 PCB, if so what did you find wrong with it?

Thanks