About Encoders...

[Al_The_Man]

Early days, BLDC had discrete hall type devices mounted somewhere on the end of the stator or housing with a magnet or similar on the rotor.
For some time now, motors have used the hall equivalent tracks on the actual encoder, it proves a bit more accurate also.
Did you compare USdigital for price? I have not used them personally but they seem to have some competitive prices.
Al.

[Jim Blackwood]

Thanks Al, I did look at US Digital but didn't find anything that was very compatible with what I already have. The EC35 looked the closest but still at least $186 each by the time you configure it, add in the connector and cable and then I still need to rework the shaft arrangement. So it's not all that much better. It amazes me to think nobody has come up with a better solution for what is essentially a 3 pole rotary switch.

Maybe here is something... What do you think of this one?

Magnetic Encoder - Low Resolution - P9123 - Phoenix America

Jim

[mactec54]

Jim Blackwood

With such a low count, it would not be of much use for CNC operation

DMM Technology make an Encoder that may suit what you need
DMM Technology Corp.

[Jim Blackwood]

Hope I'm not replying to this multiple times...

I am thinking of this Hall sensor for the commutation model, using it with the optical encoders I already have.

Jim

[Al_The_Man]

AFAIK, There is not just a hall effect add on, probably as there are so many different physical configurations of motor.
You could conceivably design your own I guess, these would be spaced every 120° per ELECTRICAL cycle.
This link is mainly geared for RC type motors, but the principle should be the same.
Rocket Science: Hall Effect Sensor Placement for Permanent Magnet Brushless DC Motors
Al.

[mike_Kilroy]

Jim, h ow much does the drive you picked (not bought yet?) cost? There are drives like our Kollmorgen ones that work without halls - we use something called 'wake and shake' or if fancier needed, 'wake no shake' to eliminate the need of halls. but you may not like the cost of these drives....

[Jim Blackwood]

Thanks Al, I just ran across that yesterday. Pretty helpful. The single biggest thing I got from it was the following tidbit:

Commutation is the means of replacing brushes to switch the windings with electronics so that a BLDC (or any AC motor, 3 phase or otherwise) can operate in a similar manner to a brushed motor.and run at non-synchronous speeds.

Phoenix does make an add-on commutation Hall effect sensor. Their new model P9221 is not on their website yet but is the one their engineer recommended. It has 3 hall outputs and is similar to the link I provided below (which also is available in a triple hall output commutation version). 3/8" thick, 2.1" diameter, the case can be opened up for a through shaft, and they have rotor magnets with multiple poles. Price should be somewhere around $50- $75 I believe, I'll know more when I talk to their sales girl.

Mike, thanks for the help but it sounds more expensive.

Jim

[Jim Blackwood]

To finish up this thread, I just ordered the Phoenix Hall commutator and rotor (magnet). The commutator is configured for 4 or 8 pole rotors, the rotor matches it to the 4 pole motors. Rotor is .210" thick, 1" diameter, housing is .375" thick and can be opened up for thru-shaft operation to a bore size of about 1" or possibly a bit more more. They should be here in a little over a week. Cost was just over $200 for 4 of them. They do have an order minimum that could make them less than practical for most DIY applications, but for our prototype machine they thankfully did not mention that until after the order was placed.

Thanks to everyone for their help, I now have a MUCH better understanding of how the system is supposed to work.

Jim

[Al_The_Man]

Good luck with setting them up.
I imagine the same method as the Renco PDF could be used to align them, unless they have other suggestions.
Al.

[Jim Blackwood]

I plan to mount the body fixed to the end cap (inside) and attach the rotor to the spiral coupler end. Then the coupler can be rotated on the motor shaft to align it. With no index pulse on the encoder, it's rotational position shouldn't matter. Have to get the leads connected correctly but that shouldn't be too difficult. Greg has a scope and I can drive the leadscrew with a drill motor to get the signals.

Jim

[Al_The_Man]

With no index pulse on the encoder, it's rotational position shouldn't matter.

Jim

The index pulse position does not usually matter anyway, but is is nice to use with the Galil precise homing routine.
Al.

[Jim Blackwood]

Got the Hall commutators. Turns out that by acknowledging their $250/item minimum I authorized them to ship 5 ea instead of 4 ea but at least they didn't treat the magnet (rotor) as a separate item so I'm fine with that. Definitely in the realm of affordable for a retrofit. At $500 maybe not, and what would you do with 25 rotor magnets anyway?

I'll try to attach some photos. Using a 1" holesaw I cut out the back side of 4 of the housings, then using the 5th to locate the mounting holes piloting it on the encoder shaft, I used the two fixed holes and attached the housing to the back plate. I slipped the rotor on the encoder shaft and used a strong but flexible adhesive between it and the flex coupler. Because the encoder shaft is 6mm I used a .007" shim for centering.

That's really about all there was to it. I can get to the other end of the coupler to adjust the angle relative to the coils.

2 down and 2 to go.

Jim

[Jim Blackwood]

They sent the wrong rotors so I had to replace them with 4 pole rotors (2 north, 2 south). After this was done commutation went smoothly using a DS203 4 trace portable digital storage scope. But beware what you do with your grounds, I fried the scope and had to order another half way through. ($180) Hope to finish up this weekend. Used the digital inputs for the Hall sensors and the analog inputs for the motor windings, turning the motor shafts with a Milwaulkee drill at about 7-800 rpm which gave a good signal. Used the Hold button on the scope to save it and adjusted the shaft coupler to center the hall signal on the side wave. Made a chart and matched and organized the leads in sequence then when I connected them in 1,2,3 order the connection seemed right. Running the test pot both ways the motor's max speed in either direction sounded the same. I didn't check the frequency with the scope, that was where I blew it up by trying to hook the scope probe ground to one of the motor legs. Bad idea. Got the motors to run using the AMC amps and balanced them. So on to the Mach3/Galil controller. I need some help there so I will start a new thread.

Jim

[harue]

Jim, h ow much does the drive you picked (not bought yet?) cost? There are drives like our Kollmorgen ones that work without halls - we use something called 'wake and shake' or if fancier needed, 'wake no shake' to eliminate the need of halls. but you may not like the cost of these drives....

Hi Mike

How much do the motors move when shaking ?

Thanks

Harald

[mike_Kilroy]

Hi Mike

How much do the motors move when shaking ?

Thanks

Harald

depends on mode; anywhere from about 1 motor pole (360degrees/(poles) count to a degree or so:



Wake and Shake Overview

Motors which have no way to automatically identify commutation must run Wake and Shake. Motors without Halls fall into this category, as well as motors whose hall effect sensors were manually installed, and not aligned to a motor phase of 0 degrees.

The AKD will determine if Wake and Shake is needed for safe operation based on feedback type. If needed, the drive will automatically arm Wake and Shake. If Wake and Shake does not complete, a fault will be generated. When that fault is cleared, Wake and Shake will be automatically armed again.



If the user attempts to manually disarm Wake and Shake on an AKD which requires it, a fault will be thrown and Wake and Shake will automatically arm again.


If Wake and Shake is armed on a motor with a feedback type that does not normally require Wake and Shake, the algorithm will still execute. If Wake and Shake fails, a fault will be generated. After the faults are cleared, Wake and Shake will automatically arm again.



If Wake and Shake has been manually armed, it can be disarmed by clicking on the "disarm" button on the Wake and Shake screen, or by executing the WS.DISARM command over telnet.


The AKD has multiple modes of completing Wake and Shake, each with different benefits for different operating conditions.

Mode 0:

Mode 0 is the fastest of the Wake and Shake methods; it executes and completes in a few seconds, and is ideal for applications that need the fastest start-up time. The tradeoff for this fast execution time is complexity in configuration. There are a lot of parameters that must be precisely tuned for this mode to operate correctly.

For more information see Using Wake and Shake Mode 0 (WS.MODE 0)

Mode 1:

Mode 1 is a more traditional method of completing Wake and Shake. It is also referred to as Commutation Alignment and Pole Locking. Mode 1 takes a little longer to execute than Mode 0, but is easier to configure. Only WS.IMAX (current to be applied) and WS.TSTANDSTILL (time to apply current) need to be configured. The default values work for most motors.

For more information see Using Wake and Shake Mode 1 (WS.MODE 1)

Mode 2:

Mode 2 is the easiest of the Wake and Shake modes. It requires no parameters to be configured (setting WS.FREQ for sine frequency excitation is optional, and the default value of 10,000 Hz will work for most cases). This Auto-Commutation finding algorithm can take as long as 30 seconds to complete. Measurements are taken by exciting the motor with a sine excitation as the motor phase is adjusted. After a series of data points have been taken a DFT analysis is done to determine the motor phase offset for commutation.

[harue]

Hi Mike

Well, that's what i guessed, since I oculdn't any information about that at kollmorgen. So I will set it up and see what happens.

Thanks

Harald

[mike_Kilroy]

Hi Mike

Well, that's what i guessed, since I oculdn't any information about that at kollmorgen. So I will set it up and see what happens.

Thanks

Harald

If you have a Kollmorgen drive, I would be happy to send you more info about how wake and shake works with it; details are a bit different on today's drives than ones from 1985...
For instance, if you have a Servostar that uses MotionLink software, we had a routine called wake NO shake; I helped refine it! We could do the encoder alignment in just a couple encoder pulses of motion! We even had word masks to mask out some of the dozen failure modes; we routinely wns direct drive rotary tables with 50,000# weight on them - you could not see or feel any motion.

Funny you revive this TODAY; we just repaired a 4 foot diam frameless motor for a customer that uses encoder initialization and I was helping them with the routines not 20 minutes before you posted last! Scary sometimes...

[harue]

Hi Mike

Not scary, but great ! To be honest, I am quite new to digital servo drives. I've been using mostly AMC brush type amplifiers , with Precision Micro Control Cards.

For my momentary project I need drives which close the position loop internally and have a step/direction input. I just got a few Schunk MLD Linear Motors. They have 2 phase 90° 1 Vpp sin/cos encoder. ( They call it hall in their manual ),which gives absolut reference to the poles. and also 1mm pitch magnetic encoder also with 1Vpp. Schunk recommended the Indradrive HCS01 drive for it, but no idea how get them to understand step/ direction.
For kollmorgen the S300 or the AKD seems to be the best choice, because they support high frequency input for both encoder and step/dir input ( X1 und X5 for the S300 and X9 and X10 for the AKD). AMC digiflex would also be an option, but you need to get the exact version, whereas the S300 has vereything in the basic version.

The gantry, i think, would take 1mm offset for the shake moment. Or is there a parallel shaking ?

I will set up some of the drives I have on the weekend and see how far I get....