Re: [Emc-users] Resolvers and 2 phase AC servos

[gene heskett]

On 1/22/23 04:54, Robin Szemeti via Emc-users wrote:

Yep, I have used steppers, servos, optical encoders with EMC etc etc in the
past, however, on this particular device, I am not going to try and hack
the mechanicals apart to start changing stuff. It's a high quality bit of
military kit that has presumably worked well enough in its previous life,
it just needs some suitable control system putting on it.  I've been
running stuff on EMC for decades as you probably know, so yeah, I know what
it can do. As it stands, it is fully waterproof, with all stainless or
aluminium parts (yes, I am aware of the issues mixing stainless and
aluminium) and the highest quality waterproof Amphenol connectors, I hope
to be able to leave it as it is.

A 2 pole resolver will typically be good to 5 minutes of arc, these on here
are some of the nicest I have seen, the military are not known for cheaping
out on their kit!  I fully expect them to acheive that level off accuracy
with the Mesa board.



Touche`. That is about the accuracy I'm getting with a clone BS-1 I 
CNC'd a year ago. Worm in the encoded gate motor into worm drive in the 
BS-1.  Too slow for active machining unless EDM'ing something. Great for 
holding tools at add angles to sharpen them with a CBN wheel. Also too 
heavy and too tall for my G0704 to carry comfortably.  Should have 
started with a BS-0.


Cheers, Gene Heskett.
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author, 1940)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Robin Szemeti via Emc-users]

Yep, I'm planning to use the Mesa resolver boards, simply because I already
have a fair bit of Mesa kit.   I expect resolution to be 14 bits and
accuracy to be 12 bits.

On Sun, 22 Jan 2023 at 16:52, Jon Elson  wrote:

> On 1/21/23 19:34, Robin Szemeti via Emc-users wrote:
> > One thing that might cause some ... well, "issues" is my resolvers are
> used
> > to determine the absolute angular position of my two rotational axes, ie
> > they are driven off the axes themselves after the gearing, not off the
> > motors before the gearing. This may prove to be a bit "interesting" ..
> but
> > since my desired movement is very slow, fractions of a degree per
> minute, I
> > suspect it is not going to be a problem and it can all be handled in a
> slow
> > task loop.
> >
> The resolver boards I make deliver 4096 quadrature counts
> for one full rev, and they do not dither when standing still.
>
> So, that is better than 1/10th of a degree.  Some other
> resolver converters can do even finer resolution.
>
> Jon
>
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Nicklas SB Karlsson]

sön 2023-01-22 klockan 10:51 -0600 skrev Jon Elson:
> On 1/21/23 19:34, Robin Szemeti via Emc-users wrote:
> > One thing that might cause some ... well, "issues" is my resolvers
> > are used
> > to determine the absolute angular position of my two rotational
> > axes, ie
> > they are driven off the axes themselves after the gearing, not off
> > the
> > motors before the gearing. This may prove to be a bit "interesting"
> > .. but
> > since my desired movement is very slow, fractions of a degree per
> > minute, I
> > suspect it is not going to be a problem and it can all be handled
> > in a slow
> > task loop.
> > 
> The resolver boards I make deliver 4096 quadrature counts 
> for one full rev, and they do not dither when standing still.
> 
> So, that is better than 1/10th of a degree.  Some other 
> resolver converters can do even finer resolution.
Up to 16 bit resolution though there is a choice between resolution and
response time.

Nicklas Karlsson



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Jon Elson]

On 1/21/23 19:34, Robin Szemeti via Emc-users wrote:

One thing that might cause some ... well, "issues" is my resolvers are used
to determine the absolute angular position of my two rotational axes, ie
they are driven off the axes themselves after the gearing, not off the
motors before the gearing. This may prove to be a bit "interesting" .. but
since my desired movement is very slow, fractions of a degree per minute, I
suspect it is not going to be a problem and it can all be handled in a slow
task loop.

The resolver boards I make deliver 4096 quadrature counts 
for one full rev, and they do not dither when standing still.


So, that is better than 1/10th of a degree.  Some other 
resolver converters can do even finer resolution.


Jon



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Robin Szemeti via Emc-users]

Hi Chris,

An interesting device, but you are making a fundamental mistake in
confusing resolution with accuracy.  If you read the datasheet, they make
no claims whatsoever about accuracy, if it was good, I am sure they would
mention it. That said, I would reasonably expect them to achieve +- 5
degrees of accuracy,  maybe better, and yes, I appreciate they have a much
higher resolution.

A good 2 pole resolver will achieve 5' of accuracy (that's 1/12th of a
degree), resolution will be limited to the resolution of the ADC's reading
the signals. Being an AC device it is relatively immune to the earths
magnetic field. The devices I have are fully waterproof high quality
resolvers costing probably thousands of dollars each ... I think I'll stick
with them.

On Sun, 22 Jan 2023 at 01:35, Chris Albertson 
wrote:

> You can simplify this if you replace the resolvers with magnetic encoders.
> They cost $4 each and come  mounted to PCBs and with a magnet.   What you
> do is super-glue the magnet to the shaft end and then face the chip at the
> magnet with an about 1mm air gap.  The chip tells you the shaft angle to
> within 1/4000 of a revolution.It is easy because it is an absolute
> encoder so you don't have to track the shaft you can query the angle
> whenever you need it.  There is both an analog and digital output
>
> It is a 4mm diameter magnet and you mount it by first making a small hole
> in the shaft with a drill if the shaft is not already center drilled.  Then
> the magnet is very powerful and you really don't need glue.  They work
> better if you get larger 6mm dia. magnets but still I get reading good to
> about 0.1 degree.
>
> With resolvers and quadrature encoders you have to track the encoders
> continuously but these are "query on demand."   With more money, for $10
> you get a better sensor that does 16,000 "ticks" per revolution, absolute.
> These are cheap enough to place them on the motor and the driven structure
> https://www.amazon.com/HiLetgo-Magnetic-Encoder-Measurement-Precision/
> <
> https://www.amazon.com/HiLetgo-Magnetic-Encoder-Measurement-Precision/dp/B09KGWC1PT/ref=sr_1_2_sspa?keywords=as5600&qid=1674349813&sr=8-2-spons&psc=1&spLa=ZW5jcnlwdGVkUXVhbGlmaWVyPUEzOThLWTdPUkkyWlpCJmVuY3J5cHRlZElkPUEwMjg2NTczMjVSOVNEOU1aTDFEUCZlbmNyeXB0ZWRBZElkPUEwMzE1MDEwMk5UUjVRNlJINkhRQiZ3aWRnZXROYW1lPXNwX2F0ZiZhY3Rpb249Y2xpY2tSZWRpcmVjdCZkb05vdExvZ0NsaWNrPXRydWU=
> >
>
> On Sat, Jan 21, 2023 at 4:59 PM gene heskett  wrote:
>
> > On 1/21/23 06:25, Robin Szemeti via Emc-users wrote:
> > > So,  in a vaguely CNC related folly I have purchased a 2 axis military
> > dish
> > > mount for radio stuff (moonbounce, if you must know)
> > >
> > > My plan is to control it from EMC, with a Mesa anything io card and the
> > > resolver interface, I already have various Mesa cards under EMC, so
> thats
> > > not an issue and I have a spare 5i23 PCI card and the isolated IO
> > duaghter
> > > card, so with the resolver card I should be able to uses the resolvers,
> > and
> > > pick up the limit switches on the IO card. I can just use some Python
> > > script to work out the moon position and generate "G1 X285.4, Y23.8" or
> > > whatever MDI motion commands every 15 seconds or so to track the moon,
> > not
> > > a problem.
> > >
> > > Now, the question is driving the two phase AC servos.   Generating vast
> > > quantities of 50 or 400Hz or whatever AC for the drives is simple
> > enough, I
> > > can just use a couple of large Class D amplifiers, no worries.
> Generating
> > > the AC low level signals to drive these  basically, it needs a
> > constant
> > > 50Hz signal on one coil of a fixed level, and a variable level sine
> wave
> > on
> > > the other coil, at either +90 or -90 degrees depending on which way it
> > > needs to go.
> > >
> > > Is it possible to use HAL or something to generate the variable level
> > 50Hz
> > > servo drive signals and output them from the PWM outputs on the
> resolver
> > > card?
> > >
> > I've not used it in that context, but take a look at man 9 siggen.
> > There are several mux's available that could be used to switch motor
> > directions. Its main use is in 1 kilohertz loop stuff for linuxcnc, but
> > if you need finer time resolution, should be happy in a 3 to 5 kilohertz
> > threads Use a halmeter to see how much time the individual module takes.
> >
> > Hal can do a lot, look the logic choices over in /usr/share/man/man9 of
> > an linuxcnc install and let your imagination out to play w/o a
> > chaperone.  You can use float vars at a 1 or even 4 kilohertz loop
> > thread, so updating a motor drive at 50HZ ought to be doable.
> >
> > For motor amps, take a look at class D audio amps, the 400 watt modules
> > are rather commonplace these days, running at ultrasonic carrier
> > frequencies. Use chokes to keep the carrier out of the motor wires, and
> > good single point(one bolt) grounds.
> >
> > Use mux's for analog switches.
> >
> > Using modern stepper drives that have curre

Re: [Emc-users] Resolvers and 2 phase AC servos

[Robin Szemeti via Emc-users]

Yep, I have used steppers, servos, optical encoders with EMC etc etc in the
past, however, on this particular device, I am not going to try and hack
the mechanicals apart to start changing stuff. It's a high quality bit of
military kit that has presumably worked well enough in its previous life,
it just needs some suitable control system putting on it.  I've been
running stuff on EMC for decades as you probably know, so yeah, I know what
it can do. As it stands, it is fully waterproof, with all stainless or
aluminium parts (yes, I am aware of the issues mixing stainless and
aluminium) and the highest quality waterproof Amphenol connectors, I hope
to be able to leave it as it is.

A 2 pole resolver will typically be good to 5 minutes of arc, these on here
are some of the nicest I have seen, the military are not known for cheaping
out on their kit!  I fully expect them to acheive that level off accuracy
with the Mesa board.

On Sun, 22 Jan 2023 at 03:21, gene heskett  wrote:

> On 1/21/23 20:20, Robin Szemeti via Emc-users wrote:
> > That looks like a useful module ... sine for one of the coils, cosine for
> > the other, and simply multiply the cosine signal by the servo drive
> signal
> > +1 goes one direction, -1 goes the other and anything in between is just
> > slower ...
> >
> > I have a company that manufactures audio amps, driver stages won't be a
> > problem.
> >
> > Just got to do some testing of the resolvers and decide which version of
> > the Mesa hard to order, I think these resolvers output 2 x the input
> > voltage, so I probably need the HV version, but we'll see.
> >
> > This will be used on the 10GHz band, with about 400W from a travelling
> wave
> > tube  there are good Python modules for moon position data,
> 'skyfield'
> > is the best of the bunch, I need roughly +- 1 degree precision, with
> just a
> > position update every 30 seconds or so .. the move doesnt move that
> fast. A
> > good resolver should easily be able to get within 0.5 degrees ...
> >
> > I'll let you know how it pans out.
>
> One of the reasons I mentioned the stepper/servo is that with its
> builtin encoder telling it where it is, the error signal is then used as
> the current control to the motor. It gives two characteristics to the
> control, the first being its accuracy in terms of rotational precision
> can be just one or two counts of the encoder, which in even the bargain
> basement models has about 14 bits of resolution per full turn of the
> motors armature. Such is the tech today, where a nema17 motor with a42
> mm thick core, develops nominally .5NM at a max rpm of about 1000,
> drives with the normal step/dir signals,and sells for about $40 a copy
> on amazon.  Needs 24 to 28 volts, and about 2 amps when working hard,
> microsteps to 256, we normally run them at 16, meaning 16 steps either
> way are one full steps move. That $5 arduino can drive them. They would
> need weather protection, but what doesn't. First stage geardown would be
> a 20 tooth gt2 pulley and 6mm wide belt turning a 150mm diameter pulley,
> which would drive another chain drive with #35 chain. Big 3 layers of
> plywood for the final pulley.  Or drive the screw in a long sat jack
> from Venture mfg in Dayton OH, but that would involve correcting for the
> non-linearity, and a full 180 scan is at best impractical.
>
> The other just as huge advantage is if it hits a stop and can't get to a
> +-2 counts or so balance between where it was told to go vs where it is,
> and if that is violated for any great amount of time, it will close a
> circuit, stopping linuxcnc in its tracks. I can on my 11x54" 80 yo
> sheldon lathe, position a jaw so the tool chip will hit the jaw, issue a
> move at an 1nch a second causing the tool's chip to hit the chuck jaw,
> It runs to hit the jaw, see's the stop, and kills linuxcnc in about a
> millisecond. The carbide chip in the tool is not damaged, nor is the
> chuck jaw marked.
>
> The mks-servo42C OTOH will blow its drivers if it can't get there.
>
> Yet that has never actually happened while doing real work. The original
> 2 phase steppers, they ran at a constant current when moving and half
> current standing still so they ran just short of burn your hand hot.
>
> Now these run maybe 10F above ambient unless working really hard at high
> speeds, a considerable power savings.
>
> [snip]
>
> Take care & stay well, Robin.
>
> Cheers, Gene Heskett.
> --
> "There are four boxes to be used in defense of liberty:
>   soap, ballot, jury, and ammo. Please use in that order."
> -Ed Howdershelt (Author, 1940)
> If we desire respect for the law, we must first make the law respectable.
>   - Louis D. Brandeis
> Genes Web page 
>
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.

Re: [Emc-users] Resolvers and 2 phase AC servos

[gene heskett]

On 1/21/23 20:20, Robin Szemeti via Emc-users wrote:

That looks like a useful module ... sine for one of the coils, cosine for
the other, and simply multiply the cosine signal by the servo drive signal
+1 goes one direction, -1 goes the other and anything in between is just
slower ...

I have a company that manufactures audio amps, driver stages won't be a
problem.

Just got to do some testing of the resolvers and decide which version of
the Mesa hard to order, I think these resolvers output 2 x the input
voltage, so I probably need the HV version, but we'll see.

This will be used on the 10GHz band, with about 400W from a travelling wave
tube  there are good Python modules for moon position data, 'skyfield'
is the best of the bunch, I need roughly +- 1 degree precision, with just a
position update every 30 seconds or so .. the move doesnt move that fast. A
good resolver should easily be able to get within 0.5 degrees ...

I'll let you know how it pans out.


One of the reasons I mentioned the stepper/servo is that with its 
builtin encoder telling it where it is, the error signal is then used as 
the current control to the motor. It gives two characteristics to the 
control, the first being its accuracy in terms of rotational precision 
can be just one or two counts of the encoder, which in even the bargain 
basement models has about 14 bits of resolution per full turn of the 
motors armature. Such is the tech today, where a nema17 motor with a42 
mm thick core, develops nominally .5NM at a max rpm of about 1000, 
drives with the normal step/dir signals,and sells for about $40 a copy 
on amazon.  Needs 24 to 28 volts, and about 2 amps when working hard, 
microsteps to 256, we normally run them at 16, meaning 16 steps either 
way are one full steps move. That $5 arduino can drive them. They would 
need weather protection, but what doesn't. First stage geardown would be 
a 20 tooth gt2 pulley and 6mm wide belt turning a 150mm diameter pulley, 
which would drive another chain drive with #35 chain. Big 3 layers of 
plywood for the final pulley.  Or drive the screw in a long sat jack 
from Venture mfg in Dayton OH, but that would involve correcting for the 
non-linearity, and a full 180 scan is at best impractical.


The other just as huge advantage is if it hits a stop and can't get to a 
+-2 counts or so balance between where it was told to go vs where it is, 
and if that is violated for any great amount of time, it will close a 
circuit, stopping linuxcnc in its tracks. I can on my 11x54" 80 yo 
sheldon lathe, position a jaw so the tool chip will hit the jaw, issue a 
move at an 1nch a second causing the tool's chip to hit the chuck jaw, 
It runs to hit the jaw, see's the stop, and kills linuxcnc in about a 
millisecond. The carbide chip in the tool is not damaged, nor is the 
chuck jaw marked.


The mks-servo42C OTOH will blow its drivers if it can't get there.

Yet that has never actually happened while doing real work. The original 
2 phase steppers, they ran at a constant current when moving and half 
current standing still so they ran just short of burn your hand hot.


Now these run maybe 10F above ambient unless working really hard at high 
speeds, a considerable power savings.


[snip]

Take care & stay well, Robin.

Cheers, Gene Heskett.
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author, 1940)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Chris Albertson]

You can simplify this if you replace the resolvers with magnetic encoders.
They cost $4 each and come  mounted to PCBs and with a magnet.   What you
do is super-glue the magnet to the shaft end and then face the chip at the
magnet with an about 1mm air gap.  The chip tells you the shaft angle to
within 1/4000 of a revolution.It is easy because it is an absolute
encoder so you don't have to track the shaft you can query the angle
whenever you need it.  There is both an analog and digital output

It is a 4mm diameter magnet and you mount it by first making a small hole
in the shaft with a drill if the shaft is not already center drilled.  Then
the magnet is very powerful and you really don't need glue.  They work
better if you get larger 6mm dia. magnets but still I get reading good to
about 0.1 degree.

With resolvers and quadrature encoders you have to track the encoders
continuously but these are "query on demand."   With more money, for $10
you get a better sensor that does 16,000 "ticks" per revolution, absolute.
These are cheap enough to place them on the motor and the driven structure
https://www.amazon.com/HiLetgo-Magnetic-Encoder-Measurement-Precision/


On Sat, Jan 21, 2023 at 4:59 PM gene heskett  wrote:

> On 1/21/23 06:25, Robin Szemeti via Emc-users wrote:
> > So,  in a vaguely CNC related folly I have purchased a 2 axis military
> dish
> > mount for radio stuff (moonbounce, if you must know)
> >
> > My plan is to control it from EMC, with a Mesa anything io card and the
> > resolver interface, I already have various Mesa cards under EMC, so thats
> > not an issue and I have a spare 5i23 PCI card and the isolated IO
> duaghter
> > card, so with the resolver card I should be able to uses the resolvers,
> and
> > pick up the limit switches on the IO card. I can just use some Python
> > script to work out the moon position and generate "G1 X285.4, Y23.8" or
> > whatever MDI motion commands every 15 seconds or so to track the moon,
> not
> > a problem.
> >
> > Now, the question is driving the two phase AC servos.   Generating vast
> > quantities of 50 or 400Hz or whatever AC for the drives is simple
> enough, I
> > can just use a couple of large Class D amplifiers, no worries. Generating
> > the AC low level signals to drive these  basically, it needs a
> constant
> > 50Hz signal on one coil of a fixed level, and a variable level sine wave
> on
> > the other coil, at either +90 or -90 degrees depending on which way it
> > needs to go.
> >
> > Is it possible to use HAL or something to generate the variable level
> 50Hz
> > servo drive signals and output them from the PWM outputs on the resolver
> > card?
> >
> I've not used it in that context, but take a look at man 9 siggen.
> There are several mux's available that could be used to switch motor
> directions. Its main use is in 1 kilohertz loop stuff for linuxcnc, but
> if you need finer time resolution, should be happy in a 3 to 5 kilohertz
> threads Use a halmeter to see how much time the individual module takes.
>
> Hal can do a lot, look the logic choices over in /usr/share/man/man9 of
> an linuxcnc install and let your imagination out to play w/o a
> chaperone.  You can use float vars at a 1 or even 4 kilohertz loop
> thread, so updating a motor drive at 50HZ ought to be doable.
>
> For motor amps, take a look at class D audio amps, the 400 watt modules
> are rather commonplace these days, running at ultrasonic carrier
> frequencies. Use chokes to keep the carrier out of the motor wires, and
> good single point(one bolt) grounds.
>
> Use mux's for analog switches.
>
> Using modern stepper drives that have current limits built in, you can
> creep steppers along at .01 hz a second, and if you spend the sheckels
> for the closed loop stepper/servo's suitably geared down, you ought to
> be able to track the moons motion within an arcsecond in real time and
> do it on 40 watts of power or less since they use the error to determine
> how hard to drive the motor. The recovery time when it goes below the
> horizon and has to pre-aim at the other horizon for when the moon comes
> back in sight at next moonrise is a problem since the active times match
> the dead times pretty closely.
>
> I've not paid that much attention but from satellite experience, you
> likely need just a little azimuth, even for polar mounts, your latitude
> above or below the equator determines that, and a large longitude.
>
> I have an old friend from back in the 70's that at that time was
> building a 12 bay steerable 2 meter get up. 16 element yagi's in a
> broadside array. But another job opening took m

Re: [Emc-users] Resolvers and 2 phase AC servos

[Robin Szemeti via Emc-users]

One thing that might cause some ... well, "issues" is my resolvers are used
to determine the absolute angular position of my two rotational axes, ie
they are driven off the axes themselves after the gearing, not off the
motors before the gearing. This may prove to be a bit "interesting" .. but
since my desired movement is very slow, fractions of a degree per minute, I
suspect it is not going to be a problem and it can all be handled in a slow
task loop.

On Sun, 22 Jan 2023 at 01:19, Robin Szemeti  wrote:

> That looks like a useful module ... sine for one of the coils, cosine for
> the other, and simply multiply the cosine signal by the servo drive signal
> +1 goes one direction, -1 goes the other and anything in between is just
> slower ...
>
> I have a company that manufactures audio amps, driver stages won't be a
> problem.
>
> Just got to do some testing of the resolvers and decide which version of
> the Mesa hard to order, I think these resolvers output 2 x the input
> voltage, so I probably need the HV version, but we'll see.
>
> This will be used on the 10GHz band, with about 400W from a travelling
> wave tube  there are good Python modules for moon position data,
> 'skyfield'  is the best of the bunch, I need roughly +- 1 degree precision,
> with just a position update every 30 seconds or so .. the move doesnt move
> that fast. A good resolver should easily be able to get within 0.5 degrees
> ...
>
> I'll let you know how it pans out.
>
> On Sun, 22 Jan 2023 at 00:57, gene heskett  wrote:
>
>> On 1/21/23 06:25, Robin Szemeti via Emc-users wrote:
>> > So,  in a vaguely CNC related folly I have purchased a 2 axis military
>> dish
>> > mount for radio stuff (moonbounce, if you must know)
>> >
>> > My plan is to control it from EMC, with a Mesa anything io card and the
>> > resolver interface, I already have various Mesa cards under EMC, so
>> thats
>> > not an issue and I have a spare 5i23 PCI card and the isolated IO
>> duaghter
>> > card, so with the resolver card I should be able to uses the resolvers,
>> and
>> > pick up the limit switches on the IO card. I can just use some Python
>> > script to work out the moon position and generate "G1 X285.4, Y23.8" or
>> > whatever MDI motion commands every 15 seconds or so to track the moon,
>> not
>> > a problem.
>> >
>> > Now, the question is driving the two phase AC servos.   Generating vast
>> > quantities of 50 or 400Hz or whatever AC for the drives is simple
>> enough, I
>> > can just use a couple of large Class D amplifiers, no worries.
>> Generating
>> > the AC low level signals to drive these  basically, it needs a
>> constant
>> > 50Hz signal on one coil of a fixed level, and a variable level sine
>> wave on
>> > the other coil, at either +90 or -90 degrees depending on which way it
>> > needs to go.
>> >
>> > Is it possible to use HAL or something to generate the variable level
>> 50Hz
>> > servo drive signals and output them from the PWM outputs on the resolver
>> > card?
>> >
>> I've not used it in that context, but take a look at man 9 siggen.
>> There are several mux's available that could be used to switch motor
>> directions. Its main use is in 1 kilohertz loop stuff for linuxcnc, but
>> if you need finer time resolution, should be happy in a 3 to 5 kilohertz
>> threads Use a halmeter to see how much time the individual module takes.
>>
>> Hal can do a lot, look the logic choices over in /usr/share/man/man9 of
>> an linuxcnc install and let your imagination out to play w/o a
>> chaperone.  You can use float vars at a 1 or even 4 kilohertz loop
>> thread, so updating a motor drive at 50HZ ought to be doable.
>>
>> For motor amps, take a look at class D audio amps, the 400 watt modules
>> are rather commonplace these days, running at ultrasonic carrier
>> frequencies. Use chokes to keep the carrier out of the motor wires, and
>> good single point(one bolt) grounds.
>>
>> Use mux's for analog switches.
>>
>> Using modern stepper drives that have current limits built in, you can
>> creep steppers along at .01 hz a second, and if you spend the sheckels
>> for the closed loop stepper/servo's suitably geared down, you ought to
>> be able to track the moons motion within an arcsecond in real time and
>> do it on 40 watts of power or less since they use the error to determine
>> how hard to drive the motor. The recovery time when it goes below the
>> horizon and has to pre-aim at the other horizon for when the moon comes
>> back in sight at next moonrise is a problem since the active times match
>> the dead times pretty closely.
>>
>> I've not paid that much attention but from satellite experience, you
>> likely need just a little azimuth, even for polar mounts, your latitude
>> above or below the equator determines that, and a large longitude.
>>
>> I have an old friend from back in the 70's that at that time was
>> building a 12 bay steerable 2 meter get up. 16 element yagi's in a
>> broadside array. But another job opening took me to NM 

Re: [Emc-users] Resolvers and 2 phase AC servos

[Robin Szemeti via Emc-users]

That looks like a useful module ... sine for one of the coils, cosine for
the other, and simply multiply the cosine signal by the servo drive signal
+1 goes one direction, -1 goes the other and anything in between is just
slower ...

I have a company that manufactures audio amps, driver stages won't be a
problem.

Just got to do some testing of the resolvers and decide which version of
the Mesa hard to order, I think these resolvers output 2 x the input
voltage, so I probably need the HV version, but we'll see.

This will be used on the 10GHz band, with about 400W from a travelling wave
tube  there are good Python modules for moon position data, 'skyfield'
is the best of the bunch, I need roughly +- 1 degree precision, with just a
position update every 30 seconds or so .. the move doesnt move that fast. A
good resolver should easily be able to get within 0.5 degrees ...

I'll let you know how it pans out.

On Sun, 22 Jan 2023 at 00:57, gene heskett  wrote:

> On 1/21/23 06:25, Robin Szemeti via Emc-users wrote:
> > So,  in a vaguely CNC related folly I have purchased a 2 axis military
> dish
> > mount for radio stuff (moonbounce, if you must know)
> >
> > My plan is to control it from EMC, with a Mesa anything io card and the
> > resolver interface, I already have various Mesa cards under EMC, so thats
> > not an issue and I have a spare 5i23 PCI card and the isolated IO
> duaghter
> > card, so with the resolver card I should be able to uses the resolvers,
> and
> > pick up the limit switches on the IO card. I can just use some Python
> > script to work out the moon position and generate "G1 X285.4, Y23.8" or
> > whatever MDI motion commands every 15 seconds or so to track the moon,
> not
> > a problem.
> >
> > Now, the question is driving the two phase AC servos.   Generating vast
> > quantities of 50 or 400Hz or whatever AC for the drives is simple
> enough, I
> > can just use a couple of large Class D amplifiers, no worries. Generating
> > the AC low level signals to drive these  basically, it needs a
> constant
> > 50Hz signal on one coil of a fixed level, and a variable level sine wave
> on
> > the other coil, at either +90 or -90 degrees depending on which way it
> > needs to go.
> >
> > Is it possible to use HAL or something to generate the variable level
> 50Hz
> > servo drive signals and output them from the PWM outputs on the resolver
> > card?
> >
> I've not used it in that context, but take a look at man 9 siggen.
> There are several mux's available that could be used to switch motor
> directions. Its main use is in 1 kilohertz loop stuff for linuxcnc, but
> if you need finer time resolution, should be happy in a 3 to 5 kilohertz
> threads Use a halmeter to see how much time the individual module takes.
>
> Hal can do a lot, look the logic choices over in /usr/share/man/man9 of
> an linuxcnc install and let your imagination out to play w/o a
> chaperone.  You can use float vars at a 1 or even 4 kilohertz loop
> thread, so updating a motor drive at 50HZ ought to be doable.
>
> For motor amps, take a look at class D audio amps, the 400 watt modules
> are rather commonplace these days, running at ultrasonic carrier
> frequencies. Use chokes to keep the carrier out of the motor wires, and
> good single point(one bolt) grounds.
>
> Use mux's for analog switches.
>
> Using modern stepper drives that have current limits built in, you can
> creep steppers along at .01 hz a second, and if you spend the sheckels
> for the closed loop stepper/servo's suitably geared down, you ought to
> be able to track the moons motion within an arcsecond in real time and
> do it on 40 watts of power or less since they use the error to determine
> how hard to drive the motor. The recovery time when it goes below the
> horizon and has to pre-aim at the other horizon for when the moon comes
> back in sight at next moonrise is a problem since the active times match
> the dead times pretty closely.
>
> I've not paid that much attention but from satellite experience, you
> likely need just a little azimuth, even for polar mounts, your latitude
> above or below the equator determines that, and a large longitude.
>
> I have an old friend from back in the 70's that at that time was
> building a 12 bay steerable 2 meter get up. 16 element yagi's in a
> broadside array. But another job opening took me to NM from NE so I lost
> track of him. Employment brought me to WV 38 years ago, and I'm now 21
> years retired, and still here.
>
> And tell us how you did it when it working.
>
> Cheers, Gene Heskett.
> --
> "There are four boxes to be used in defense of liberty:
>   soap, ballot, jury, and ammo. Please use in that order."
> -Ed Howdershelt (Author, 1940)
> If we desire respect for the law, we must first make the law respectable.
>   - Louis D. Brandeis
> Genes Web page 
>
>
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https

Re: [Emc-users] Resolvers and 2 phase AC servos

[gene heskett]

On 1/21/23 06:25, Robin Szemeti via Emc-users wrote:

So,  in a vaguely CNC related folly I have purchased a 2 axis military dish
mount for radio stuff (moonbounce, if you must know)

My plan is to control it from EMC, with a Mesa anything io card and the
resolver interface, I already have various Mesa cards under EMC, so thats
not an issue and I have a spare 5i23 PCI card and the isolated IO duaghter
card, so with the resolver card I should be able to uses the resolvers, and
pick up the limit switches on the IO card. I can just use some Python
script to work out the moon position and generate "G1 X285.4, Y23.8" or
whatever MDI motion commands every 15 seconds or so to track the moon,  not
a problem.

Now, the question is driving the two phase AC servos.   Generating vast
quantities of 50 or 400Hz or whatever AC for the drives is simple enough, I
can just use a couple of large Class D amplifiers, no worries. Generating
the AC low level signals to drive these  basically, it needs a constant
50Hz signal on one coil of a fixed level, and a variable level sine wave on
the other coil, at either +90 or -90 degrees depending on which way it
needs to go.

Is it possible to use HAL or something to generate the variable level 50Hz
servo drive signals and output them from the PWM outputs on the resolver
card?

I've not used it in that context, but take a look at man 9 siggen. 
There are several mux's available that could be used to switch motor 
directions. Its main use is in 1 kilohertz loop stuff for linuxcnc, but 
if you need finer time resolution, should be happy in a 3 to 5 kilohertz 
threads Use a halmeter to see how much time the individual module takes.


Hal can do a lot, look the logic choices over in /usr/share/man/man9 of 
an linuxcnc install and let your imagination out to play w/o a 
chaperone.  You can use float vars at a 1 or even 4 kilohertz loop 
thread, so updating a motor drive at 50HZ ought to be doable.


For motor amps, take a look at class D audio amps, the 400 watt modules 
are rather commonplace these days, running at ultrasonic carrier 
frequencies. Use chokes to keep the carrier out of the motor wires, and 
good single point(one bolt) grounds.


Use mux's for analog switches.

Using modern stepper drives that have current limits built in, you can 
creep steppers along at .01 hz a second, and if you spend the sheckels 
for the closed loop stepper/servo's suitably geared down, you ought to 
be able to track the moons motion within an arcsecond in real time and 
do it on 40 watts of power or less since they use the error to determine 
how hard to drive the motor. The recovery time when it goes below the 
horizon and has to pre-aim at the other horizon for when the moon comes 
back in sight at next moonrise is a problem since the active times match 
the dead times pretty closely.


I've not paid that much attention but from satellite experience, you 
likely need just a little azimuth, even for polar mounts, your latitude 
above or below the equator determines that, and a large longitude.


I have an old friend from back in the 70's that at that time was 
building a 12 bay steerable 2 meter get up. 16 element yagi's in a 
broadside array. But another job opening took me to NM from NE so I lost 
track of him. Employment brought me to WV 38 years ago, and I'm now 21 
years retired, and still here.


And tell us how you did it when it working.

Cheers, Gene Heskett.
--
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author, 1940)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Robin Szemeti via Emc-users]

Sure, I could do it on an ESP32 ... and the two axes of resolver feedback,
with the reference generator for the resolvers? and of course I'll need
some sort of graphical interface to show the current position, and an
interface to an external keypad or somesuch for intial setup and
establishing "zero" etc ...



On Sat, 21 Jan 2023 at 18:58, Chris Albertson 
wrote:

> This is a job best done on a $5 microcontroller, like the ESP32.   You can
> even program the thing in Python to do what is needed.   You can generate
> the AC as a 3-volt signal to an analog pin and then amplify it.
>
> Microcontrollers have come a long way.  ESP32 is a dual-core 32-bit CPU
> with hardware floating point, and yes the entire thing sells for $5 and it
> can self-host a python system.  All you need to program it is a terminal
> window on a PC.
>
> Yes it is fast enough.  On my desk now I an using one to commutate the
> coils on a BLDC motor using FOC control.  It computes the sinusoidal
> voltage that goes the each of the three phases to keep the magnetic field
> pointing the right way.   Quite a lot of calculations this is not simple
> 6-step algorithm.
>
> Here is an demo-project.   An ESP32 is used to compute the voltage on each
> of the three phases of a small motor so as to keep the ball on top of the
> wheel.There is a second microcontroller with a built-in camera that
> tells the ESP32 how far the ball is from the center point.   Again a $5
> computer is doing this and if you look there is not 100 lines of code in
> the loop.   The control is dramatically better then you can do with a Linux
> PC and 10x simpler  and 100x cheaper.
> https://youtu.be/fXxd8guAY7g
>
> Tracking a green ball is a servo system is a lot like tracing the Moon,
> except the Moon is easier to track
>
>
>
>
>
>
>
>
> On Sat, Jan 21, 2023 at 3:26 AM Robin Szemeti via Emc-users <
> emc-users@lists.sourceforge.net> wrote:
>
> > So,  in a vaguely CNC related folly I have purchased a 2 axis military
> dish
> > mount for radio stuff (moonbounce, if you must know)
> >
> > My plan is to control it from EMC, with a Mesa anything io card and the
> > resolver interface, I already have various Mesa cards under EMC, so thats
> > not an issue and I have a spare 5i23 PCI card and the isolated IO
> duaghter
> > card, so with the resolver card I should be able to uses the resolvers,
> and
> > pick up the limit switches on the IO card. I can just use some Python
> > script to work out the moon position and generate "G1 X285.4, Y23.8" or
> > whatever MDI motion commands every 15 seconds or so to track the moon,
> not
> > a problem.
> >
> > Now, the question is driving the two phase AC servos.   Generating vast
> > quantities of 50 or 400Hz or whatever AC for the drives is simple
> enough, I
> > can just use a couple of large Class D amplifiers, no worries. Generating
> > the AC low level signals to drive these  basically, it needs a
> constant
> > 50Hz signal on one coil of a fixed level, and a variable level sine wave
> on
> > the other coil, at either +90 or -90 degrees depending on which way it
> > needs to go.
> >
> > Is it possible to use HAL or something to generate the variable level
> 50Hz
> > servo drive signals and output them from the PWM outputs on the resolver
> > card?
> >
> > --
> >
> > Best regards,
> >
> > Robin Szemeti
> >
> > Redpoint Consulting Limited
> >
> > E: ro...@redpoint.org.uk
> > T: +44 (0) 1299 405028
> > M: +44 (0) 7971 883371
> >
> > CONFIDENTIALITY NOTICE
> > The information contained in this e-mail is intended only for the
> > confidential use of the above named recipient. If you are not the
> > intended recipient or person responsible for delivering it to the
> > intended recipient, you have received this communication in error
> > and must not distribute or copy it.
> > Please accept the sender's apologies, notify the sender immediately
> > by return e-mail and delete this communication.
> >
> > Thank you.
> >
> > ___
> > Emc-users mailing list
> > Emc-users@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/emc-users
> >
>
>
> --
>
> Chris Albertson
> Redondo Beach, California
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Chris Albertson]

This is a job best done on a $5 microcontroller, like the ESP32.   You can
even program the thing in Python to do what is needed.   You can generate
the AC as a 3-volt signal to an analog pin and then amplify it.

Microcontrollers have come a long way.  ESP32 is a dual-core 32-bit CPU
with hardware floating point, and yes the entire thing sells for $5 and it
can self-host a python system.  All you need to program it is a terminal
window on a PC.

Yes it is fast enough.  On my desk now I an using one to commutate the
coils on a BLDC motor using FOC control.  It computes the sinusoidal
voltage that goes the each of the three phases to keep the magnetic field
pointing the right way.   Quite a lot of calculations this is not simple
6-step algorithm.

Here is an demo-project.   An ESP32 is used to compute the voltage on each
of the three phases of a small motor so as to keep the ball on top of the
wheel.There is a second microcontroller with a built-in camera that
tells the ESP32 how far the ball is from the center point.   Again a $5
computer is doing this and if you look there is not 100 lines of code in
the loop.   The control is dramatically better then you can do with a Linux
PC and 10x simpler  and 100x cheaper.
https://youtu.be/fXxd8guAY7g

Tracking a green ball is a servo system is a lot like tracing the Moon,
except the Moon is easier to track








On Sat, Jan 21, 2023 at 3:26 AM Robin Szemeti via Emc-users <
emc-users@lists.sourceforge.net> wrote:

> So,  in a vaguely CNC related folly I have purchased a 2 axis military dish
> mount for radio stuff (moonbounce, if you must know)
>
> My plan is to control it from EMC, with a Mesa anything io card and the
> resolver interface, I already have various Mesa cards under EMC, so thats
> not an issue and I have a spare 5i23 PCI card and the isolated IO duaghter
> card, so with the resolver card I should be able to uses the resolvers, and
> pick up the limit switches on the IO card. I can just use some Python
> script to work out the moon position and generate "G1 X285.4, Y23.8" or
> whatever MDI motion commands every 15 seconds or so to track the moon,  not
> a problem.
>
> Now, the question is driving the two phase AC servos.   Generating vast
> quantities of 50 or 400Hz or whatever AC for the drives is simple enough, I
> can just use a couple of large Class D amplifiers, no worries. Generating
> the AC low level signals to drive these  basically, it needs a constant
> 50Hz signal on one coil of a fixed level, and a variable level sine wave on
> the other coil, at either +90 or -90 degrees depending on which way it
> needs to go.
>
> Is it possible to use HAL or something to generate the variable level 50Hz
> servo drive signals and output them from the PWM outputs on the resolver
> card?
>
> --
>
> Best regards,
>
> Robin Szemeti
>
> Redpoint Consulting Limited
>
> E: ro...@redpoint.org.uk
> T: +44 (0) 1299 405028
> M: +44 (0) 7971 883371
>
> CONFIDENTIALITY NOTICE
> The information contained in this e-mail is intended only for the
> confidential use of the above named recipient. If you are not the
> intended recipient or person responsible for delivering it to the
> intended recipient, you have received this communication in error
> and must not distribute or copy it.
> Please accept the sender's apologies, notify the sender immediately
> by return e-mail and delete this communication.
>
> Thank you.
>
> ___
> Emc-users mailing list
> Emc-users@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/emc-users
>


-- 

Chris Albertson
Redondo Beach, California

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

Re: [Emc-users] Resolvers and 2 phase AC servos

[Andy Pugh]

> On 21 Jan 2023, at 12:26, Robin Szemeti via Emc-users 
>  wrote:
> 
> Is it possible to use HAL or something to generate the variable level 50Hz
> servo drive signals and output them from the PWM outputs on the resolver

Updating the analog output at 1khz will probably give a good enough sine wave 
at 50hz. Maybe with some smoothing in the amp. 
The BLDC component and the Mesa 3phasePWM only know how to do 120 degreee phase 
shifts, so this is probably a job for a custom HAL component. 



___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users

[Emc-users] Resolvers and 2 phase AC servos

[Robin Szemeti via Emc-users]

So,  in a vaguely CNC related folly I have purchased a 2 axis military dish
mount for radio stuff (moonbounce, if you must know)

My plan is to control it from EMC, with a Mesa anything io card and the
resolver interface, I already have various Mesa cards under EMC, so thats
not an issue and I have a spare 5i23 PCI card and the isolated IO duaghter
card, so with the resolver card I should be able to uses the resolvers, and
pick up the limit switches on the IO card. I can just use some Python
script to work out the moon position and generate "G1 X285.4, Y23.8" or
whatever MDI motion commands every 15 seconds or so to track the moon,  not
a problem.

Now, the question is driving the two phase AC servos.   Generating vast
quantities of 50 or 400Hz or whatever AC for the drives is simple enough, I
can just use a couple of large Class D amplifiers, no worries. Generating
the AC low level signals to drive these  basically, it needs a constant
50Hz signal on one coil of a fixed level, and a variable level sine wave on
the other coil, at either +90 or -90 degrees depending on which way it
needs to go.

Is it possible to use HAL or something to generate the variable level 50Hz
servo drive signals and output them from the PWM outputs on the resolver
card?

-- 

Best regards,

Robin Szemeti

Redpoint Consulting Limited

E: ro...@redpoint.org.uk
T: +44 (0) 1299 405028
M: +44 (0) 7971 883371

CONFIDENTIALITY NOTICE
The information contained in this e-mail is intended only for the
confidential use of the above named recipient. If you are not the
intended recipient or person responsible for delivering it to the
intended recipient, you have received this communication in error
and must not distribute or copy it.
Please accept the sender's apologies, notify the sender immediately
by return e-mail and delete this communication.

Thank you.

___
Emc-users mailing list
Emc-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-users