4 Attachment(s)
Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Hello, about 2 years ago I acquired an old cnc mill which is equipped with servo source and runs off of DOS. It has been a pain in the ass to keep this thing and its dinosaur age electrical components working and according to my buddy who has been helping me with this and who has spoken to Everett Reeve a few times(the guy in charge of Servosource who we have to send parts to for pricey fixing, and from what Ive gathered from the few other mentions of this company of the web, designer of the whole system) sounds like he might die at any moment and has no other technicians under his wing. This thing is unreliable from generic communication errors randomly, losing its limits on startup, to its latest hobby of reporting that the Y axis has moved over 1000 inches away. Yesterday it decided to not stop the Y axis when retracting the Z to move past a wall, didnt catch the error and halt the operation like those other times, and crashed the tool into the part snapping a 1/2 cutter. Im thinking retrofitting might be the best option
An idea I had awhile ago was to just swap out the control system and drivers reusing the "servos" hooked up to something running like LinuxCNC Mach3 ect until they eventually broke, assuming the motors labeled as "servo II" were you know actually servos. Come to find out not only are they closed loop stepper motors but are also special designed "VR stepper motors" which according to another thread are impossible to find a driver for. Great. Dr. DFlow over on youtube has a series and guide on converting a PM-833TV to a cnc mill. https://www.drdflo.com/pages/Projects/CNC-Mill.html (In hindsight one of these things would have been better sigh) Im mostly stuck on the kind of servo's I would need. He list the Nema 34 for the X and Y and Nema 42 for the Z. With my mill only the quill moves for the z, not the whole head so I dont think id need as powerful of a servo for the Z. (he also has it with a brake to resist gravity but dont know if I need that?) The PM-833TV however also just weighs less then the supermax, under 800lbs with the supermax being over 2000 so Im not really sure what amount of torque Id need. The servo II's have a torque of 115 in/lb or nearly 13 nm according to a converter i found. I see that the Nema 42 has a Instantaneous max. torque of 15 nm and Flow states on his guide "AC Servo motors have a low rated torque compared to a similar size stepper motor. However, they possess a high instantaneous torque that can be used intermittently, which makes up for the difference." so does that mean I could use Nema 42's for all the axis and get a comparable feed rate to what I have now? Ive been told the bed can move up to 100 inch/sec but I never program above 60 because it sounds terrible and tends to lead to (more) communication errors. I dont know how translate nm to inches per min though to find what nm id need though and that intermittently thing confuses me. (ive only ever worked with steppers before) I saw some somebody bring up DC servo motors are best for conversions. Another conversion thread for a bridegport somebody suggested Clearpath servos. Would closed loop stepper motors be okay for this kind of weight while maintaining accuracy? I see theres kits like Acorn CNC but those looks like theyre for smaller machines and plus Id kinda like to not have to rely on a company for service. Thank your for any information you can provide.
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
I think a 400W servo on the Z and 750W on the X&Y would be more than enough on that machine.
Your machine is about the same size as mine, with a 10x54 table. My X&Y brushed servos are 32 in/lb (3.61Nm) rated torque, rated at about 750W. My Z axis motor is actually a NEMA 34 stepper, but it runs as an analog servo, the controller thinks it's running an analog servo motor, this is a special and very rare case. I suspect your machine weighs around 3500 lb as mine does, pretty sure they are the same base frame size.
Having said that, modern AC servos are the way to go. My preference is Automation Direct SureServo2 (Delta) or LS Servo line. I have used both on projects with good results. Both have excellent documentation, and local (USA) support.
https://www.automationdirect.com/selectors/sureservo2
https://www.automationdirect.com/selectors/ls-servo
While I have used both ClearPath and DMM servos on projects, I no longer recommend them. I'm not going to further explain that statement.
The only reason I am using DC servos on my machine is that they are the original motors and have been working fine for the last 35 years and I saw no reason to change them out when I did the controls upgrade. If one fails, I will be going with the SureServo2 750W units. No point in installing 50 year old technology when you need to change out the motors anyway.
A brake is not needed on the Z. My rapid feed rates will easily hit over 200 IPM, and most likely around 300 IPM. But I limit it to 100 IPM rapids because the machine is not bolted to the floor and it actually will move the machine at high feed rates. Nm does not translate to IPM, they are not really related. Acceleration is a function of torque, but once up to speed it requires very little torque to maintain the speed, just friction losses plus tool load.
Acorn is fine, so is Linux CNC, or Mach4, or any number of other of other systems out there. The servo systems I linked to above are compatible with virtually any system. Acorn is an open loop system as is Mach4, so the loop would be closed at the drive, this works fine but my preference is to close the loop at the controller using analog control and 1 micron magnetic linear scales on the axes. This pretty much automatically compensates for any backlash or ballscrew error, but there is no substitute for a mechanically tight machine. I typically see accuracy of +/- 0.0002'' or better with my machine.
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Hi,
I use 750W Delta B2 series servos, Delta's entry level series that have a 160,000 count/encoder. They have a rated torque of 2.4NM at 3000rpm, and a peak torque of 7.1Nm and a max speed of 5000rpm.
They work well. The price difference between 400W and 750W was only $40USD, with the 750W B2 kit (servo motor/servo drive/cables) costing $438USD, so naturally I got the 750W versions. I also
got one with a brake for the Z axis, but in truth I don't really need the brake. The Z axis ballscrew is 5mm pitch and its propensity to back-drive is low.
You might think that 2.4Nm is fairly low torque, and compared to a stepper motor of the same size it is. The real winner is the peak overload torque which easily matches any stepper. When a stepper
gets overloaded it stall or misses steps, a servo just 'digs in' and does the business......they eat any stepper ever made, and seem to perform very much better than their specs suggest.
Delta is a Taiwanese brand made in China.
DMM is a Canadian brand made in China.
Both are good quality, performance, documentation and most importantly have free set-up and tuning software, at fair prices. If this is your first foray into servos you'll want the set-up software. There
are cheaper Chinese brands and while they seem to work and are reasonable quality the crap documentation and no set-up software will make them a real headache for a newcomer to servos.
Clearpath are good quality but Teknics have made so many compromises to make them seem attractive to first time buyers progressing from steppers that they are poor value for money.
Firstly they have one one digital output, for fault reporting for instance. Other brands have six, eight or more outputs. Clearpath has a 800 cpr encoder as standard and 6000 at special order
whereas Delta (entry level) has 160,000 cpr. The last and most damning thing about them is cost. A 400W Clearpath costs $517 whereas a 750W Delta costs $438, so twice the power for less money.
I believe that as Jim has pointed out Delta are calle SureServo in the US but otherwise the same.
I use Mach4 and it works well. Again as Jim has pointed out Mach4 is an 'open loop controller' so the loop is closed at the servo. This relies on good ballscrews and no backlash. I use C5 ground THK ballscrews
which are superb quality, and as a result get something like 5um repeatable resolution. Jim uses linear scales to close the position loop and that accommodates any ballscrew non-linearity for even better resolution,
at the cost of some complexity.
Just as a matter of interest the next model up in the Delta range is the A2 series. That has two encoder inputs, the regular rotary encoder (with 1,280,000 cpr resolution) but also an auxillary encoder,
a linear scale for example, to close the position loop. This means that Mach4, or one of the other open loop control solutions, can still have 'load sensing position control'. An A2 series servo cost about
$75 more than a B2 series. I struggle to use all the performance of the B2 as is withouit worrying about even more with the A2 series!
Craig
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Check this EtherCAT Servo Motor Kit with Mach4, 750W , It will work good too
You can see this video for your reference
https://www.youtube.com/watch?v=tiVW...ature=youtu.be
https://www.automationtechnologiesin...h4-750w-3-axis
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Thank you for the sugjestions! I was actually in the middle of watching Tony's conversion series. I like his weird mpg cube monstrosity cus something I did like about the servosource system was being able to do manual control when I set tool height, used an indicator, or whatever. I see automationtechnologiesinc also has ac servo kits which are cheaper then sureservo2 or LS Servo line so im wondering what the main differences are
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Hi everyone,
I’m an applications engineer from Teknic. I noticed some comments that are factually incorrect. As an employee, I obviously am biased. I won’t dispute opinions that one product is “poor value for [the] money” – but I can present facts so you can make informed decisions. I’ve addressed the following sections below: (1) ClearPath output capability, (2) ClearPath encoder information, (3) Accurate cost comparison, and (4) Data omission that may be misleading
1) “Firstly, they have one one (sic) digital output” – The ClearPath SD series (commonly used in CNC applications) has one configurable digital output. For most applications, this provides sufficient feedback. However, other ClearPath products allow you to query 100’s of different data points from the servo (but more data is not often needed).
2) “Clearpath has an 800 cpr encoder as standard and 6000 at special order whereas Delta (entry-level) has 160,000 cpr.” –All DC ClearPath servos have a 12,800 count per revolution optical encoder. However, command resolution varies based on firmware selection (800 or 6400 counts per revolution standard option). The optimal selectable command resolution typically depends on the step rate limits of the controller.
With encoders, larger numbers don’t necessarily equate to better. Here’s a useful segment from Motion Control Tips, a leading motion control magazine (unaffiliated from Teknic, Delta, and DMM): “It’s important to note that higher resolution does not mean higher accuracy. Consider two encoders – one with 100 PPR resolution and one with 10,000 PPR resolution, but both with the same accuracy specification. The lower resolution (100 PPR) encoder can report a movement of 90 degrees just as accurately as the higher resolution (10,000 PPR) model. The higher resolution encoder just has the ability to break up that 90 degree movement into much smaller increments.” The DC ClearPath repeatability is 0.03 degrees (Delta’s encoder repeatability spec does not appear to be published). https://www.motioncontroltips.com/en...he-difference/
3) “The last and most damning thing about them [Teknic] is cost.” This statement is factually wrong. When comparing Teknic ClearPath with the SureServo2 directly from Automation Direct, the out-the-door cost of ClearPath on a Multi-axis system is notably less money.
100-Watt Sure Servo2 system example:
•Individual components:
o $372 drive
o $280 100 Watt motor
o $122 Breakout Board Kit
o $59 3m Encoder Feedback cable
o $39.50 3m motor power cable
•Total 1-axis cost: $872.50. Total 3-axis cost: $2617.50
•Web link: https://www.automationdirect.com/sureservo2
100-Watt Teknic system example:
•Individual components:
o $257 motor/drive
o $19 power cable
o $23 Controller Cable
o $248 Power Supply (powers multiple axes)
o $14 AC Power Cable
•Total 1-axis cost: $561. Total 3-axis cost: $1,159.
•Web link: https://teknic.com/products/clearpat...arpath-models/
Availability:
•Teknic’s US-built servos are available on our eCommerce site with a 3-day lead time.
•SureServo2 product availability varies. While some models are available to ship in 2 days, others are back ordered until April and May.
4) Data Omission: There are also several omissions in the information that could be misleading to users unfamiliar with Teknic:
•Teknic servos are all designed, manufactured, sold, and supported in the United States.
•Teknic also provides free set-up and tuning software (and free factory-direct technical support).
•The ClearPath is an integrated servo, a different product than separate servo drives and servo motors. Teknic manufactures and sells separate drives and motor products. In business since 1985, in our experience, the integrated approach has several advantages including lower total installed cost and reduced cabling (which is the largest source of motion-related failures on automated machines).
-Aaron B. Teknic Servo Systems Engineer
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Hi,
Quote:
However, other ClearPath products allow you to query 100’s of different data points from the servo (but more data is not often needed).
Certainly you can monitor many different variables, but only output one. If you wish to use that output as a fault indicator, that's fine, you can do so, but then you cannot also have
an 'In Position' indicator. Neither do you have any analog outputs for power monitoring for instance, neither do you have an auxiliary encoder outputs. Teknics have made some severe
compromises in order to make their products appealing to first time servo users, and this is one of those compromises. One, and only one digital output.....
Quote:
–All DC ClearPath servos have a 12,800 count per revolution optical encoder. However, command resolution varies based on firmware selection (800 or 6400 counts per revolution standard option). The optimal selectable command resolution typically depends on the step rate limits of the controller.
I stand corrected on this matter. If however you are saying that a low resolution encoder is as good as a high resolution encoder I call BS. I certainly agree even a modest encoder can to all intents and purpose
adequately resolve the angular position of the shaft. But the smoothness and discrimination the internal algorithm by which the servo operates is much improved by increased resolution and for that purpose I would
consider the Delta B2 160,000 cpr as adequate at best. The 20 bit and 24 bit encoders of higher spec models are now the 'sweet spot' of servos....any suggestion that 1280 cpr is competitive with either 160,000 cpr
or higher is disingenuous.
Quote:
3) “The last and most damning thing about them [Teknic] is cost.” This statement is factually wrong. When comparing Teknic ClearPath with the SureServo2 directly from Automation Direct, the out-the-door cost of ClearPath on a Multi-axis system is notably less money.
You are cherry picking data to suit your argument.
FACT: I have in the last week taken delivery of a 750W Delta B2 series servo kit, consisting of the servo, the drive and the cables for $438USD plus $150USD three day Fedex to New Zealand, total $588USD delivered.
Happy to post the invoice if you require confirmation.
I bought a 750W servo for less (excluding shipping) than you want for a 100W servo. You are choosing to price SureServo from AutomationDierct but that is far from the best deal available.
Quote:
Availability:
•Teknic’s US-built servos are available on our eCommerce site with a 3-day lead time.
•SureServo2 product availability varies. While some models are available to ship in 2 days, others are back ordered until April and May.
Tecknics does enjoy very good reputation for support and supply, no question. I can't speak to SureServo availability but my Delta supplier got a braked version of a servo that I wanted in three days and
was shipped to me four days after placing the order. My evidence is that availability of Delta products is good, even if not as good as Teknic's.
Quote:
•Teknic also provides free set-up and tuning software (and free factory-direct technical support).
•The ClearPath is an integrated servo, a different product than separate servo drives and servo motors. Teknic manufactures and sells separate drives and motor products. In business since 1985, in our experience, the integrated approach has several advantages including lower total installed cost and reduced cabling (which is the largest source of motion-related failures on automated machines).
As I said earlier Tecknics has an enviable reputation for sales and support, and I'm sure that reputation is well earned.
Quote:
•Teknic servos are all designed, manufactured, sold, and supported in the United States.
This may once have been a mark of quality but that reputation has slipped badly. 'Made in USA' is increasingly becoming 'Mad in USA'.
Craig
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Hi joeavaerage,
I worry a little from the tone of your response that you took my feedback in a way not intended. I did not mean to be disingenuous, say anything that is BS, or cherry-pick examples to support any points. Teknic’s goal in participating in these forums is to offer customer support and advice, and if we come across anything that is objectively incorrect about Teknic products, to try to correct the errors.
There are many high quality servos on the market, and you have made it clear that you value certain features that ClearPath (SD-series) doesn’t have. You have reiterated this many times, on many forums (often unsolicited). That’s fine, that’s your right. If Delta or DMM servos meet your needs, and after checking out ClearPath specs and prices you have concluded that it is not the optimal solution for you (your previous posts establish that you have no first-hand experience with ClearPath), we will not try to convince you otherwise. I will just clarify/correct some technical points:
It is true that there is only one physical status output on ClearPath (on the ClearPath-SD series). But it is incorrect to say that we can output only one variable (at a time). For example, by modulating the output with a PWM signal, we can simultaneously provide actual torque usage, MoveDone, In-position, and Fault status. Download the ClearPath user manual to see how this works.
Your inference that we made design choices to make ClearPath appealing to first-time users is also incorrect. Teknic primarily sells to large OEM corporations in the medical, semiconductor, packaging, and CNC industries. Our average customer is an engineer with many years of motion control experience, working for a company that buys hundreds or thousands of servos per year from us.
In addition to high performance and value, these engineers place a very high value on system reliability. Our experience over more than thirty years of technical support has proven that the biggest cause of malfunction in machines using servos is electrical corruption of the encoder signals, or intermittent or bad connection of one or more encoder wires. We, and our customers, have found that overall system reliability is greatly enhanced by reduced wiring, especially the elimination of encoder wiring, given their susceptibility to noise and their criticality to the proper functioning of the machine.
All of our servo drives designed prior to ClearPath have encoder outputs, and the number of machine malfunctions attributable to faulty wiring, grounding, or impedance issues associated with those outputs is astounding. On the other hand, most applications don’t need to see the encoder data given that the machine software always knows the commanded position, and the high level feedback signal from ClearPath confirms the tracking accuracy of the servo. Given that Teknic has thousands of customers who use ClearPath this way, including OEMs in the CNC industry, it’s fair to say that this is a good architecture for many serious users, if not for you.
Despite the fact that Teknic products are designed for the OEM market (see the About Us page on our website), a side effect of the reduced wiring problems and the ease with which a ClearPath servo system can be implemented, is that many corporate end-users and hobbyists have been attracted to ClearPath. And even though we didn’t set out to sell to end-users and hobbyists, we have put in a great deal of effort to carry our reputation for excellent service and support into these markets. Thank you for recognizing that our reputation is well-earned.
While we’re on the subject of encoders, let’s talk about resolution. I responded not too long ago to a post you made on another forum claiming that DMM had “vastly better encoders” than ClearPath. I’ll repeat what I wrote there for the benefit of people on this forum, because it’s important for users to understand that servo performance does depend on excellent encoder performance, and also that resolution is far from the only encoder metric that matters.
A good analogy for resolution in encoders is megapixels in digital cameras. Inexperienced photographers often believe that the quality of a camera is dictated by the number of pixels on its sensor. Camera manufacturers take advantage of this ignorance by touting extraordinarily high pixel counts. But the number of pixels is only one factor of many, many others that determine the quality of a photograph.
The camera on the Samsung Galaxy S22 Ultra cell phone has 108 megapixels! It’s a great camera—for a phone—but the photo quality is downright poor compared to a Canon 1DX Mark III high-end professional camera that has less than one-fifth of the pixels (20.1 megapixels). There’s many reasons, however, that sports photographers are not using their cell phones to take photos for Sports Illustrated despite the fact that their phones often have many more megapixels. Even a quick glance at photos from the two cameras will convince anyone that the Canon camera is, by far, the better camera of the two despite having vastly fewer megapixels.
DMM motors use magnetic encoders; the native resolution of magnetic encoders is generally quite low. According to the DMM website, the DMM encoder uses a 2-pole magnet. As the motor shaft turns, hall-effect sensors pick up the changing magnetic field from the encoder disk and convert it into an analog signal. One revolution generates one sinewave. So the raw resolution is one sinewave per rev—very low. The actual encoder counts are all generated by interpolation. In other words, the encoder estimates the motor position based on the amplitude of the analog sinewave. This interpolation has to be accomplished in the face of errors such as shaft run-out, noise, DC offset, short- and long-term drift, sensor hysteresis, non-linearities, temperature effects, etc. Although these errors can be somewhat compensated for, you can see on DMM’s website, that the accuracy of their encoder is only 0.1 degree (i.e., 1 part in 3,600). This means that because of the extreme interpolation, the accuracy is nowhere near their resolution figure (the accuracy is 18 times worse than the resolution!).The accuracy of the DC ClearPath servos is 3.56 times greater and the accuracy of the AC ClearPath servos is 17.78 times greater than the DMM encoders.
On top of this, in order for these “high resolution” encoders to remove enough noise from the analog signal to be able to do even a moderate amount of interpolation, the signal needs to be filtered. This adds a delay to the encoder data of 2-10 milliseconds typically. This introduces a dynamic error that reduces the servo performance. Even worse, because the encoder is a serial encoder, the servo gets position updates at a fixed clock rate (every 100 microseconds, per the DMM website). This means that at 3,000 RPM, you would have a positional change of 327 counts before the servo is even updated. The servo cannot respond to data it does not have no matter how fine-grained it may be.
Even if the high resolution of these encoders came with reasonably proportional accuracy, it would still not help the servo performance appreciably, but when you consider the amount of errors induced by the extreme interpolation, it is actually harmful. Internally, these servos have to throw away much of their interpolated (i.e., inaccurate and noisy) data or average it. Either way, the servos are not actually getting any benefit from their purported high resolution—except a marketing benefit. These extreme resolutions are pure specsmanship.
We know this well because during ClearPath’s development we seriously considered using a 16-bit magnetic encoder because it was actually cheaper than the optical encoder we ultimately chose, but considering all of the above disadvantages, we decided that the cost savings of using a magnetic encoder were not worth the performance degradation. We would rather educate users about the fallacy of focusing on these encoder specs, than play the specsmanship game and deliver lower performance.
We both agree that even a modest encoder [resolution] can “adequately resolve the angular position of the shaft”. But you feel that, “the smoothness and discrimination the internal algorithm by which the servo operates is much improved by increased resolution” and that “160,000 cpr is adequate, at best”. This is the “more must be better” fallacy.
The reality is that with a high quality optical encoder, you reach diminishing returns once you are using a 12-14-bit encoder (assuming a rotary motor; the analysis is a bit different for a linear motor). With a very well designed servo, you can squeak out a little more performance with a 16-bit encoder (~65,000 cpr). And this assumes that this higher resolution brings proportional accuracy. You are much better off with a high-fidelity, lower resolution encoder, than a higher-resolution encoder with a lot of error and non-linearity.
The empirical fact is that there is no truth whatsoever to the statement that the smoothness or discrimination of the servo algorithm is even marginally improved by resolutions anywhere near 160,000 cpr. More is not better at this level, and certainly not if more resolution is at the expense of accuracy, noise, drift, and non-linearities.
If you’re still not convinced, here is one more way to prove this: Let’s say the play (and compliance) in the motor bearing and bore combine to be 1 ten-thousandths of an inch (that’s 2.5 microns, so it’s an extreme assumption, but let’s assume we can achieve this, to prove the point). This means that with a one inch diameter encoder disk, you will have random, mechanically induced encoder noise (and uncertainty) of 1 part in about 32,000. But realistically, the bearing play alone (assuming a high-quality, deep-groove ball bearing) in a motor of the size we’re talking about, if you specify a “tight” clearance bearing, will typically be about 3 tenths. That makes the mechanically induced encoder noise about 1 part in 11,000 (just from bearing play alone). So, you can see that most of the resolution in an encoder above, say 32-64,000 cpr is pure noise, and not helpful for servo control. Empirical testing proves this out.
Best regards,
Aaron B- Teknic Servo Systems Engineer
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Hi,
Quote:
I worry a little from the tone of your response that you took my feedback in a way not intended. I did not mean to be disingenuous, say anything that is BS, or cherry-pick examples to support any points. Teknic’s goal in participating in these forums is to offer customer support and advice, and if we come across anything that is objectively incorrect about Teknic products, to try to correct the errors.
Perhaps my use of the term 'disingenuous' is too strongly emotional for this discussion. I certainly do not take any exception to a reasoned and balanced discussion, and your participation has been exactly that,
balanced and reasonable, and defending the products your produce natuarly enough.
There is no doubt that you have only one digital output, thus its not possible to distinguish between an overload fault or a Following Error fault for example where all fault conditions are combined. I understand that you have made that
choice to simplify the IO arrangement but that is a consequence.
Quote:
While we’re on the subject of encoders, let’s talk about resolution. I responded not too long ago to a post you made on another forum claiming that DMM had “vastly better encoders” than ClearPath.
I claimed no such thing, although I do claim that Delta encoders are vastly higher resolution, I am familiar with Delta products, not DMM.
Quote:
The empirical fact is that there is no truth whatsoever to the statement that the smoothness or discrimination of the servo algorithm is even marginally improved by resolutions anywhere near 160,000 cpr.
This is where we part company....I am by training and qualification a control engineer, and I know full well that discrimination and smoothness is improved by a high resolution encoder. Indeed the ultimate resolution
is if you like, analog, a perfectly smooth and continuous variable. A digital encoder is only ever at best an approximation to a smoothly continuous variable, therefore the higher the resolution the better.
Please note that the smoothness I refer to is numerical.
You posit that the mechanical limitations of a servo far exceed the advantages of a numerically smooth control algorithm, and that is fair enough. I propose that the the control algorithm needs be at least as
good as any mechanical limitations and for that purpose high resolution encoders are required.
Craig
Re: Looking to retrofit a Supermax YCM-40, how do I figure out what I need
Hi Joeaverage,
Quote:
Originally Posted by
Teknic_Servo
I worry a little from the tone of your response that you took my feedback in a way not intended. I did not mean to be disingenuous, say anything that is BS, or cherry-pick examples to support any points. Teknic’s goal in participating in these forums is to offer customer support and advice, and if we come across anything that is objectively incorrect about Teknic products, to try to correct the errors.
Quote:
Originally Posted by
joeavaerage
Perhaps my use of the term 'disingenuous' is too strongly emotional for this discussion. I certainly do not take any exception to a reasoned and balanced discussion, and your participation has been exactly that,
balanced and reasonable, and defending the products your produce natuarly enough.
I’m glad you understand that we are trying to have productive and honest discussions about these fairly complex technical points. Our participation in forums is strictly limited to answering questions related to our products, and to correcting factual errors. In fact, if you made an incorrect statement that made ClearPath look better than it is, or you claimed that it could do something it couldn’t, we would be just as quick to point out that error. We are educating the user base more than “defending” our products.
Quote:
Originally Posted by
joeavaerage
There is no doubt that you have only one digital output, thus its not possible to distinguish between an overload fault or a Following Error fault for example where all fault conditions are combined. I understand that you have made that
choice to simplify the IO arrangement but that is a consequence.
You are correct, if there is a servo fault, the HLFB output will not tell you the exact fault condition without querying ClearPath via the MSP software. Most of the time however, if you get a fault (especially multiple faults), you have to take a closer look at the situation anyway, so it doesn’t matter. And just to be completely accurate, this point only pertains to ClearPath SD- and MC-series. The SC-series of ClearPath can be queried under software control and can also provide running statistics about motion parameters (e.g., average and peak tracking error, average torque usage versus time, etc.) That said, most non-OEM CNC users are still better off buying the SD-series.
Quote:
Originally Posted by
Teknic_Servo
While we’re on the subject of encoders, let’s talk about resolution. I responded not too long ago to a post you made on another forum claiming that DMM had “vastly better encoders” than ClearPath.
Quote:
Originally Posted by
joeavaerage
I claimed no such thing, although I do claim that Delta encoders are vastly higher resolution, I am familiar with Delta products, not DMM.
It was only a few months ago (12/4/2022), in your post on the IndustryArena forum, that you wrote that the forum member “1832vin” should:
“Research Delta (Taiwanese brand made in China) or DMM (Canadian brand made in China). Both are good quality with vastly better encoders…”
See: https://en.industryarena.com/forum/s...stcount2532098
There would be no issue with you saying that certain Delta (or DMM) motors have higher resolution. That is a fact according to their specifications. But it is misleading (even if unintentional) if you imply or state that this resolution improves performance in any material way. This is not in the best interest of the forum participants.
Quote:
Originally Posted by
Teknic_Servo
The empirical fact is that there is no truth whatsoever to the statement that the smoothness or discrimination of the servo algorithm is even marginally improved by resolutions anywhere near 160,000 cpr.
Quote:
Originally Posted by
joeavaerage
This is where we part company....I am by training and qualification a control engineer, and I know full well that discrimination and smoothness is improved by a high resolution encoder.
The issue is not whether higher resolution improves servo performance (it does up to a point), it’s about how much resolution you need before something else becomes the limiting factor in system performance. It’s also about the fact that resolution (whether high or low) is by itself a good figure of merit for an encoder without regard to other important encoder quality metrics.
Quote:
Originally Posted by
joeavaerage
Indeed the ultimate resolution is if you like, analog, a perfectly smooth and continuous variable. A digital encoder is only ever at best an approximation to a smoothly continuous variable, therefore the higher the resolution the better.
Please note that the smoothness I refer to is numerical.
Again, this is an oversimplified and theoretical “more is always better” argument. Your premise is correct, but the conclusion is wrong. You are starting to reach the point of diminishing returns around the 8,000 cpr point (assuming a high quality optical encoder). Once the encoder reaches about 32,000 to 64,000 cpr (the exact value is dependent on the quality of the other components and the mechanical system), you have hit the limit where no amount of increased resolution buys you any improvement in performance.
As for the analog is better than digital argument, this is a theoretical argument that ignores many important elements of the overall system, e.g., the noise in an analog system, or the quality of the ADC in a digital system. An old Victrola record player reproduces audio using a continuous analog system (a needle wiggling back and forth in a continuous groove). I’m sure we can agree that it doesn’t sound better than digital audio recorded and played back today with a 16-bit, 48 kHz audio system. (Similarly, a high quality analog system can be better than a poor quality digital system.) So an analog-is-better-than-digital argument is not valid without considering the full context.
To take the analogy further, if you compare the above mentioned 16-bit, 48 kHz audio against audio sampled at 24 bits and 96kHz, the latter will sound very marginally better (played back in a quiet environment with excellent audio equipment). This is analogous to moving from an 8,000 cpr optical encoder to a 64,000 cpr optical encoder. The latter will give you a subtle, but measurable performance advantage in a very well designed and built mechanical system.
Now, to complete the analogy, an encoder with 160,000 cpr, would be like recording audio at 48 bits at a 1 MHz sample rate—it would be pointless. And given that we’re talking about encoders that achieve their high resolution with extreme interpolation (making the resolution spec spurious), it’s really analogous to recording audio at an unnecessarily high bit depth and sample rate, and also using a cheap microphone and a noisy preamp.
If you’re not convinced at this point by the technical explanation, I think the only way you will ever be convinced is to do empirical testing (like Teknic does). Buy a good servo drive that can support a 24-bit serial encoder (almost 17 million counts per rev, lol!) as well as a quadrature encoder. Buy two of the same motors: one with a 24-bit encoder and one with an 8,000 count quadrature (optical) encoder. Tune them both for optimal performance and compare the tracking response. You will see negligible difference, if any. Switch the 8,000 cpr encoder to a 64,000 cpr (optical) encoder, and you will see zero difference in performance versus the 24-bit encoder. Then you will be convinced.
Quote:
Originally Posted by
joeavaerage
Please note that the smoothness I refer to is numerical. You posit that the mechanical limitations of a servo far exceed the advantages of a numerically smooth control algorithm, and that is fair enough. I propose that the the control algorithm needs be at least as good as any mechanical limitations and for that purpose high resolution encoders are required.
The mechanical (and other) limitations in a system should always be considered during the design of its components. And I agree that it’s good if the servo control algorithm isn’t the limiting factor, but that doesn’t mean you need orders of magnitude of overkill in encoder resolution to get there.
-Aaron B. Teknic Servo Systems Engineer
