Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations 3DDave on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Magnetic flux vs True Sensorless Vector by Mitsubishi...What gives? 1

Status
Not open for further replies.

fastline12

Aerospace
Jan 27, 2011
306
We are looking to use a sensorless drive in a machine tool spindle that is currently only using a v/f drive. Anything would be better. What basically happens is the rpms drop in a cut, the vfd does not know it, and to load to the tool goes up because of it. This compounds until a stall occurs. We simply need that drive to recognize there is a load change and to respond to it.

Mitsubishi indicated their older drives use "magnetic flux vector" control while the new ones use "true sensorless vector". I have a feeling this is just a coined phrase and both are nearing the same technology but the newer stuff just has faster response.

We are also trying to achieve much improved accel/decel from the spindle so we are purposely oversizing the drive by once size to give even more start current and to get a bigger regen transistor. The motor is thermally protected.

Does anyone know enough about the A500 and A700 Mits units to know the differences and any deficiencies of the A500?
 
Replies continue below

Recommended for you

Are stuck with Mitsubishi?

There are other drives that keep speed much better and also have full torque at very low speed. The Invertek Optidrive drives is one example.


Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
 
Because this is a machine tool spindle, slow is really not in the vocabulary so spending more to get that extra 1hz of grunt on the bottom will offer zero return. There is also a gear box involved so anything under 1800 rpm at the spindle nose is underdrive through the box. The actual ratio escapes me but it is a pretty fair assumption to say the motor will not run much below 100rpm other than to spool up to speed.

Does that extra low end grunt really make an accel difference? Can we quantify that? I am open to any good drive but I really thought Mitsubishi made an outstanding product and I do like the fact that I can get a tech on the phone pretty quick. Siemens and ABB are a different story. I know Jraef really likes the ABB stuff but the Mits was offered at a value.

I am still trying to find specs here to see the reaction ratings of the A700 and A500. Seems every drive is SUPER concerned about low speed performance but I am most worried about getting everything I can at the top end, around 200hz...
 
Sorry that I mentioned the zero speed torque. It may have led you to think that low speed is its only virtue. It is also very good at keeping speed constant at higher speeds. The control algorithm is 'crisp' and compensates for parasitic resistance and inductance in the stator in a way that standard PWM inverters don't. Study the Invertek site to understand it better.

But, at 200 Hz, most drives will choke because that usually is way above base speed and voltage doesn't correspond to frequency any more (if you didn't design 200 Hz to be top speed and laid out voltages and motor accordingly).


Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
 
I think I am still trying to quantify the elements that make a drive "better". Are we talking about the control speed in rad/sec? Or slip at certain rpm and loads? I would think there would be some standard tests that could prove one is better than another. These are not cheap options but will admit the A500 available to us would be cheap. that being said, I do not want to put a bandaid on our problem. Obviously true encoder feedback would be best but not really an option here.
 
Gross rule of thumb:

Scalar drives (i.e. non-vector) in machine tool spindles are able to maintain accurate speed only to about 50% speed, after which the drive torque variations begin to be a problem. So if you are always doing high speed work, you may never know the difference.

But in my salad days of applying countless Yaskawa and Teco scalar drives to machine tool spindles, one of mine ended up at a Boeing plant and I learned a very expensive lesson about low speed torque accuracy with a scalar drive. When they were machining aluminum (it turns out that happens a lot in the aircraft biz, go figure), they were slogging through some thick cuts at about 250RPM and chewing through bits at a terrible rate, all the while taking ugly chunks out of their stock material. At what was essentially 15% speed at a proscribed feed rate, the VFD could not react to the change in slip and the torque dropped off precipitously, which changed the tool bite. Not good. The solution at that time was to go back to the back gear and start with a lower spindle speed so we could keep the motor speed higher. That now could have been done with even the least expensive vector drive.

Mitsi makes good drives, I have nothing against them, I just don't have a good source for them where I live. But I would stick with the vector versions, scalar had it's day.

"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376
 
Jraef, as I understand it, scalar is v/f or true open loop operation? I had no intention of using something like that. They just do not belong on a machine tool spindle. I think my main questions were regarding the performance aspect of different sensorless build methods. Everyone seems to have "the better design". What values should I be looking at to quantify that "betterness"?

Also, regarding braking, I am a little green there and need to do a bit of research on braking. I hear regen, DC injection, and across the line braking thrown around but not quite sure what will take to slow this machine spindle from 9000 to 0 in 1-2 seconds. Is DC braking where the braking transistor diverts DC buss power right to 2 legs of the motor? Just how is regen braking initiated within the drive?
 
Update: I got a chance to educate a bit on the commonly available braking. It really seems like most of my braking needs to be done via dynamic braking to reduce heating of an already loaded up motor. All the high speed starts and heavy loads will certainly tax the motor.

What I think I need to determine is IF and WHEN I should use the DC braking side to help at some point? I am not sure if it is common in machine tool drives or not due to the extra heat. The spindle usually needs to come to a dead stop and I know that as the speed decreases, the dynamic brake loses effectiveness.
 
Hello fastline.

The dynamic braking doesn't reduce heat in the motor. You shall definitely forget about DC injection. Uncontrolled and not optimal. Delivering braking energy to a resistor bank (also called 'dynamic braking' in some quarters or sending it back to the grid (regeneration) is a matter of how often you brake. Once in a while or start/stop is what determines that. The braking is usually exactly the same with regeneration as it is with braking resistors. The only thing that differs is how you get rid of the energy - 'burn' vs 'sell back to utility'. In a DC drive, there is a difference, but not in a VFD.

The criteria for 'betterness'that you ask about are the criteria you put down yourself. Of course, there are static and dynamic accuracy, step response, overshoot and all those parameters but if you want us to scrutinize all those parameters and present a table with data and how well they correspond to your needs - I'm afraid you will not get it here. There are very experienced drives engineers available that can do that for you.

By the way, we have no data on that drive. Motor? Spindle inertia? You gave us speed. Duty cycle? Cycle time?

Did you really check the Optidrive? Or didn't you because you think I am their salesman? I am not. They are just better.



Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
 
I have looked at the Optidrive and they came recommended in the past. I guess my only questions though is, if it IS better, what makes it better? I am certainly not discounting the Optidrive at all, they are just more expensive and not sure why. As for the A500, it would come at a whopping 1000% cheaper than the Optidrive so we do have incentive to try it.
 
Sorry for misinterpreting your question earlier.

I'm not really clear as to what that "magnetic flux vector" term really means. I briefly skimmed a white paper put out by Mitsi on it and I still can't quite figure out what it was, because the white paper was more about how much better their version of "sensorless vector" is by comparison. And in reading it, I got the impression that they had a sensor in the sensorless version! It wasn't well written in my opinion, at least not for quick skimming as I did. Maybe if I were more interested I'd waste more time on it and it would be clear.

As another generality though, companies like to make up their own terminology sometimes because it forces customers to ask questions rather than lump everyone together in a "me too" conglomeration. But the reality is, that "me too" lump is a fairly accurate model. For all their perceived differences, it's more like taking different roads to get to the same city. If any of these different technologies were vastly inferior, the product would be non-viable in a very competitive market like what exists today for VFDs. Bottom line, they are all pretty good and if you are going from a scalar (V/Hz) drive to something with ANY kind of vector control, you are going to see a somewhat dramatic increase in performance. Beyond that you start splitting hairs. Are you going to notice the difference between an velocity response rate of 30rad/s vs 120rad/sec? As dramatic as that sounds, in 99% of applications you may never see a tangible difference.

Re: braking.
Dynamic braking is where the kinetic energy is taken from the motor as a generator and burned off in a resistor. DC Injection Braking is where you pump DC into an AC motor and creaet a stationary field in it. You actually need both. DCIB is very hard on the motor, but DB suffers from a "law of diminishing returns" problem. With DB, the faster the motor is going, the more braking energy you can take advantage of. But as it slows, that energy is at the same time diminishing, so at some point near the end, the DB can't quite "finish the job". So what the VFD mfrs do is allow you to automatically switch in the DCIB once the motor is almost stopped already. So most of the kinetic energy is already removed from the motor and mass by the DB, making it so that the effects of DCIB on the motor are minimal.

An alternative is Regenerative Braking and there are some decent options for that available now. I work for an integrator and we have been using the Sinamics G120 drive with their new low-cost regen power unit, I'm impressed. It's not the same as a full regen VFD like the ABB ACS800, but it also is not twice the rice of a non-regen drive either. I have yet to apply one to a spindle however.

But if you are only considering the Mitsi's, I'd go for the lower cost one if it were my money. I have customers who have put in some el-cheapo Automation Destruct Dura-Pulse vector drives on spindles to replace some of the older Yaskawas I did years ago for them, they are tickled pink. Mitsubishi makes a MUCH better quality drive than what Auto-Destruct is selling in my opinion. But like I said, even the worst is still pretty good. My general statement on "best" judgment calls is to find a good local supplier whom you trust, because that will ultimately have much more to to with your overall satisfaction that little minor differences in technology.

"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376
 
Hear, hear!

Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
 
You guys have a huge amount of help here. Jraef, you certainly have put my mind at ease. After all, if I really need 0hz performance or orient of spindle, I would really only consider a true encoder system for that. This is an older machine and the only real issues we have are response to load, accel, and decel rates.

Realizing that I cannot provide accurate mass or inertial calculations, in general, in a machine tool spindle, at what point could you safely induce DCIB to finish off the braking? The machine in question goes through a gear box, then belted to the spindle cartridge. General assumptions might have to be made here.
 
I would add that the OE system settings are setup to induce DCIB at 1.5hz and 20% brake power. I might want to reach up to maybe 10hz but if this will largely affect motor heating or wear, I might want to reconsider.
 
10Hz is the factory default for most drives. It won't make much of a difference in motor stress at that point unless you still had a lot of inertia left in the load at 10Hz, and it would have to be one heck of a spindle for that to be the case... I do centrifugal separator drive systems for one client, we switch on the DCIB at 2Hz for him because there still is a lot of inertia until then. But that's a totally different application from what you want to do.

"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376
 
Great! thanks so much. I have the Mits A500 on the way. Right from the mfg with new caps and tested out. Guess we will give it a shot.

The manual indicates the DCIB voltage is programmable from 0-30%, should I just run the default setting there or go ahead and use the max setting? I probably will bring it in at 10hz and see how things go. I am trying to stop this load in 1-2sec so probably will try and sneak a little less resistance on the DB brake to get a touch more from it as well as use the DCIB.
 
OH, I also want to ensure that I have this right, regarding DB braking, the only way to achieve more braking is the lower the resistance of the DB resistor? Basically the driven motor becomes a generator and the resistor is the load so when you lower than resistance, you are increasing load or demand to the motor thus slowing it faster?

I do know that I already have the target 20ohm resistor pack in this machine that matches the new drive specs. If this is the case, I can expect a whopping zero improvement in braking other than to use the DCIB more. Mits did indicate that we can lower the resistance slightly to get better braking. I am hoping that since the drive is over sized for the motor, there will be an extra layer of protection there for that DB transistor.

How is the power calculated through that transistor by the way?
 
I think you need to understand the difference between DC injection braking (no resistor involved) dynamic braking (usually only DC machines), inverter controlled deceleration with a braking resistor (to dissipate energy that would otherwise pump up DC link voltage) and inverter controlled braking with regeneration (no resistor).

If I have understood you correctly, you are addressing two things: 1) poor speed regulation at high spindle speeds and 2) ability to come to a fast stand-still.

The poor regulation at high speed is usually a result of reduced V/Hz and since the torque is reduced with V/Hz squared, it can be extremely difficult to do anything without doing the complete design over again and make sure that the torque you need is available at the speeds in question. Start with a comparison between nameplate data and actual inverter output frequency and voltage at the highest operation speed.

The slow stopping is a direct function of reduced torque at low frequencies. I have mentioned before that there are drives that are almost perfect in that respect and that means that you do not need to spend extra time to wait for a complete stop. Your reaction was "Why?". I cannot give you a complete course on drive technology and the various algorithms used to achieve good torque performance near zero. Sorry. It has taken me 20 - 30 years to gain a rather good understanding of these things and it is not something one can transfer in a few lines (OK, not so few now) of text. Just have a look at the technology and convince yourself.



Gunnar Englund
--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...
 
I've come in a little late on this conversation but something is concerning me a little but maybe I'm not too clear.
If this is a spindle motor designed to operate at 200Hz then typically the spindle is designed for that higher operating speed. This means that the rotor and cooling fan are designed with low inertia in mind. If it has an integral cooling fan that is.
A low inertia cooling fan is usually very much focused on providing cooling air when spinning at speeds=200Hz (9000rom you indicate).
However, at low speeds, the ability of the fan to cool the spindle are very much limited and you have the potential to overheat the spindle. Lower limit settings for spindle motors running at 200Hz are often set no less than typically 100Hz.
Having used VFD's on high speed spindles, the use of a VFD is purely to get to the high speed, not for low speed running.

Having said this, it might not be a problem as it depends on the spindle motor and overall cooling arrangement.

 
Every spindle motor I have ever been around use a separate motor controlled fan that can provide max cooling at all rpm. IMO, coupling a fan to the output shaft on a motor that will operate at a wide range of speeds is asking for heat problems.
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor