[ckm]

Found also a Rutex 200V/40A drive priced at $456.00. I wonder how many of you would require higher than 20 Amps output current.

Kreutz.

Well, it would be useful as a spindle drive on a mill or a lathe. My current mill has a 6hp servo for a spindle (160v @ 40amps) and it's a small mill. A lathe is even bigger.

Why use a servo rather than an AC drive with a VFD? Because with a servo, you can position the spindle for a tool change and you have precise control for rigid tapping....

Chris.

[kreutz]

Well, it would be useful as a spindle drive on a mill or a lathe. My current mill has a 6hp servo for a spindle (160v @ 40amps) and it's a small mill. A lathe is even bigger.

Why use a servo rather than an AC drive with a VFD? Because with a servo, you can position the spindle for a tool change and you have precise control for rigid tapping....

Chris.

Thank You, Here I learn something new everyday.

Kreutz.

[L98FIERO]

Count me in on three of the new drives if they can preform as well as expected, my motors are the same(I think) as Tenmetalman's, SEM motors from an old Crusader M control, so if they work for him they should work for me as well. I'm impressed at how it's looking and the work that has been put into the development with little or no expectation of profit, I really appreciate the work you have done on this project, my hat is off to the three of you!!!!! When you consider the cost of the parts there is not much left in there for the labour involved in assembly and testing.

I originally bought the high power drives from Smarbaga and have posted the results of that.... fireworks, but don't have too much of a problem with the product that Garry tried to make available as I sold them without a significant loss so I'm pretty much OK with the whole experience.

Since then I have purchased the Granite VSD-A drives but haven't installed them because I saw that Kreutz was working on the new UHU that will have the capability of driving the motors to thier capacity as well as adding the ability to read differential encoders. The Granite drives will probably be for sale soon, too. I had also considered replacing the motors and drives with AC servos but if you have priced that option out you will find that you can replace the three existing drives and probably add a fourth axis to a mill for the price of one new AC drive and motor.

LaVerne

[H.O]

Hi LaVerne,

Since then I have purchased the Granite VSD-A drives but haven't installed them because I saw that Kreutz was working on the new UHU that will have the capability of driving the motors to thier capacity as well as adding the ability to read differential encoders.

Just to clarify something: The VSD-A do have support for differential encoder. I read your message as if they didn't.

[L98FIERO]

Hi LaVerne,

Just to clarify something: The VSD-A do have support for differential encoder. I read your message as if they didn't.

You are right, the VSD-A do support differential encoders but the original UHU didn't. The other thing I noticed is that the tuning program _appeared_ to be a little easier to operate though I haven't actually used it.

[H.O]

Ahh, you meant the original UHU, sorry 'bout that.

Yes, the VSD-A looks like a really nice drive (true torque loop among other things) and as I said before, if it wasn't for the 10A limit I would've bought them already. Now I'm really hoping the HP UHU boards will 'cut it'.

[contactirfu]

HI all

Just an update, most of the parts for the first few kits are already purchased. Paul has ordered the American parts and I have purchased the Indian parts which will be on way to Paul next Monday after I repack and label them. I believe if all goes well, we can start taking orders for unassembled kits once the Indian parts reach Paul.

I am working on the wiring of my router to get to attach the motors right ahead.
http://www.cnczone.com/forums/showth...229#post405229

mean while I was not able to get tickets for my flight to US, I may just be able to get thru this week end at home.

RGDS
Irfan

[rokag3]

hello,

150 us$

with or without Uli controller?

there is surface mount component can it be welded on the board ,eyes problems

it is no problem if you do not send the big capacitor or the cooler it weight, take volume and it's easy for almost every one to buy also if you work at 185 V you will prefer to buy a 350 V capacitor !!!
but the controller this is not so easy
i think that the board + the components (no cooler no capacitor) might be put in a small envelope and send as sample so it will be with no V.A.T. and due to the weight it will be cheaper

can you give me the price in this configuration
SMC in place
no capacitor no cooler
idea of transport fee for europe
thank lucien

[MrWild]

I have a diagram for a DPDT relay that gives power to the supply and shunts the caps through a resistor to short the caps quickly during an E-stop. This circuit is from Maris and is said to ground out the caps in a half a second to stop any chance of a servo moving in an E-stop condition,

Even at less than 50 volts, I notice a large ZAP, and can see some wear to the relay's contacts. This circuit isn't really designed for 150 volts as the relay contacts could weld themselves together. In another post reguarding bleed down resistors in high power systems however, I see how an SCR is incorporated to handle the surge and keep the relay contacts clean.

Is this a good idea for the HP UHU boards, or will the on board safety be enough and a plain bleed down resistor will be sufficient for cap discharge?

[contactirfu]

Hi Guys.

I am trying my level best to get my controller to finish i am still lagging behind. Seems it is taking a lot of effort and time than i originally estimated.

take a look
http://www.cnczone.com/forums/showthread.php?p=407878

RGDS
IRfan

[kreutz]

I have a diagram for a DPDT relay that gives power to the supply and shunts the caps through a resistor to short the caps quickly during an E-stop. This circuit is from Maris and is said to ground out the caps in a half a second to stop any chance of a servo moving in an E-stop condition,

Even at less than 50 volts, I notice a large ZAP, and can see some wear to the relay's contacts. This circuit isn't really designed for 150 volts as the relay contacts could weld themselves together. In another post reguarding bleed down resistors in high power systems however, I see how an SCR is incorporated to handle the surge and keep the relay contacts clean.

Is this a good idea for the HP UHU boards, or will the on board safety be enough and a plain bleed down resistor will be sufficient for cap discharge?

I don't recommend a power resistor across the relay contacts as a main source of power bleeding, it might be used complementing an electronic regenerative energy dumping like the one Mariss posted on his site, although, it was designed for 80 Vdc power supply it will probably need to be modified too for 150Vdc or over.

I don't have a CNC machine and my new 3KVA (0-185 or 0-360 Vdc) power supply (the old one was lost when I lent it to a friend to take it overseas for a repair), uses a couple of 120V 150W lamps (in series) for the same purpose. I did a commercial design, not long ago, that used a micro-controller to do electronic soft-start and regenerative power bleeding among other safety related functions, it works like a charm, but I can't disclose the schematics or code.

[chf525]

Servo controller developed in Germany.

Since I have seen a couple of very interesting threads here that I have read with great pleasure and from which I have learned a great deal I thought I’d pass trough some things I have learned elsewhere. On the cncecke.de (German CNC forum) there have been developments that should not be made available to just German reading enthousiasts.

Many of you will have seen the pages made by some Japanese guy that made a simple Servo controller complete with schematic for the driver board and source code for it’s Atmel processor. This project seemed to have a lot of potential and a couple of guys went to work on it. The result is a controller that is able to work with DC Motors with ~200 to 500 CPR encoders and with up to about 3000 RPM.
The controller board can be adapted to drive DC servo’s with power in excess is 1 kW!!!
It has a serial port which can be used to alter PID parameters, debug and finetune the controller. There is a dedicated (DOS) program that can be used for this, that has a little graph showing the position of the servo in realtime (for fine tuning and reducing overshoot). It is controlled with normal Step/Dir signals just like nearly every other Stepper driver. It has a current limiter.

Safety measure include the following.
Daisy chaining controllers to ensure shutdown of all controllers in case a single one has an error. Emergency stop can also be wired to this.
Current limiter
Adjustable position error shutdown (if for some reason the position error exceeds a preset value the controller is shut down.
If the controller hangs the output stage is shut down so no irretic behaviour can occur.

About the amount of power this thing can supply. Yesterday there was a post of a guy that had 1.6kW servo’s attached and while trying to determine the correct PID parameters it started to oscilate. Afterwards his son asked him why the house had vibrated to stangely…. Another has replaced his industrial Haidenhain controllers with these.

The project is not completely open source since the author of the software for the controller keeps the code to himself. However controllers can be bought from him at cost price and the entire project is called beerware. Repay him his costs (2 euro per controller (programmed) and postage and send him a beer or something else usefull (couple of milling bits, centre drills, or whatever) and he’s completely content.
I have built my first controller and I’m impressed.

This can take home cnc’ing to a whole new level. Imaging the amount of power and accuracy this will give your machine not to mention the increase in speed.
At about 40 Euro (~$50) (depending on where you get them) in parts this cannot be called expensive.

Erik Jan
Layout and schematic
Controller software (DOS) and manual (also in English)

Edit: Corrected link.

I thought your post interesting to give it a look see. Get back if anything comes of it!

[kreutz]

I thought your post interesting to give it a look see. Get back if anything comes of it!

What came from it was the original UHU drive.

[ulihuber]

What came from it was the original UHU drive.

Yes and no.
Yes, the article refers to the UHU Project.
No, the UHU is not just a modified SMC. The SMC was a starting point to learn how things stick together in the transfer from theorie to practice.
But the UHU shares no code with the SMC. It was two times completely reingeneered, uses a complex interupt structure instead of the encoder polling. The EEPROM routines are secured by forward error correction, there is the tracking error display, the enhanced PIDH control loop, watchdog, error daisy chain, clear text parameterizing, encoder error detection, 10bit PWM and quite a lot of other things the SMC did not even touch.

By the way, there are new features in beta like controlled decelleration in case of an error shutdown and others.

Best regards
Uli

[kreutz]

Yes and no.
Yes, the article refers to the UHU Project.
No, the UHU is not just a modified SMC. The SMC was a starting point to learn how things stick together in the transfer from theorie to practice.
But the UHU shares no code with the SMC. It was two times completely reingeneered, uses a complex interupt structure instead of the encoder polling. The EEPROM routines are secured by forward error correction, there is the tracking error display, the enhanced PIDH control loop, watchdog, error daisy chain, clear text parameterizing, encoder error detection, 10bit PWM and quite a lot of other things the SMC did not even touch.

By the way, there are new features in beta like controlled decelleration in case of an error shutdown and others.

Best regards
Uli

Oops, my mistake, I should have clarified that point, I knew it.

Thanks for all your effort on the UHU project! The new beta features are going to be helpful, the controlled deceleration in case of error is really needed.

Regards,

Kreutz

[Jay C]

Yes and no.
Yes, the article refers to the UHU Project.
No, the UHU is not just a modified SMC. The SMC was a starting point to learn how things stick together in the transfer from theorie to practice.
But the UHU shares no code with the SMC. It was two times completely reingeneered, uses a complex interupt structure instead of the encoder polling. The EEPROM routines are secured by forward error correction, there is the tracking error display, the enhanced PIDH control loop, watchdog, error daisy chain, clear text parameterizing, encoder error detection, 10bit PWM and quite a lot of other things the SMC did not even touch.

By the way, there are new features in beta like controlled decelleration in case of an error shutdown and others.

Best regards
Uli

Uli, is this the correct place to ask for more details of your forward error correction? I am using an ATMega128 for my robot with a PD loop (no integration component) running at 50Hz with interrupt driven encoders (2000ppr) joined to the output wheels. I set up a wireless back channel via bluetooth to print out current readings and allow system tuning after plotting the output in Excel.
The process is slow and I am working on a Java UI to allow more "real-time" plots. My speed is controlled plus or minus 1 tick per 20ms. Highr functions use these values to calculate distance and angle so the bot knows how far and what direction it's gone.
This is only an experimental platform for me to play with ideas and test myself. If this is not the right forum, could I exchange ideas with you via email to see if I can benefit from some of your ideas. At the minimum I'd like to modify my serial stream to be compatible with your tuning program if you'd be so kind. My blog on my project is http://jaydmdigital.blogspot.com if you are interested.

Sorry for the run on and possible hijack.

Jay

[ulihuber]

Jay,

I'd like to anwer your posting as well as the email you sent me right here.
To make it short: I do not share code. There are several reasons for this decision and I am going to name the three most important ones:
- Code, especially in applications where space is a challenge, makes an essential difference to the product. The optimization took a long time to develop and I just see no need to shorten development time for alternative products.
- My code is of course documented, but not for external use. I have neither the time to redo these comments nor the passion to go into following discussions about details. This is even more valid for extremely compressed code as it is in the UHU controller.
- The PIDH control loop is not as easy as I thought when I started to implement it. There are certain details I could not fully explain with a mathematical approach and which were developed over hundreds of subreleases in a way of step by step optimization. The result is the control loop stability you get with the UHU controller even for difficult setups without increasing tracking error. I see this part as intellectual property.

On the other hand, the protocol the UHU controller uses for the output of the tracking error is public. Besides the software I provide, there is at least one public and one non-public implemetation of terminal software using this protocol. It is easy by the way and uses the fact that characters do not use bit 7. Hence all data sent by the controller with bit 7 set is position error data. This gives a range of 128 to 255, out of which 191 reflects the center (zero error). The transmission of error signal starts after character 'A' is sent to the controller and stops with 'Z' or 'X' dependant on the language version of the controller.

Best regards
Uli

[subculture]

Hello,

i guess most UHU users are running their machines with a variation of Mach.

I wondered what step pulse settings you are running (in micro seconds)?

Thanks

Andy

[H.O]

Hi,

(OK, this got really, really long and I apologize for that but it needs to be discussed)

Over the last week or so I’ve been exchanging PM’s with Kreutz regarding the if, when, why and how to use an extra resistor (sometimes called a braking resistor) in series with the motor armature in order to limit the high currents that apparently can occur during hard deceleration, E-stop condition or a crash.

The issue is discussed in the Mystique2.pdf document published by Rutex regarding their drives. According to Kreutz the same issues applies to the HP-UHU and this is something that future users needs to be aware of.

Despite Kreutz’s (thank you Kretuz) endless attempts trying to explain exactly how and why this is a problem I can’t say I understand it fully but if he says it needs to be considered then I’m positive that’s true. So I will just try my best to show you what I’ve grasped from the discussion with Kreutz and if anyone sees anything wrong I appreciate if you jump in.

Here we go…..

To determine if you need a series resistor you need to know the motors armature resistance, the voltage constant of the motor, the max speed you intend to use and the power supply voltage. Here, I will use my motors as an example:

Motor voltage constant: 0.47Vs/Rad
Armature resistance: 0.24ohm
Max speed: 2000rpm
Max cont. current: 19A

So my motors max speed is 2000 rpm and I definitely want to be able to use that. The back EMF (BEMF) at that speed is calculated as 2 * Pi * VoltConst * MaxRPM / 60 which in my case is 2 * Pi * 0.47 * 2000 / 60 = 98.5V. The continuous current rating of the motor is 19A and the armature resistance is 0.24ohm so utilizing full cont. torque will result in a voltage drop across the armature resistance of 4.6V. On top of that the max PWM duty-cycle of the UHU chip is 87% so in order for me to actually GET 2000rpm from the motor (at full torque) my power supply would need to be at least around 120V. 120V * 0.87 – 4.6 = 99.8V that’s close enough.


[Sidenote]
Kreutz says that Power Supply Voltage * 0.87 + BEMF should be kept below 200, I haven’t understood exactly why since that will render the HP-UHU unable to fully utilize a motor with a voltage rating above 100V, or wouldn’t it?
[/Sidenote]


What determines the need of a series resistor is the current rating of the diode inside each MOSFET. (The Rutex document says to calculate with the continuous rating of the diode but Kretuz says that the pulsed rating will be fine). The HP-UHU apparently uses the IRFP264N and its diode’s pulsed rating is 170A. A safety margin of 20% should be used so the number to use for our calculation is 136A.

Now, the total resistance, as “seen” by the drive, should be at least BEMF + 0.87 * PowerSupplyVoltage / 136 or in my case 98.5 + 0.87 * 120 / 136 = 1.5ohm. The motor armature itself is 0.24ohm and let’s say the wiring is another 0.26ohm. Then I’d need a 1ohm resistor in series with the motor. 1ohm may not seem much but considering I want to (be able to) pull 19A thru it, it will have to be a pretty a big one since it will produce 360W of heat. On top of that I will loose a bit “on the corner” of the speed/torque curve as described in the Mystique2.pdf document.


This is getting very long now but let’s take another quick example. HomeShopCNC sells a pretty nice Servo motor. The specs on the site are a bit “messed up” but I believe this should be right:

Terminal Voltage: 72V
Max Cont Current: 8.2A
Voltage constant: 15.71V/1000rpm
Armature resistance: 0.85ohm
Max Speed: 4200rpm.

Here, the manufacturer already provides us with the terminal voltage so we don’t need to calculate it. ( I did anyway and got 73V, that’s close enough). Since the max PWM dutycycle is 87% we need a power supply voltage of around 85V.

The total resistance should then be 85 * 0.87 + 72 / 136 = 1.07ohm. The datasheet says that the armature resistance is 0.85ohm and on top of that we have the wiring, connectors etc so we would probably be fine without any extra resistor. But this “smallish” motor seems to be the limit of what can be driven without any form of extra protection.

As I said at the beginning: I don’t fully understand how the regenerative current can be so large so that it destroys the diodes in the MOSFETS but the expertise says that is a fact. So my only hope now is that someone jumps in and tells me my understanding of the issue is all wrong.

If you read this, thanks for hanging in all the way thru!

Best regards
/Henrik Olsson.

[contactirfu]

Good Post henrik, a lot of things understood, and a lot left still to ponder upon.