[cnc2]

hello.
looking for a good cheap encoder

http://catalog.digikey.com/scripts/d...me=102-1307-ND

it works perfect and gives you the opportunity to change the resolution with 4 microchips

also the axe system is very smart and it cost 30 us$ for 48 to 2048 lines !!!

on my hp board
Replaced C28 with 4.7uF/25V tantalum cap.
Replaced R54 with jumper wire
Placed 1N4148 between R54 and GND, cathode facing GND
my motor is 180v 8Amp cont. should I replace r21 also ?

Hi,

Great info about the encoders & great price !

Thanks for sharing !

cnc2.

[kreutz]

...............

Hmmm, EUREKA!, I think I've got it....Will make some tests as soon as possible...nothing to do with U9, the PCB, motors or currents....it should be the recovery time of the Slow limit circuit...

Kreutz.

Here is what I think is happening. If I am right, Henrik's new tests bypassing the Slow limit, as per the recommended modification, will give the same torque in both directions.

As I see it, the "apparent problem" shows when you move the axle in any direction from the rest position while holding it trying to overcome the motor movement. At first, it will try to return quickly back to the rest position because it is tapping the intermittent torque region, so the current limit will be the VREF_OC value, as noted by Henrik. If you immediately try to move it the opposite direction, then the Slow circuit integrator capacitor is still charged and quickly the nominal torque will kick in, limiting the current to the VREF, so the torque seems to be different in the opposite direction, while is actually the expected behavior of the protection circuit.

The intended effect of the Slow/Fast current limit circuit is to provide extra torque (as proven on the first step) during the acceleration phase but only for a short time, after that, it is necessary to wait some time until the capacitor C28 discharges (and the armature internal temperature drops) before having that extra potential torque available in whatever direction you chose to move. It is possible to shorten the recovery time (if required), I will let you know how to do it.

I have been looking at HJozsi's videos again. During that review I noticed that he repeatedly tested the motor on the lathe at over 30 amps (peak) Slow limit without smoking the board, even with the high inertia load of the lathe chuck. I don't recommend going over the 25 Amps design limit of 0.25V at VREF_OC (measured at U9 pin #10), it is necessary to be aware and measure that limit voltage (U9 pin 10) when using the Slow /Fast circuit at over 11 Amps (0.11 volts measured at VREF, test point J3), because the VREF_OC voltage increases when increasing VREF with the potentiometer.

[contactirfu]

Finally it all makes sense Kreutz, and I did buy a Fluke 87V multimeter over ebay.

[kreutz]

.... I did buy a Fluke 87V multimeter over ebay.

Congratulations.

I have been using a Fluke 87 and a Fluke 89 multimeters for many years without problems.

Kreutz.

[kreutz]

Hi,
With U9 removed the current limit LED turns ON as soon as the 15V powerspupply is turned on. It's quite dim though but that's probably due to the 1k series resistor. Put U9 back in and it behaves as before.

/Henrik.

That is normal, maybe if the LED is replaced by a high brightness LED (or you solder a 1K resistor in parallel with R32) it will be easier to see it lighting up when the Slow Limit kicks in.

Kreutz.

[Color]

Hi guys,

This is my very first writing in a forum, so please be patient with me (and with my english).
Soon I will start to work with my own approach to servo-controlled motors for a xyz table; table which should be born in a near future.

So, I read almost all posts in this thread and I have a question in respect to the circuit proposed as a regenerative power dump.
Considering that a great percentage of power supplies used by DIY CNC people are not regulated, and with a possible input line voltage variation of +/- 10% transferred to the output (more or less), what happen with this circuit, which have his threshold fixed, when is adjusted in a certain moment of low line voltage (or normal line voltage) and lather the line voltage goes up five or ten percent???.

I have an answer but in text, not in circuit yet and may be a long and boring text for the first post.
Somebody else have considered this point?

By the way, I find the place interesting and useful and I wish to say thanks all for that.

Gabriel

[H.O]

Hi Gabriel,
When the circuit senses a voltage higher than the setpoint that it's set to "trigger" at it will activate. It doesn't know, or care, if it's due to regenerative power or higher line voltage.

It all depends on how much fluctuations you have on your incoming power. The powerdump will safely handle a short surge (seconds) of overvoltage but if the voltage goes up and down over time you should set the dump circuit acording to the highest voltage.

It's really not that critical untill you aproach the max rating of the drive. If you are using a 120V powersupply, setting the dump circuit to trip at 140V should be fine.

/Henrik.

[H.O]

Hello
I want to know if the output transistor of the power dump circuit needs any additional cooling and how much it heats after some long period of operating?
I am designing my own PCB and need to know if I have to leave some place for a cooler to mount the FET on. Even not for a cooling purpose I will use some aluminum elbow because it just doesn't look good having this size of transistor waving alone in the air only on its legs. I could also bend it 90deg and bolt it on the pcb.
I may post some pictures when I am ready with the design later.
Todor

Hi Todor,
It depends on how much current you are dumping and how often. Under normal conditions I dare to guess that you don't need any extra cooling. The RdsON of the IRFP264N MOSFET is 0.06ohm so dumping 20A will only produce 1.2W of heat, plus some losses during switching etc (which is rather slow due to the BC557) but since it only operates in short bursts during hard braking my guess is that you'll be fine.

/Henrik.

[HJozsi]

That is normal, maybe if the LED is replaced by a high brightness LED (or you solder a 1K resistor in parallel with R32) it will be easier to see it lighting up when the Slow Limit kicks in.

Kreutz.

Hi Kreutz,

I've just made a test for slow current limit and Led. I turned off the fast current limit, pulling out the right side foot of R41 and soldered to gnd with a wire. Replaced R32 with 470 ohm, and the Led to a high brightness one. R21 remained 6.8kohm and I have now 470pf C28. Stall torque of the motor become like buffalo, apart from the Vref setting... Even I turned down Vref to 2mV it allowed higher current, the torque become not weaker. The led can now turn on... Runnig the motor, the led light bright on low Vref setting and less or off on higher Vref setting. I tested with Vref up to 0.25V, the Led was mostly off...

Regards,
Jozsi

[kreutz]

Hi Kreutz,

I've just made a test for slow current limit and Led. I turned off the fast current limit, pulling out the right side foot of R41 and soldered to gnd with a wire. Replaced R32 with 470 ohm, and the Led to a high brightness one. R21 remained 6.8kohm and I have now 470pf C28. Stall torque of the motor become like buffalo, apart from the Vref setting... Even I turned down Vref to 2mV it allowed higher current, the torque become not weaker. The led can now turn on... Runnig the motor, the led light bright on low Vref setting and less or off on higher Vref setting. I tested with Vref up to 0.25V, the Led was mostly off...

Regards,
Jozsi

Hello;

If the LED lights at low VREF means that the circuit is working, what happens is that it is necessary to have the current over the VREF for a few seconds before it limits the current, most of the time the motor current decreases within that time, so it really does not kick in often. If you disable the Fast limit, then actually there is no current limit during the C28- R40 integration period, hence the buffalo like torque.

The LED does not light up when the Fast limit is active. With the Slow Limit Bypass modification (and leaving the fast limit untouched), the Slow limit does not work as such anymore, it becomes a backup safety limit (because of the higher time constant) in case of fast limit circuit malfunctioning. In that case, both limits respond to VREF, but the Slow part of the circuit has more filtering (delay).

The proper way to test the Slow limit it is by having a controlled load and increase the load (so to slowly increase the current until the voltage drop in R36/R37 goes over VREF without reaching the VREF_OC level) and holding it for a few seconds. It will help if you use an Small motor, since it won't break your arms while braking the load manually, or you turn down Vref so to have less peak current in the motor.

It is important to note that the current pulses measured at R37/R36 are not a continuous current but have a repeticion frequency of 23 Khz and the width is dependent on the speed/position error, so is the voltage across them(their parallel combination) that gets compared to VREF. That is why the output of U9 that drives the LED is not a continuous level but rather a narrow pulse at 23Khz when the current has been over VREF for longer than the time constant of the integrator circuit, so the need for a high brightness LED in order to see it working. The integrator voltage R40/C28 decreases partially during a PWM cycle, since the current decreases during part of the PWM period, and the output of U9B (pin2) goes to Ground. So the time constant is actually higher than the theoretical RC constant.

Tonight, as soon as I get home, I will reply to your e-mail.

Regards,


Kreutz.

[H.O]

Hi Jozsi,
I'm not really following you here.... Are you saying that you now, by turning down the Vref can get the LED to turn on by "loading" the motor but that, when the LED lights, there's still no change in motor torque?

When you experienced the "strong in one direction, weak in the other phenomenon", was it always strong in the same direction? I'm 99.9% sure that in my case it has always been in the same direction.

Thanks!
/Henrik.

[HJozsi]

Hi Jozsi,
I'm not really following you here.... Are you saying that you now, by turning down the Vref can get the LED to turn on by "loading" the motor but that, when the LED lights, there's still no change in motor torque?

When you experienced the "strong in one direction, weak in the other phenomenon", was it always strong in the same direction? I'm 99.9% sure that in my case it has always been in the same direction.

Thanks!
/Henrik.

Hi Henrik,

yes, but I turned off the fast limit track with pulling out R41 leg from U9 13pin and connected to gnd. The slow limit track seems work, and the led is turning on. I've just read Kreut's explanation how it works ( or must work in normal way). So my experience was that whatever Vref I set, the motor had high torque in both direction. I went down to 1mV and up to 250mV the torque was as I felt the same. Probably I couldn't wait enough ( my tumb has pain already, because of the buffalo ) until the real timeconstant last...
so the power was the same.... On the other hand if I turned down Vref the Led started to flash and light... It was light (flash) while motor running according to Vref level...
In that experiment I left the original Ref1 setting ( R21 6k8). After that I put 10k parallel to R21 ( ~ 4k as a result ) the result was similar ...

BUT, soldering back the fast limit, I could reproduce the assimmetric power of the motor... The Vref was set low current and my experience was also that always the same direction was weak ( the motors are not 100% the same in both direction I think, even they have different tone...).The output pwm signal was different, good for the strong direction and jumping and missing period in the opposit. I turned up the Vref, it helped a bit, but the problem was solved when I take out the parallel 10k so R21 was 6k8 again.
On my board R20 is out, and C28 is 470 pF...
Well in this case the OC LED never light to me..
Still just a bit confused where to set Vref... I think I have to set a bit higher than the theoretical continous ...
You can try this setting...

Regards,

Jozsi

[kreutz]

On my board R20 is out, and C28 is 470 pF...
Well in this case the OC LED never light to me..

In that case it is not supposed to light up unless the Fast limit part of the circuit fails.
You can reverse the behavior by replacing R41 by 330K, R40 by 820 ohm, C29 by 470 pf and C28 by 100 pf. So the LED will light up whenever the now "Fast" circuit limits the current.

I did not want to recommend doing so many changes because it is confusing. Maybe on the next revision it will be that way.

[HJozsi]

In that case it is not supposed to light up unless the Fast limit part of the circuit fails.
You can reverse the behavior by replacing R41 by 330K, R40 by 820 ohm, C29 by 470 pf and C28 by 100 pf. So the LED will light up whenever the now "Fast" circuit limits the current.

I did not want to recommend doing so many changes because it is confusing. Maybe on the next revision it will be that way.

Well, the board will be built into the lathe housing so I leave it as it is...

[jcdammeyer]

Hi,

My http://www.kelinginc.net/KL34-180-90.pdf just arrived but I do need to order an encoder from US Digital. The servo is rated at 3200 RPM. What's a good encoder line count for the UHU chip?

500 lines * 4 = 2000 lines per rev. * 3200 RPM divided by 60 = 107kHz. Is that pushing it for the UHU?

Thanks

John

[H.O]

Hi John,
I've seen two different values on max encoder frequency on the UHU. The datasheet says 100kHz but in the FAQ-document on Uli's web-site it says that he's got all the way up to ~300kHz under optimal conditions but that you can expect 150Khz 'for sure'. My guess is that you'll be fine with 107kHz but I can't promise.

Even if the limit really is 100kHz you'll still get 3000rpm - I'd go for the 500 lines. If you're ordering new ones make sure to get differential outputs, it will make it less sensitive to noice.

/Henrik.

[jcdammeyer]

Even if the limit really is 100kHz you'll still get 3000rpm - I'd go for the 500 lines. If you're ordering new ones make sure to get differential outputs, it will make it less sensitive to noice.

/Henrik.

Thanks Henrik,

Yes. Differential for sure. I've got both an HP UHU on the way and one from Manjeet. In either case I'll get something working properly.

John

[cnc2]

Hi,

My http://www.kelinginc.net/KL34-180-90.pdf just arrived but I do need to order an encoder from US Digital. The servo is rated at 3200 RPM. What's a good encoder line count for the UHU chip?

500 lines * 4 = 2000 lines per rev. * 3200 RPM divided by 60 = 107kHz. Is that pushing it for the UHU?

Thanks

John

Hi John,

I think you can get a 1000 line encoder for your UHU, here's what Uli Huber said about it:http://gsst.wikispaces.com/UHUservomotors (Look at the end of the page).
- encoder 2 channel, 5V TTL , 250-1000 lines/rev, Uli said.

And her's what he says in the UHU FAQ:

Q: What encoders are suitable ?
A: These are incremental rotation encoders with two channels and 90 deg. phase shift. Usually these channels
are named 'A' and 'B'. Encoders with 250 to 1.000 lines per revolution are best. By decoding all edges of the
signals, the controller resolution will be 1000 to 4000 steps per revolution.
The electrical parameters depend on the circuit design. The common PCB needs a 5V TTL Signal.
Other designs may use 5V differential, 11uA or 1Vpp.

Q: What are the speed limits of the UHU ?
A: The maximum proven speed was 303 kHz with activated analyzer function. Using a 512 line encoder this is
about 9.000 rev/min. Such a speed however cannot be guaranteed for all operating conditions and is dependent
on the recalculation cycle value and the step multiplier.
150.000 steps per second is what you can expect for sure. The speed of the machine is dependent on the
mechanical setup (gear, lead screw).

I hope it helped !
Good Luck !

cnc2.

[jcdammeyer]

Hi John,

I think you can get a 1000 line encoder for your UHU, here's what Uli Huber said about it:http://gsst.wikispaces.com/UHUservomotors (Look at the end of the page).
- encoder 2 channel, 5V TTL , 250-1000 lines/rev, Uli said.

And her's what he says in the UHU FAQ:

cnc2.

Thanks. I have so many docs downloaded in my DCServo folder and I went through all the UHU ones I could find. Don't have a copy of the FAQ. Should have thought of that. Sorry for the trouble.

500 line differential it is then.

John

[tenmetalman]

Hello All,
Uli & I talked about this quite awhile back. He suggested for a machine with any real rapid/feed movement, I'm thinking 100ipm, 500 line encoders were max. Otherwise with three axis's problems would arise !. Kreutz refined the differential encoder interface on the UHU Highpower drive after my asking................. just recently there have been some new ABZ encoders for $30.00 from Digi-Key listed on the list. Am I wrong in thinking of incremental encoders as being A,A-,B,B-,C,C-,Z,Z-. The $30.00 new encoder is such a savings I'm wondering if the noise issue can be handled to make the these encoders feasible. any input ?
Paul (tenmetalman)