DIY Servo controller

[smarbaga]

uhu layout redrawn
i have not thickened the traces yet
i have changed the screw terminals to 0.2" centers, from 0.156"
i think i will put pads for faston connectors as well
if anybody does look at this layout and notices an error or a change needed
let me know, i would apreciate it.
i plan on getting some of these boards made with silk screen and soldermask
so any assistance will be greatly apreciated.
i don't know what the words are for the 10 pin terminal, but i would like to screen them on the pcb , in english

[Mariss Freimanis]

1) My biggest comment is the power section should be done with a 'copper pour' instead of traces. This means copper should cover the entire area with only 0.025" etched seperation lines between the copper 'islands'. Reason? Anyplace where FR4 green shows is an area that could have been a conductor. Di/dt will exceed 100A/uS, dv/dt will exceed 3 to 5 V/nS; this is for all effects an RF board layout. Trace inductance must be kept to a minimum particularly in the MOSFET source terminal area.

2) The BYV27 rectifiers are completely unnecessary. The MOSFET intrinsic drain to source diodes have the same current rating as the MOSFET, are probably faster and being at the same temperature as the MOSFET will conduct 99% of the reverse current anyway due to the neg tempco of silicon. Those BYV27s, being colder will conduct nada.

3) Do not use 1N4007s anywhere. Their reverse recovery time is measured in microseconds, not the needed nanoseconds. They are hopelessly slow; use high-speed or ultra high-speed diodes.

4) I don't know about the servo design here but the PWM duty cycle must not be allowed to go to 0% or 100%. This means a MOSFET must never be allowed to be continuously on or off. This circuit uses bootstrap charging of the top MOSFET gate drive power supply. A continuously 'on' top MOSFET will drain-off the 220nF bootstrap cap and fail the circuit, potentially catastrophically.

5) There is no low-inductance bypass cap across the main bridge DC bus. Without it the di/dt and dv/dt effects of an operating bridge under power will cause severe ringing on the DC bus. These voltage excursions are almost certain to take the MOSFET drain to source voltage into avalanche failure. You need at least several uF of non-inductive film or ceramic cap bypass. The 'lytics just don't cut it HF-wise.

Just my 2 cents worth.

Mariss

[smarbaga]

thanks for the advice mr freimanis.
right now the traces are just as a reference of routing and conectivity.
the power area will be copper pours.
this is a re-drawing of the uli's controller layout using his recommended parts, and his schematic, as the downloaded pdf( from this sight ) seems a little sloppy as it looks like an auto router did his board with no manual cleanups.
perhaps i will change the 4007's to d1,d2,d3 ....., as u say the 4007's are just rectifyers (blockers).
i am also wondering is anyone has received any controller chips from uli yet, and has one of his designs and layouts up and running , since the first post on here.
thanks all .

[ejkoeze]

The PCB design was made to be as flexible as possible. The BYV27 diodes are there to be used with other than the IRF540's and IRFP260N's.
A couple of people have built the controller with IGBT's and they do need them.

As far as I recall the PWM runs up to 80%.

Current calculations as shown here are peak currents. The normal or rated current for a servo motor is much lower. For short bursts the dissipation of the resistors may be higher than 5 watts (or 25watts for 5 in parallel) since they also need time to heat up.
Fact is people have run 1kw and more from this controller with NO modifications other than IGBT's and matching diodes and capacitors. I'll ask Stefan (the designer of the baord exactly how things were determined.

Erik Jan

Edit: They made no modifications to the board

[vroemm]

Originally Posted by ejkoeze Current calculations as shown here are peak currents. The normal or rated current for a servo motor is much lower. For short bursts the dissipation of the resistors may be higher than 5 watts (or 25watts for 5 in parallel) since they also need time to heat up.

I understand that.
I know it won't burn it down that quickly.

But why take the risk ?
What is so bad about changing the R17 and R16 so that the power in the 0.1 ohm is never bigger then 5 watt ?

And then there is the unclear question how many 0.1 ohm one should place.
The original question was, what to do with the fact that at one place there is mention of 3 resistors of 0.1 ohm and in a other place 5 resistors of 0.1 ohm.
There has never been a clear answer on this.

With my calculations you have a answer for this (I think)
For each X ampere of the servo motor 1 resistor of 0.1 ohm.
But then the risk of 13.5 watt on the 5 watt resistor is increased, and adjustment of the R17 and R16 become more usefull.

Allthough i must say, seems nobody has had any problems with the original R17 and R16.
Yet ?

[vroemm]

A way to set the Current Limiter of the UHU servo controller.

The Current Limiter can be set with potmeter R12.
By setting the voltage across R16.

U = Voltage over R16.
I = the maximum current you want through the servo motor.
R = the resistance of the current sensing resistors.
n = number of R resistors in parallel on your board.
Rp = the resistance value create by the n resistors R in parallel.

Rp= 1/( 1/Ra + 1/Rb + ..... 1/Rn)

If R = 0.1 ohm.
With 1 x 0.1 resistor Rp = 0.1 ohm.
With 2 x 0.1 resistor Rp = 0.05 ohm.
With 3 x 0.1 resistor Rp = 0.0333 ohm.
With 4 x 0.1 resistor Rp = 0.025 ohm.
With 5 x 0.1 resistor Rp = 0.02 ohm.

U = I x Rp

Set the voltage over R16 to U volt with potmeter R12.

Example:

You want to limit the current through your servo motor to 10 ampere.
And you have 3 resistors 0.1 ohm.

I = 10 ampere.
R = 0.1 ohm
n = 3
Rp = 0.0333 ohm.

U = I x Rp = 10 x 0.0333 = 0.333 Volt.

Turn R12 until the voltage over R16 is 0.333 Volt.

Now the current through the servo motor should be limited to 10 ampere.

[ulihuber]

Hi Mariss,

thanks for your comments. In parts I do not know if you reference to the 'classic' board or the new smd board of Jose which is in the design phase.

Just a reply from my perspective:

1: Using as much copper as possible is obviously the right approach. Nevertheless the classic PCB did not show any problems like heating up there. Nor did we see any EMC problems.

2: Yes, if using the proposed MOSFETS with integrated diodes the external diodes are of no effect. They can just be left out.

3: Personally I use fast diodes which are not even more expensive. This should be changed in the parts list.

4: Of course the duty cycle is not 0% and 100%. This would be an unallowed condition for the IR2184.

5: On the classic PCB we have non-inductive capacitors for all supply buses. I Know that Jose is going to add these in his design too.

Uli

[ejkoeze]

Explanation of the current limiting circuit. The voltage over the measuring resistors is compared to the reference voltage from Pot R12 by IC3. Wenn the reference voltage is exceeded (Pin2/IC3) the Flip/Flop IC8 will be triggered. This shuts down the pulsetrain to the powerstage. The calculation by Vroemm is theoretically correct, meaning that when R12 is maxed out a motor that takes more that 35A continuous can smoke out the resistors. BUT: during normal use a motor will use significatly less than 35A and will only use this high current in acceleration. If the current limiter is set to act just above the starting current you can easily set 50A for a motor with a nominal current of 30A. With Vroemm's resistor proposal this is not possible. Gruß Stefan

I translated this post from CNCecke.de and hope to have to have answered your questions about the current limiting circuit. Since neither English nor German is my first language the translation may be a bit rough.

If you have any more questions, don't hesitate

Erik Jan

[zoltan]

Where is posted the full schematics (including driver) and pcb in order to download?

Thank you,

Zolta

[smarbaga]

page 1 post 1

[vroemm]

Originally Posted by ulihuber 3: Personally I use fast diodes which are not even more expensive. This should be changed in the parts list.

Which diodes exactly are needed ?
If we have a type number then it can be put in the partslist.

Thanks..

[jlcortes]

i still have a couple of minor changes to make, but it is the definitive shape of my design.

I have placed 4 non-inductive capacitors just under mosfets, I think the inpedance of lines capacitors-mosfets are very very low. capacitors value can be select for used voltage, example 4 x 1uF 100v or 4 x 0.22uF 250v

Mariss has said about 0.025" separation between the copper pour, i can optimize this point.

what is the current conmutation frequency? how many power do you think could me managed without heatsink?

Jose Luis