[BCwanderer]

Credit for this link goes to slp_prlzys who posted it to General Electronics Discussion
This is posted to Open Source Controller boards as I hope it will be picked up and developed quickly

Allegro has a new Bipolar Stepper driver chip that drives N-FETs in a bridge configuration.
You can buy them now at Digikey for about $5.00
The chip includes high and low side drivers. - No IR interface chips required.
Look here
http://www.allegromicro.com/sf/3986/index.asp

A3986
Dual Full-Bridge MOSFET Driver with Microstepping Translator
Features

* 2-wire step and direction interface
* Dual full-bridge gate drive for N-channel MOSFETs
* Operation over 12 to 50 V supply voltage range
* Synchronous rectification
* Cross-conduction protection
* Adjustable mixed decay
* Integrated sinusoidal DAC current reference
* Fixed off-time PWM current control

Description

The A3986 is a dual full-bridge gate driver with integrated microstepping translator suitable for driving a wide range of higher power industrial bipolar 2-phase stepper motors (typically 30 to 500 W). Motor power is provided by external N-channel power MOSFETs at supply voltages from 12 to 50 V.

This device contains two sinusoidal DACs that generate the reference voltage for two separate fixed-off-time PWM current controllers. These provide current regulation for external power MOSFET full-bridges.

Motor stepping is controlled by the two-wire step and direction interface, providing complete microstepping control at full-, half-, quarter-, and sixteenth-step resolutions. The fixed-off time regulator has the ability to operate in slow-, mixed-, or fast-decay modes, which results in reduced audible motor noise, increased step accuracy, and reduced power dissipation.

The translator is the key to the easy implementation of this IC. Simply inputting one pulse on the STEP input drives the motor one step (full, half, quarter, or sixteenth depending on the microstep select input). There are no phase-sequence tables, high frequency control lines, or complex interfaces to program. This reduces the need for a complex microcontroller.

The above-supply voltage required for the high-side N-channel MOSFETs is provided by a bootstrap capacitor. Efficiency is enhanced by using synchronous rectification and the power FETs are protected from shoot-through by integrated crossover control and programmable dead time.

In addition to crossover current control, internal circuit protection provides thermal shutdown with hysteresis and undervoltage lockout. Special power-up sequencing is not required.

This component is supplied in an 38-pin TSSOP (package LD). The optional lead (Pb) free package (suffix –T) has 100% matte tin leadframe plating.

[Azrael]

This is a tough one!!! In a combination with IRF540 brings up to 10A at 50V which is quite enough power even for bigger steppers. Next week I will start prototyping my board. The only thing that is still missing is current limiter to approx. 25-30% when no step signal is detected.
Cheers

[JohnDH]

This is a tough one!!! In a combination with IRF540 brings up to 10A at 50V which is quite enough power even for bigger steppers. Next week I will start prototyping my board. The only thing that is still missing is current limiter to approx. 25-30% when no step signal is detected.
Cheers

You control current through the VREF pin. You can use a retriggerable 0ne-Shot (Half of 74HC123) or you can use the PWM output of a very small microprocessor (ATtiny13). The input for both circuits is the Step signal after the opto-coupler. For the tiny13 solution you set up the timer to give 1mS ticks and count ticks until you reach the timeout value. (100mS to 500mS). You set up two PWM values, one for Run and one for Hold. Switch to the Run value any time there is a step, and to Hold if the time-out occurs. For the HC123 method you set up a resistor chain to feed VREF and the open collector output of HC123 to clamp one point in the chain to ground.

[waldek]

Hi All!
This is my first post in this CNC forum. At the beginning, I’d like to thank to Phil (pminmo), Kreutz and Mark (markomp77). In his post no. 526. Phil showed me that I didn’t do anything wrong in my circuit: in the video he provided http://video.google.com/videoplay?do...937&pr=goog-sl, the steps of the stepper motor were not uniform – I observed the same behaviour in my system. Mark in one in his first posts advised me to use a MOSFET transistor to reduce the voltage from 35V to supply save voltage for 7805 voltage regulator.
I’m only interested in the quarter-step working mode of the A3986 (I need 800 steps per rotary) and I have noticed that the 1st step is very long (175% of the nominal step length), the 2nd one is very short (25%), the 3rd one is long (125%), the 4th one is short (75%) and so on. This is a big disadvantage of A3986. I chose this IC because its application circuit is simple – it isn't necessary to generate steps with an external microprocessor. At the
beginning of my project I broke two A3986s when I incremented the supply
voltage from 18V to 35V on VDD, which made the ICs heat too much. I don’t know why. Now I have manually soldered two PCBs and tested them. They don’t work satisfactorily: in some step position loud noises can be heard and the steps are not of equal length.

I have two questions about the incorrect operation of the A3986 in its quarter mode:
1. How can I make the step lengths more uniform?
2. Do you know if the producer of the circuit, Alegromicro, is going to correct it in its next version?

In my project I have used a circuit with a retriggerable monostable multivibrator 4538; by the transistor Q10, this circuit reduces the current in both phases of the stepper motor to about 30% of the value set by HEX switch when no STEP signals arrive for 0.8 s.
By the way, at the moment when the currents are reduced, the shaft of the stepper motor is moved by a little angle for some step position.
To produce VREF I used the following circuit:


Rsense1 = Rsense2 = 0R033
Where Vcc = 5V, R21 = 5k1, R23 = 51k, R15 = 130k, R14 = 68k, R17 = 36k and R16 = 18k.
I set the current by the HEX switch S1:
HEX switch position Rhref[Ω] Vref[V] Vsense[V] Iphasemax [A]
0 51000 0,455 0,0569 1,724
1 36630 0,611 0,0764 2,315
2 29143 0,745 0,0931 2,821
3 23806 0,882 0,1103 3,342
4 21103 0,973 0,1216 3,685
5 18156 1,096 0,1370 4,152
6 16105 1,203 0,1504 4,558
7 14330 1,312 0,1640 4,970
8 13304 1,386 0,1733 5,252
9 12069 1,485 0,1856 5,624
A 11127 1,571 0,1964 5,952
B 10250 1,661 0,2076 6,291
C 9714 1,721 0,2151 6,518
D 9039 1,804 0,2255 6,833
E 8500 1,875 0,2344 7,103
F 7978 1,950 0,2438 7,388


Rsense [Ω] R21= Rlref [Ω]
0,033 5100
R23 [Ω] R15 [Ω] R14 [Ω] R17 [Ω] R16 [Ω]
51000 130000 68000 36000 18000

[waldek]

Hi All!
This is my first post in this CNC forum. At the beginning, I’d like to thank to Phil (pminmo), Kreutz and Mark (markomp77). In his post no. 526. Phil showed me that I didn’t do anything wrong in my circuit: in the video he provided [ame]http://video.google.com/videoplay?docid=-9106592029775083937&pr=goog-sl[/ame], the steps of the stepper motor were not uniform – I observed the same behaviour in my system. Mark in one in his first posts advised me to use a MOSFET transistor to reduce the voltage from 35V to supply save voltage for 7805 voltage regulator.
I’m only interested in the quarter-step working mode of the A3986 (I need 800 steps per rotary) and I have noticed that the 1st step is very long (175% of the nominal step length), the 2nd one is very short (25%), the 3rd one is long (125%), the 4th one is short (75%) and so on. This is a big disadvantage of A3986. I chose this IC because its application circuit is simple – it isn't necessary to generate steps with an external microprocessor. At the
beginning of my project I broke two A3986s when I incremented the supply
voltage from 18V to 35V on VDD, which made the ICs heat too much. I don’t know why. Now I have manually soldered two PCBs and tested them. They don’t work satisfactorily: in some step position loud noises can be heard and the steps are not of equal length.

I have two questions about the incorrect operation of the A3986 in its quarter mode:
1. How can I make the step lengths more uniform?
2. Do you know if the producer of the circuit, Alegromicro, is going to correct it in its next version?

In my project I have used a circuit with a retriggerable monostable multivibrator 4538; by the transistor Q10, this circuit reduces the current in both phases of the stepper motor to about 30% of the value set by HEX switch when no STEP signals arrive for 0.8 s.
By the way, at the moment when the currents are reduced, the shaft of the stepper motor is moved by a little angle for some step position.
To produce VREF I used the following circuit (file Fragment.PNG):

Rsense1 = Rsense2 = 0R033
Where Vcc = 5V, R21 = 5k1, R23 = 51k, R15 = 130k, R14 = 68k, R17 = 36k and R16 = 18k.
I set the current by the HEX switch S1 (HEX switch position.doc):

Below is a photo of my project - dimensions 4.88 in x 2.63 in, 124mm x 67 mm (Plytki 25.PNG):

The top layer (Stepper top.PNG):

The bottom layer (Stepper bot.PNG):

The schematic diagram (Stepper_sch.PNG):

Thanks,
Waldek

[Xerxes]

Below is a photo of my project - dimensions 4.88 in x 2.63 in, 124mm x 67 mm (Plytki 25.PNG):

The top layer (Stepper top.PNG):

The bottom layer (Stepper bot.PNG):

The schematic diagram (Stepper_sch.PNG):

Layout looks quite nice.

Have you tried using different RC filter values for current sense inputs? 100nF and 100 ohms makes 10us time constant which is quite slow. I would try 1nF and 100 ohms instead.

[waldek]

Have you tried using different RC filter values for current sense inputs? 100nF and 100 ohms makes 10us time constant which is quite slow. I would try 1nF and 100 ohms instead.

Hi Xerxes!
Today I changed C11* and C13* in my project to 1nF and I tested it. I saw that the steps are still not of equal length and I think that the noises are more loud.

[pminmo]

This is the part I mentioned a earlier in the year I was going to wait for. I've been working a compatable design compatable with the other designs on my website for about 2 weeks. Have the 3986 parts in hand, those tssops are small. Most all smds. Cant see a non soldermasked board as buildable. Layout is a bear!

[DR-Motion]

Nice chip, just finished a layout for the first prototype
with room left for a midband resonance compensation
circuit. My plan is to test the driver board separately
and switch in the compensation circuit afterward to save
on prototyping costs.
A hint for the curious... the compensation circuit consists
of a bandpass filter and a voltage controlled pwm 555.

more to follow,Gary

[BCwanderer]

Here is an update on a A3986 board
All components are regular thru hole except the A3986 TSSOP
I am gathering parts so my final may change depending on the parts I am able to buy
I use Target layout - I don't have eagle files
Any improvements or ideas appreciated.
FYI

Any other good ideas appreciated

[DR-Motion]

with your placement of the fets... it will be
difficult to use heatsinks.

regards

[BCwanderer]

with your placement of the fets... it will be
difficult to use heatsinks.

regards

I have a bag of clip-on heat sinks that slip on the TO220 chips
The clip on heat sinks are very versatile.

I would like to use the back to back FET setup like Gecko but the prototype board was set up the way it is to use the simplest and most direct connection from the Allegro chip and the FETs.

After I get the chip up and running, refinements will follow.
Cheers

[pminmo]

Among others, your FET current paths and sense resistor arrangement is a problem.

[BCwanderer]

Among others, your FET current paths and sense resistor arrangement is a problem.

Please explain your improvement ideas.
I would like to fix it on paper before I commit it to the boardmaker.
Thanks

[pminmo]

You have to take into consideration high current paths and where to sense the voltage developed across the sense resistor. Same for grounds, motor supply, clamping........... low impedence signal drives, noise immunity.

Right now I need to finish mine, it's laid out, passes all my rules, just need to spend some time reviewing everything.

[Azrael]

@DR Motion.
Can you please explain the things about midband resonance compensation, I would be glad.
Cheers

[markcomp77]

explain the things about midband resonance compensation

Here explanation is...
http://www.cnczone.com/forums/showpo...0&postcount=30

---

Here, starts of my projects



Add something else ?

[tmobile12345]

Add something else ?[/QUOTE]

I think, diodes missing on h-bridge for each mosfet.

[Anodizer]

This is the first time I have seen this ic. Looks very interesting.

In my experimenting in my shop at Virginia Tech Physics Department, I found it easier to make a bridge array on a separate board with leads and a plug to terminate them. This way, when I come up with a new idea, I already have
the bridges made up. Also, it allows me to use 2 ounce print and wide traces
to carry the current.

I am going to look up the lead specs to see if my QC7000 can mill a proto board.
I am currently designing a pan and tilt unit for a small laser for sensor calibration.

Fred:idea:

[markcomp77]

I think, diodes missing on h-bridge for each mosfet.

MOSFET have diode...
and
A3986 have SR - synchronous rectification mode....

...eliminates the need for additional schottky diodes