AVR-Chat

>
> +AVRDrive +5V +12V +5V
> -------. | | |
> | | | |
> R10k | R1k COIL
> | |¯¯¯¯¯| | |
> 300mV o-----(+)| | | |--' IRF530
> | | LM |---o---|--. IRL530N
> R620 | 339 | |<-| IRLZ24
> | .-(-)| | | w/clamp diode
> _|_ | |_____| |
> | |
> '---------------------o
> 300mV |
> R0.05
> ; |
> _|_
>
> R620 could be a trimpot adjusting full span current of 6A
> over R0.05.
> One LM339 houses 4 comparators, so one chip per unipolar
> motor (4 phases & FETs).
>
> /* So you're going to have 4 trimpots? */

Not really, the current limit doesn't need to be that precise. Limiting at 5.9A or 6.1A will not change that much in safety for the transistor or the motor, not even for the power supply. The trimpot idea is just in case to find out the better R620 value. Sometimes the problem is not with the voltage divider R10k/R620, but mostly with the R0.05 part and connections. A not so good soldering or terminal oxidation can change easily those tens of milliohm and that will be a problem. Those expected 300mV can change drastically to 400mV with a simple increase in 20 milli ohm in the shunt resistor. Then, beign prepared for the worst, I guess a R1k small trimpot in place of the R620 is not that bad idea. So, answering, yes, probably will have 4 trimpots per motor.

>
> Any known dedicated IC to substitute the LM339 with
> advantages?
>
> /* LM339 is OK. This might also work and you only need one
> trimpot and fewer resistors:
> AVRdrive must be high at the same time the comparator is
> turned OFF for the FET to turn on.
>
> +AVRDrive
> |
> +5V | +5V +5V
> | | | |
> | | | |
> | 1k0 COIL 10k
> |¯¯¯¯¯| | | |
> Vref-----(+)| | | |--' IRF530 |
> | LM |---o--120R--|--. IRL530N o---Vref
> ; | 339 | |<-| IRLZ24 | (4
> places_
> +-(-)| | | |
> | |_____| | |
> | ; | 1k0<-----o
> '--------------------------o ; | |
> | | |
> R0.05 _|_
> _|_
> |
> _|_
> */

Yes, the above is an idea.

I used something similar with the Alegro 2398 part, a common Vref supplied by a trimpot and 339 comparing and feeding the enable/disable pin of the 2398s, the AVR drive comes via the 339 releasing the feedback signal. Implementing the idea in this circuit, to block the motor, the AVR simply overflows the feedback with 5V (4.3V) and it cuts completely the FET.

+5V +5V +12V +5V
| | | |
| | | |
R10k | R1k COIL
| |¯¯¯¯¯| | |
0.3VREF o-----(+)| | | |--' IRF530
Common | | LM |---o-R120-|<-. IRL530N
>R1k | 339 | |--| IRLZ24
| .-(-)| | | w/clamp diode
_|_ | |_____| |
| |
(XA)o---------R10k-----------o
; _|_ ; 300mV |
A 1N4148 |
-AVRDrive | R0.05

o----------' |
_|_

Here, the LM339 will cut the FET whenever it feeds more than 6A to the COIL, or when the AVR supplies a positive level to the (-) pin via de diode. The only bad part here is to ensure that at power on, before the AVR resets and assert the Drive port pins, the motors stay not running erratic. One way to do that would include an extra "NotErratic" circuit, that will be connected to the (X) point at the above circuit.

Perhaps a general "+MotorHold" circuit like that:

+5V

|

|

R2k2

| 1N4148

o---->|---o Phase XA

|

o---->|---o Phase XB

|

|

o----'

+AVR_MotorHold

The problem is related to the fact that during RESET, the AVR takes its port pins to high impedance state, and that will not take XA point to +5V to ensure motors holding. They can run erratic. They will not run erratic if all phases will suck power at the same time, but anyway it is not desireable to have 24A being pushed into each motor for few moments. What about if the AVR got stuck during reset? Will have a bad power supply soon enough.

The +AVR_MotorHold above applies aprox 3.4V to XA (XB, XC and XD) whenever the AVR is offline (port pins in high impedance). It means the FETs will be cut as long as the AVR does not supply a negative level to the +AVR_MotorHold. As soon the AVR feels it is safe to release motor operation, just apply a negative level to +AVR+MotorHold, and everything will be ok.

That means, each motor will require:

4 x IRLZ44 or similar FET units

4 x 1N4007 (clamp diode)

1 x LM339

8 x 1N4148

1 x R1k

1 x R120

1 x R0.05

1 x R10k

The common VREF being made with;

1 x R10k

1 x R1k Trimpot

Or an extra 1N4148 and a R divider to

ensure better VREF stability.

You see, initialy it seems to be few components, but then, when thinking about everything (and something will always be missing) the coponents population increases significantly.

This is why I thought to use some ready made FET gate driver circuit with current limit and everything else, like enable, input, pwm counter, etc.

Wagner.