LF: Simple 4 channel analog to PWM IC. Does it exist?

[[email protected]]

Hi All,

I need to produce 4 PWM signals to input to 4 power MOSFETs that are
driving 4 DC motors. The MOSFET side of things is done/tested/working.
Currently I'm producing the PWM signals with a computer I/O interface,
however since I'm going with an embedded solution I want to give the
PWM generation to the hardware to take some burden off of the embedded
controller as it has other things to do.

I've looked at ICs like the Tl494 or the UC1824 for PWM generation,
but I'm not sure they are the best for motor control, and I've had a
really hard time finding some simple application notes on these 2 ICs.
Basically, I'm looking for a single IC solution that will allow me to
give 4 analog signals to the chip and will produce 4 unique PWM
signals with minimal external components. The PWM frequency is
flexible.

If a 4 channel Voltage-to-PWM IC is not available can anyone suggest
the simplest single channel Voltage-to-PWM IC that will require
minimal components for my application?

If anyone has any suggestions, please let me know.

Thanks,
wind-it-up

 

[Joerg]

Hi All,

I need to produce 4 PWM signals to input to 4 power MOSFETs that are
driving 4 DC motors. The MOSFET side of things is done/tested/working.
Currently I'm producing the PWM signals with a computer I/O interface,
however since I'm going with an embedded solution I want to give the
PWM generation to the hardware to take some burden off of the embedded
controller as it has other things to do.

I've looked at ICs like the Tl494 or the UC1824 for PWM generation,
but I'm not sure they are the best for motor control, and I've had a
really hard time finding some simple application notes on these 2 ICs.
Basically, I'm looking for a single IC solution that will allow me to
give 4 analog signals to the chip and will produce 4 unique PWM
signals with minimal external components. The PWM frequency is
flexible.

If a 4 channel Voltage-to-PWM IC is not available can anyone suggest
the simplest single channel Voltage-to-PWM IC that will require
minimal components for my application?

If anyone has any suggestions, please let me know.

Thanks,
wind-it-up

Easiest would probably be to spring for a little uC that does only PWM.
If it really has to be analog chips it'll become more expensive: You can
build that around a LM331 which is a precision V/F converter.

 

[John Barrett]

Hi All,

I need to produce 4 PWM signals to input to 4 power MOSFETs that are
driving 4 DC motors. The MOSFET side of things is done/tested/working.
Currently I'm producing the PWM signals with a computer I/O interface,
however since I'm going with an embedded solution I want to give the
PWM generation to the hardware to take some burden off of the embedded
controller as it has other things to do.

I've looked at ICs like the Tl494 or the UC1824 for PWM generation,
but I'm not sure they are the best for motor control, and I've had a
really hard time finding some simple application notes on these 2 ICs.
Basically, I'm looking for a single IC solution that will allow me to
give 4 analog signals to the chip and will produce 4 unique PWM
signals with minimal external components. The PWM frequency is
flexible.

If a 4 channel Voltage-to-PWM IC is not available can anyone suggest
the simplest single channel Voltage-to-PWM IC that will require
minimal components for my application?

If anyone has any suggestions, please let me know.

Thanks,
wind-it-up

Look at the Atmel AVR microcontrollers -- they have versions with up to 3
PWM outputs, and most all of them have A/D converters with 8 channel
multiplexed inputs.... or just use 4 of them -- I've used the Atmega168 for
PWM motor control and it had plenty of horepower for dealing with encoder
inputs, PWM generation, and a serial configuration interface... it comes in
an 18 pin dip and other packacging and is more than up to the single channel
task with minimal programming... there are probably 8 pin atmel or pic
controllers that can do the job if you want to hold down board space

RE: minimal components -- the Atmega168 has onboard system clock (1mhz if I
recall correctly.... external crystal up to 20mhz optional) so about the
only external components needed are power-on-reset and power supply
filtering, plus anything your specific application requires.

 

[[email protected]]

Easiest would probably be to spring for a little uC that does only PWM.
If it really has to be analog chips it'll become more expensive: You can
build that around a LM331 which is a precision V/F converter.

--
Regards, Joerg

http://www.analogconsultants.com- Hide quoted text -

- Show quoted text -

Thanks for the suggestions.

The main reason I was looking for an analog chip solution is to avoid
having to program a uC which requires additional interface hardware,
and time etc. I'm not worried about the expensive of using analog
chips so much as this project is a single prototype only.

*Do you have any suggestions for a small uC that would be easy to
program to generate 4 PWM from voltage?

*Can you offer any additional info on where to find application/design
notes on using an LM331 to generate a PWM signal? I searched a bit and
can't find much.

Thanks,
wind-it-up

 

[Joerg]

Thanks for the suggestions.

The main reason I was looking for an analog chip solution is to avoid
having to program a uC which requires additional interface hardware,
and time etc. I'm not worried about the expensive of using analog
chips so much as this project is a single prototype only.

*Do you have any suggestions for a small uC that would be easy to
program to generate 4 PWM from voltage?

Not really. I am only familiar with some 80C51 uC and some MSP430. The
larger MSP430 might have enough CCRs in them but the smaller ones only
contain 2-3 of them so that would not be enough for four PWMs unless you
play some software tricks.

*Can you offer any additional info on where to find application/design
notes on using an LM331 to generate a PWM signal? I searched a bit and
can't find much.

In my cases the duty cycle of the LM331 didn't matter, and the data
sheet is unfortunately quite silent about that. What you can do is hook
a one-shot to its output. Now you'd have a true PWM with great
linearity. You could also voltage-control a simple one-shot such as the
555 timer directly but then linearity goes to pots.

 

[linnix]

The main reason I was looking for an analog chip solution is to avoid
having to program a uC which requires additional interface hardware,
and time etc. I'm not worried about the expensive of using analog
chips so much as this project is a single prototype only.

Sometimes, you waste more time avoiding the right solution.

*Do you have any suggestions for a small uC that would be easy to
program to generate 4 PWM from voltage?

The lm3s828 is a 48pins TQFP with 6 PWM channels. It can also A2D
eight channels of 150,000 samples per second simultaneously. I know
you don't need all the capabilities. but there are rooms to grow. You
only need add a 6Mhz to 8MHz crystal for the uC.

Your function can be implemented in less than 50 lines of codes. For
example, this is alll you need to set up the auto sequenced A2D
interrupts.

ADCSequenceConfigure(ADC_BASE, 0, ADC_TRIGGER_TIMER, 0);
ADCSequenceStepConfigure(ADC_BASE, 0, 0, ADC_CTL_CH0 );
...
ADCSequenceStepConfigure(ADC_BASE, 0, 7, ADC_CTL_CH7 | DC_CTL_IE |
ADC_CTL_END);
ADCSequenceEnable(ADC_BASE, 0);
ADCIntEnable(ADC_BASE, 0);
IntEnable(INT_ADC0);

If you want it, I can post the complete solution.

 

[linnix]

OK, here is the code for a 4 channels Analog to PWM converter.
The maximum sample rate is 125,000 per second for four
channels. The chip is probably 90% idle waiting for interrupts.
Of course, you can use it for something else to kill the idle time.
One suggestion is to have a 920,000 baud serial link for debug and
status reports.

Preprogrammed chips are available for $10 (minimum 10) plus shipping.


ADCIntHandler(void)
{
int adc;

ADCIntClear(ADC_BASE, 0);
ADCSequenceDataGet( ADC_BASE, 0, &val);
PWMPulseWidthSet(PWM_BASE, PWM_OUT_0, adc);
ADCSequenceDataGet( ADC_BASE, 0, &val);
PWMPulseWidthSet(PWM_BASE, PWM_OUT_1, adc);
ADCSequenceDataGet( ADC_BASE, 0, &val);
PWMPulseWidthSet(PWM_BASE, PWM_OUT_2, adc);
ADCSequenceDataGet( ADC_BASE, 0, &val);
PWMPulseWidthSet(PWM_BASE, PWM_OUT_3, adc);
}

main()
{
// Divide 50 MHz system clock to 125,000 interrupts per second
TimerConfigure(TIMER1_BASE, TIMER_CFG_32_BIT_PER);
TimerLoadSet(TIMER1_BASE, TIMER_A, SysCtlClockGet() / 400);
TimerControlStall(TIMER1_BASE, TIMER_A, true);
TimerControlTrigger(TIMER1_BASE, TIMER_A, true);
TimerEnable(TIMER1_BASE, TIMER_A);

// Auto sequence A2D
ADCSequenceConfigure(ADC_BASE, 0, ADC_TRIGGER_TIMER, 0);
ADCSequenceStepConfigure(ADC_BASE, 0, 0, ADC_CTL_CH0 );
ADCSequenceStepConfigure(ADC_BASE, 0, 1, ADC_CTL_CH1 );
ADCSequenceStepConfigure(ADC_BASE, 0, 2, ADC_CTL_CH2 );
ADCSequenceStepConfigure(ADC_BASE, 0, 3, ADC_CTL_CH3 | ADC_CTL_IE
| ADC_CTL_END);
ADCSequenceEnable(ADC_BASE, 0);
ADCIntEnable(ADC_BASE, 0);
IntEnable(INT_ADC0);

// wait for interrupt
while(1)
;
}

 

[[email protected]]

Hey,

Thanks everyone for all of the reply's, and thanks Linnix for posting
that code.

I still haven't decided on what to go with yet. I may just go with an
embedded solution that is powerful enough to do these 4 PWM channels,
and all other A/D D/A and computations that I need it to do.

Thanks again for the suggestions.
wind-it-up

 

[whit3rd]

I need to produce 4 PWM signals to input to 4 power MOSFETs that are
driving 4 DC motors.

Why not just make a sawtooth generator (like with a '555) and
use four comparators (one LM339)?