PWM output as a signal to input of another thing

Ya, a filter to remove the carrier frequency.

If you don't need much speed, a few RC's cascaded can remove a very good slice of it indeed.

If you want to get high-falootin', you could use like ten or twenty pole LC filter and get like -100dB at the carrier while preserving the modulated signal. Something tells me you don't need quite that though ;-)

...What's wrong with on-or-off though?

Tim

-- Deep Fryer: a very philosophical monk. Website:

A little preface, my SCR firing system is controlled by input voltage.

Right now, they are controlled by simple pots (works fine for arc welding).

I will keep that, for my first try with the microcontroller. That means as I switch from one arc process to another, I would still have to adjust voltage and current with pots, even though it is redundant. (for instance, for arc welding I know I need 80V voltage, for MIG welding I need max current, IOW, sometimes voltage control or current controls are fixed and not inputted). That's redundant, but I want to have a working system.

After a while, I want to change this so that the microcontroller sends signals to the SCR firing board telling it what is its desired voltage and current. (possibly I would then switch to keypad entry for relevant parameters). But that would be later.

That's why I want to find out how to send PWM modulated signals, to control the SCR firing board with a micro.

If I get my "alpha version" control panel working, I may be able to build in the microcontroller into the welder this weekend.

I know that I am wasting time moving from one prototype to another, but it is a better alternative to trying to create the perfect system from the first time.

i

Yes, all I want is to send voltage to the SCR firing board, to tell the board what is my desired voltage level.

Well, I may not be replying to what you were trying to say, but if I understand what you said correctly, I shoudl say that PWM frequency is set by a PWM command, and then the PWM would run regardless of any timing issues of the microcontroller.

I am not trying to output pulses for SCR gates -- I am trying to output signal for the SCR firing board to say what is my desired voltage.

My SCR firing board is a regulated power supply board, but it needs to be told what is the current and voltage I want.

ui

--
LOL, Have you not been reading?

He has a microntroller he can use to generate a PWM signal.

He has a power supply and can control its output voltage and current
with DC input control voltages.

He wants to know how to change the PWM pulses into the DC control
voltages that he can use to control the output voltage and current
of his supply.

IOW, he wants to do something like this:

PWM>---[R]--+--->DC 
            |
           [C]
            |
GND>--------+--->GND

They're called 'low pass filters', and a single RC is the simplest kind.

You need to know how fast you want to be able to change the signal to the SCR board, how fast your PWM will go, and how much ripple you can tolerate -- with these three numbers you'll be able to decide how complex a filter you need, if indeed a filter will work for you at all.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Posting from Google?  See http://cfaj.freeshell.org/google/

"Applied Control Theory for Embedded Systems" came out in April.
See details at http://www.wescottdesign.com/actfes/actfes.html

SCR's need to be fired in phase with the supply to control the output voltage. you really shouldn't be turning on an SCR at random assuming that the source to the SCR's are AC ? so i can only assume that what your looking for is a form of variable output voltage ? in which case you would need to have the MPU monitor the AC line for timing.. to stay in sync with in a 3 hz plus and minus, also you need to have the MPU keep the timing free running for a few cycles at the last known accurate lock frequency in case there is noise on the line that causes a sudden timing shift which you do not want your MPU to to adjust to. if the signal remains non-syncable after about 3 or 4 cycles the MPU should stop pulsing the SCR's the output pulse to the SCR's is only a short and narrow with via a optical coupling or Pulse xformer. the key is timing!. using the timing method as i stated above, you can decide at what moment the scr should be turned as the sin drops near the zero point. as you most likely already know, SCR's latch on as long as they are in their holding current window. when the supply voltage gets below the effected holding current the scr will turn off. if your trying to make a regulated supply on the output which is what i think you should do, then use an ADC input on the MPU to monitor a filtered output of the scr's and this can help make timing assessments when the pulse is needed.

--
Real Programmers Do things like this.
http://webpages.charter.net/jamie_5

Yes, that's what I want. Any pointers to how to pick R and C. I can pick frequency, I suppose the higher, the merrier.

i

I am not even sure. The only thing I know is that the changes would not need to occur very fast.

i

Oh OK, so it's a variable thing.

Hmm, well that gives a clue to the response time then I guess- within 10ms or so (about a line cycle, maybe sooner for triphase switching) would probably be the fastest useful time, since the board can't very well turn off the SCRs mid-cycle.

So a cutoff frequency around 100Hz wouldn't be untoward. That should pretty well clean up the PWM, depending on how high it is (10kHz? 100k? Clock (1M+)?).

Tim

--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms

I will double check the frequency, I think that it is about 100k max.

i

Then use John Fields's filter circuit; for component values, just figure out which values will give a steady enough DC to not confuse the regulator; that will depend on the fundamental frequency of the PWM clock.

Good Luck! Rich

Thanks Rich, I will do just what you (and John) say. I will also verify output on the oscilloscope.

i

Are you a hacker? What would happen if you just bypassed the SCR supply's regulating/firing circuitry, and drove the gates right from the PWM? (through optos, of course.) You could have a "start PWM pulse" signal derived from the line, and just fire the SCRs at an appropriate time. I did this once with a 68HC11, for a 24V 40A voltage and current limited or regulated lead-acid battery charger. I even wrote in a little emergency shutdown if either the voltage or current went off-scale on any sample - my code just refused to fire the SCR any more. It's kind of kewl, seeing a

24V, 40A charger just stop when its output is shorted. Didn't even blow the fuse! :-)

Good Luck! Rich

I am not that kind of a hacker, and I am not sure what is the point.

I have a really nice SCR driver, why wuold I want to make some sort of a kludge?

That's cool and I will have that sort of protection in software. I already have some of it.

i

"Wasting Time"???!?!?!?!?!!!?? Dood, you're learning by doing! Even I've been learning from you. Sorry about pooh-poohing you like I did when you first showed up - you clearly have more amtition/energy than I have. ;-)

Then it should be trivially simple to pick component values - the period of a 100KHz signal is, lessee - 10 microseconds per cycle. (1/100000). So, pick a time constant that's more than, say, 25 us, and less than, say, a millisecond. That should give you a lot of leeway - the frequency you need to filter is a thousand times higher than what you're trying to control; I'd pick a resistor based on imput and output impedances and stuff.

Good Luck! Rich

Thank you, I doubt that I ever said anything worthwhile, but yes, I am learning as I go. (which is a very effective form of learning).

i

Yes, Rich, it ought to be easy.

I will try to do it tonight.

For the first try, I will direct connect the pots controlling voltage and current (mounted on the control board) to the SCR firing system. I feel that it is more idiot proof.

However, I want to wire the board so that it does have those PWM outputs, available to the SCR firing system, for later reconnection.

i