By tying the charge to the PWM it gives the problem you just mentioned. The way my circuit is, the charge is actually taking place longer than one clock period. This way the caps are sure to be fully charged.

Also having the flip-flop plus 'and' gate was the only way I found of having both alternating discharges and full control over the other elements I needed. By other I mean the high and low cap voltage comparators.

I read an App Note on the UC3842/43. It shows that the osc pulse goes low as the PWM output goes high. I'll add the app note to this post. Page 11 shows the timing. It also shows using a 555 for the clock, instead of a RC .

 

The way my circuit is, the charge is actually taking place longer than one clock period. This way the caps are sure to be fully charged.

Why do they need to be charged in more than one clock cycle? If the charge is under 90V (90V is almost charged, right) you disable the discharge circuit anyway... So why not charge them fast?

 

I didn't (know how) to calculate charging times. You are correct, as long as the caps are charged enough to turn on the comparator that would be enough. The comparators are an important part of the whole system though. The total amount of capacitance with all of them switched on (for a rough cut) would be 26uF.

Would you show me how to figure charging time calculation? I know I'm asking a lot of you.

 

You can use this nice calculator if you want to know after what time the capacitor voltage will be a certain value with a resistor X in series.
http://www.csgnetwork.com/rctimecalc2.html

It's after about 2x the time constant that you arrive at 90% (90V in your case).

Apparently I didn't think this through before writing my other post.
At 50kHz and 90% PWM duty cycle we would only have 2us left to charge the cap.
In order to charge 26uF from 0 to 90V from a 100V source in 2us we would need a resistor of 0.034Ohm in series. That would mean initially 2800Amps. Not what you want.

But then I see that the capacitor cannot really be discharged to 0V... At least not at 50kHz...

 

Thank you for the calculator link! I'll have a play with it to do some scenarios of different cap values and time.

The caps should never discharge to 0V. The lower comparator would shut down discharge at 10V.

When using the full 26uF the frequency would be closer to the 5kHz level and the PWM would be near 50%. Let me do some playing with the capacitance/time calculator.

There are so many things involved with the working of an EDM like this it is hard to give all of the scenarios and other things. Its really more of an art than a science. The newest generations of them are all CNC, all of the controls for the sparks are done by a self learning system, from the first couple sparks in a cut. All feed-backed closed loop. What I'm trying is really old technology second generation stuff. But theres not much free information out there on it.

 

I'm still working with calculations. Life has gotten in the way again, d@mn lawn mower!

 

Although this is a different calculator that you linked to, I'm having the same problem that I had a couple of years ago when working on this project. The calc says, not going to work. But I actually ran this type of machine in real life.

The only difference in the ones at work was it had only one capacitor bank. What I'm trying to do should make it easier to do. By using two cap banks, the charge time on the caps should be double.

Going to step away and look at this again. Thanks for all of the help you guys have given, I have learned a lot and that's a good thing.

 

It would help if more parameters were known. For example how much does the bank discharge each cycle? I never heard about this machine before I read this thread.

I suppose the caps do not discharge very much each cycle because then there would be no time to recharge them in one cycle. But if it goes down to 10% (where the PWM stops) how could you ever recharge it in a few us?

 

The 10V was just a safe point in the circuit. To prevent a short in the electrode to work gap. The actual gap voltage and servo voltage (to keep the gap distance) is in the 20-25V area. The 10V was to shut down the pulse in case of a failure or lag in servo movement.

Is there a difference in the characteristics of a cap when used in a more industrial context instead of a 'electronic' context?

 

@Praondevou, is the calculator you linked to the time for one RC period or does it need to be multiplied by 5?

Also what is the instantaneous voltage in the calculator? Does that mean the possible voltage present on the cap at the start of charging?

 

@Praondevou, is the calculator you linked to the time for one RC period or does it need to be multiplied by 5?

Also what is the instantaneous voltage in the calculator? Does that mean the possible voltage present on the cap at the start of charging?

The calculator shows you the voltage at any given time when charging a capacitor through a resistor from a voltage source you define.

Values you have to put in are:
- voltage of the supply
- resistance
- capacitance
- time

The result will be the voltage to which the capacitor is charged in that time.
EDIT: I just saw that it calculates whatever value is missing. Just leave the one you need to calculate blank and fill in values for the other 4.

Is there a difference in the characteristics of a cap when used in a more industrial context instead of a 'electronic' context?

I'm not sure I understand the question. If a capacitor is suitable depends on the applications requirements and the capacitors characteristics. Depending on the application the capacitor ripple current, frequency, ESR etc may have more or less impact.

 

The calculator shows you the voltage at any given time when charging a capacitor through a resistor from a voltage source you define.

Values you have to put in are:
- voltage of the supply
- resistance
- capacitance
- time

The result will be the voltage to which the capacitor is charged in that time.
EDIT: I just saw that it calculates whatever value is missing. Just leave the one you need to calculate blank and fill in values for the other 4.

Thank you. I was using it wrong.

Is there a way to figure out the time to charge when there is an initial voltage on the cap? Say starting at 20V and charging to 90V. Or is that what the instantaneous voltage is used for?

 

I'm not sure I understand the question. If a capacitor is suitable depends on the applications requirements and the capacitors characteristics. Depending on the application the capacitor ripple current, frequency, ESR etc may have more or less impact.

I meant like the pulse power supplies for things like a cutting laser or an inverter style TIG welder. They are operating at comparable frequencies and don't seem to have problem with recharge times.

I thought maybe at higher voltages and amperage caps worked different.

 

Thank you. I was using it wrong.

Is there a way to figure out the time to charge when there is an initial voltage on the cap? Say starting at 20V and charging to 90V. Or is that what the instantaneous voltage is used for?

You could calculate the time it takes until it's at 20V and then the time it takes to get to 90V. Then just use the difference between both values.