ED

If I was using a micro controller (which I'm not) a pulse would probably do. But a steady voltage to a comparator will be used to keep SW3 in my schematic turned off during the on time sequence of events, to keep the spark ignition voltage from reverse biasing the capacitor discharge into the gap. Or at least that's what my thinking is.

Just to be clear, we are talking about sensing the current at the ground node to determine if the spark has started, not voltage as you've stated. So we need to look at how current can be measured. My previous suggestion was to use an inductive pickup for this purpose, and you correctly determined that would only sense the beginning (and end) of the flashover. Your diodes actually do provide an indication of the current in the ground leg per the equation Vd = Vt*ln(Id/Is) where Vt = 25mv at root temp and Is is a parameter of the diodes, typically not specified and usually measured. This equation will vary greatly over temperature. I can think of two other ways to measure current. One way is using a shunt as shown in some of Max's posts. Alternatively, the shunt can be just a low value resistor. This would be simple and probably give acceptable results. The other way is to use a hall effect device. Keep in mind the current will be very low, and so whatever method used needs to have good resolution down to the mA range.

I really don't know anything about EDM and so I'm out of my league. But this is just my 2c.

 

ED



Just to be clear, we are talking about sensing the current at the ground node to determine if the spark has started, not voltage as you've stated. So we need to look at how current can be measured. My previous suggestion was to use an inductive pickup for this purpose, and you correctly determined that would only sense the beginning (and end) of the flashover. Your diodes actually do provide an indication of the current in the ground leg per the equation Vd = Vt*ln(Id/Is) where Vt = 25mv at root temp and Is is a parameter of the diodes, typically not specified and usually measured. This equation will vary greatly over temperature. I can think of two other ways to measure current. One way is using a shunt as shown in some of Max's posts. Alternatively, the shunt can be just a low value resistor. This would be simple and probably give acceptable results. The other way is to use a hall effect device. Keep in mind the current will be very low, and so whatever method used needs to have good resolution down to the mA range.

I really don't know anything about EDM and so I'm out of my league. But this is just my 2c.

Now I see what is bothering you.
He's not trying to measure current only to detect that the arc has started. Something else is needed to measure the current.

Shortbus, help me out here. I'm still trying to determine current and voltage. I have read that the arc resistance is about 1 ohm. That would suggest a current, at least initially of 100 amps. So I'm guessing there are some big resistors in series that limit the maximum current to about 25 or 30 amps so the cutting voltage would be 25 or 30 volts. So maybe a 25 ohm resistor. I'm also guessing that you can set the voltage the caps are charged to? Then if you want higher current you can switch in a bigger cap and make the resistor smaller?? That way the voltage stays pretty constant like your pictures?? Then the only thing you need to worry about is keeping the repletion rate low enough that your power supply can charge the cap back up?

 

Now I see what is bothering you.
He's not trying to measure current only to detect that the arc has started. Something else is needed to measure the current.

What I'm saying is that by sensing current, he will know when the arc has started and when it ends. That's the only way I know of to tell outside some kind of optical sensor.

 

Before the arc there will be zero volts across the diodes. As the arc begins the voltage will go up to 1.4 volts. He can then look at the voltage with a comparator set at .7 volts and know the arc started. He could just sense current, but it might be slower or more expensive, but could be done.

 

Now I see what is bothering you.
He's not trying to measure current only to detect that the arc has started. Something else is needed to measure the current.

Shortbus, help me out here. I'm still trying to determine current and voltage. I have read that the arc resistance is about 1 ohm. That would suggest a current, at least initially of 100 amps. So I'm guessing there are some big resistors in series that limit the maximum current to about 25 or 30 amps so the cutting voltage would be 25 or 30 volts. So maybe a 25 ohm resistor. I'm also guessing that you can set the voltage the caps are charged to? Then if you want higher current you can switch in a bigger cap and make the resistor smaller?? That way the voltage stays pretty constant like your pictures?? Then the only thing you need to worry about is keeping the repletion rate low enough that your power supply can charge the cap back up?

Your correct on the arc resistance, I've seen both 1 and 1.5 ohms as gap resistance. There will be a comparator on the cap bank that will be keeping track of voltage on it. There will be a wire wound (corn cob?) power resistor to limit the current draw on the transformer, and eventually there will be extra resistors that can be switched to set current levels. But was going to wait on that until I can get it running and try controlling by switching different caps on the bank. The supply cap on the rectifier side is really big, 47,000uF if I remember correctly.

The normal Ohm's law doesn't seem to work well in EDM. Think it has to do with the nature of the way the energy is flowing in the gap. The gap itself is complex, not a pure resistance, it is a mix of resistance and capacitance. And the transfer is more of a plasma/electron flow. No physical contact between the work and electrode. Since all of the energy is coming from the caps, the voltage will drop fast to the gap working voltage of 25 to 35 volts. A TIG or stick welder does the same thing, fairly high open circuit voltage, quickly dropping to a low voltage high amperage arc.

I know it doesn't make sense with all normal electronics theory applied, but there are many of these out there that are working, with just an RC oscillator. A big resistor charging a cap to the flash over of the gap, and starting again. The not conforming to normal rules and laws of electronics is why, until this thread I've gotten little help with this.

 

ED



Just to be clear, we are talking about sensing the current at the ground node to determine if the spark has started, not voltage as you've stated. So we need to look at how current can be measured. My previous suggestion was to use an inductive pickup for this purpose, and you correctly determined that would only sense the beginning (and end) of the flashover. Your diodes actually do provide an indication of the current in the ground leg per the equation Vd = Vt*ln(Id/Is) where Vt = 25mv at root temp and Is is a parameter of the diodes, typically not specified and usually measured. This equation will vary greatly over temperature. I can think of two other ways to measure current. One way is using a shunt as shown in some of Max's posts. Alternatively, the shunt can be just a low value resistor. This would be simple and probably give acceptable results. The other way is to use a hall effect device. Keep in mind the current will be very low, and so whatever method used needs to have good resolution down to the mA range.

I really don't know anything about EDM and so I'm out of my league. But this is just my 2c.

I really don't care about what volts or current at that particular point. Just trying to detect that an arc is there, to shut off the spark ignition voltage and allow the start of an on time timer and the flow of energy from the cap bank. Since the ignition voltage is higher, it reverse biases the diode on the cap bank to the gap, shutting it off allows the cap energy into the gap.

I'm doing this to make sure that a spark starts, then lasts the same amount of on time, each sequence. Without this, some sparks may be delayed then the working(on time) of each spark varies. With this the higher ignition voltage should make sure a spark starts and then it will still have the cap energy there when it does. The commercial machines do this but none of the DIY machines do.

 

Incidentally what are you using for your servo system and electrode material, copper or graphite?
Max.

My ram/servo is a stepper motor driving a ball screw. The ram is built using a modified PHD air slide actuator. It has Thompson rails and bearings. The air cylinder I replaced with the stepper and ball screw. Real smooth and solid. The machine is an old Enco drill mill that I took the head off of and made brackets to mount the ram assemble on. The work tank started life as a electric control panel that I cut down to suit the mill table size. The pump for the dielectric is originally for a parts washer tank.

Depending on what I'm burning the electrodes will be Poco graphite. For burning out taps I'll use K-S brass tubing of the correct size.

https://www.phdinc.com/product/?product=linear-slides&series=sg

 

A small coil and a diode detector will give you a negative or positive pulse on start of spark.

No connection or modification to current mechanism necessary.

But no other EDM in area.

 

I think I got it.
So see how this sounds.
Charge working cap to 70 volts. Allows for some sag between cycles from the line. Some current limiting required, but not much since you want it fast.
Always charge boost cap. Limit its current to some fairly low value.
Discharge both supplies to the work limiting current.
Once arc is sensed turn off boost supply.
If no arc is sensed in 150 usec. light open arc led and move in. Repeat.
If short is detected shut off both and move out. Repeat.
If arc is sensed, start timer (adjustable?)
Delay 15 - 20 usec. and measure voltage.
To high - move in, to low, move out.
After time out turn off work supply and recharge working cap.
Allow some time if move is required.
Rinse and repeat.

Do you already have the big cap and resistors?
This would like to be a micro, but it can be done with hardware without to much "stuff". Noise may be the biggest stumbling block.

 

I really don't care about what volts or current at that particular point. Just trying to detect that an arc is there, to shut off the spark ignition voltage and allow the start of an on time timer and the flow of energy from the cap bank.

I've tried numerous times to explain that's exactly what sensing current will do for you. But I won't continue to beat a horse that won't run. Good luck.

 

I think I got it.
Ron, I'll try to answer these by putting your questions in quotes, and then my answer.

"Charge working cap to 70 volts. Allows for some sag between cycles from the line. Some current limiting required, but not much since you want it fast." I was going to control charge voltage with a comparator on the cap bank. Shutting off charge switch at 90V.

"Always charge boost cap. Limit its current to some fairly low value." If I am understanding this correctly, wouldn't this then be a RC relaxation circuit?

"Discharge both supplies to the work limiting current. Once arc is sensed turn off boost supply." I had asked about this in a previous thread. And was told that the higher voltage would dominate the lower, not add. because they are two separate supplies. They would need to be in series for what your saying to work, wouldn't they?

"If arc is sensed, start timer (adjustable?)" Yes the timer will be adjustable.

"Do you already have the big cap and resistors?" Yes, I have three of the caps and numerous and varied values of 250W and higher resistors.

" This would like to be a micro, but it can be done with hardware without to much "stuff". Noise may be the biggest stumbling block" Some in the hobby EDM community have attempted using a micro, but like you said the "noise" and RF has been a big stumbling block. Don't know how the commercial machines get around it. Plus I'm struggling with just this part, don't know if I have enough lifetime left to also learn micro's and coding. I am going to try with just comparators and logic gates to control the gate drives of mosfets.

You do have the ram electrode movements with regard to gap voltages correct. But the only moves are strictly the ones controlling the gap voltage/distance. The off/recharge time is to recharge the cap(s) and allow the dielectric oil to reform from a gas to liquid and flush the debris out of the gap.

 

I've tried numerous times to explain that's exactly what sensing current will do for you. But I won't continue to beat a horse that won't run. Good luck.

I don't know this stuff very well, not trying to disrespect you at all. I'm not sure where I would find a 0.001 ohm ~500W resistor to measure current though. But diodes for the voltage and amperage are common and a voltage comparator I understand. Again, no disrespect meant, just an old dumb guy in over his head trying to realize a dream.

 

I don't know this stuff very well, not trying to disrespect you at all. I'm not sure where I would find a 0.001 ohm ~500W resistor to measure current though. .

500W would be extreme overkill. .001ohm, 2W would be more than enough.

 

You won't be able to get the full 90 volts all the time because the transformer only recharges the big caps every 8 ms. So in between these peaks the voltage will sag. How much depends on the caps and the transformer. Maybe you can nail down the caps and measure the primary and secondary resistance of the transformer. It may be hard to measure the transformer as the resistance will be low, but you can try.

The boost and working supply would work just like you have them drawn. As soon as the arc starts the boost can be turned off. Yes, the boost voltage will drop quite a bit, but the arc is already going so I don't think it matters.

What is the interface to the stepper. Can this circuit just supply move in or move out signals to it or does it need something more?

Getting close now.

 

You won't be able to get the full 90 volts all the time because the transformer only recharges the big caps every 8 ms. So in between these peaks the voltage will sag. How much depends on the caps and the transformer. Maybe you can nail down the caps and measure the primary and secondary resistance of the transformer. It may be hard to measure the transformer as the resistance will be low, but you can try.

The boost and working supply would work just like you have them drawn. As soon as the arc starts the boost can be turned off. Yes, the boost voltage will drop quite a bit, but the arc is already going so I don't think it matters.

What is the interface to the stepper. Can this circuit just supply move in or move out signals to it or does it need something more?

Getting close now.

Thanks again Ron!. By measure the primary and secondary resistance you do mean just measure leads with a DMM? Can and will do that. And will get one of the caps out and make sure what the value is, they are about the size of a beer/soda can. Iwas thinking that a 5V drop would be within reason, the DCV output is ~95V.

I thought you meant that the working and boost would combine, sorry. That other tread a while back is what lead me to this idea.

The stepper inter face is through an adjustable window comparator. With a 5V window opening. Don't know about my terms. By the 5V window I mean, if voltage is higher than 30V electrode moves closer. If below 25V it moves away. And I think I need a small value cap between comparator and stepper driver. To keep the electrode from backing away during the off time, when voltage drops due to no spark. But haven't done the calculations for it yet.

Thanks to you, I may just get this working!!

 

500W would be extreme overkill. .001ohm, 2W would be more than enough.

Like I said I'm pretty dumb about this. Just figured the shunt resistor would be the circuit wattage. Working circuit levels ; 35V * 15A = 525W

 

Like I said I'm pretty dumb about this. Just figured the shunt resistor would be the circuit wattage. Working circuit levels ; 35V * 15A = 525W

That's why you always need to do the math. P = I^2*R = 15^2*.001 = .225W or 1/4W.

 

That's why you always need to do the math. P = I^2*R = 15^2*.001 = .225W or 1/4W.

I think it might be pretty hard to dig a 10 mv signal out when you just put 30 or 40 amps into the ground and a 100 volt arc is above your head.

Shortbus, Yes try it with your meter if it has a good low scale. If not time will tell us.

I think the way to do the window is just to check it say 20 usec after the arc starts.

Do you already have some of this built?

 

I think it might be pretty hard to dig a 10 mv signal out when you just put 30 or 40 amps into the ground and a 100 volt arc is above your head.

Use techniques that eliminate the ground as a reference I.E. differential, shielded amplifier. Also, at 30 - 40A, that would be 30 - 40mV, and not 10mV.

BTW, each diode drops .6 - .7V and dissipates 9 - 10.5W. Use appropriate power diodes.

 

That's why you always need to do the math. P = I^2*R = 15^2*.001 = .225W or 1/4W.

See, told you I was dumb! I always use the V*A = W, only one shown usually with an Ohms law chart.