EDM - Electrical discharge machine

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
@AnalogKid , @dl324 , @Gophert(when you're allowed back), thank all of you for the help so far getting me to where I am. And anyone else that wants to join in.

I'm building a circuit for an EDM. https://en.wikipedia.org/wiki/Electrical_discharge_machining What is called a sinker EDM. There is another form called a wire EDM, but a sinker is what I'm building. After running them when still working I'm trying to come up with one that will work like a brand name one called an Eltee Pulsatron. I ran many different machines from old, old ones using vacuum tubes to the newest style using CNC controls. And the Eltee was the easiest to use. It was self regulating, by just switching different values of capacitors on the controls it set it's own frequency of sparking. The sparking between an electrode and the work piece is what cuts or burns the desired shape in any metal or conductive material, no matter how hard or soft.

Never being able to find a schematic for the controls in the machine I started to learn about electronics, and am still learning, thanks mostly to the members on this site and a few others. There are a few DIY versions of these machines and I have bought most of the plans I could find over the years. But they are all lacking in how the controls work. Like most of the older machines I ran they have too many parameters in the controls that need to be set by an operator. And most people that want a DIY machine have never ran a real one. So they buy a set of plans and spend money to build one, only to have it not do what they expect it to do. Not because the plans a really flawed, but because the controls are so hard to set to get it to work. I've been working for many years to get to the point I'm at, and hope this will be one that works.

What I'm doing is what I've seen called, "step by step" logic. No clock as such involved in the logic. Many will say it's unconventional use of logic, and to most standards it is. But it looks to my self taught eyes to do what is needed. It uses voltages in a few parts of the circuit, checked by comparator's, to toggle on and off mosfets to control the output to the electrode. By switching different capacitors in or out of the power circuit, the time and frequency of the discharge is automatically set to suit. Instead of the operator hoping they get it correct. Other than the capacitor switches there is only two other controls, that are not as important to the operation, gap voltage and ram servo speed. Those are not as precise to the machining process and listening to the "sizzle" of the sparking will tell you when it's close to optimal on those controls.

So after all of that explanation, I'd like you to look at the schematics. I'll explain step by step how I see it working, so ask any question you have. This is my first design of anything like this so tell me where it looks wrong, if you will please. Thanks.edm pulse power139.jpg edm pulse logic138.jpg
 

AnalogKid

Joined Aug 1, 2013
10,987
Since the two schematics are part of the same circuit, strongly consider a single set of reference designators across both schematics.

I see 8 or 9 diodes. Part numbers?

The risetime of a 393 output is slower than that of a 40106. Also, the positive-going input transition level of a 40106 os higher than that of a 4071. These combine such that Ull will respond to its B input before U13 will respond to its A input.

What is the purpose of D1 and D2 in the 2nd schematic.

Lower schematic D4 cathode needs a large value resistor to GND. Without some current through the diode you will not get the 0.6 V voltage window.

Please post larger images. The site doesn't like large files, but your schematics are only around 90Kb so crank them up by 50%-100% to make them more readable.

ak
 

LesJones

Joined Jan 8, 2017
4,174
I would have thought that you would need a low value resistor to limit the capacitor charging current through Q1 to within it's peak current rating. Also I would think the same protection would be required Q4 to limit the peak spark current. Am I correct in assuming that Q2 and Q3 are only required so the bootstrapping in the driver ICs works correctly. I will watch this thread with interest as I have been to mean to buy Ben Fleming's pulse EDM book.

Les.
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
Since the two schematics are part of the same circuit, strongly consider a single set of reference designators across both schematics.
I thought I did that? The outputs from the first drawing are letters, that correspond with the letters on the inputs of the second drawing. And the outputs of the second drawing are numbers that go to the inputs of the first drawing. How do you like them to be? Like I've said before I'm totally a novice at this.


I see 8 or 9 diodes. Part numbers?
The schematic is just a "proof of concept" drawing. It was done with DipTrace and most things are just generic parts. This is so hard for me(no real training and 70years old), that I didn't want to spend the time to do any real part values in it. Since there is no available information on one of these machines out there some of it will need to be built and voltages measured to get real values.

What is the purpose of D1 and D2 in the 2nd schematic.
I got that from Lancaster's "Cmos Cookbook". The diodes and the resistors, R1 and R4, are to help "square up" the pulses, or so it says in the book.

Lower schematic D4 cathode needs a large value resistor to GND. Without some current through the diode you will not get the 0.6 V voltage window.
OK I thought the input resistance of the comparator would be enough to take care of that. The idea behind that diode is to give a ~5V window in the comparators. This is one of the things that is know in the scheme of things on how these machines work. Also one of the problems with most of the existing DIY builds. Most use two pots that need to be adjusted to give the 5V window, and I came up with this. The ~100V from the cap bank will be scaled in the voltage divider to give the voltage drop over diode D4 a ~5V equivalent.

Please post larger images. The site doesn't like large files, but your schematics are only around 90Kb so crank them up by 50%-100% to make them more readable.
I tried to make them bigger but guess I made them too small. Most other forums want the at 600 x 800 and that's what these were. I'll try to make them larger and repost. What size do you suggest?

The risetime of a 393 output is slower than that of a 40106. Also, the positive-going input transition level of a 40106 os higher than that of a 4071. These combine such that Ull will respond to its B input before U13 will respond to its A input.
Not sure I understand that? Being an 'OR' U11 only only uses the "A" input to start the system. After the U1 and U2 time out from the power up. Then on the low going pulse from that, U3 goes high and times out and stays low, taking U11 A out of the system, starting the cascade of the whole system until it gets stopped by a shutdown. Or at least that was my idea behind what I drew.
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
I would have thought that you would need a low value resistor to limit the capacitor charging current through Q1 to within it's peak current rating. Also I would think the same protection would be required Q4 to limit the peak spark current. Am I correct in assuming that Q2 and Q3 are only required so the bootstrapping in the driver ICs works correctly. I will watch this thread with interest as I have been to mean to buy Ben Fleming's pulse EDM book.

Les.
Thanks for looking at my circuit.
To make this easier to follow I left the power supply for Q1 out of the schematic. There is a 20A transformer,bridge rectifier, and a very high value (don't have it in front of me and don't remember of hand) cap and a dropping resistor to limit current in case of a short circuit while 'burning' a part.

Both Q1 and Q4 are going to be industrial type Mosfets that are from a Tig welder, in the Isotop type package. Limiting current through Q4 shouldn't be a problem since the mosfets are way higher in rating than what this circuit will see. Again don't have the parts in front of me to give the numbers.

And you are correct Q2 and Q3 are only a path to ground to charge the bootstrap caps.

I've been a member of Flemming's Yahoo EDM group since he started it, not his new Pulse site but the old book related site. In fact a picture of the mechanical parts of my machine are in the pulse book. I'm not a big fan of some of the things he does in his circuits or builds. That is one of the reasons I started coming up with my own circuit. He is also the only one in all of the DIY builds that uses a brushed DC gear motor on his ram. When a EDM is running the ram/motor is constantly reversing when it's cutting, and all of the other builds use a stepper. The more controlled movement of a stepper keeps that reversal down to a very minimum, when the ram is reversing the electrode isn't removing metal. And that is the whole idea behind this. He also uses a window comparator that is 'high' when it is in the window instead of low like normal. Then he comes up with caps to keep the ram from reversing while the cutting caps recharge. Me and him went round and round about that years ago, but it's his circuit so he wouldn't think of making it simpler.

Another thing about getting his pulse book or the first book, he does updates to the circuit to fix problems and doesn't update the books. So then people are stuck with buying his PCB to get the latest fix. He won't even give the updates on his forum, which I can see his side of it, he wants to make money to pay for his time, but he should have worked out the bugs before selling a book, in my opinion.

To me the best book out there that gives the most information on how the whole system of EDM works is the one by Langolis. It is published by Home Shop Machinist magazine. A lot of my understanding on this comes from that book.
 

AnalogKid

Joined Aug 1, 2013
10,987
Refdes - there is a D1 in both schematics. and D2, and D3, etc. It is normal for a schematic of one assembly to cover multiple pages, but the reference designators must be unique across the entire design. It save clutter not having to say "D2 on page 1" instead of just D2.

Diodes - I just wanted to make sure that none of them were specialty parts, such as zeners or shottkeys.

D1 and D2 - CMOS today is better than it was when Don wrote that book. Plus, your circuits do not need any special cleanup. Delete R1, D1, and R4. Keep D2 reverse but it. There already is a protection diode built into the inverter input stage, But depending on the size of C2 it might present enough energy to damage the U3 input over time.

U4 inputs - As a starting point, assume all opamp and comparators have an infinite input impedance on both inputs.

U4 inputs - Just so we're clear, the window created by one signal diode will be approx 0.6 V at the comparator pins, not 5 V.

ak
 

AnalogKid

Joined Aug 1, 2013
10,987
It looks like the circuit around each ff steers the clock input to a different comparator depending on the ff state. So it is set by one input voltage level and reset by a different voltage level. If this is the intent, you are recreating a hysteresis circuit. https://en.wikipedia.org/wiki/Hysteresis#Electronic_circuits

The "normal" approach would be to implement the hysteresis in one comparator, and then follow that with stuff so you get a positive clock edge on both edges of the comparator output. In your case, the net parts savings would be almost zero and you would lose the ease of setting the window, so I'd say to stay with your design.

ak
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
U4 inputs - Just so we're clear, the window created by one signal diode will be approx 0.6 V at the comparator pins, not 5 V.
Probably didn't explain about that as good or clearly as I should. My thinking is 0.6V divided by 5 = 0.12V, so that makes 0.12V = to 1V scaled. And that makes 100V = 12V scaled. Like many things about this I don't know the correct way of doing or saying it.
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
It looks like the circuit around each ff steers the clock input to a different comparator depending on the ff state. So it is set by one input voltage level and reset by a different voltage level. If this is the intent, you are recreating a hysteresis circuit. https://en.wikipedia.org/wiki/Hysteresis#Electronic_circuits
Not how I see it or want it to work. Each ff is only activating a mosfet due to what is being measured in that part of the circuit, to toggle that mosfet on or off when that part of the circuit reaches a set voltage. The ff are "D" type but wired as a "T" type. And the clock signal is coming from the comparators. Thus what I was calling "step by step logic".
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
D1 and D2 - CMOS today is better than it was when Don wrote that book. Plus, your circuits do not need any special cleanup. Delete R1, D1, and R4. Keep D2 reverse but it. There already is a protection diode built into the inverter input stage, But depending on the size of C2 it might present enough energy to damage the U3 input over time.
This is the kind of stuff I don't know. Though 4xxx series Cmos was all the same after they switched to the "B" type
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
Refdes - there is a D1 in both schematics. and D2, and D3, etc. It is normal for a schematic of one assembly to cover multiple pages, but the reference designators must be unique across the entire design. It save clutter not having to say "D2 on page 1" instead of just D2.
Again I misunderstood. I thought you meant the inputs and outputs between pages didn't match. After I get some more comments on this I'll redraw. More of what I don't know.
 

AnalogKid

Joined Aug 1, 2013
10,987
Don't worry about what you don't know; we'll just keep chipping away at it. You've already cleared the single greatest hurdle in this project:

In attempts to improve your character, know what is in your power and what is beyond it.
- Francis Thompson.

A man's got to know his limitations.
- Harry Callahan

CHAP. XVII. The Master said, 'Yu, shall I teach you what knowledge is? When you know a thing, to hold that you know it; and when you do not know a thing, to allow that you do not know it;-- this is knowledge.
- Confucius (September 28, 551 – 479 BC)

Translated into Americanese:

Know what you know, and know that you don't know what you don't know.

ak
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
I'm glad to hear you are making traction with this. It's the whole reason you came here almost a decade ago, right?
Yes it is! I've had the mechanical part done for a long, long time. Was going to go with the most common, DIY, RC/relaxation oscillator way with the electronics, but decided I wanted something better, like the ones at work.
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
I know this is messy but thought maybe explaining the circuit block by block would be good and this was the fastest way of doing it.

1. The switch is turned on block#1 and a rising pulse sets both flipflops to the -Q state

2. That pulse then goes low and the falling edge sets a high in the OR gate of block #3. This triggers the Q output high to the FAN7390 in block8 to turn on Q1 and start charging the cap bank.

3. The cap bank reaches it's set point and then comparator 5 turns on the FAN7390 in block #9 and Q4 conducts.

4. The spark gap ionizes and comparator block #4 causes Block #3 to turn off Q1. Q1 is left on only to help with the gap ionization, then the cap bank can discharge it's full value through the gap, when Q1 shuts off.

5. Block #7 reads the value in the discharge path and signals the ram to go up or down depending on the voltage. A high voltage means it advances down a low voltage means it goes back up. The low voltage will also trigger block #6 to turn off preventing the electrode from shorting out in the gap, an undesired condition.

6 The process then starts over again when the comparator in block #4 goes low and the inverter retriggers block #3edm pulse power142.jpg edm pulse power143.jpg

EDIT changed #6, was represented wrong, now correct.
 
Last edited:

LesJones

Joined Jan 8, 2017
4,174
Hi SB,
Your description of the sequence of events makes perfect sense. I will comment on two points. The first is I think you will need a resitor in parallel with D4 & 5 (In section 10) to better define the gap current detection threashold to decide it is ionised. As it is a very small leakage currrent through the dialectric fluid would cause false triggering. The second point is that I think a pull down resistor is required on the cathode of D4 (Section 7) so there is enough current through it (More than the input leakage of the Lm339) to give the 0.6 (or so) volts drop which defines the difference between the lower and upper threasholds of the window detector. It might be better in future to number components for the whole circuit rather than just in each section. (Hence my comments in brackets after the diode numbers.)

Les.
 

Thread Starter

shortbus

Joined Sep 30, 2009
10,045
Thanks Les. Yes I am going to redraw this, AK pointed the same thing out to me about using a continuing number system for two drawings. If I can figure out how to do it in Diptrace. I'm new to all of this including the schematic drawing in CAD, Diptrace was the one I chose because it was the one that is easiest for me, know most use Eagle but could never get past the learning curve for it.

As far as D4&D5 in block 10, I don't think it will mater much in this. Others ciruits I've seen don't use it and suffer from lost or unused pulses in the burn. By keeping the power on for even a short while and then shutting down Q1, it should give the cap bank a better chance of putting all of it's energy into the actual burn. Instead of like most do, letting the cap bank be responsible for starting the ionization of the dielectric. A company called Charmilles uses something similar(I think that's how they're doing it) they call it "Isopulse". Makes every energy pulse as close to the same as possible. But all of this stuff is a very closely guarded secret, not even the low level company techs that serviced the machines at work would let me see the circuit schematics when they came in to service machines.

Nice to find some one else that is interested in this stuff.
 

LesJones

Joined Jan 8, 2017
4,174
Hi SB,
I first played around with EDM about 50 years ago in the mid 1960s but did not have much succes. I was mainly interested in removing broken taps etc at the time. Then a couple of years ago there was some discussion on the subject on the Model Engineer Magazine forum (A UK magazine.) So I had a play round with it again and built an RC design based on Ben Flemmings design. There was something about the window comparitor design in his circuit that I did't like so I modified that. I also added an actve rectifier circut to the input from the electrode so that it work with either polarity to the electrode. (I had read that for certain situations reversing the polarity had an advantage.) It worked quite well putting holes in high speed steel and tungsten carbide tool tips. I was interested in his pulse design but never found any schematics. One thing I thought of trying was to feed the electrode via a high current diode and charge the capacitors to a bit less than the breakdown voltage. Another diode would also be connected to the electrode and a high voltage (A few hundred volts.) would be pulsed via this diode to initiate the breakdown. This current from this pulse would not need to be very high. Onece the breakdown had been initiated the main capacitor bank would discharge through the gap. One problem I can see is that there will always be a power loss from the resistors that limit the charging current of the capacitor bank. The only way I can see round this would be to use a switch mode constant current supply to charge the capacitors.

Les
 
Last edited:
Top