[QUOTE="shortbus, post: 869976, member: 6184
@shortbus
In response to your point #1, I thought by using a SPST toggle the line would not be floating but stay at supply (12V) level. I was under the assumption that the S/RS would only cycle once and then stay at the level that results from that.

That's true when the switch is at reset, but when it is in start (as shown) it is not at 12 volts or ground, but what I call floating. A little bit of noise would make it look like another reset / start pulse.[/QUOTE]

That double switch symbol was my fault trying to learn Diptrace. Couldn't find a toggle switch but could find a push button. So used a push button then drew the toggle by hand before scanning and posting. Sorry.

My thoughts were to use one switch to do both things, that's why the two different times in the edge detectors. A short one for reset, and a longer one for start. Then both would stay at 12V (no longer detecting an edge) for the duration of the burn, due to switch being a toggle.

 

Speaking of dividers......
So if we are ok on the first 2 changes I have a question about U13.1 & .2. I think you want the thresholds to be 95 and 20 when the timer is not fired????. Then when there is an arc the timer fires and you want the thresholds to change. What do you want them to change to???

The comparators in question are just used to toggle the F/F's. The one that says 90V to turn on both the discharge and boost 130V. Then when the second one gets to 20V it will shut down the discharge mosfet. There is no "arc timer", nor is one needed, these two comparators will control the arc time. When different caps are selected they will discharge either shorter or longer times. And the 20V comparator shuts the circuit down when that level is reached. Now both of those voltages may have to be "played with" once it is built, to make it work. But those should give a good starting point.

The timer in the pulse logic is an "off timer". It's out put is tied to the 90 and 130V comparator reference voltage, with it low there is no reference voltage on those comparators. So they can't trigger their F/F's. When the arc shuts off it allows the gap to flush debris out and the charge F/F to turn on and recharge the caps.----- Then when the 90V comparator goes high the whole thing starts again.

The other 555 with no values is just for the stepper driver. It should not even be on this schematic. It and the stepper logic are a separate circuit to it's self.

 

="shortbus, post: 870128, member: 61844"]

My thoughts were to use one switch to do both things, that's why the two different times in the edge detectors. A short one for reset, and a longer one for start. Then both would stay at 12V (no longer detecting an edge) for the duration of the burn, due to switch being a toggle.

Yes, I saw that. You can try it without the 1k pull down. If the wires to the switch pick up noise and it acts squirrely you can always add it.

The comparators in question are just used to toggle the F/F's. The one that says 90V to turn on both the discharge and boost 130V. Then when the second one gets to 20V it will shut down the discharge mosfet. There is no "arc timer", nor is one needed, these two comparators will control the arc time. When different caps are selected they will discharge either shorter or longer times. And the 20V comparator shuts the circuit down when that level is reached. Now both of those voltages may have to be "played with" once it is built, to make it work. But those should give a good starting point.

I thought that was your intent. Just wasn't sure. I'll have a look.

The timer in the pulse logic is an "off timer". It's out put is tied to the 90 and 130V comparator reference voltage, with it low there is no reference voltage on those comparators. So they can't trigger their F/F's. When the arc shuts off it allows the gap to flush debris out and the charge F/F to turn on and recharge the caps.----- Then when the 90V comparator goes high the whole thing starts again.

Haven't got that far yet, but did find a little buglet. :=) The wire going to the bottom or block on the FF needs to go to the not Q output or the charge voltage is still on during the arc.

The other 555 with no values is just for the stepper driver. It should not even be on this schematic. It and the stepper logic are a separate circuit to it's self.

Please add it. I want to try and understand it.

 

Yes, I saw that. You can try it without the 1k pull down. If the wires to the switch pick up noise and it acts squirrely you can always add it.

Haven't got that far yet, but did find a little buglet. :=) The wire going to the bottom or block on the FF needs to go to the not Q output or the charge voltage is still on during the arc.

Please add it. I want to try and understand it.

Thank you for trying to understand my unlogical logic, and all your help Ron. I'll put the pulldown in , just didn't know it would be necessary.

About the 'buglet', I'll have to look closer to try and understand what you mean. After the tread with Brownout about 4013's, the subcircuit I came up with is what's used in all three of the "switches". The Q and notQ wrap around to the input AND gates. This is so only the one that is ready to toggle is the only one that can toggle. Know I'm not explaining it correctly, but it's to prevent any type of 'race' condition or out of sequence toggling. And still be able to use the 'clock' pin for toggling.

The intention was to keep the cap bank 'live' and able to charge up until the arc is fully started. And prevent any loss of charge on the caps during the time that the 95V output diode starts to conduct. A small time and probably a small amount of charge but that's my reasoning.

 

OK, added the components to the 555 astable, and a copy of the page from the stepper driver manual. The astable is how it was in the Langlois book, that's the values he used for his motor.

The 555 just stays running the whole time, and is sent to the driver as needed by the 'or' gate. One direction just uses the timer. The opposite direction uses the timer and grounds the direction. Didn't show the pull ups on the 40107 outputs because they are shown on the stepper driver page. Did show that the 40107 outputs are 'open drain' though.

 

="shortbus, post: 870284, member: 61844"]Thank you for trying to understand my unlogical logic, and all your help Ron. I'll put the pulldown in , just didn't know it would be necessary.

About the 'buglet', I'll have to look closer to try and understand what you mean. After the tread with Brownout about 4013's, the subcircuit I came up with is what's used in all three of the "switches". The Q and notQ wrap around to the input AND gates. This is so only the one that is ready to toggle is the only one that can toggle. Know I'm not explaining it correctly, but it's to prevent any type of 'race' condition or out of sequence toggling. And still be able to use the 'clock' pin for toggling.

Here is a better fix for the buglet. It makes the 4538 timer correct as well. I think maybe you meant to use the non inverting input to the 339 instead of the inverting input.

The thought just occurred to me that maybe you don't know how to interpret the waveforms. Can you read them?

 

OK, added the components to the 555 astable, and a copy of the page from the stepper driver manual. The astable is how it was in the Langlois book, that's the values he used for his motor.

The 555 just stays running the whole time, and is sent to the driver as needed by the 'or' gate. One direction just uses the timer. The opposite direction uses the timer and grounds the direction. Didn't show the pull ups on the 40107 outputs because they are shown on the stepper driver page. Did show that the 40107 outputs are 'open drain' though.

Is the OR gate an OR gate or an exclusive OR like the other schematic.

 

Is the OR gate an OR gate or an exclusive OR like the other schematic.

This was drawn before the other logic and I just added the 555 values, the 'or' can be the extra 'exclusive or' left over from the pulse logic. When I said they were separate circuits, they can be on/will be on the same logic board. It's just that they aren't really tied together. They need to act separately and independent of each other.

 

="shortbus, post: 870284, member: 61844"]Thank you for trying to understand my unlogical logic, and all your help Ron. I'll put the pulldown in , just didn't know it would be necessary.



Here is a better fix for the buglet. It makes the 4538 timer correct as well. I think maybe you meant to use the non inverting input to the 339 instead of the inverting input.

The thought just occurred to me that maybe you don't know how to interpret the waveforms. Can you read them?

I have looked at my schematic unknown number of times and never noticed that I didn't connect the notQ to the 'D' input on the discharge and boost F/F's. What a dummy!

I don't see any "fix"? What wave forms? If you mean the ones with the spice files, no not really. Kind of see what some are saying but not all of them.

I'm pretty sure that the 4538 is correct. According to the data sheet it is set up as a positive edge trigger timer, but knowing me,it may be wrong. -- The start edge detector gives a high on one pin of the 'and' when it times out. Then when the gap volt detector goes low and inverts that gives the other 'and' pin a high also. That give a high to the '+' input of the 4538 and the timer starts. Using the notQ output of the timer it goes from high to low for the time period, then back to high, allowing the 90V and 20V to operate again. And the cycle starts over.

 

@shortbus

I hate it when I forget the attachment.

 

@ronv, did my explanation of how the 4538 was to work make sense? Or am I still missing something?

 

@shortbus
I'm having a little trouble deciding where to start. There are several things going on. So lets stay on the first problem. When you push start U2.1 toggles and charge goes plus So name the line going to the 2110 that turns on the charge FET +charge. When this line is positive the charge FET is on. OK so this part works ok. Now. Lets assume everything else works as planned - not sure if it does or not, but lets say it does and the discharge 115 FET gets turned on and we get an arc. What we would like to have happen is for charge to get turned off. But it does not. It stays on. So lets see why. When the arc starts the top of the arc sense diodes goes plus. The output of the comparator inverts so it goes to ground or 0. That signal gets inverted by U10.1 and is fed to the AND gate U4.1. But the Q output is now minus so it can't get thru to turn off charge. The opposite is true of the top AND gate U5.2. The Q not is high but when the arc starts the other input is minus. The net result is that the charge FET stays on until the 40,000 Ufd cap gets discharged which is never. So we need to fix this one first.

 

Thanks Ron I may have a drawing error. I'll look and see.

 

@ronv, OK, must be dumber than I thought, I don't see the problem? With your 'fix' of my drawing error in post #30, it seems OK to me.

Reset puts notQ of U2-1 high, and turns onU8 (ir2110-charge), and makes pin 5 of U5-2 high.

When arc starts, U13-3 goes high and so does pin 6 of U5-2 toggling F/F U2-1. This shuts off U8 and makes pin 2 on U4-1 high.

When arc stops, U13-3 goes low and U10-1 inverts. This starts timer U1.1 and makes pin 1 of U4 high and U2-1 toggles turning U8 on starting charge again.

When start happens, no arc is present so U13-3 is low, U10-1 is high and U4-1 pin 1 is high. But Q of U2-1 is lo and so is U4-1 pin 2.

This is my take on it, so don't see a problem? Other than my poor drawing skills. I'll redraw this and try not to make the components so out of sequence this time. And implement your fixes into it. Was my first real try at using the CAD and didn't do so well.

Again can't thank you enough for putting up with me!!

 

Another question, what should the reference voltage be on the arc detect comparator? If I'm redrawing I want to do some figuring and put in some resistor values. I'll do the math so you won't need to just need to know what voltage to shoot for.

Will these diodes work for the arc detector? Bought them long ago when planing to build this another way. Thanks again for your help.

 

Another question, what should the reference voltage be on the arc detect comparator? If I'm redrawing I want to do some figuring and put in some resistor values. I'll do the math so you won't need to just need to know what voltage to shoot for.

Will these diodes work for the arc detector? Bought them long ago when planing to build this another way. Thanks again for your help.

Yes, I think with the fix in #30 you are ok.
Those diodes should be ok. Each diode will drop about 0.9 volts so at the top of the two there should be about 1.8 volts when the arc is taking place. So you can set the threshold to .7 volts or so and it should be ok. You will also need a resistor from the top of the 2 diodes to ground so they are not floating when there is no arc or they will pick up noise. You can also use a diode for the reference if you like. Kind of like the one I posted way back when. This will also set the reference at about .7 volts. Either way works ok. Keep the values kind of low to make it less sensitive to noise as it's in a bad spot.

 

Thanks Ron. If I remember correctly you just had a diode and resistor to ground on the other one you showed. no connection to any voltage, didn't see how that would work? I'll have to go back and look. The voltage on top of the two diodes is what I thought but wasn't sure if it was like that or the the arc voltage minus the diode drop.

Edit. went back and looked and was wrong. I'll do it like you showed.

 

OK! Worked out an easy way for the timer to disable the comparators for the 90/20. The 339 is an open collector output so it needs a pull up to +12 for the output to go high. But we can also add a diode from the timer output to each comparator output to disable them while the caps charge. The anodes go to the timer and the cathode of each one goes to each comparator. We can discuss how long the timer needs to be later.
Right now how about when there is no arc. Right now it just waits for one. Any ideas?
@shortbus

 

@ronv OK, what I think your saying is instead of a normal pull up you would use a diode replace the pull up. That sounds like it would work! But would a resistor still be needed with the diode? Or being a logic level from the timer out put the resistor is not needed?

While the 'wait for an arc' seems like a bad thing, it really isn't. If the arc 'should' happen but doesn't it would usually mean the gap conditions have a problem. By waiting it gives the dielectric time to flush/clean the gap, which is the most common reason for no arc.

 

Another question while talking about diodes. Will the 1N5817 diodes you showed in post#16 (I think) work for all of the small diodes? If so I'll get a bunch of them.