Re: [Emc-users] Watchdogs
On Mon, 2008-04-28 at 21:07 -0400, Greg Michalski wrote: ... snip improve it please feel free to post :) ) Are you running that directly off a parallel port pin or is it buffered/opto-isolated? ...snip My charge pump is being feed by a parallel port buffer which is a 54HCT541. There isn't much to see but it's shown here: http://www.wallacecompany.com/machine_shop/Shizuoka/00011-1a.jpg The breadboard on the left has the buffers and opto-couplers. The left board has the charge pump and a circuit from a totally different project. Other Shizuoka pictures are here: http://www.wallacecompany.com/machine_shop/Shizuoka/ -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
On Mon, 2008-04-28 at 21:08 -0400, John Kasunich wrote: ... snip After I wrote my reply I looked closer at the schematic and saw the how big your caps are. I also noticed that you are using an LS14. I used much smaller caps, probably 0.1uF for the filter and 0.01uF or smaller for the pumps. I used proportionally higher resistors to get the time constants I wanted, but the net result is less loading on the '14 sections that are driving the pump. I started with .1 uF pump caps but could not get enough voltage. When I put the 2.2 uF caps, I got what I wanted and went on to other issues. One of these days, it would be nice to learn spice and do a proper job of design. In your case, the input current requirements of the LS14 make it interesting. IIRC, LS inputs go high when open - you need to make sure your load resistor R5 is low enough to ensure a logic zero when the pump is off. If you've socketed the chip, I'd strongly consider replacing it with CMOS one, like an HC or HCT. I checked for proper output (low) for steady high and steady low input signals. I used the LS because that is what I found in a pile of telecomm boards I use for my own little RadioShack. I have never paid much attention to the technology types, so I appreciate your advise and will take it into consideration when I put my Digikey order together. For the current circuit it might be helpful if I got the scope out and looked at the signals along the way. Since the CMOS parts have very high DC input impedance, you could put 10K to 50K between the top of R5 and the input to the chip. That would greatly limit any current that might flow into internal protection diodes, and make the clamp issue go away. And since the signal is DC at this point, switching speed is not an issue, so the resistor would not have any down side(?). The tradeoff between HC and HCT is driven mostly by the input from the PC (pin 5). HC is better for the output section, since I think the HC thresholds are about mid-rail. HCT thresholds are lower, and are a good choice if the parport might be driving only 3.3V. Regards, John Kasunich -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
On Mon, 2008-04-28 at 21:47 -0400, Kenneth Lerman wrote: Does anyone use a single chip solution consisting of a single 8 pin microprocessor? An Atmel atTiny13 is $1.40 in an 8 pin dip package. These chips power up nicely, require no external oscillator, and are easy to program. If you don't trust them, get two. Use one to drive the high side of the SSR relay and the other to drive the low side. The paranoid among us can take the drive signals from the chips and feed them back as inputs to the main processor. It would then check that the chips are in the no signal state prior to starting the charge pump. It wouldn't do to have a failure of the failsafe circuit be undetected. Ken Who's checking the failsafe circuit checker? If you can do the programming easily, these eight pin processors certainly are cheap. I find it a little frustrating to consider using them because I always try to think of other functions I can put in while I am there. I am the only one here, so no one shoots the engineer to finish the project. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Here is what I breadboarded that seems to work. http://www.wallacecompany.com/machine_shop/Shizuoka/watchdog-1b.png I used John's circuit and added the two inverters to the output because the EMI from the mechanical relay engaging would cycle the solid-state relay which would cycle the mechanical relay. Fortunately, it was a degenerative oscillation. Adding the buffer to the output cleaned up the problem. The component values are a combination of guessing, parts that happen to be on the bench, and cut and try. I welcome any improvements anyone might have. I loaded the charge-pump component to my .hal file and ran it in the servo thread. Does the thread determine the pump frequency? I believe I got 500 Hz out. I used these connections in the .hal file: ... # create a signal for the estop loopback net estop-loop iocontrol.0.user-enable-out iocontrol.0.emc-enable-in charge-pump.enable net cp charge-pump.out parport.1.pin-04-out ... -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Kirk Wallace wrote: Here is what I breadboarded that seems to work. http://www.wallacecompany.com/machine_shop/Shizuoka/watchdog-1b.png I used John's circuit and added the two inverters to the output because the EMI from the mechanical relay engaging would cycle the solid-state relay which would cycle the mechanical relay. Fortunately, it was a degenerative oscillation. Adding the buffer to the output cleaned up the problem. The component values are a combination of guessing, parts that happen to be on the bench, and cut and try. I welcome any improvements anyone might have. I loaded the charge-pump component to my .hal file and ran it in the servo thread. Does the thread determine the pump frequency? I believe I got 500 Hz out. I used these connections in the .hal file: ... # create a signal for the estop loopback net estop-loop iocontrol.0.user-enable-out iocontrol.0.emc-enable-in charge-pump.enable net cp charge-pump.out parport.1.pin-04-out ... Using a schmidt trigger to clean up the output is good. However, there is one potential problem. The two stage charge pump can produce more than 5V, which the '14 might not like. Maybe a clamp diode to the supply rail? Or maybe use just one stage - remove C2, short D3. You'd have to make sure that one stage generates enough voltage to reliably trigger the '14. Regarding the frequency - yes, the period depends on the thread speed. It toggles the output each time it runs, so a 1KHz run rate = 500Hz out. Regards, John Kasunich - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
On Mon, 2008-04-28 at 19:54 -0400, John Kasunich wrote: ... snip Using a schmidt trigger to clean up the output is good. However, there is one potential problem. The two stage charge pump can produce more than 5V, which the '14 might not like. Maybe a clamp diode to the supply rail? Or maybe use just one stage - remove C2, short D3. You'd have to make sure that one stage generates enough voltage to reliably trigger the '14. ... snip John Kasunich You are right. I got 5.5 Volts on the pump output going into the first buffer. I figured it might be okay. I like the clamp diode idea. I see the datasheet indicates that I am at the max. TI's LS14 shows a zener on the input already, so I guess I don't really know how hard I am driving the input. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Here is what I breadboarded that seems to work. http://www.wallacecompany.com/machine_shop/Shizuoka/watchdog-1b.png That looks nice and simple. If I have troubles with what I've got on hand (the CNC4PC charge pump board) I might try your circuit out. (So if you improve it please feel free to post :) ) Are you running that directly off a parallel port pin or is it buffered/opto-isolated? Regarding the frequency - yes, the period depends on the thread speed. It toggles the output each time it runs, so a 1KHz run rate = 500Hz out. So in order to get my CNC4PC charge pump board to work correctly (it relies on a 12.5kHz constant signal) I am going to need to create it's own thread in addition to the servo thread and the base thread? I am thinking this is possible to do, are there any negatives to creating a third thread? I'm still assembling my control enclosure so I can't fire things up to experiment, the backplane isn't quite ready to mount the MOBO yet as I'm still drilling tapping to mount some parts. Would help if my ebay score of contactors showed up...I'd get more motivation to get it done. - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Kirk Wallace wrote: On Mon, 2008-04-28 at 19:54 -0400, John Kasunich wrote: ... snip Using a schmidt trigger to clean up the output is good. However, there is one potential problem. The two stage charge pump can produce more than 5V, which the '14 might not like. Maybe a clamp diode to the supply rail? Or maybe use just one stage - remove C2, short D3. You'd have to make sure that one stage generates enough voltage to reliably trigger the '14. ... snip John Kasunich You are right. I got 5.5 Volts on the pump output going into the first buffer. I figured it might be okay. I like the clamp diode idea. I see the datasheet indicates that I am at the max. TI's LS14 shows a zener on the input already, so I guess I don't really know how hard I am driving the input. After I wrote my reply I looked closer at the schematic and saw the how big your caps are. I also noticed that you are using an LS14. I used much smaller caps, probably 0.1uF for the filter and 0.01uF or smaller for the pumps. I used proportionally higher resistors to get the time constants I wanted, but the net result is less loading on the '14 sections that are driving the pump. In your case, the input current requirements of the LS14 make it interesting. IIRC, LS inputs go high when open - you need to make sure your load resistor R5 is low enough to ensure a logic zero when the pump is off. If you've socketed the chip, I'd strongly consider replacing it with CMOS one, like an HC or HCT. Since the CMOS parts have very high DC input impedance, you could put 10K to 50K between the top of R5 and the input to the chip. That would greatly limit any current that might flow into internal protection diodes, and make the clamp issue go away. The tradeoff between HC and HCT is driven mostly by the input from the PC (pin 5). HC is better for the output section, since I think the HC thresholds are about mid-rail. HCT thresholds are lower, and are a good choice if the parport might be driving only 3.3V. Regards, John Kasunich - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Greg Michalski wrote: So in order to get my CNC4PC charge pump board to work correctly (it relies on a 12.5kHz constant signal) I am going to need to create it's own thread in addition to the servo thread and the base thread? I am thinking this is possible to do, are there any negatives to creating a third thread? Does it need _exactly_ 12.5 KHz, or a _minimum_ of 12.5KHz? My bet is on the latter. In that case, you could put the charge pump in the base thread. If your base period is 40uS or shorter, you will get 12.5 KHz or more. I would try to avoid adding another thread - if the new thread is faster than your base thread it could actually increase the jitter in your step pulses, which is not something you want to do. Regards, John Kasunich - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Does anyone use a single chip solution consisting of a single 8 pin microprocessor? An Atmel atTiny13 is $1.40 in an 8 pin dip package. These chips power up nicely, require no external oscillator, and are easy to program. If you don't trust them, get two. Use one to drive the high side of the SSR relay and the other to drive the low side. The paranoid among us can take the drive signals from the chips and feed them back as inputs to the main processor. It would then check that the chips are in the no signal state prior to starting the charge pump. It wouldn't do to have a failure of the failsafe circuit be undetected. Ken - Original Message - From: John Kasunich [EMAIL PROTECTED] To: Enhanced Machine Controller (EMC) emc-users@lists.sourceforge.net Sent: Monday, April 28, 2008 9:11 PM Subject: Re: [Emc-users] Watchdogs Greg Michalski wrote: So in order to get my CNC4PC charge pump board to work correctly (it relies on a 12.5kHz constant signal) I am going to need to create it's own thread in addition to the servo thread and the base thread? I am thinking this is possible to do, are there any negatives to creating a third thread? Does it need _exactly_ 12.5 KHz, or a _minimum_ of 12.5KHz? My bet is on the latter. In that case, you could put the charge pump in the base thread. If your base period is 40uS or shorter, you will get 12.5 KHz or more. I would try to avoid adding another thread - if the new thread is faster than your base thread it could actually increase the jitter in your step pulses, which is not something you want to do. Regards, John Kasunich - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
[Emc-users] Watchdogs
I have every thing controlled by parport pins on my Shizuoka which makes for an inexpensive conversion, but one of the problems is that if I have my main breaker on while booting the EMC2 PC the mill will come up in an unsafe state. One of the solutions I was thinking about is, I seem to recall a watchdog frequency output from from the estop component that I could connect to a watchdog sensor connected to my main power relay. A quick look did not reveal the information about the estop watchdog. Should I keep looking? Does anyone have a link to a watchdog sensor circuit that would output high at the set frequency and low in all other conditions? Is there a better way to do this? -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
On Fri, Apr 25, 2008 at 07:47:32AM -0700, Kirk Wallace wrote: I have every thing controlled by parport pins on my Shizuoka which makes for an inexpensive conversion, but one of the problems is that if I have my main breaker on while booting the EMC2 PC the mill will come up in an unsafe state. One of the solutions I was thinking about is, I seem to recall a watchdog frequency output from from the estop component that I could connect to a watchdog sensor connected to my main power relay. A quick look did not reveal the information about the estop watchdog. Should I keep looking? Does anyone have a link to a watchdog sensor circuit that would output high at the set frequency and low in all other conditions? Is there a better way to do this? There is a charge pump component that toggles its output every time its function is run... - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Chris Radek wrote: On Fri, Apr 25, 2008 at 07:47:32AM -0700, Kirk Wallace wrote: I have every thing controlled by parport pins on my Shizuoka which makes for an inexpensive conversion, but one of the problems is that if I have my main breaker on while booting the EMC2 PC the mill will come up in an unsafe state. One of the solutions I was thinking about is, I seem to recall a watchdog frequency output from from the estop component that I could connect to a watchdog sensor connected to my main power relay. A quick look did not reveal the information about the estop watchdog. Should I keep looking? Does anyone have a link to a watchdog sensor circuit that would output high at the set frequency and low in all other conditions? Is there a better way to do this? There is a charge pump component that toggles its output every time its function is run... That HAL component makes the charge pump signal. I think Kirk is looking for the real circuit that uses the charge pump signal to turn on a relay. I built one for my Shoptask, unfortunately I think I just drew the circuit on a piece of paper and don't have it in electronic form. It is on the board described in this blog posting (the board also contains two relays with drivers and three optically isolated digital inputs): http://jmkasunich.com/cgi-bin/blosxom/shoptask/breakout-board-12-19-07.html If I remember when I get home I'll try to draw up a little schematic for a charge pump and post it. Regards, John Kasunich - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
On Fri, 2008-04-25 at 11:47 -0400, John Kasunich wrote: ... snip That HAL component makes the charge pump signal. I think Kirk is looking for the real circuit that uses the charge pump signal to turn on a relay. Thats correct. I built one for my Shoptask, unfortunately I think I just drew the circuit on a piece of paper and don't have it in electronic form. It is on the board described in this blog posting (the board also contains two relays with drivers and three optically isolated digital inputs): http://jmkasunich.com/cgi-bin/blosxom/shoptask/breakout-board-12-19-07.html If I remember when I get home I'll try to draw up a little schematic for a charge pump and post it. Regards, John Kasunich Thanks John. If the the 7414 does the whatchdog sensing, I could probably reverse engineer your pictures. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Kirk Wallace wrote: On Fri, 2008-04-25 at 11:47 -0400, John Kasunich wrote: ... snip That HAL component makes the charge pump signal. I think Kirk is looking for the real circuit that uses the charge pump signal to turn on a relay. Thats correct. I built one for my Shoptask, unfortunately I think I just drew the circuit on a piece of paper and don't have it in electronic form. It is on the board described in this blog posting (the board also contains two relays with drivers and three optically isolated digital inputs): http://jmkasunich.com/cgi-bin/blosxom/shoptask/breakout-board-12-19-07.html If I remember when I get home I'll try to draw up a little schematic for a charge pump and post it. Regards, John Kasunich Thanks John. If the the 7414 does the whatchdog sensing, I could probably reverse engineer your pictures. I just looked at the photos again, and it was enough to jog my memory. Two sections of the '14 are used for the watchdog, the other sections are used for the relays. The horizontally oriented single-inline-pin component below the relays is a MOSFET array with four FETs in it. I'm using two of them to drive the relays and the third for the watchdog (the watchdog output is open-drain, the watchdog relay is off-board). It looks like I might have paralleled the third and fourth FETs for a lower on resistance, and there is a zener there to clamp turn-off spikes. The charge pump input goes to the input of one section of the '14. The output of that section (call it output 1) goes to the input of another section. So output 1 is inverted, and output 2 is non-inverted. The pump itself is the three diodes in the lower left. They are in series, and the bottom of the diode stack is grounded. The top feeds a filter capacitor (probably a 0.1uF or so), a resistor (to bleed it down, with a time constant of a few tens of milliseconds), and the gate of the output MOSFET. Output 1 of the '14 is connected to the junction of the bottom and middle diodes, through a small pump capacitor - I don't recall the value, but it is 1/10 or less of the value of the output filter cap. Output 2 is connected to the junction of the middle and top diodes, with the same size pump cap. This is actually a bit more complex than the simplest charge pump, but it can deliver more voltage. A single-stage charge pump has only two diodes and one pump cap, and can deliver a voltage that is roughly the logic output swing minus two diode drops - maybe 3.5 volts on a good day. This three-diode version with two pump caps running out of phase can deliver more. I want to say it is 2x the logic swing minus 4 diode drops, but I'm not sure if that is exactly right. I do know it is over 5V, enough to nicely saturate a logic level FET, and would probably do OK with a normal FET. If you need more details don't hesitate to ask. Regards, John Kasunich - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
On Fri, 2008-04-25 at 17:02 -0400, John Kasunich wrote: ... snip exactly right. I do know it is over 5V, enough to nicely saturate a logic level FET, and would probably do OK with a normal FET. If you need more details don't hesitate to ask. Regards, John Kasunich I guess I need a little more help than I thought. Here is my first pass at your board schematic: http://www.wallacecompany.com/machine_shop/Shizuoka/watchdog-1a.png I guessed at the FET array pins and some bits are missing but I think I got the watchdog part. If the input is constantly low the signal to the FET will be low. If the input is constantly high, the signal to the FET will be the division of R2 and R5? What I need is; 0 to x Hz = low, x Hz to high = low, x Hz = high, or a band-pass filter. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
Kirk Wallace wrote: I guess I need a little more help than I thought. Here is my first pass at your board schematic: http://www.wallacecompany.com/machine_shop/Shizuoka/watchdog-1a.png That looks correct. I guessed at the FET array pins and some bits are missing but I think I got the watchdog part. You guessed right. If the input is constantly low the signal to the FET will be low. Yep. If the input is constantly high, the signal to the FET will be the division of R2 and R5? Nope. If the input is constantly high, the signal to the FET will be low. One end of the diode string is grounded, and the other end is connected to ground through a resistor that requires a DC current to develop any voltage across it. Neither C1 nor C2 will pass DC. The only way to get any voltage on the FET is to have an AC voltage on C1 and C2 that the diodes can rectify. What I need is; 0 to x Hz = low, x Hz to high = low, x Hz = high, or a band-pass filter. I'm having trouble parsing that ;-) I don't think I've ever seen a charge pump that was a bandpass. They turn on as long as the input is above some minimum frequency, and off for DC (either steady high or steady low). In this design, the minimum frequency is determined by the values of C1, C2, and R5. The size of C1 and C2 determine how much charge is pumped into C3 on each cycle, and R5 determines how fast that charge bleeds off and must be replenished. C3 is less critical - but it needs to be at least 10x the value of C1 and C2, to make sure a single edge as might be generated during boot or driver loading can't charge it up enough to turn things on. If C3 is too big, it will take a long time (relatively) to turn off when the input stops toggling. Regards, John Kasunich - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] Watchdogs
On Fri, 2008-04-25 at 23:17 -0400, John Kasunich wrote: Kirk Wallace wrote: ... snip If the input is constantly high, the signal to the FET will be the division of R2 and R5? Dooh. It's as plain as day that R2 is grounded. Somehow I visualized current going through R2, the diodes, R5 then ground. Any current in R2 goes to ground only. Nope. If the input is constantly high, the signal to the FET will be low. One end of the diode string is grounded, and the other end is connected to ground through a resistor that requires a DC current to develop any voltage across it. Neither C1 nor C2 will pass DC. The only way to get any voltage on the FET is to have an AC voltage on C1 and C2 that the diodes can rectify. I see the basics of it now. The math is beyond me but I can breadboard it to get it working. What I need is; 0 to x Hz = low, x Hz to high = low, x Hz = high, or a band-pass filter. I'm having trouble parsing that ;-) I don't think I've ever seen a charge pump that was a bandpass. They turn on as long as the input is above some minimum frequency, and off for DC (either steady high or steady low). Yes, right again. A band pass would do what its name suggests, which is pass the selected frequency through unchanged, instead of converting it to a digital indicator. In this design, the minimum frequency is determined by the values of C1, C2, and R5. The size of C1 and C2 determine how much charge is pumped into C3 on each cycle, and R5 determines how fast that charge bleeds off and must be replenished. C3 is less critical - but it needs to be at least 10x the value of C1 and C2, to make sure a single edge as might be generated during boot or driver loading can't charge it up enough to turn things on. If C3 is too big, it will take a long time (relatively) to turn off when the input stops toggling. Regards, John Kasunich I probably don't have any 7414's in hand. For now, I wonder if I can use a couple of NPN's with the load and pull up resistors on the collectors and the emitters tied to ground to make them inverters. I guess I'll find out tomorrow. Thanks for your help. It is fairly dramatic when the main relay unexpectedly kicks in if I forget to throw the disconnect during boot or shutdown. -- Kirk Wallace (California, USA http://www.wallacecompany.com/machine_shop/ Hardinge HNC/EMC CNC lathe, Bridgeport mill conversion, doing XY now, Zubal lathe conversion pending Craftsman AA 109 restoration Shizuoka ST-N/EMC CNC) - This SF.net email is sponsored by the 2008 JavaOne(SM) Conference Don't miss this year's exciting event. There's still time to save $100. Use priority code J8TL2D2. http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
