Re: [Emc-users] [OT] Charge Pump
On Thu, 2010-12-09 at 20:06 -0800, Kirk Wallace wrote: I blew out the only 74HC (74HC04 NAND) part in my junk box, so I cut my charge pump circuit down to this: ~ _ _ Buffer Card Output _| |_| |_ Charge pump in servo thread = 1kHz 0-5V -+ | === C1 .01uF | O'Scope +-|--+--|--+-+--- || | |C2 ===\ | .01uF | / R1 15k || \ V (gnd) V V ~ I played with different caps I had on hand and got the best results with .01 uF for both C1 and C2 and 15k for R1. Even with R1 removed the output would decay pretty rapidly, so I had to bump C1 up enough to offset the leakage. Anything larger for C2 didn't seem to fully charge. Then I sized R1 to get a sub 1 second drop, so would probably trip the output at 200 to 300ms after losing the input signal. I suppose a CMOS input and the scope should be a similar load, so the timing should not change much if I add a Schmidt buffer to the output? I also wired this up: +- O'Scope _ _ | _| |_| || C1 .01uF ---\/\/-+---||--+ R1 15k | V (gnd) If T=RC, T=15000x.1=.15 ? On the scope, I got a shark fin like signal where most of the signal change occurred by 200ms, so I guess this confirms the .15 for T above? I didn't quite understand what I read on Wikipedia -- something like RC is the value where the signal change is at around 63% of the signal swing? I suppose I could use this circuit to measure caps with C=T/R with T@63% of signal? I'll have to play with this some more. Hi, (Hope is not to late...) Why don't you use one of Mariss Freimanis circuits ?!? http://www.artofcnc.ca/ChargePumpSafety.pdf or this from EDN Hardware watchdog timer accepts range of frequencies http://www.edn.com/file/25543-Hardware_watchdog_timer_accepts_range_of_frequencies_PDF.pdf Florin -- Colocation vs. Managed Hosting A question and answer guide to determining the best fit for your organization - today and in the future. http://p.sf.net/sfu/internap-sfd2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Mon, 2011-03-21 at 17:00 +0200, fi wrote: ... snip (Hope is not to late...) Why don't you use one of Mariss Freimanis circuits ?!? http://www.artofcnc.ca/ChargePumpSafety.pdf or this from EDN Hardware watchdog timer accepts range of frequencies http://www.edn.com/file/25543-Hardware_watchdog_timer_accepts_range_of_frequencies_PDF.pdf Florin Thanks for the links FLorin. What I have done so far is here: http://wiki.linuxcnc.org/emcinfo.pl?About_Charge_Pumps -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Enable your software for Intel(R) Active Management Technology to meet the growing manageability and security demands of your customers. Businesses are taking advantage of Intel(R) vPro (TM) technology - will your software be a part of the solution? Download the Intel(R) Manageability Checker today! http://p.sf.net/sfu/intel-dev2devmar ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
Kirk Wallace wrote: On Sat, 2010-12-11 at 12:28 -0500, John Kasunich wrote: Sorry to be so chatty, but... ... snip But if the cylinder is too small, pumped too slowly, or the leak is too big, then the pumping might not be able to keep up with the leak. In that case, make the leak smaller or pump faster if you can. Making the pump cylinder bigger should be the last resort. Note that if you make the leak too small, then it will take a long time for the bucket to empty when you stop pumping. It's all about the tradeoffs... John Kasunich I have been limiting myself to dealing with the 1kHz signal because it is a common servo-period frequency and if one doesn't have a base thread, a base thread would need to be added just for the charge pump. But, that may be a reasonable requirement. I don't think this is a concern. People who have hardware that removes the need for a base thread very likely also have a watchdog feature in their hardware. I know the Mesa and Pico products do, and I believe the Pluto does as well. (I'm not sure about Motenc and STG) The only real audience for this circuit is people who use simple I/O, such as a parallel port or 8255-based card, with software step/PWM generation. - Steve -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
Am 13.12.2010 15:56, schrieb Stephen Wille Padnos: I don't think this is a concern. People who have hardware that removes the need for a base thread very likely also have a watchdog feature in their hardware. I know the Mesa and Pico products do, and I believe the Pluto does as well. (I'm not sure about Motenc and STG) Motenc has a watchdog circuit barking at 8 or 16ms (selectable) if you don't write DACs or the wd register. - Steve --- Ulf Dambacher i...@dambacher-retrofit.de --- -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On 12/12/2010 01:49 PM, emc-users-requ...@lists.sourceforge.net wrote: There is then a lot to be said for removing IC1C, so the slow input is discriminated only once. If pins 1 3 are then connected, any small same-chip threshold variation will have negligible effect, because here the input flank is rather vertical. Dooh, of course. I'll make the change. http://www.wallacecompany.com/machine_shop/EMC2/Charge%20Pump/chpmp-5c.png If you're trying to make this circuit more universal, you could put IC1E on the output of IC1B, and then include a jumper on board so that the two buffer gates could select inverting or non-inverting polarity to signal CP-GOOD. I like to include a trace on board selecting one or the other as standard, and if you want to change it, you cut the trace, and add a jumper wire. Three extra plated through holes to pay for, but more universal. -- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
Stephen Wille Padnos wrote: (I'm not sure about Motenc and STG) STG does not have a complete watchdog. Years ago I used the STG. It produces a pulse that cycles when there is input to the various DAC channels. You need to provide an external circuit of the missing pulse detector sort to detect the cessation of these pulses. I suppose a charge pump scheme would have worked as well. Jon -- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
I don't know if it's been pointed out, but those caps on the regulator are in the wrong place. (on the diagram below) and a trivial point; I don't know how you selected which gate where, but when it comes to laying the board, change inverter number 1-6 to suit the mechanics, and choose tie-up or tie-down on unused inputs according to the closest rail input. Regards Roland On 13 December 2010 19:59, cogoman cogo...@optimum.net wrote: On 12/12/2010 01:49 PM, emc-users-requ...@lists.sourceforge.net wrote: There is then a lot to be said for removing IC1C, so the slow input is discriminated only once. If pins 1 3 are then connected, any small same-chip threshold variation will have negligible effect, because here the input flank is rather vertical. Dooh, of course. I'll make the change. http://www.wallacecompany.com/machine_shop/EMC2/Charge%20Pump/chpmp-5c.png If you're trying to make this circuit more universal, you could put IC1E on the output of IC1B, and then include a jumper on board so that the two buffer gates could select inverting or non-inverting polarity to signal CP-GOOD. I like to include a trace on board selecting one or the other as standard, and if you want to change it, you cut the trace, and add a jumper wire. Three extra plated through holes to pay for, but more universal. -- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Mon, 2010-12-13 at 20:39 +0200, Roland Jollivet wrote: I don't know if it's been pointed out, but those caps on the regulator are in the wrong place. (on the diagram below) Oops. That's such a bad mistake, I made myself laugh. Err... I was just testing to see if anyone would notice. This should be slightly better(may need Reload button on browser): http://www.wallacecompany.com/machine_shop/EMC2/Charge%20Pump/chpmp-5c.png I don't know if a real board will ever be made, but being mindful of the pin layout is a good point. Thanks Roland. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Mon, 2010-12-13 at 16:36 +0100, dambacher-retrofit.de wrote: Am 13.12.2010 15:56, schrieb Stephen Wille Padnos: I don't think this is a concern. People who have hardware that removes the need for a base thread very likely also have a watchdog feature in their hardware. I know the Mesa and Pico products do, and I believe the Pluto does as well. (I'm not sure about Motenc and STG) Motenc has a watchdog circuit barking at 8 or 16ms (selectable) if you don't write DACs or the wd register. - Steve Just thinking out loud. I see the PWM controller has this in the doc's: Jumper JP4 can be selected to enable or disable the watchdog timer on the board. When enabled, the watchdog will force the board into Emergency Stop mode when the rate generator registers have not been written to in 20 mS. I guess one issue is that the integrator will need to make sure that the PWM controller e-stop signal out is connected to all of the e-stop slaves and not just EMC2, because if the controller initiates a watchdog fault, we can't rely on EMC2 to handle it. A proper e-stop system would commonly be set up this way, so it shouldn't be a problem. Also, this watchdog only validates EMC2. The Pluto doc's are a little more interesting: Before configuration and after properly exiting emc2, all Pluto-P pins are tristated with weak pull-ups (20kΩ min, 50kΩ max). If the watchdog timer is enabled (the default), these pins are also tristated after an interruption of communication between emc2 and the board. The watchdog timer takes approximately 6.5ms to activate. However, software bugs in the pluto_servo firmware or emc2 can leave the Pluto-P pins in an undefined state. An independent wathcdog might be prudent. With the Motec one can pet the dog internally with dac-write (from EMC2) or from an external source (using watchdog-reset as an input rather than output)? I found this for Mesa's Hostmot2: ... Watchdog The HostMot2 firmware may include a watchdog Module; if it does, the hostmot2 driver will use it. The watchdog must be petted by EMC2 periodically or it will bite. When the watchdog bites, all the board’s I/O pins are disconnected from their Module instances and become high-impedance inputs (pulled high), and all communication with the board stops. The state of the HostMot2 firwmare modules is not disturbed (except for the configuration of the IO Pins). Encoder instances keep counting quadrature pulses, and pwm- and step-generators keep generating signals (which are *not* relayed to the motors, because the IO Pins have become inputs). Resetting the watchdog resumes communication and resets the I/O pins to the configuration chosen at load-time. If the firmware includes a watchdog, the following HAL objects will be exported: Pins: (bit in/out) has_bit: True if the watchdog has bit, False if the watchdog has not bit. If the watchdog has bit and the has_bit bit is True, the user can reset it to False to resume operation. Parameters: (u32 read/write) timeout_ns: Watchdog timeout, in nanoseconds. This is initialized to 1,000,000,000 (1 second) at module load time. If more than this amount of time passes between calls to the pet_watchdog() function, the watchdog will bite. Functions: pet_watchdog(): Calling this function resets the watchdog timer and postpones the watchdog biting until timeout_ns nanoseconds later. ... It seems when the watchdog bites, the controller goes and hides in the dog house (stops communication and tri-states IO pins) until it's told that the coast is clear. It may be that an independent watchdog might be handy here too? Another use for a stand alone charge-pump/watchdog is to validate other hardware like limit switches by having an oscillator at the switch. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
Back in the '80s, we used to do something similar that also took care of the reset pulse width. (See attached). The values of the C's and Rs set the reset pulse width and wtdog time. We needed to have an ultra reliable system, but problem was that some power glitches could hang up processors. Wtdog counters could also get hung up - and we would see it happen with the early built-in uP wtdogs. Today they have the bugs out of it for the most part and if you have a good power supply and bypassing you won't have problems, yet stray alpha particles can still sometimes flip a bit. While you can get away with it most of the time, be aware that you can sometime cause parts to draw a lot of current if the input isn't supposed to see analog. You can get single gate schmitt triggers these days. If you are having problems with your system hanging, most likely you have a ground path problem. Home run grounding with a beefy pavilion ground bus is the best way. Karl Schmidt EMail k...@xtronics.com Transtronics, Inc. WEB http://xtronics.com 3209 West 9th Street Ph (785) 841-3089 Lawrence, KS 66049 FAX (785) 841-0434 Never put off till to-morrow what you can do the day after to-morrow just as well. -- Mark Twain attachment: resetgen.png-- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
Am 13.12.2010 22:52, schrieb Kirk Wallace: On Mon, 2010-12-13 at 16:36 +0100, dambacher-retrofit.de wrote: Am 13.12.2010 15:56, schrieb Stephen Wille Padnos: I don't think this is a concern. People who have hardware that removes the need for a base thread very likely also have a watchdog feature in their hardware. I know the Mesa and Pico products do, and I believe the Pluto does as well. (I'm not sure about Motenc and STG) Motenc has a watchdog circuit barking at 8 or 16ms (selectable) if you don't write DACs or the wd register. - Steve Just thinking out loud. I see the PWM controller has this in the doc's: Jumper JP4 can be selected to enable or disable the watchdog timer on the board. When enabled, the watchdog will force the board into Emergency Stop mode when the rate generator registers have not been written to in 20 mS. I guess one issue is that the integrator will need to make sure that the PWM controller e-stop signal out is connected to all of the e-stop slaves and not just EMC2, because if the controller initiates a watchdog fault, we can't rely on EMC2 to handle it. A proper e-stop system would commonly be set up this way, so it shouldn't be a problem. Also, this watchdog only validates EMC2. All commecial systems I saw were wired this way. if the control fails, the estop circuit gets triggered, main relay goes off, no power on drives. New VFDs have the possibility of doing a controlled emergency stop. They need power to do this. so you neeed addititonal logic to shut down power correctly, e.g. timed after estop. The Pluto doc's are a little more interesting: Before configuration and after properly exiting emc2, all Pluto-P pins are tristated with weak pull-ups (20kΩ min, 50kΩ max). If the watchdog timer is enabled (the default), these pins are also tristated after an interruption of communication between emc2 and the board. The watchdog timer takes approximately 6.5ms to activate. However, software bugs in the pluto_servo firmware or emc2 can leave the Pluto-P pins in an undefined state. An independent wathcdog might be prudent. With the Motec one can pet the dog internally with dac-write (from EMC2) or from an external source (using watchdog-reset as an input rather than output)? Only internal source (dac-write or direct write to wd-register) can reset the software watchdog. The Motenc handbook sais about function: The watchdog output is Active low. If the PC software enables the watchdog circuit and fails to reset the watchdog timer within the configured amount of time, the watchdog circuit resets the entire board and sets the watchdog output to logic 0. All DAC outputs, Digital outputs and Encoder counter are reset to zero. The watchdog output pin can be used to monitor this state by an external controller if needed. This pin is available on Header J8 pin 1. Software can also read the state of this pin at register location 6 bit 22. So besides triggering the watchdog it triggers internal reset and sets all pins/DACs to zero. If you wire correctly, your machine stops immediately. bye ulf -- --- Ulf Dambacher i...@dambacher-retrofit.de --- -- Lotusphere 2011 Register now for Lotusphere 2011 and learn how to connect the dots, take your collaborative environment to the next level, and enter the era of Social Business. http://p.sf.net/sfu/lotusphere-d2d ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sun, 2010-12-12 at 18:08 +1100, Erik Christiansen wrote: ... snip there is much greater risk of slightly different input thresholds causing the outputs to contest, thereby trying to short +5v to ground, when the input is slowly changing. (Counted in nanoseconds) There is then a lot to be said for removing IC1C, so the slow input is discriminated only once. If pins 1 3 are then connected, any small same-chip threshold variation will have negligible effect, because here the input flank is rather vertical. Dooh, of course. I'll make the change. http://www.wallacecompany.com/machine_shop/EMC2/Charge%20Pump/chpmp-5c.png I need to add a voltage regulator and filter caps, but there is no magic there. I've noticed Matt's circuit is leaner because it doesn't need one. My circuit might be improved with some I/O protection, but that will make it more complex. For just one 74HC14, a resistor and 5.1v zener diode will do. I'm used to a Zener on a gate input to limit the input voltage, but the circuit input is an opto-isolator which is not so delicate. I noticed an extra diode on my Bandit drivers opto's. My guess is that the opto diode has a lower reverse break-down voltage and the additional diode shores it up? I'm concerned about the buffer output, which could have a pull-up, pull-down, and or current limit resistor. I suppose a Zener on the output could help with over-voltage or reverse voltage, but the buffer is supposed to be driving an input not a source, but an inductive load could be a source, but... I usually put off the output components until I know what I'm going to hook it up to. Though, this doesn't help in coming up with a more generic consumer grade product. I've ordered some 74HC14's in order to build my latest try, but I'm getting the itch to move on. A bigger ratio between C1 and C2 would make it easier to reject a single input spike. But I can understand the attraction of seeing what the laternatives can do. Erik The present C1/C2 ratio seemed to give the best compromise, but with adding the 74HC14 and probing the buffer output instead of C2 may change that. I'll order a range of caps to play with. Thanks again. I'll post the results of testing, but it may be a while. I hope I'll remember all this, I'd like to do a write-up for the wiki if it isn't already there. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sun, Dec 12, 2010 at 10:49:33AM -0800, Kirk Wallace wrote: I need to add a voltage regulator and filter caps, but there is no magic there. I've noticed Matt's circuit is leaner because it doesn't need one. My circuit might be improved with some I/O protection, but that will make it more complex. For just one 74HC14, a resistor and 5.1v zener diode will do. I'm used to a Zener on a gate input to limit the input voltage, but the circuit input is an opto-isolator which is not so delicate. Sorry for confusing by placing the zener comment one sentence too late. It was intended to suggest using the zener in place of a voltage regulator IC. If there's little else requiring +5v, then a resistor and 400 mW zener will do to supply tens of milliamps at 5.1v. (Perhaps 5.2v when the load is low, and the zener has to shunt more.) I noticed an extra diode on my Bandit drivers opto's. My guess is that the opto diode has a lower reverse break-down voltage and the additional diode shores it up? The reverse breakdown voltage of a LED is not high, so the normal input voltage applied across LED and series resistor can easily be enough to destroy the LED, if accidentally reversed. (I avoid putting even 5v reverse polarity on them) The series diode blocks that. (My habit is to connect any old diode backwards across the LED, to clamp reverse bias at 0.7 or so. Any reversed drive is then dropped across the series resistor.) I'm concerned about the buffer output, which could have a pull-up, pull-down, and or current limit resistor. I suppose a Zener on the output could help with over-voltage or reverse voltage, but the buffer is supposed to be driving an input not a source, but an inductive load could be a source, but... I usually put off the output components until I know what I'm going to hook it up to. Though, this doesn't help in coming up with a more generic consumer grade product. That's tricky. You never know what people will do with an exposed output. Perhaps a small relay, with flywheel diode across the coil, even if a transistor is used to drive it, would both provide a robust output, and be easy to include in an Estop chain? Erik -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
Hello gentlemen, lately, I saw a lot of attempts to get schematic drawings of circuitry over the email line using ordinary sign characters, e.g. backslash or underscore or the like. Please keep in mind that these signs may make up a sensible sketch on your screen as you send it down the line, but on others like mine there appears a crazy heap of lines and signs making no sense at all (without redrawing it all, trying to understand what has been meant). It all depends on the font, character set and size, tab- and line spacing the different mail programs are using. You can't be sure that aligned characters stay aligned when they are displayed on a different system. The main problem with this is the use of national and especially proportional fonts. So, please use picture (graphic) formats for mailing drawings instead of this seemingly simple way. You can draw on scratch paper with a pencil which is easier than constructing pictures by means of the normal character set. Thank you Peter Blodow Erik Christiansen schrieb: On Fri, Dec 10, 2010 at 08:51:51PM -0800, Kirk Wallace wrote: On Thu, 2010-12-09 at 22:26 -0800, Kirk Wallace wrote: On Thu, 2010-12-09 at 20:06 -0800, Kirk Wallace wrote: ... snip I also wired this up: +- O'Scope _ _ | _| |_| || C1 .01uF ---\/\/-+---||--+ R1 15k | V (gnd) ... snip To provide a window comparator function, similar to the effect of a PLL tone detector like the NE567, or a suitably programmed AVR, just connect the two comparators of an LM393 like so: +5v |\| +5v-Ru--|-|+\ 2k2 GND-Rv--| e.g. 3v| \ | | \_| # It's open collector | / | # so the outputs make | / | # a wired AND here. From C2 ---|--|-/ | | |/| | LM393 | | |\| |--|+\ | | \ | | \_|__ High = 1v VC2 3v | / GND-Ry--| e.g. 1v| / +5v-Rx--|-|-/ |/ N.B. Input is to inverting input on the high threshold comparator, and non-inverting on the other. -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
Peter, As long as the list accepts attachments, that would work. Anybody have the definitive answer on whether the list is set up to accept attachments? Mark At 06:29 AM 12/11/2010, you wrote: Hello gentlemen, lately, I saw a lot of attempts to get schematic drawings of circuitry over the email line using ordinary sign characters, e.g. backslash or underscore or the like. Please keep in mind that these signs may make up a sensible sketch on your screen as you send it down the line, but on others like mine there appears a crazy heap of lines and signs making no sense at all (without redrawing it all, trying to understand what has been meant). It all depends on the font, character set and size, tab- and line spacing the different mail programs are using. You can't be sure that aligned characters stay aligned when they are displayed on a different system. The main problem with this is the use of national and especially proportional fonts. So, please use picture (graphic) formats for mailing drawings instead of this seemingly simple way. You can draw on scratch paper with a pencil which is easier than constructing pictures by means of the normal character set. Thank you Peter Blodow Erik Christiansen schrieb: On Fri, Dec 10, 2010 at 08:51:51PM -0800, Kirk Wallace wrote: On Thu, 2010-12-09 at 22:26 -0800, Kirk Wallace wrote: On Thu, 2010-12-09 at 20:06 -0800, Kirk Wallace wrote: ... snip I also wired this up: +- O'Scope _ _ | _| |_| || C1 .01uF ---\/\/-+---||--+ R1 15k | V (gnd) ... snip To provide a window comparator function, similar to the effect of a PLL tone detector like the NE567, or a suitably programmed AVR, just connect the two comparators of an LM393 like so: +5v |\| +5v-Ru--|-|+\ 2k2 GND-Rv--| e.g. 3v| \ | | \_| # It's open collector | / | # so the outputs make | / | # a wired AND here. From C2 ---|--|-/ | | |/| | LM393 | | |\| |--|+\ | | \ | | \_|__ High = 1v VC2 3v | / GND-Ry--| e.g. 1v| / +5v-Rx--|-|-/ |/ N.B. Input is to inverting input on the high threshold comparator, and non-inverting on the other. -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Fri, 10 Dec 2010 20:51:51 -0800 Kirk Wallace kwall...@wallacecompany.com wrote: Here is my latest stab at an all analog charge pump: http://www.wallacecompany.com/machine_shop/EMC2/chpmp-5a.png I'm thinking the opto-isolator on the input might be a good thing, but using an IC for one gate seemed kind of silly so I was thinking a FET on the output might work. I have no idea if the output circuit makes any sense. Comments are welcome. Here is a charge pump board I currently build: http://www.mattshaver.com/cp2estop/cp2estop.htm The web page isn't finished yet, but there is a schematic showing the use of an optoisolator. The board sells for $47.50, but I gather you are pretty much intending on a DIY course of action ;) Thanks, Matt -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sat, 11 Dec 2010 12:29 +0100, Peter Blodow p.blo...@dreki.de wrote: Hello gentlemen, lately, I saw a lot of attempts to get schematic drawings of circuitry over the email line using ordinary sign characters, e.g. backslash or underscore or the like. Please keep in mind that these signs may make up a sensible sketch on your screen as you send it down the line, but on others like mine there appears a crazy heap of lines and signs making no sense at all (without redrawing it all, trying to understand what has been meant). It all depends on the font, character set and size, tab- and line spacing the different mail programs are using. You can't be sure that aligned characters stay aligned when they are displayed on a different system. The main problem with this is the use of national and especially proportional fonts. Email is a pure text medium. Not HTML, not rich-text, not formatted text, just plain text. That includes using non-' proportional fonts. If your email client is using proportional fonts, you should be able to tell it not to. If you can't tell it to use a mono-spaced font, then it is broken. Are we also not allowed to show information in columns, because your proportional fonts will muck up the columns? For a complex drawing, by all means feel free to do it in a CAD program of your choice, export as a pdf or gif, post it on a webpage, and include a link in your message. But for a tiny and simple sketch, there is absolutely nothing wrong with ASCII art in a mono-spaced font. So, please use picture (graphic) formats for mailing drawings instead of this seemingly simple way. I'm not 100% sure if the list allows attachments or not. It definitely does NOT allow large attachments (or very large messages). Sending a 500K pdf or jpeg to every list member to show something that fits in a few lines of ASCII art is just rude. You can draw on scratch paper with a pencil which is easier than constructing pictures by means of the normal character set. And then what? How does the scratch paper sketch get into an email? I suppose you could scan it or take a digital photo, but both of those are a pain for a simple sketch. As an added bonus, the ASCII art is in-line with the rest of the message, while a graphical file needs to be opened in another tab or window, and you need to switch your attention back and forth from image to text as you read. So the ASCII version is more convenient for the reader - if they have their email client configured correctly. John Kasunich resident grouchy old traditionalist -- John Kasunich jmkasun...@fastmail.fm -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sat, 2010-12-11 at 17:53 +1100, Erik Christiansen wrote: ... snip Adding a pull-down resitor to OK1's emitter, followed by the 3 spare CMOS gates, wired in parallel, will provide the required push-pull drive to C1. (And remove the problem of wasted gates.) Thanks Erik. I to study what posted, but it seems to put me back on the right track. ... snip the input shorted to a supply. The .1uF C2 cap would show a ~2V spike when I made the +5V DC connection, I get nothing with the .33uF C2. I'm not sure that I follow the details of this, but the basic charge pump already checks for DC input (i.e. loss of oscillation), including short to either rail. If C1 is not actively driven high and low, then there is no output. The circuit works properly to detect non-AC input. It's just when the +5 input is connected, there is a one pump charge that gets through which is large enough to bring the output cap up a couple of volts in a spike. I am concerned that when the machine power is turned on and there might be a glitch on the output. A slightly larger output cap absorbs the glitch, but cap also charges up well enough with real AC input (or more than one pump in a row). -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sat, 2010-12-11 at 12:29 +0100, Peter Blodow wrote: Hello gentlemen, lately, I saw a lot of attempts to get schematic drawings of circuitry over the email line using ordinary sign characters, e.g. backslash or underscore or the like. The ASCII art is just me getting lazy. I was hoping that blocking the art and selecting Preformat would fix it, but I guess not. It would be nice to to be able to attach files, but I can just post a link. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sat, 11 Dec 2010 09:12 -0800, Kirk Wallace kwall...@wallacecompany.com wrote: The circuit works properly to detect non-AC input. It's just when the +5 input is connected, there is a one pump charge that gets through which is large enough to bring the output cap up a couple of volts in a spike. I am concerned that when the machine power is turned on and there might be a glitch on the output. A slightly larger output cap absorbs the glitch, but cap also charges up well enough with real AC input (or more than one pump in a row). -- This is exactly why the pump cap is ideally a lot smaller than the reservoir cap. That way it takes quite a few cycles to build up enough voltage in the reservoir cap and turn on the output. But you can only get away with a small pump cap when the pumping frequency is much faster than the output time constant. I like the water pump analogy that someone posted earlier. Continuing that analogy: the pump cap is the cylinder of your pump. The output cap is the bucket the pump empties into. The load resistor is a leak in the bucket. If you have a big pump cylinder, one stroke of the pump can put a lot of water in the bucket - you don't want that. A small cylinder needs a lot more strokes to fill up the bucket, which is good. But if the cylinder is too small, pumped too slowly, or the leak is too big, then the pumping might not be able to keep up with the leak. In that case, make the leak smaller or pump faster if you can. Making the pump cylinder bigger should be the last resort. Note that if you make the leak too small, then it will take a long time for the bucket to empty when you stop pumping. It's all about the tradeoffs... John Kasunich -- John Kasunich jmkasun...@fastmail.fm -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sat, Dec 11, 2010 at 11:37:03AM -0500, John Kasunich wrote: Email is a pure text medium. Not HTML, not rich-text, not formatted text, just plain text. That includes using non-' proportional fonts. If your email client is using proportional fonts, you should be able to tell it not to. If you can't tell it to use a mono-spaced font, then it is broken. Are we also not allowed to show information in columns, because your proportional fonts will muck up the columns? ... John Kasunich resident grouchy old traditionalist +1 small grump Being also old enough to honour hysterical raisins, I also submit that it is not reasonable to demand that use of the prime email format (plain text, fixed width) be banned. Everyone can use this fundamental email format, even if they don't use it exclusively, as some of us do. While I have incoming HTML automatically converted to plain text (fixed width font) for viewing, some do not accept it at all, as can be seen in the sig, below. So, no apologies offered for offering help in a long-standing standard format. /small grump Incidentally, I succeeded in sending a small circuit as a PDF attachment on this list on the second. (But the response didn't warrant a repeat of the effort.) Erik -- HTML is not email, and email doesn't contain HTML, so please turn HTML formatting OFF in your email client. We have filters in place that will reject your message if your posting contains HTML. - http://gpl-violations.org/mailinglists.html -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sat, Dec 11, 2010 at 09:12:24AM -0800, Kirk Wallace wrote: On Sat, 2010-12-11 at 17:53 +1100, Erik Christiansen wrote: ... snip Adding a pull-down resitor to OK1's emitter, followed by the 3 spare CMOS gates, wired in parallel, will provide the required push-pull drive to C1. (And remove the problem of wasted gates.) Thanks Erik. I to study what posted, but it seems to put me back on the right track. A quick and practical way to check it out could be to try your circuit as-is. It'll quickly become apparent that OK1 can charge C1, but not discharge it, so the pump won't function. A pull-down on OK1's emitter would allow C1 to discharge, but only slowly, since the value cannot be too low or OK1 can't pull C1 high any more. However, add the spare gates as buffer, and you're cooking with gas. ... snip the input shorted to a supply. The .1uF C2 cap would show a ~2V spike when I made the +5V DC connection, I get nothing with the .33uF C2. I'm not sure that I follow the details of this, but the basic charge pump already checks for DC input (i.e. loss of oscillation), including short to either rail. If C1 is not actively driven high and low, then there is no output. The circuit works properly to detect non-AC input. It's just when the +5 input is connected, there is a one pump charge that gets through which is large enough to bring the output cap up a couple of volts in a spike. I am concerned that when the machine power is turned on and there might be a glitch on the output. A slightly larger output cap absorbs the glitch, but cap also charges up well enough with real AC input (or more than one pump in a row). Ah, now I see. Yes, the circuit needs to be dimensioned to provide that discrimination, if it is to work. One pulse must never be enough to reach the detection threshold on the schmitt trigger or other comparator on the output. So C2 can be made significantly larger, and without going to electrolytics, which someone deprecated. 2uF ceramic SMD caps are tiny and cheap. Bang several in parallel, if desired. The discharge resistor controls how high a pulse frequency is needed, more than single pulse immunity. That's controlled by the C1/C2 ratio. Consider the charge Q tranferred from C1 to C2 on one pulse: C1*V1 = Q = C2*V2 If C2 = 10*C1, then V2 = V1/10 , minus diode drops. So you could use schottky diodes, if inclined, do reduce diode drops. But if you use CMOS, which has a much larger output voltage swing than the LS device I think I've seen mentioned (And I'd not consider using here), then ordinary silicon diodes will suffice. Hopefully that'll further help avoid the need to figure it all out the hard way. (Just yell out if too many details are spoiling the investigative fun. :-) Erik -- Most OpenBSD mailing lists strip messages of MIME content before sending them out to the rest of the list. If you don't use plain text your messages will be reformatted or, if they cannot be reformatted, summarily rejected. - http://www.openbsd.org/mail.html -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sun, 2010-12-12 at 15:59 +1100, Erik Christiansen wrote: ... snip That's controlled by the C1/C2 ratio. Consider the charge Q tranferred from C1 to C2 on one pulse: C1*V1 = Q = C2*V2 If C2 = 10*C1, then V2 = V1/10 , minus diode drops. I found that even without a resistor across the bucket cap the bucket would leak pretty rapidly. I don't know if the leak was into the scope probe or something else. The 1X/10X probe setting didn't seem to make a difference. I would think this unknown leak would make the calculations more difficult, unless one could model the leak. The .1 to .3 uF C1/C2 ratio seemed to pass all tests on the bench. ... snip Hopefully that'll further help avoid the need to figure it all out the hard way. (Just yell out if too many details are spoiling the investigative fun. :-) Erik When I was much younger, I was pretty decent at math and pursued mechanical engineering because I was fascinated by the thought of being able the mathematically model a physical system and actually have the real system match. I dropped out, but I still like to try to play with math. Although, most times it's easier to guess and just get-r-done. I revised my circuit: http://www.wallacecompany.com/machine_shop/EMC2/Charge%20Pump/chpmp-5b.png I need to add a voltage regulator and filter caps, but there is no magic there. I've noticed Matt's circuit is leaner because it doesn't need one. My circuit might be improved with some I/O protection, but that will make it more complex. I've ordered some 74HC14's in order to build my latest try, but I'm getting the itch to move on. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Sat, Dec 11, 2010 at 10:06:29PM -0800, Kirk Wallace wrote: On Sun, 2010-12-12 at 15:59 +1100, Erik Christiansen wrote: ... snip That's controlled by the C1/C2 ratio. Consider the charge Q tranferred from C1 to C2 on one pulse: C1*V1 = Q = C2*V2 If C2 = 10*C1, then V2 = V1/10 , minus diode drops. I found that even without a resistor across the bucket cap the bucket would leak pretty rapidly. I don't know if the leak was into the scope probe or something else. The 1X/10X probe setting didn't seem to make a difference. I would think this unknown leak would make the calculations more difficult, unless one could model the leak. The .1 to .3 uF C1/C2 ratio seemed to pass all tests on the bench. Well if the probe is 1Mohm, then the timeconstant is of the order of a third of a second, with the probe in place. Since the input resistance for a CMOS gate can be 10^14 Ohms (I haven't checked a schmitt gate, though), it can be used as a very high impedance monitor. If C2 is charged by briefly connecting +5v, after D2 R2 are disconnected, then CP GOOD should remain high for several months. After checking that it stays high for minutes, you can reconnect D2, and repeat the experiment, to see if it is leaky. If it is, then D1 is also, if C1 is OK. If nothing makes much difference, then C2 is leaky. If it's e.g. a greencap, out of the junkbox, then that's eminently possible. ... snip Hopefully that'll further help avoid the need to figure it all out the hard way. (Just yell out if too many details are spoiling the investigative fun. :-) Erik When I was much younger, I was pretty decent at math and pursued mechanical engineering because I was fascinated by the thought of being able the mathematically model a physical system and actually have the real system match. I dropped out, but I still like to try to play with math. Although, most times it's easier to guess and just get-r-done. Yeah, just enough maths to get the componentry dimensioned, then build it and see how it goes, suits me too. I was just average, so specialised in digital electronics, back in the early '70s. There was less maths and all the career scope of a nascent technology. I revised my circuit: http://www.wallacecompany.com/machine_shop/EMC2/Charge%20Pump/chpmp-5b.png I'm sorry that I always seem to find something to critique on those circuits, but whereas parallelling a couple of schmitt gates should work OK on a rapidly changing input, there is much greater risk of slightly different input thresholds causing the outputs to contest, thereby trying to short +5v to ground, when the input is slowly changing. (Counted in nanoseconds) There is then a lot to be said for removing IC1C, so the slow input is discriminated only once. If pins 1 3 are then connected, any small same-chip threshold variation will have negligible effect, because here the input flank is rather vertical. I need to add a voltage regulator and filter caps, but there is no magic there. I've noticed Matt's circuit is leaner because it doesn't need one. My circuit might be improved with some I/O protection, but that will make it more complex. For just one 74HC14, a resistor and 5.1v zener diode will do. I've ordered some 74HC14's in order to build my latest try, but I'm getting the itch to move on. A bigger ratio between C1 and C2 would make it easier to reject a single input spike. But I can understand the attraction of seeing what the laternatives can do. Erik -- Universities are places of knowledge. The freshmen each bring a little in with them, and the seniors take none away, so knowledge accumulates. -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Thu, 2010-12-09 at 22:26 -0800, Kirk Wallace wrote: On Thu, 2010-12-09 at 20:06 -0800, Kirk Wallace wrote: Some corrections: 74HC04=Hex Inverter, I tried a 2in NOR, but it's all the same. I blew out the only 74HC (74HC04 NAND) part in my junk box, so I cut my ... snip I also wired this up: +- O'Scope _ _ | _| |_| || C1 .01uF ---\/\/-+---||--+ R1 15k | V (gnd) ... snip I used this to measure the caps I had on hand and found that I really need to invest in a cap kit. I have just a few different values which isn't enough for tuning as circuit up or down 10 or 20%. Here is what I have on the bread board now: http://www.wallacecompany.com/machine_shop/EMC2/chpmp-4a.png I changed C2 from .1uF to .33uF. The 1kHz pumped voltage is now a little higher, ~4 V with very little ripple, and the decay rate is about the same. So I want to go to the next level. Here is my latest stab at an all analog charge pump: http://www.wallacecompany.com/machine_shop/EMC2/chpmp-5a.png I'm thinking the opto-isolator on the input might be a good thing, but using an IC for one gate seemed kind of silly so I was thinking a FET on the output might work. I have no idea if the output circuit makes any sense. Comments are welcome. I'm still working on a tone detector and AVR version. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Fri, 2010-12-10 at 20:51 -0800, Kirk Wallace wrote: ... snip I changed C2 from .1uF to .33uF. The 1kHz pumped voltage is now a little higher, ~4 V with very little ripple, and the decay rate is about the same. ... snip Oops, I forgot to mention, the larger cap helped with the DC high input signal. Part of my testing is to check the output for a 0V DC input to test for no signal or shorted to ground. Then I tie the input to +5V for the input shorted to a supply. The .1uF C2 cap would show a ~2V spike when I made the +5V DC connection, I get nothing with the .33uF C2. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
Re: [Emc-users] [OT] Charge Pump
On Fri, Dec 10, 2010 at 08:51:51PM -0800, Kirk Wallace wrote: On Thu, 2010-12-09 at 22:26 -0800, Kirk Wallace wrote: On Thu, 2010-12-09 at 20:06 -0800, Kirk Wallace wrote: ... snip I also wired this up: +- O'Scope _ _ | _| |_| || C1 .01uF ---\/\/-+---||--+ R1 15k | V (gnd) ... snip Here is what I have on the bread board now: http://www.wallacecompany.com/machine_shop/EMC2/chpmp-4a.png I changed C2 from .1uF to .33uF. The 1kHz pumped voltage is now a little higher, ~4 V with very little ripple, and the decay rate is about the same. So I want to go to the next level. Here is my latest stab at an all analog charge pump: http://www.wallacecompany.com/machine_shop/EMC2/chpmp-5a.png Kirk, that circuit has a small problem. Since OK1 can charge C1, but not discharge it, the circuit is hard put to produce any output. (If you refer back to the fluid pump analogy, you can drive the piston up, but not down.) Adding a pull-down resitor to OK1's emitter, followed by the 3 spare CMOS gates, wired in parallel, will provide the required push-pull drive to C1. (And remove the problem of wasted gates.) If you're ordering some parts, then an IC socket is useful while mucking about. I'm thinking the opto-isolator on the input might be a good thing, but using an IC for one gate seemed kind of silly so I was thinking a FET on the output might work. I have no idea if the output circuit makes any sense. Comments are welcome. The original approach, using schmitt trigger gates, was better, because that provides a clean transition from bad to good, over a narrow input voltage range. The FET will gradually sag from one state to the other as the voltage on C2 decays. To provide a window comparator function, similar to the effect of a PLL tone detector like the NE567, or a suitably programmed AVR, just connect the two comparators of an LM393 like so: +5v |\| +5v-Ru--|-|+\ 2k2 GND-Rv--| e.g. 3v| \ | | \_| # It's open collector | / | # so the outputs make | / | # a wired AND here. From C2 ---|--|-/ | | |/| | LM393 | | |\| |--|+\ | | \ | | \_|__ High = 1v VC2 3v | / GND-Ry--| e.g. 1v| / +5v-Rx--|-|-/ |/ N.B. Input is to inverting input on the high threshold comparator, and non-inverting on the other. Since the voltage references are ratiometrically derived from the power supply, this'll track supply induced variation in the voltage on C2. It will detect variation in mark-space ratio, even if the input frequency does not also change. In detecting changed software behaviour, that can be an asset. It may even work reasonably well for you, with thresholds set to (not too closely) bracket the normal voltage on C2. :-) I'm still working on a tone detector and AVR version. One benefit of that is that window detection is not subject to component tolerance, or drift with ageing, so long as we don't rely on the on-board RC oscillator. (If a crystal is being used, its drift will be very much smaller than the allowed input variation.) and also On Fri, Dec 10, 2010 at 09:08:54PM -0800, Kirk Wallace wrote: On Fri, 2010-12-10 at 20:51 -0800, Kirk Wallace wrote: ... snip I changed C2 from .1uF to .33uF. The 1kHz pumped voltage is now a little higher, ~4 V with very little ripple, and the decay rate is about the same. ... snip Oops, I forgot to mention, the larger cap helped with the DC high input signal. Part of my testing is to check the output for a 0V DC input to test for no signal or shorted to ground. Then I tie the input to +5V for the input shorted to a supply. The .1uF C2 cap would show a ~2V spike when I made the +5V DC connection, I get nothing with the .33uF C2. I'm not sure that I follow the details of this, but the basic charge pump already checks for DC input (i.e. loss of oscillation), including short to either rail. If C1 is not actively driven high and low, then there is no output. If some of the components for experimentation are coming off old circuit boards, then it's a great idea to test them first, as you're doing. Erik -- Oracle to DB2 Conversion Guide: Learn learn about native support for PL/SQL, new data types, scalar functions, improved concurrency, built-in packages, OCI, SQL*Plus, data movement tools, best practices and more. http://p.sf.net/sfu/oracle-sfdev2dev
Re: [Emc-users] [OT] Charge Pump
On Thu, 2010-12-09 at 20:06 -0800, Kirk Wallace wrote: Some corrections: 74HC04=Hex Inverter, I tried a 2in NOR, but it's all the same. I blew out the only 74HC (74HC04 NAND) part in my junk box, so I cut my ... snip I also wired this up: +- O'Scope _ _ | _| |_| || C1 .01uF ---\/\/-+---||--+ R1 15k | V (gnd) If T=RC, T=15000x.1=.15 ? .01uF = .01 x .01 = .0001F .0001 x 15000 = .00015 On the scope, I got a shark fin like signal where most of the signal change occurred by 200ms, should be .2ms = .0002ms so I guess this confirms the .15 for T above? I didn't quite understand what I read on Wikipedia -- something like RC is the value where the signal change is at around 63% of the signal swing? I suppose I could use this circuit to measure caps with C=T/R with t...@63% of signal? I'll have to play with this some more. It's confusing enough without goofing up the decimals. -- Kirk Wallace http://www.wallacecompany.com/machine_shop/ http://www.wallacecompany.com/E45/index.html California, USA -- ___ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users