On Wednesday, December 9, 2015 at 12:59:30 PM UTC, Tauno Voipio wrote:> On 9.12.15 14:02, steve wrote: > > On Tuesday, December 8, 2015 at 4:38:24 PM UTC, rickman wrote: > >> On 12/8/2015 9:54 AM, steve wrote: > >>> On Monday, December 7, 2015 at 5:03:19 PM UTC, Rob Gaddi wrote: > >>>> steve wrote: > >>>> > >>>>> I have a requirement were I need to get a processor up and running very quickly. 10ms in fact. Most power on reset circuits have a large variation on reset time of 5 to 15ms. I am looking for something that is accurate to say 20% and simple and cheap. Threshold of around 2.7V. I do have a 32kHz clock that appears when power above 2.8V which could possible be used, any ideas would be appreciated. > >>>> > >>>> 10ms from what? A power supply being plugged in? A physical power > >>>> switch being turned on? A wake-on-LAN message? And how quickly are > >>>> your power supplies going to go from 0 to ready? > >>>> > >>>> I ask because all that stuff bakes in additional ambiguity. ESPECIALLY > >>>> anything mechanical, which might chatter for milliseconds just on its > >>>> own. Speaking of mechanical things, how reliable is the start time of > >>>> your 32 kHz clock? > >>>> > >>>> If this requirement really is managable, I'm going to guess that your > >>>> best answer will be something with a shunt VREF, an open-drain > >>>> comparator, and an RC. The LM393 predates cave paintings, but is dirt > >>>> cheap and a fine choice for this; half to make a ramp and half to > >>>> square it up. > >>> > >>> 10ms from the 3V rail reaching 2.7V, the 32KHz clock is from an RTC chip that is permanently running. Its 32KHz output appears when the power rail hits 2.8V when its switches from battery to 3V power. The psu reaches 2.7V in less than 5ms. An external device triggers the PSU to start up and 30ms later it sends its data to be captured by us. Hoipe that makes sense > >> > >> Just curious, how much searching did you do? There are tons of reset > >> supervisors out there in small packages and low prices. > >> > >> How about the MAX6394? It has a 2.9 volt version with a 1 to 2 ms > >> timeout. > >> > >> -- > >> > >> Rick > > > > Rick I need 10 ms +/- 20%, this chip does 14ms min 28ms max which is the usual problem of a 3:1 min to max ratio > > > What prevents you from being ready before 10ms - 20% ? > > -- > > -TVIt is possible 5ms minimum 10ms max. 5ms is the amount of time the processors fast clock takes to start. From experimentation 5ms was ok 3ms was not, so 10ms seems a could place
Accurate fast start up power on reset
Started by ●December 7, 2015
Reply by ●December 9, 20152015-12-09
Reply by ●December 9, 20152015-12-09
On 12/9/2015 7:02 AM, steve wrote:> On Tuesday, December 8, 2015 at 4:38:24 PM UTC, rickman wrote: >> On 12/8/2015 9:54 AM, steve wrote: >>> On Monday, December 7, 2015 at 5:03:19 PM UTC, Rob Gaddi wrote: >>>> steve wrote: >>>> >>>>> I have a requirement were I need to get a processor up and running very quickly. 10ms in fact. Most power on reset circuits have a large variation on reset time of 5 to 15ms. I am looking for something that is accurate to say 20% and simple and cheap. Threshold of around 2.7V. I do have a 32kHz clock that appears when power above 2.8V which could possible be used, any ideas would be appreciated. >>>> >>>> 10ms from what? A power supply being plugged in? A physical power >>>> switch being turned on? A wake-on-LAN message? And how quickly are >>>> your power supplies going to go from 0 to ready? >>>> >>>> I ask because all that stuff bakes in additional ambiguity. ESPECIALLY >>>> anything mechanical, which might chatter for milliseconds just on its >>>> own. Speaking of mechanical things, how reliable is the start time of >>>> your 32 kHz clock? >>>> >>>> If this requirement really is managable, I'm going to guess that your >>>> best answer will be something with a shunt VREF, an open-drain >>>> comparator, and an RC. The LM393 predates cave paintings, but is dirt >>>> cheap and a fine choice for this; half to make a ramp and half to >>>> square it up. >>> >>> 10ms from the 3V rail reaching 2.7V, the 32KHz clock is from an RTC chip that is permanently running. Its 32KHz output appears when the power rail hits 2.8V when its switches from battery to 3V power. The psu reaches 2.7V in less than 5ms. An external device triggers the PSU to start up and 30ms later it sends its data to be captured by us. Hoipe that makes sense >> >> Just curious, how much searching did you do? There are tons of reset >> supervisors out there in small packages and low prices. >> >> How about the MAX6394? It has a 2.9 volt version with a 1 to 2 ms >> timeout. >> >> -- >> >> Rick > > Rick I need 10 ms +/- 20%, this chip does 14ms min 28ms max which is the usual problem of a 3:1 min to max ratioI didn't understand. I thought you just needed it to come up *within* 10 ms. You need an exact time delay? Someone suggested a 555 timer. But that is 10 ms from reaching 2.7 volts? That will require a comparator circuit with a good reference. The 555 has the comparator if you add a good reference. But that will need its own supply since it has to be up first I think. You said the 32 kHz output is enabled at 2.8 volts, so if you need a time from 2.7 volts that won't help. -- Rick
Reply by ●December 9, 20152015-12-09
Den onsdag den 9. december 2015 kl. 15.17.11 UTC+1 skrev steve:> On Wednesday, December 9, 2015 at 12:59:30 PM UTC, Tauno Voipio wrote: > > On 9.12.15 14:02, steve wrote: > > > On Tuesday, December 8, 2015 at 4:38:24 PM UTC, rickman wrote: > > >> On 12/8/2015 9:54 AM, steve wrote: > > >>> On Monday, December 7, 2015 at 5:03:19 PM UTC, Rob Gaddi wrote: > > >>>> steve wrote: > > >>>> > > >>>>> I have a requirement were I need to get a processor up and running very quickly. 10ms in fact. Most power on reset circuits have a large variation on reset time of 5 to 15ms. I am looking for something that is accurate to say 20% and simple and cheap. Threshold of around 2.7V. I do have a 32kHz clock that appears when power above 2.8V which could possible be used, any ideas would be appreciated. > > >>>> > > >>>> 10ms from what? A power supply being plugged in? A physical power > > >>>> switch being turned on? A wake-on-LAN message? And how quickly are > > >>>> your power supplies going to go from 0 to ready? > > >>>> > > >>>> I ask because all that stuff bakes in additional ambiguity. ESPECIALLY > > >>>> anything mechanical, which might chatter for milliseconds just on its > > >>>> own. Speaking of mechanical things, how reliable is the start time of > > >>>> your 32 kHz clock? > > >>>> > > >>>> If this requirement really is managable, I'm going to guess that your > > >>>> best answer will be something with a shunt VREF, an open-drain > > >>>> comparator, and an RC. The LM393 predates cave paintings, but is dirt > > >>>> cheap and a fine choice for this; half to make a ramp and half to > > >>>> square it up. > > >>> > > >>> 10ms from the 3V rail reaching 2.7V, the 32KHz clock is from an RTC chip that is permanently running. Its 32KHz output appears when the power rail hits 2.8V when its switches from battery to 3V power. The psu reaches 2.7V in less than 5ms. An external device triggers the PSU to start up and 30ms later it sends its data to be captured by us. Hoipe that makes sense > > >> > > >> Just curious, how much searching did you do? There are tons of reset > > >> supervisors out there in small packages and low prices. > > >> > > >> How about the MAX6394? It has a 2.9 volt version with a 1 to 2 ms > > >> timeout. > > >> > > >> -- > > >> > > >> Rick > > > > > > Rick I need 10 ms +/- 20%, this chip does 14ms min 28ms max which is the usual problem of a 3:1 min to max ratio > > > > > > What prevents you from being ready before 10ms - 20% ? > > > > -- > > > > -TV > > It is possible 5ms minimum 10ms max. 5ms is the amount of time the processors fast clock takes to start. From experimentation 5ms was ok 3ms was not, so 10ms seems a could placecan't the processor wake up on the slow clock and switch to the fast when it is ready? -Lasse
Reply by ●December 10, 20152015-12-10
On Wednesday, December 9, 2015 at 6:11:54 PM UTC, lasselangwad...@gmail.com wrote:> Den onsdag den 9. december 2015 kl. 15.17.11 UTC+1 skrev steve: > > On Wednesday, December 9, 2015 at 12:59:30 PM UTC, Tauno Voipio wrote: > > > On 9.12.15 14:02, steve wrote: > > > > On Tuesday, December 8, 2015 at 4:38:24 PM UTC, rickman wrote: > > > >> On 12/8/2015 9:54 AM, steve wrote: > > > >>> On Monday, December 7, 2015 at 5:03:19 PM UTC, Rob Gaddi wrote: > > > >>>> steve wrote: > > > >>>> > > > >>>>> I have a requirement were I need to get a processor up and running very quickly. 10ms in fact. Most power on reset circuits have a large variation on reset time of 5 to 15ms. I am looking for something that is accurate to say 20% and simple and cheap. Threshold of around 2.7V. I do have a 32kHz clock that appears when power above 2.8V which could possible be used, any ideas would be appreciated. > > > >>>> > > > >>>> 10ms from what? A power supply being plugged in? A physical power > > > >>>> switch being turned on? A wake-on-LAN message? And how quickly are > > > >>>> your power supplies going to go from 0 to ready? > > > >>>> > > > >>>> I ask because all that stuff bakes in additional ambiguity. ESPECIALLY > > > >>>> anything mechanical, which might chatter for milliseconds just on its > > > >>>> own. Speaking of mechanical things, how reliable is the start time of > > > >>>> your 32 kHz clock? > > > >>>> > > > >>>> If this requirement really is managable, I'm going to guess that your > > > >>>> best answer will be something with a shunt VREF, an open-drain > > > >>>> comparator, and an RC. The LM393 predates cave paintings, but is dirt > > > >>>> cheap and a fine choice for this; half to make a ramp and half to > > > >>>> square it up. > > > >>> > > > >>> 10ms from the 3V rail reaching 2.7V, the 32KHz clock is from an RTC chip that is permanently running. Its 32KHz output appears when the power rail hits 2.8V when its switches from battery to 3V power. The psu reaches 2.7V in less than 5ms. An external device triggers the PSU to start up and 30ms later it sends its data to be captured by us. Hoipe that makes sense > > > >> > > > >> Just curious, how much searching did you do? There are tons of reset > > > >> supervisors out there in small packages and low prices. > > > >> > > > >> How about the MAX6394? It has a 2.9 volt version with a 1 to 2 ms > > > >> timeout. > > > >> > > > >> -- > > > >> > > > >> Rick > > > > > > > > Rick I need 10 ms +/- 20%, this chip does 14ms min 28ms max which is the usual problem of a 3:1 min to max ratio > > > > > > > > > What prevents you from being ready before 10ms - 20% ? > > > > > > -- > > > > > > -TV > > > > It is possible 5ms minimum 10ms max. 5ms is the amount of time the processors fast clock takes to start. From experimentation 5ms was ok 3ms was not, so 10ms seems a could place > > can't the processor wake up on the slow clock and switch to the fast when > it is ready? > > -LasseThe processor needs at least 5ms of slow clock, which is 32768 whilst held in reset. It could be done using the 32768 clock, this is always running but its output is enabled when supply reaches 2.8V. This could drive a divide by 256 ripple counter, if such a thing existed in a very small package other wise an accurate time reset circuit, maybe using a voltage detector and a 555 type circuit.
Reply by ●December 10, 20152015-12-10
On 12/10/2015 4:17 AM, steve wrote:> On Wednesday, December 9, 2015 at 6:11:54 PM UTC, lasselangwad...@gmail.com wrote: >> Den onsdag den 9. december 2015 kl. 15.17.11 UTC+1 skrev steve: >>> On Wednesday, December 9, 2015 at 12:59:30 PM UTC, Tauno Voipio wrote: >>>> On 9.12.15 14:02, steve wrote: >>>>> On Tuesday, December 8, 2015 at 4:38:24 PM UTC, rickman wrote: >>>>>> On 12/8/2015 9:54 AM, steve wrote: >>>>>>> On Monday, December 7, 2015 at 5:03:19 PM UTC, Rob Gaddi wrote: >>>>>>>> steve wrote: >>>>>>>> >>>>>>>>> I have a requirement were I need to get a processor up and running very quickly. 10ms in fact. Most power on reset circuits have a large variation on reset time of 5 to 15ms. I am looking for something that is accurate to say 20% and simple and cheap. Threshold of around 2.7V. I do have a 32kHz clock that appears when power above 2.8V which could possible be used, any ideas would be appreciated. >>>>>>>> >>>>>>>> 10ms from what? A power supply being plugged in? A physical power >>>>>>>> switch being turned on? A wake-on-LAN message? And how quickly are >>>>>>>> your power supplies going to go from 0 to ready? >>>>>>>> >>>>>>>> I ask because all that stuff bakes in additional ambiguity. ESPECIALLY >>>>>>>> anything mechanical, which might chatter for milliseconds just on its >>>>>>>> own. Speaking of mechanical things, how reliable is the start time of >>>>>>>> your 32 kHz clock? >>>>>>>> >>>>>>>> If this requirement really is managable, I'm going to guess that your >>>>>>>> best answer will be something with a shunt VREF, an open-drain >>>>>>>> comparator, and an RC. The LM393 predates cave paintings, but is dirt >>>>>>>> cheap and a fine choice for this; half to make a ramp and half to >>>>>>>> square it up. >>>>>>> >>>>>>> 10ms from the 3V rail reaching 2.7V, the 32KHz clock is from an RTC chip that is permanently running. Its 32KHz output appears when the power rail hits 2.8V when its switches from battery to 3V power. The psu reaches 2.7V in less than 5ms. An external device triggers the PSU to start up and 30ms later it sends its data to be captured by us. Hoipe that makes sense >>>>>> >>>>>> Just curious, how much searching did you do? There are tons of reset >>>>>> supervisors out there in small packages and low prices. >>>>>> >>>>>> How about the MAX6394? It has a 2.9 volt version with a 1 to 2 ms >>>>>> timeout. >>>>>> >>>>>> -- >>>>>> >>>>>> Rick >>>>> >>>>> Rick I need 10 ms +/- 20%, this chip does 14ms min 28ms max which is the usual problem of a 3:1 min to max ratio >>>> >>>> >>>> What prevents you from being ready before 10ms - 20% ? >>>> >>>> -- >>>> >>>> -TV >>> >>> It is possible 5ms minimum 10ms max. 5ms is the amount of time the processors fast clock takes to start. From experimentation 5ms was ok 3ms was not, so 10ms seems a could place >> >> can't the processor wake up on the slow clock and switch to the fast when >> it is ready? >> >> -Lasse > The processor needs at least 5ms of slow clock, which is 32768 whilst held in reset. It could be done using the 32768 clock, this is always running but its output is enabled when supply reaches 2.8V. This could drive a divide by 256 ripple counter, if such a thing existed in a very small package other wise an accurate time reset circuit, maybe using a voltage detector and a 555 type circuit.It's not clear how much time happens between the power supply crossing 2.7 volts and crossing 2.8 volts, but a simple counter chip will allow the 32 kHz to be used for accurate timing from the 2.8 volt crossing... assuming the clock chip has an *accurate* 2.8 volt detector. Which clock chip? http://www.intersil.com/content/dam/Intersil/documents/cd40/cd4020bms-24bms-40bms.pdf The three counters in the above data sheet give you binary divides from the 32 kHz. The reset would need to be asserted until power is up enough to work which may require a reset detector chip with no delay or the one I specified with minimum delay might do. You get binary multiples of slightly less than 1 ms. The CD4020 would do well and there is an HCT equivalent. http://www.nxp.com/documents/data_sheet/74HC_HCT4020.pdf So this might be 2 chips, but the reset chip is pretty minimal. The 74HC(T) is available in a QFN so it's durn small and probably cheap as well. -- Rick
Reply by ●December 11, 20152015-12-11
rickman wrote:> On 12/8/2015 9:54 AM, steve wrote: >> On Monday, December 7, 2015 at 5:03:19 PM UTC, Rob Gaddi wrote: >>> steve wrote: >>> >>>> I have a requirement were I need to get a processor up and running >>>> very quickly. 10ms in fact. Most power on reset circuits have a >>>> large variation on reset time of 5 to 15ms. I am looking for >>>> something that is accurate to say 20% and simple and cheap. >>>> Threshold of around 2.7V. I do have a 32kHz clock that appears when >>>> power above 2.8V which could possible be used, any ideas would be >>>> appreciated. >>> >>> 10ms from what? A power supply being plugged in? A physical power >>> switch being turned on? A wake-on-LAN message? And how quickly are >>> your power supplies going to go from 0 to ready? >>> >>> I ask because all that stuff bakes in additional ambiguity. ESPECIALLY >>> anything mechanical, which might chatter for milliseconds just on its >>> own. Speaking of mechanical things, how reliable is the start time of >>> your 32 kHz clock? >>> >>> If this requirement really is managable, I'm going to guess that your >>> best answer will be something with a shunt VREF, an open-drain >>> comparator, and an RC. The LM393 predates cave paintings, but is dirt >>> cheap and a fine choice for this; half to make a ramp and half to >>> square it up. >> >> 10ms from the 3V rail reaching 2.7V, the 32KHz clock is from an RTC >> chip that is permanently running. Its 32KHz output appears when the >> power rail hits 2.8V when its switches from battery to 3V power. The >> psu reaches 2.7V in less than 5ms. An external device triggers the PSU >> to start up and 30ms later it sends its data to be captured by us. >> Hoipe that makes sense > > Just curious, how much searching did you do? There are tons of reset > supervisors out there in small packages and low prices. > > How about the MAX6394? It has a 2.9 volt version with a 1 to 2 ms timeout. >MAXIM???? Are you nuts? They have been known for ages to announce things they (for practical purposes) do not make, drop "product" like a hot potato with no notice, etc. Any other manufacturer, including none is better and far more reliable.
Reply by ●December 11, 20152015-12-11
On 11.12.2015 г. 09:21, Robert Baer wrote:> .... >> >> How about the MAX6394? It has a 2.9 volt version with a 1 to 2 ms >> timeout. >> > MAXIM???? Are you nuts? They have been known for ages to announce > things they (for practical purposes) do not make, drop "product" like a > hot potato with no notice, etc. > Any other manufacturer, including none is better and far more reliable. > >Well this is not entirely fair. Much of it true may be, but Maxim have also had some good products for some years. And they have always been helpful with samples to me when they could not deliver small quantities. For example, they (still?) are the source of various fast comparators, nobody else could match them when I needed them over the years (not quite recently but I used some again 2-3 years ago). Then they used to have good references - nowadays ADI beat them (and ADI just do not drop products, which is truly great) but I have used theirs for years (in small quantities indeed). More to the point - reset circuits - I'd probably agree with you though. Many years ago I used a MAX695 - which worked as specified but I misused it somewhat (in a sane way), and all my misuse went wrong (unlike other times when I have misused parts). Was no drama but was not nice either. Dimiter ------------------------------------------------------ Dimiter Popoff, TGI http://www.tgi-sci.com ------------------------------------------------------ http://www.flickr.com/photos/didi_tgi/
