Post by thread:[PIC]: Excessive PIC power at 32KHz

At 09:52 15/06/00 +0100, you wrote: >We use PICs running on 3.7V lithium cells at 32KHz for long battery life. >Typically our circuits operate between 12 and 18uA. We have used a number of >processors 16LC58A, 16LC558, 923, 924 etc. Always LC versions for 32KHz and >Industrial temperature range. > >The 58A and the 558 are products which Microchip are superseding with newer >parts: 58B, 622A (like a 558 but with comparators thrown in). However our >experience on Microchip's latest die shrinks is that the power consumption >at 32KHz has gone through the roof - typically x2 or x3. > >Has anyone else come across this problem? >Has anyone got any ideas? > >James Wilson Running at 32KHz is sometime just not the good solution even for current saving. Another approach that save cost and power is to run your microcontroller at maximum frequency using internal RC oscillator. As running the MCU at 32Khz is surely to count time, you may generate an interrupt, say each 10ms using an external circuitery running at 32Khz. A good and cheap one is a 4060 that is a counter, it integrate also an oscillator that might run with a 32Khz crystal. So an interrupt can be geenrated at each edge of one of the counter output, the rest of the time the MCU will enter in sleep mode saving maximum of current. Just make a little calcul for PIC16C622A RC osc. mode, 5V operation: Icc @ 4Mhz is 3,3 mA (max) Ipd is 20 uA (max) If you need, say, 1000 clock cycles to process your interrupt and do what you have to do, you will need: 1000 x 4 x 0,25us = 1ms @ 3,3mA if an interrupt occur each 100ms then you will spend 1ms @ 3,3mA and 99ms in power down mode @ 20uA, so finally the average will give: [1x3,3mA]/100 + [99x20uA]/100 = 33uA + 19,8uA = 52.8 uA This average must be majored by the needed time for the oscillator to restart, and then it need to add a "IccStart" value that can be quantified by experiment. Do not forget to add the current needed to run the 4060 ! This just show a high speed can save power in some case and allow high communication speed when really needed. Sure the calcul show we are still far from 18uA, but we take only MAX value, an experiment will surely show lowest current. Regards, Philippe. +--------------------------------------------------------+ | Virtual Micro Design | | UMPS: Universal assembler/disassembler/simulator | | for AVR, PIC, HC05, HC11, 8051, ST6, HOLTEK ... | | | | E-Mail: spam_OUTinfoTakeThisOuTvmdesign.com | | URL: http://www.vmdesign.com | +--------------------------------------------------------+

> power down mode @ 20uA, so finally the average will give: > [1x3,3mA]/100 + [99x20uA]/100 = 33uA + 19,8uA = 52.8 uA > > This average must be majored by the needed time for the oscillator to restart, > and then it need to add a "IccStart" value that can be quantified by experiment. > Do not forget to add the current needed to run the 4060 ! Which is about 40uA for an HC part @ 3V IIRC. 40uA + 53uA = 93uA vs. about 15uA in my application. Assuming the batteries last that long, we get 3 years out of them. With the 4060 solution that would be 6 months. Not much difference in microamps but a big difference in the product market place. It's a good technique but when you are at this level of power consumption all the wrong things draw current. For example, any reset generator will cost a year or more of product life. Steve. ====================================================== Steve Baldwin Electronic Product Design TLA Microsystems Ltd Microcontroller Specialists PO Box 15-680, New Lynn http://www.tla.co.nz Auckland, New Zealand ph +64 9 820-2221 email: .....stevebKILLspam@spam@tla.co.nz fax +64 9 820-1929 ======================================================

> However our > experience on Microchip's latest die shrinks is that the power consumption > at 32KHz has gone through the roof - typically x2 or x3. > > Has anyone else come across this problem? > Has anyone got any ideas? Yes. I had exactly the same problem with 54's and you'll also notice that the minimum operating voltage tends to creep up with die shrinks too. Not what I was expecting. I talked to one of the Microchip production guys about it at ESC last year. I can't remember the exact reasoning but it was quite reasonable. The upshot was that a die shrink gives them less production tolerance so they have to adjust parameters to make sure the yield stays within economic reality. Most devices are used at 5V so those specs are the ones that must stay the same or get better. If something has to suffer then it is better (for them) that it is the applications on the edge of the operating envelope, than those in the middle. The good news is that they don't discontinue the earlier versions although they cost more than the later ones. Steve. ====================================================== Steve Baldwin Electronic Product Design TLA Microsystems Ltd Microcontroller Specialists PO Box 15-680, New Lynn http://www.tla.co.nz Auckland, New Zealand ph +64 9 820-2221 email: stevebKILLspamtla.co.nz fax +64 9 820-1929 ======================================================

Steve Baldwin wrote: >> However our >> experience on Microchip's latest die shrinks is that the power consumption >> at 32KHz has gone through the roof - typically x2 or x3. >> >> Has anyone else come across this problem? >> Has anyone got any ideas? > >Yes. I had exactly the same problem with 54's and you'll also notice >that the minimum operating voltage tends to creep up with die >shrinks too. Not what I was expecting. >I talked to one of the Microchip production guys about it at ESC last >year. I can't remember the exact reasoning but it was quite >reasonable. The upshot was that a die shrink gives them less >production tolerance so they have to adjust parameters to make >sure the yield stays within economic reality. Most devices are used .......... The other thing you will notice is that the Vdd max rating is also moving in - was 6v, is 5.5v. So you have to watch it on that side, too. [2 Li batteries - no, no]. I just talked to Mchp about this 2 days ago, and they used much the same rationale as stated above - shrinking dies, "traces closer together", blah/blah/blah, .... - DanM ======

<x-rich>Steve Baldwin wrote: <color><param>0000,0000,0000</param>> It's a good technique but when you are at this level of power > consumption all the wrong things draw current. For example, any reset > generator will cost a year or more of product life. </color>Sorry, can't help on the current consumption problem but Seiko have some extremely low current parts that make good reset generators. Check these out. http://www.seiko-usa-ecd.com/intcir/html/power/s808.html <color><param>0100,0100,0100</param>á Very-low current consumption: : 1.3µA typ. (VDD=1.5V) When detection voltage is 1.4Vtyp. or less. : 0.8µA typ. (VDD=3.5V) When detection voltage is 1.5Vtyp. or more. á Accuracy: ±2.0% á Operating voltage range: : 0.7V to 5.0V When detection voltage is 1.4Vtyp. or less. : 0.95V to 10.0V When detection voltage is 1.5Vtyp. or more. á Hysteresis characteristics 5% typ. á Detection voltage : 0.8V to 6.0V (0.1V step) á Output forms: : Nch open drain active "L" output : CMOS active "L" output á Super-small package SC-82AB, TO-92, SOT-89-3, SOT-23-5</color> Brent Brown Electronic Design Solutions 16 English Street Hamilton, New Zealand Ph/fax: +64 7 849 0069 Mobile: 025 334 069 eMail: .....brent.brownKILLspam.....clear.net.nz </x-rich>

I currently have a product that uses the Watchdog sleep mode. So we wake up every 2.5 seconds for about a ms or so (we have a 4MHz oscillator) and decide whether we really want to be awake or not. It works well. We are getting a projected several years out of our lithium cells. I believe I set my steady state power budget at 100uA, with a bucnh of external circuitry. At 11:46 AM 6/15/00 +0200, you wrote: {Quote hidden}>At 09:52 15/06/00 +0100, you wrote: >>We use PICs running on 3.7V lithium cells at 32KHz for long battery life. >>Typically our circuits operate between 12 and 18uA. We have used a number of >>processors 16LC58A, 16LC558, 923, 924 etc. Always LC versions for 32KHz and >>Industrial temperature range. >> >>The 58A and the 558 are products which Microchip are superseding with newer >>parts: 58B, 622A (like a 558 but with comparators thrown in). However our >>experience on Microchip's latest die shrinks is that the power consumption >>at 32KHz has gone through the roof - typically x2 or x3. >> >>Has anyone else come across this problem? >>Has anyone got any ideas? >> >>James Wilson > >Running at 32KHz is sometime just not the good solution even for current >saving. Another approach that save cost and power is to run your >microcontroller >at maximum frequency using internal RC oscillator. As running the MCU >at 32Khz is surely to count time, you may generate an interrupt, say each 10ms >using an external circuitery running at 32Khz. A good and cheap one is a 4060 >that is a counter, it integrate also an oscillator that might run with a 32Khz >crystal. So an interrupt can be geenrated at each edge of one of the counter >output, the rest of the time the MCU will enter in sleep mode saving maximum >of current. >Just make a little calcul for PIC16C622A RC osc. mode, 5V operation: > Icc @ 4Mhz is 3,3 mA (max) > Ipd is 20 uA (max) > >If you need, say, 1000 clock cycles to process your interrupt and do what >you have to do, you will need: > 1000 x 4 x 0,25us = 1ms @ 3,3mA >if an interrupt occur each 100ms then you will spend 1ms @ 3,3mA and 99ms in >power down mode @ 20uA, so finally the average will give: > [1x3,3mA]/100 + [99x20uA]/100 = 33uA + 19,8uA = 52.8 uA > >This average must be majored by the needed time for the oscillator to restart, >and then it need to add a "IccStart" value that can be quantified by >experiment. >Do not forget to add the current needed to run the 4060 ! > >This just show a high speed can save power in some case and allow high >communication speed when really needed. Sure the calcul show we are still >far from 18uA, but we take only MAX value, an experiment will surely show >lowest current. > >Regards, > Philippe. > > +--------------------------------------------------------+ > | Virtual Micro Design | > | UMPS: Universal assembler/disassembler/simulator | > | for AVR, PIC, HC05, HC11, 8051, ST6, HOLTEK ... | > | | > | E-Mail: EraseMEinfospam_OUTTakeThisOuTvmdesign.com | > | URL: http://www.vmdesign.com | > +--------------------------------------------------------+ Erik Reikes Senior Software Engineer Xsilogy, Inc. ereikesspam_OUTxsilogy.com ph : (858) 535-5113 fax : (858) 535-5163 cell : (858) 663-1206