Reply by Fred Bartoli April 29, 20042004-04-29
"Spehro Pefhany" <speffSNIP@interlogDOTyou.knowwhat> a &#2013265929;crit dans le message
news: ua5290dg7186hr5091lj2um8h60fmbr9ae@4ax.com...
> > >If you have a power hungry uC and plenty of unused CPU time then you have > >die temperature control for (almost) free. > >Fred. > > Assuming you have an on-chip temperature sensor, just sleep the CPU to > maintain a constant die temperature? Or perhaps load some outputs > heavily? How to turn a complex mixed-signal microcontroller into a > LM199? > >
One have plenty of temperature sensors on a uC. Don't they have lots of available inputs ? (just use the clamping diodes). That way you could even have temperature mapping of the die, well at least at the periphery. Taking it a bit further then you can fine tune the code (which die area gets to switch) so as to draw an IR image on the die. Oh yes, that *slow* scan IR TV could well be as entertaining as ordinary programmes. Thanks, Fred.
Reply by Spehro Pefhany April 29, 20042004-04-29
On Thu, 29 Apr 2004 16:29:03 +0200, the renowned "Fred Bartoli"
<fred._canxxxel_this_bartoli@RemoveThatAlso_free.fr_AndThisToo> wrote:

> >"Spehro Pefhany" <speffSNIP@interlogDOTyou.knowwhat> a &#2013265929;crit dans le message >news: s7e090hmiiu7kj8aup5debfmhoc5o6u27f@4ax.com... >> On Thu, 29 Apr 2004 10:56:05 +1200, the renowned Jim Granville >> <no.spam@designtools.co.nz> wrote: >> > >> >Vref stability is a matter for the fine print. ALL Vrefs will differ >> >from unit to unit, and also vary over temperature. >> > The key point is, by how much ? >> > >> >[They quote 2.44V, or 2.43V Typical as Vref, so 'about 2.4401' does not >> >sound too terrible ? ] >> > >> > Their spec also says typ 15ppm/'C tempco. >> >That's not the very best, but Maxim sell plenty of Vrefs worse than >> >that. Remember this is an ON-CHIP, Vref. >> >> An on-chip Vref has the disadvantage that it is at the die temperature >> for the micro. No big deal if the micro is running at a couple of mA >> and not switching much, but if it is running hot you can get >> objectionable (perhaps on cosmetic grounds) drift during warm-up. >> > > >If you have a power hungry uC and plenty of unused CPU time then you have >die temperature control for (almost) free. >Fred.
Assuming you have an on-chip temperature sensor, just sleep the CPU to maintain a constant die temperature? Or perhaps load some outputs heavily? How to turn a complex mixed-signal microcontroller into a LM199? Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
Reply by Fred Bartoli April 29, 20042004-04-29
"Spehro Pefhany" <speffSNIP@interlogDOTyou.knowwhat> a &#2013265929;crit dans le message
news: s7e090hmiiu7kj8aup5debfmhoc5o6u27f@4ax.com...
> On Thu, 29 Apr 2004 10:56:05 +1200, the renowned Jim Granville > <no.spam@designtools.co.nz> wrote: > > > >Vref stability is a matter for the fine print. ALL Vrefs will differ > >from unit to unit, and also vary over temperature. > > The key point is, by how much ? > > > >[They quote 2.44V, or 2.43V Typical as Vref, so 'about 2.4401' does not > >sound too terrible ? ] > > > > Their spec also says typ 15ppm/'C tempco. > >That's not the very best, but Maxim sell plenty of Vrefs worse than > >that. Remember this is an ON-CHIP, Vref. > > An on-chip Vref has the disadvantage that it is at the die temperature > for the micro. No big deal if the micro is running at a couple of mA > and not switching much, but if it is running hot you can get > objectionable (perhaps on cosmetic grounds) drift during warm-up. >
If you have a power hungry uC and plenty of unused CPU time then you have die temperature control for (almost) free. Fred.
Reply by Bob Stephens April 29, 20042004-04-29
On Wed, 28 Apr 2004 20:17:12 -0400, Ben Bradley wrote:

> In sci.electronics.design,comp.arch.embedded, Bob Stephens > <stephensyomamadigital@earthlink.net> wrote: > > >>The only Cygnal part I have used is the '060 - I can't speak to the rest. >>The data sheet specs as regards power consumption, and internal Vref >>performance are extremely misleading at best. We are using it in a battery >>powered device, and with the analog peripherals running it is an extreme >>power hog. Vdd typical with CPU active is stated as 18 mA, "Power supply > ^^^^^^^ >>Current (each ADC)is given as 4.0 mA typical. > ^^^^^^^ > > <SOAPBOX,MODE="ON"> > > <ahem> > > DO NOT DESIGN WITH "TYPICAL" VALUES. > > Every chip can be at the stated minimum or maximum (whichver is > worst case) and still be within spec. If they don't have a max current > or power condumption spec, then that's bad, because it can pull five > amperes and (as long as everything else functions within guaranteed > specs) still be "within spec." > If there is some minumum or maximum spec that the chip isn't > meeting (as long as there's no asterisk where it says "these are just > approximate, we don't guarantee them" or other such weasel words), > then you can call the manufacturer and say they have defective parts, > and reasonably ask that they be replaced. > OTOH, every data sheet says specs 'may change at any time' so they > might just change the specs when you complain, but if any > manudfacturer did things like that there would be a LOT of people > complaining about it on Usenet and elsewhere, > > Bob Pease has given this lecture at least once, I forget if it was > in his ED column or 'live' as part of is National tour ... ISTR he > suggested running white-out down the 'typical' column of data sheets - > regardless, that's a good idea. > > >>Well, I'm measuring around >>85mA total for Vdd and AVdd with one converter running. Also "The >>independent, temperature stable 1.25 V bandgap voltage reference generator, >>with an output buffer amplifier which multiplies the bandgap reference by >>2" comes in at about 2.4401 volts except when it's warm or cold or ... and >>varies from device to device. I've had to add an external precision Vref to >>get any kind of stability out of the converters - not a big deal if planned >>up front but a thumping PITA to tack on to a densely packed board after the >>fact. > > ----- > http://mindspring.com/~benbradley
What you say is of course true. The problem in this case is that the "datasheet" is over 300 pages long and contradicts itself all over the place and as regards maximum and minimum specs the most common value for this part is "TBD"... Bob <SOAPBOX_MODE &= 0x00;> //you forgot to turn it off -- "Just machines that make big decisions programmed by fellas with compassion and vision." -D. Fagen (remove yomama)
Reply by Ben Bradley April 28, 20042004-04-28
In sci.electronics.design,comp.arch.embedded, Bob Stephens
<stephensyomamadigital@earthlink.net> wrote:


>The only Cygnal part I have used is the '060 - I can't speak to the rest. >The data sheet specs as regards power consumption, and internal Vref >performance are extremely misleading at best. We are using it in a battery >powered device, and with the analog peripherals running it is an extreme >power hog. Vdd typical with CPU active is stated as 18 mA, "Power supply
^^^^^^^
>Current (each ADC)is given as 4.0 mA typical.
^^^^^^^ <SOAPBOX,MODE="ON"> <ahem> DO NOT DESIGN WITH "TYPICAL" VALUES. Every chip can be at the stated minimum or maximum (whichver is worst case) and still be within spec. If they don't have a max current or power condumption spec, then that's bad, because it can pull five amperes and (as long as everything else functions within guaranteed specs) still be "within spec." If there is some minumum or maximum spec that the chip isn't meeting (as long as there's no asterisk where it says "these are just approximate, we don't guarantee them" or other such weasel words), then you can call the manufacturer and say they have defective parts, and reasonably ask that they be replaced. OTOH, every data sheet says specs 'may change at any time' so they might just change the specs when you complain, but if any manudfacturer did things like that there would be a LOT of people complaining about it on Usenet and elsewhere, Bob Pease has given this lecture at least once, I forget if it was in his ED column or 'live' as part of is National tour ... ISTR he suggested running white-out down the 'typical' column of data sheets - regardless, that's a good idea.
>Well, I'm measuring around >85mA total for Vdd and AVdd with one converter running. Also "The >independent, temperature stable 1.25 V bandgap voltage reference generator, >with an output buffer amplifier which multiplies the bandgap reference by >2" comes in at about 2.4401 volts except when it's warm or cold or ... and >varies from device to device. I've had to add an external precision Vref to >get any kind of stability out of the converters - not a big deal if planned >up front but a thumping PITA to tack on to a densely packed board after the >fact.
----- http://mindspring.com/~benbradley
Reply by Spehro Pefhany April 28, 20042004-04-28
On Thu, 29 Apr 2004 10:56:05 +1200, the renowned Jim Granville
<no.spam@designtools.co.nz> wrote:
> >Vref stability is a matter for the fine print. ALL Vrefs will differ >from unit to unit, and also vary over temperature. > The key point is, by how much ? > >[They quote 2.44V, or 2.43V Typical as Vref, so 'about 2.4401' does not >sound too terrible ? ] > > Their spec also says typ 15ppm/'C tempco. >That's not the very best, but Maxim sell plenty of Vrefs worse than >that. Remember this is an ON-CHIP, Vref.
An on-chip Vref has the disadvantage that it is at the die temperature for the micro. No big deal if the micro is running at a couple of mA and not switching much, but if it is running hot you can get objectionable (perhaps on cosmetic grounds) drift during warm-up. Best regards, Spehro Pefhany -- "it's the network..." "The Journey is the reward" speff@interlog.com Info for manufacturers: http://www.trexon.com Embedded software/hardware/analog Info for designers: http://www.speff.com
Reply by Jim Granville April 28, 20042004-04-28
Bob Stephens wrote:

> On 28 Apr 2004 12:39:08 -0700, Jeff Fox wrote: > > >>Bob Stephens <stephensyomamadigital@earthlink.net> wrote in message news:<vc0kdgkyxomt$.pswrgw8bn5wk.dlg@40tude.net>... >> >>>Actually, the Cygnal part (in *Theory*) provides all of these features and >>>more. If only they were a wee bit more honest in their datasheets... >> >>What do you consider to be misleading in the datasheets? Peak >>advertizing mips for artificial problems? That is pretty much >>common practice. Or is it something else? > > > <repost of reply to yet another post> > > The only Cygnal part I have used is the '060 - I can't speak to the rest. > The data sheet specs as regards power consumption, and internal Vref > performance are extremely misleading at best. We are using it in a battery > powered device, and with the analog peripherals running it is an extreme > power hog. Vdd typical with CPU active is stated as 18 mA, "Power supply > Current (each ADC)is given as 4.0 mA typical. Well, I'm measuring around > 85mA total for Vdd and AVdd with one converter running. Also "The > independent, temperature stable 1.25 V bandgap voltage reference generator, > with an output buffer amplifier which multiplies the bandgap reference by > 2" comes in at about 2.4401 volts except when it's warm or cold or ... and > varies from device to device. I've had to add an external precision Vref to > get any kind of stability out of the converters - not a big deal if planned > up front but a thumping PITA to tack on to a densely packed board after the > fact.
That sounds Atypical - you have, of course, asked Cygnal about it ? Vref stability is a matter for the fine print. ALL Vrefs will differ from unit to unit, and also vary over temperature. The key point is, by how much ? [They quote 2.44V, or 2.43V Typical as Vref, so 'about 2.4401' does not sound too terrible ? ] Their spec also says typ 15ppm/'C tempco. That's not the very best, but Maxim sell plenty of Vrefs worse than that. Remember this is an ON-CHIP, Vref. 'Best in class' on External Vrefs approaches a few ppm, but they will cost as much as the smaller uC :) If the tempco matters, you'd probably also want to do better than the 1.25% Vref tolerance. eg : Maxim can give Vref's with 3ppm and 0.04% Cals. The Icc numbers sound 'way out' - did you check their EVB Icc, and a number of samples ? What did Cygnal say ? I have not used their 060, but their 330 came in right where expected on Icc. -jg
Reply by Bob Stephens April 28, 20042004-04-28
On Wed, 28 Apr 2004 16:35:40 -0400, rickman wrote:

> Bob Stephens wrote: >> >> On Wed, 28 Apr 2004 17:47:47 +0200, Dejan Durdenic wrote: >> >>>> Actually, the Cygnal part (in *Theory*) provides all of these features and >>>> more. If only they were a wee bit more honest in their datasheets... >>>> >>>> >>>> Bob >>>> -- >>>> "Just machines that make big decisions >>>> programmed by fellas with compassion and vision." >>>> -D. Fagen >>>> (remove yomama) >>> >>> I'm working a lot with Cygnal devices and I'm curious what did you mean by >>> "If only they were a wee bit more honest in their datasheets..." ? >>> >>> - Dejan >> >> The only Cygnal part I have used is the '060 - I can't speak to the rest. >> The data sheet specs as regards power consumption, and internal Vref >> performance are extremely misleading at best. We are using it in a battery >> powered device, and with the analog peripherals running it is an extreme >> power hog. Vdd typical with CPU active is stated as 18 mA, "Power supply >> Current (each ADC)is given as 4.0 mA typical. Well, I'm measuring around >> 85mA total for Vdd and AVdd with one converter running. Also "The >> independent, temperature stable 1.25 V bandgap voltage reference generator, >> with an output buffer amplifier which multiplies the bandgap reference by >> 2" comes in at about 2.4401 volts except when it's warm or cold or ... and >> varies from device to device. I've had to add an external precision Vref to >> get any kind of stability out of the converters - not a big deal if planned >> up front but a thumping PITA to tack on to a densely packed board after the >> fact. > > That is sad to hear. I have not used any of the Cygnal parts, but I had > the impression from others that they were solid, well designed parts. > If you are seeing these problems over a batch of parts, is it possible > that they were not made well? Often internal parameters are not tested > and it is assumed that if the rest of the chip is good, then these > parameters should be good "by design". You might want to contact Cygnal > about it. > > One part that I know I have heard good things about on the analog side > is the MSP430 family. They are not 5 volt tolerant, but they seem to > come with a wide variety of peripherals and can run on very, very low > Idd. If you can deal with a 3.3 volt part on your board, this might be > a better choice.
Actually, my board is 3.3V. Thanks Rick, I'll take a look at the MSP. -- "Just machines that make big decisions programmed by fellas with compassion and vision." -D. Fagen (remove yomama)
Reply by rickman April 28, 20042004-04-28
Bob Stephens wrote:
> > On Wed, 28 Apr 2004 17:47:47 +0200, Dejan Durdenic wrote: > > >> Actually, the Cygnal part (in *Theory*) provides all of these features and > >> more. If only they were a wee bit more honest in their datasheets... > >> > >> > >> Bob > >> -- > >> "Just machines that make big decisions > >> programmed by fellas with compassion and vision." > >> -D. Fagen > >> (remove yomama) > > > > I'm working a lot with Cygnal devices and I'm curious what did you mean by > > "If only they were a wee bit more honest in their datasheets..." ? > > > > - Dejan > > The only Cygnal part I have used is the '060 - I can't speak to the rest. > The data sheet specs as regards power consumption, and internal Vref > performance are extremely misleading at best. We are using it in a battery > powered device, and with the analog peripherals running it is an extreme > power hog. Vdd typical with CPU active is stated as 18 mA, "Power supply > Current (each ADC)is given as 4.0 mA typical. Well, I'm measuring around > 85mA total for Vdd and AVdd with one converter running. Also "The > independent, temperature stable 1.25 V bandgap voltage reference generator, > with an output buffer amplifier which multiplies the bandgap reference by > 2" comes in at about 2.4401 volts except when it's warm or cold or ... and > varies from device to device. I've had to add an external precision Vref to > get any kind of stability out of the converters - not a big deal if planned > up front but a thumping PITA to tack on to a densely packed board after the > fact.
That is sad to hear. I have not used any of the Cygnal parts, but I had the impression from others that they were solid, well designed parts. If you are seeing these problems over a batch of parts, is it possible that they were not made well? Often internal parameters are not tested and it is assumed that if the rest of the chip is good, then these parameters should be good "by design". You might want to contact Cygnal about it. One part that I know I have heard good things about on the analog side is the MSP430 family. They are not 5 volt tolerant, but they seem to come with a wide variety of peripherals and can run on very, very low Idd. If you can deal with a 3.3 volt part on your board, this might be a better choice. -- Rick "rickman" Collins rick.collins@XYarius.com Ignore the reply address. To email me use the above address with the XY removed. Arius - A Signal Processing Solutions Company Specializing in DSP and FPGA design URL http://www.arius.com 4 King Ave 301-682-7772 Voice Frederick, MD 21701-3110 301-682-7666 FAX
Reply by Bob Stephens April 28, 20042004-04-28
On 28 Apr 2004 12:39:08 -0700, Jeff Fox wrote:

> Bob Stephens <stephensyomamadigital@earthlink.net> wrote in message news:<vc0kdgkyxomt$.pswrgw8bn5wk.dlg@40tude.net>... >> >> Actually, the Cygnal part (in *Theory*) provides all of these features and >> more. If only they were a wee bit more honest in their datasheets... > > What do you consider to be misleading in the datasheets? Peak > advertizing mips for artificial problems? That is pretty much > common practice. Or is it something else?
<repost of reply to yet another post> The only Cygnal part I have used is the '060 - I can't speak to the rest. The data sheet specs as regards power consumption, and internal Vref performance are extremely misleading at best. We are using it in a battery powered device, and with the analog peripherals running it is an extreme power hog. Vdd typical with CPU active is stated as 18 mA, "Power supply Current (each ADC)is given as 4.0 mA typical. Well, I'm measuring around 85mA total for Vdd and AVdd with one converter running. Also "The independent, temperature stable 1.25 V bandgap voltage reference generator, with an output buffer amplifier which multiplies the bandgap reference by 2" comes in at about 2.4401 volts except when it's warm or cold or ... and varies from device to device. I've had to add an external precision Vref to get any kind of stability out of the converters - not a big deal if planned up front but a thumping PITA to tack on to a densely packed board after the fact. -- "Just machines that make big decisions programmed by fellas with compassion and vision." -D. Fagen (remove yomama)