1. Home
  2. Electronics Design
  3. basal body temperature thermometer

basal body temperature thermometer

Hi sed!

I want to build a BBTT to chart the BBT [1] which fluctuates by less than 1 degC. Long term (1 month) precision is essential - accuracy is not required.

My first designs were around an instrumentation op amp and 2 resistor voltage dividers where one serves as voltage reference and the other has an NTC. The amplified difference would be digitized by a uC.

Now I stubled across another possibility. I used an ICM7555 timer (low power cmos version of NE555) with a small 10K NTC as timing resistor. See [2] for circuit. I suspended the NTC in a water/ice mixture. The frequency drifted and I was not sure if this was due to the ice water warming up or other effect. I was able to clearly detect when I hit the NTC with a 5 mW laser pointer. The frequency would also drift on power on until the NTC temperature had stabilized due to self heating.

Do you think this might be a suitable concept? Fewer components, less power, possibly easier data aquisition.

I guess the power on drift could be handled by counting the number of oscillation cycles for 1 ms from a consistent cold state. Waiting for the resistor temperature to stabilize would draw too much current.

I'm haven't yet decided on a power source. Could be a battery, a EDLC or a plain electrolytic. I don't think I'll do passive RFID. With an EDLC I thought of making a small SMPS to charge an electrolytic for each measurement. With only an electrolytic the voltage would drop during each night. This would also require a uC with very low drain sleep mode.

Hmm, I don't remember the Vdd dependence of the circuit.

Bernhard

[1] The BBT is the low body temperature during sleep. When one get's up/active the body temperature rises. The BBT may be measured internally for good precision, e.g. in the vagina, or externally, e.g. on the belly, which may be more convenient. The BBT fluctuates with the menstrual cycle and shall be charted to detect ovulation. [2]
formatting link
Reply to
Bernhard Kuemel
Loading thread data ...

You might like to consider the TI TMP275. This has a pre-calibrated accuracy of +0.1/-0.2 degC at body temperature and a resolution of

0.0625 degC. It has an i2c/smbus interface and is very easy to use. The quiescent current is about 50uA and doesn't change much during low- speed temperature acquisitions. Supply voltage can be between 2.7 and 5.5V.

A different approach could be to obtain an implantable identity chip of the type used for tagging pets. Some of these now have inbuilt temperature sensors. It should be easy to find a convenient location for such a small device - the fun would be in obtaining or building a suitable reader.

John

Reply to
John Walliker

An NTC in iced water should have a thermal time constant of a few seconds, unless you've buried it in a considerable thermal mass of insulation and protection.

A well-stirred ice bath is stable to about 1 milli-degree.

You can keep the self-heating constant by putting the NTC in series with a resistor which has the same resistance as the NTC at body temperatures, and driving both from temperature-stable voltage source, otherwise known as a voltage reference. There are lots of roughly

1.23V voltage references around - the LT1004-1.2 comes to mind

formatting link

though it may be better than you need.

If your "10k" NTC has a resistance of 10k at 25C, it will be closer to

6.2k at 37C, and would dissipate some 50uW when in series with a 6k2 resistor returned to 1.2V. 10mW/K is a ball-park thermal resistance to ambient for a small NTC, so 50uW is perhaps 5mK of self-heating, and it's going to be pretty stable in the vicinity of 37C.

It's a 555 and you are essentially comparing the voltage generated across the NTC with the voltage generated across a voltage divider created inside the 555 with a bunch of "resistors" created by diffusing stuff into silicon, which makes pretty crappy resistors.

You've also got to figure in the temperature stability if the timing capacitor, which is rarely as good as the stability of a decent metal film resistor - 15ppm/C parts are readily available from every broad- line distributor

-- Bill Sloman, Nijmegen

Reply to
Bill Sloman

You might like to consider the TI TMP275. This has a pre-calibrated accuracy of +0.1/-0.2 degC at body temperature and a resolution of

0.0625 degC. It has an i2c/smbus interface and is very easy to use. The quiescent current is about 50uA and doesn't change much during low- speed temperature acquisitions. Supply voltage can be between 2.7 and 5.5V.

A different approach could be to obtain an implantable identity chip of the type used for tagging pets. Some of these now have inbuilt temperature sensors. It should be easy to find a convenient location for such a small device - the fun would be in obtaining or building a suitable reader.

John

========================================

Are you sure about the accuracy of the TMP275?

From the data sheet : "The TMP275 is a 0.5°C accurate, Two-Wire, serial output

temperature sensor available in an MSOP-8 or an SO-8

package. The TMP275 is capable of reading temperatures

with a resolution of 0.0625°C."

I'm looking for a digital sensor with a precalibrated accuracy of 0.2C or better.

Reply to
Dennis

It all depends on the temperature range you are interested in.

So yes, I am sure in the context of measuring body temperature. The data sheet has graphs of temperature error vs temperature. The headline +/-0.5 deg C specification is for -20 to +100 deg C at 3.3V.

Reading from accuracy vs temperature graph, at 37 deg C the maximum error is better than +0.1 -0.2 deg C with a 5V supply.

Also from the data sheet: Accuracy -20 to +100 deg C 3.3V typ +/- 0.0625 max +/- 0.5 deg C Accuracy 0 to +100 deg C 3.0 to 3.6V typ +/- 0.0625 max +/- 0.75 deg C Accuracy -40 to +125 deg C 3.0 to 3.6V typ +/- 0.0625 max +/- 1.0 deg C Accuracy +25 to +100 deg C 3.3 to 5.5V typ +/- 0.2 max +/- 1.5 deg C

I have yet to find a "better" pre-calibrated temperature sensor at a reasonable price. If you find one, please let me know!

Calibrating temperature sensors in a production environment is a real pain, so I really like these devices. They "just work".

John

Reply to
John Walliker

If you're only looking for 98oF +/- 5o then a reverse biased Schottky diode driven by a constant current of 1-10uA will have 100's millivolts/oC sensitivity. You can't beat that for low power and low parts count. The output will be analog.

Reply to
Fred Bloggs

.

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D

or=20

Do everybody else a favor, fix the settings on your sorta news reader so that it quotes correctly. It can be done.

?-)

Reply to
josephkk

Do everybody else a favor, fix the settings on your sorta news reader so that it quotes correctly. It can be done.

?-)

====================================

Tell me more & I'll do it.

Reply to
Dennis

Dennis,

From the Tools, Options, Send tab, set news sending format to "plain text". Press the Plain Text Settings button and check the box for "Indent the original text with" then from the pull-down select ">"

Also, you might want to take a look here as Outlook Express is broken with regards to quoting properly even with the above settings:

formatting link

Reply to
JW

Hmmm, those are my current settings.

I have noted that occaisionally it does go screwy as it did in this thread.

I'll check out the oe-quotefix link. Thanks John.

Reply to
Dennis

which is silly, because, for whatever reason, OE indeed refuses to quotify "quoted-printable" text.

From your headers:

And what I sort-of mean by fault is, statistics. For whatever reason, most newsreaders default to a format which OE does quotify correctly. Your posts, however, do not, either by default (whatever Agent does) or by choice (in which case, you made a poor choice I guess).

At a glance, many posters use no Content-Transfer-Encoding tag; others use "7bit" or "8bit". Mine shows "7bit", with a longer "Content-Type" string. The next most common seems to be the problematic "quoted-printable".

At any rate, in OE, it's quite simple to change one's formatting. I expect, as a power user equipped with Agent, you know the equivalent settings.

Unfortunately, Google Groups also posts as quoted-printable. I haven't noticed if Google provides any control whatsoever over one's headers. However, perhaps Mr. Walker and others could be persuaded to adopt one of the many newsreaders available? If you don't get news with your ISP, AIOE and Eternal September are two free services I know of off hand. Simple to set up, no you can't exactly use it from any computer in the world like Google Groups, but the persistence and robust interface available in a proper newsreader is well worth it.

Tim

--
Deep Friar: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
Reply to
Tim Williams

which is silly, =

because,=20

text.

I do not get how MS incorrect implementation is my fault.

most=20

choice=20

use=20

string.=20

expect,=20

of=20

AIOE=20

to=20

Reply to
josephkk

I'll think about it. I like the search capabilities in GG and the ability to use it anywhere, but I do understand that there are some "not so good" aspects.

John

Reply to
John Walliker

of

I have tried GG's search capabilities from time to time, never found them all that useful. It is embedded in google search anyway, which is way more productive.

?-)

Reply to
josephkk

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.