MAX6642
±1°C, SMBus-Compatible Remote/
Local Temperature Sensor with
Overtemperature Alarm
Maxim Integrated | 10www.maximintegrated.com
III. A thermal diode on the substrate of an IC is normally
a PNP with its collector grounded. Connect the anode
(emitter) to DXP and the cathode to GND of the
MAX6642.
If a sense transistor with an ideality factor other than
1.008 is used, the output data is different from the data
obtained with the optimum ideality factor. Fortunately,
the difference is predictable.
Assume a remote-diode sensor designed for a nominal
ideality factor n
NOMINAL
is used to measure the tem-
perature of a diode with a different ideality factor n
1
.
The measured temperature T
M
can be corrected using:
where temperature is measured in Kelvin and
n
NOMIMAL
for the MAX6642 is 1.008.
As an example, assume you want to use the MAX6642
with a CPU that has an ideality factor of 1.002. If the
diode has no series resistance, the measured data is
related to the real temperature as follows:
For a real temperature of +85°C (358.15K), the mea-
sured temperature is +82.91°C (356.02K), an error of
-2.13°C.
Effect of Series Resistance
Series resistance in a sense diode contributes addition-
al errors. For nominal diode currents of 10µA and
100µA, the change in the measured voltage due to
series resistance is:
V
M
= R
S
(100µA - 10µA) = 90µA R
S
Since +1°C corresponds to 198.6µV, series resistance
contributes a temperature offset of:
Assume that the diode being measured has a series
resistance of 3. The series resistance contributes an
offset of:
The effects of the ideality factor and series resistance
are additive. If the diode has an ideality factor of 1.002
and series resistance of 3, the total offset can be cal-
culated by adding error due to series resistance with
error due to ideality factor:
1.36°C - 2.13°C = -0.77°C
for a diode temperature of +85°C.
In this example, the effect of the series resistance and
the ideality factor partially cancel each other.
Discrete Remote Diodes
When the remote-sensing diode is a discrete transistor,
connect its collector and base together. Table 7 lists
examples of discrete transistors that are appropriate for
use with the MAX6642.
The transistor must be a small-signal type with a rela-
tively high forward voltage; otherwise, the A/D input
voltage range can be violated. The forward voltage at
the highest expected temperature must be greater than
0.25V at 10µA, and at the lowest expected tempera-
ture, the forward voltage must be less than 0.95V at
100µA. Large power transistors must not be used. Also,
ensure that the base resistance is less than 100. Tight
specifications for forward current gain (50 < ß <150, for
example) indicate that the manufacturer has good
process controls and that the devices have consistent
V
BE
characteristics.
Manufacturers of discrete transistors do not normally
specify or guarantee ideality factor. This is normally not
a problem since good-quality discrete transistors tend
to have ideality factors that fall within a relatively narrow
3 0 453 1 36Ω×
°
=+ ° ..
C
C
90
198 6
0 453
µ
µ
°
=
°
V
V
C
C
.
.
TT
n
n
T
T
ACTUAL M
NOMINAL
M
M
=
=
=
1
1 008
1 002
1 00599
.
.
( . )
TT
n
n
M ACTUAL
NOMINAL
=
1
MANUFACTURER MODEL NO.
Central Semiconductor (USA) CMPT3906
Rohm Semiconductor (USA) SST3906
Samsung (Korea) KST3906-TF
Siemens (Germany) SMBT3906
Zetex (England) FMMT3906CT-ND
Table 7. Remote-Sensor Transistor
Manufacturers
Note: Discrete transistors must be diode connected (base short-
ed to collector).
MAX6642
±1°C, SMBus-Compatible Remote/
Local Temperature Sensor with
Overtemperature Alarm
Maxim Integrated | 11www.maximintegrated.com
range. We have observed variations in remote tempera-
ture readings of less than ±2°C with a variety of dis-
crete transistors. Still, it is good design practice to
verify good consistency of temperature readings with
several discrete transistors from any manufacturer
under consideration.
ADC Noise Filtering
The integrating ADC used has good noise rejection for
low-frequency signals such as 60Hz/120Hz power-sup-
ply hum. In noisy environments, high-frequency noise
reduction is needed for high-accuracy remote mea-
surements. The noise can be reduced with careful PCB
layout and proper external noise filtering.
High-frequency EMI is best filtered at DXP with an
external 2200pF capacitor. Larger capacitor values can
be used for added filtering, but do not exceed 3300pF
because excessive capacitance can introduce errors
due to the rise time of the switched current source.
Nearly all noise sources tested cause the temperature
conversion results to be higher than the actual temper-
ature, typically by +1°C to +10°C, depending on the
frequency and amplitude (see the
Typical Operating
Characteristics
).
PCB Layout
Follow these guidelines to reduce the measurement
error of the temperature sensors:
1) Connect the thermal-sense diode to the MAX6642
using two traces—one between DXP and the
anode, the other between the MAX6642’s GND and
the cathode. Do not connect the cathode to GND at
the sense diode.
2) Place the MAX6642 as close as is practical to the
remote thermal diode. In noisy environments, such
as a computer motherboard, this distance can be
4in to 8in (typ). This length can be increased if the
worst noise sources are avoided. Noise sources
include CRTs, clock generators, memory buses,
and ISA/PCI buses.
3) Do not route the thermal diode lines next to the
deflection coils of a CRT. Also, do not route the
traces across fast digital signals, which can easily
introduce a 30°C error, even with good filtering.
4) Route the thermal diode traces in parallel and in
close proximity to each other, away from any higher
voltage traces, such as +12VDC. Leakage currents
from PCB contamination must be dealt with careful-
ly since a 20M leakage path from DXP to ground
causes about +1°C error. If high-voltage traces are
unavoidable, connect guard traces to GND on
either side of the DXP trace (Figure 4).
5) Route through as few vias and crossunders as pos-
sible to minimize copper/solder thermocouple
effects.
6) When introducing a thermocouple, make sure that
both the thermal diode paths have matching ther-
mocouples. A copper-solder thermocouple exhibits
3µV/°C, and it takes about 200µV of voltage error at
DXP to cause a +1°C measurement error. Adding a
few thermocouples causes a negligible error.
7) Use wide traces. Narrow traces are more inductive
and tend to pick up radiated noise. The 10-mil
widths and spacing recommended in Figure 4 are
not absolutely necessary, as they offer only a minor
improvement in leakage and noise over narrow
traces. Use wider traces when practical.
8) Add a 47 resistor in series with V
CC
for best noise
filtering (see the
Typical Operating Circuit
).
9) Copper cannot be used as an EMI shield; only fer-
rous materials such as steel work well. Placing a
copper ground plane between the DXP-DXN traces
and traces carrying high-frequency noise signals
does not help reduce EMI.
Twisted-Pair and Shielded Cables
Use a twisted-pair cable to connect the remote sensor
for remote-sensor distances longer than 8in or in very
noisy environments. Twisted-pair cable lengths can be
between 6ft and 12ft before noise introduces excessive
errors. For longer distances, the best solution is a
shielded twisted pair like that used for audio micro-
phones. For example, Belden #8451 works well for dis-
tances up to 100ft in a noisy environment. At the
device, connect the twisted pair to DXP and GND and
the shield to GND. Leave the shield unconnected at the
remote diode.
For very long cable runs, the cable’s parasitic capaci-
tance often provides noise filtering, so the 2200pF
capacitor can often be removed or reduced in value.
MINIMUM
10 mils
10 mils
10 mils
10 mils
THERMAL DIODE CATHODE/GND
THERMAL DIODE ANODE/DXP
GND
GND
Figure 4. Recommended DXP PC Traces
MAX6642
±1°C, SMBus-Compatible Remote/
Local Temperature Sensor with
Overtemperature Alarm
Maxim Integrated | 12www.maximintegrated.com
Cable resistance also affects remote-sensor accuracy.
For every 1 of series resistance, the error is approxi-
mately 1/2°C.
Thermal Mass and Self-Heating
When sensing local temperature, this device is intend-
ed to measure the temperature of the PCB to which it is
soldered. The leads provide a good thermal path
between the PCB traces and the die. Thermal conduc-
tivity between the die and the ambient air is poor by
comparison, making air temperature measurements
impractical. Because the thermal mass of the PCB is far
greater than that of the MAX6642, the device follows
temperature changes on the PCB with little or no per-
ceivable delay.
When measuring temperature of a CPU or other IC with
an on-chip sense junction, thermal mass has virtually
no effect; the measured temperature of the junction
tracks the actual temperature within a conversion cycle.
When measuring temperature with discrete remote sen-
sors, smaller packages, such as SOT23s, yield the best
thermal response times. Take care to account for ther-
mal gradients between the heat source and the sensor,
and ensure that stray air currents across the sensor
package do not interfere with measurement accuracy.
Self-heating does not significantly affect measurement
accuracy. Remote-sensor self-heating due to the diode
current source is negligible. For the local diode, the
worst-case error occurs when autoconverting at the
fastest rate and simultaneously sinking maximum cur-
rent at the ALERT output. For example, with V
CC
=
+5.0V, at an 8Hz conversion rate and with ALERT sink-
ing 1mA, the typical power dissipation is:
5.0V x 450µA + 0.4V x 1mA = 2.65mW
ø
J-A
for the 6-pin TDFN package is about +41°C/W, so
assuming no copper PCB heat sinking, the resulting
temperature rise is:
T = 2.65mW x 41°C/W = +0.11°C
Even under nearly worst-case conditions, it is difficult to
introduce a significant self-heating error.

MAX6642ATT92+T

Mfr. #:
Manufacturer:
Maxim Integrated
Description:
Board Mount Temperature Sensors SMBus-Compatible Temperature Sensor
Lifecycle:
New from this manufacturer.
Delivery:
DHL FedEx Ups TNT EMS
Payment:
T/T Paypal Visa MoneyGram Western Union