Dallas 1 wire isolation

[Adrian Jansen]

Hi All,

I am considering using a Dallas DS18S20 temp sensor for a remote (
outside air temp ) sensor. Sensor will be at most a few feet from the
processor doing the measurement but on a 2 wire ( ground and signal )
cable plugged into the instrument. I would like some way of protecting
the processor in case someone connects the 1-wire line either to ground,
or worse, +12 volt main supply. I cant see any nice easy way of doing
this given the drive constraints of the 1-wire system ( 1K5 pullup to +5
volt supply, processor pin pulls low to communicate on bus ).

Any ideas welcome.

--
Regards,

Adrian Jansen adrianjansen at internode dot on dot net
Design Engineer J & K Micro Systems
Microcomputer solutions for industrial control
Note reply address is invalid, convert address above to machine form.

 

[[email protected]]

I am considering using a Dallas DS18S20 temp sensor for a remote (

outside air temp ) sensor. Sensor will be at most a few feet from the
processor doing the measurement but on a 2 wire ( ground and signal )
cable plugged into the instrument. I would like some way of protecting
the processor in case someone connects the 1-wire line either to ground,
or worse, +12 volt main supply. I cant see any nice easy way of doing
this given the drive constraints of the 1-wire system ( 1K5 pullup to +5
volt supply, processor pin pulls low to communicate on bus ).

Use some zener and normal diodes to protect.

 

[Mochuelo]

Hi All,

I am considering using a Dallas DS18S20 temp sensor for a remote (
outside air temp ) sensor. Sensor will be at most a few feet from the
processor doing the measurement but on a 2 wire ( ground and signal )
cable plugged into the instrument. I would like some way of protecting
the processor in case someone connects the 1-wire line either to ground,
or worse, +12 volt main supply. I cant see any nice easy way of doing
this given the drive constraints of the 1-wire system ( 1K5 pullup to +5
volt supply, processor pin pulls low to communicate on bus ).

Any ideas welcome.

See Figure 4 on the datasheet? Right where it says "1-Wire Bus" insert
a 120 ohm, 0.5 W resistor. Then, add a 5.1 V, 0.5 W zener with anode
to ground and cathode to the left side of the resistor you just
inserted (so, drain of the PMOS). The connector (processor-cable) is
to the right of your 120 ohm resistor. That will allow you to connect
the 1-wire line to 12 V, without blowing anything (not even the
zener). To tell you whether your PMOS will die when you connect the
1-wire line to ground, you need to tell us which PMOS you have. It
needs to stand Ids=-42 mA and P=0.21 W (in the worst case).

Best,

 

[Joel Kolstad]

Use some zener and normal diodes to protect.

Or better yet, a transzorb -- which is electrically roughly the same as a
Zener diode, but specifically designed to generally fail short-circuit,
whereas Zeners typically fail open-circuit.

 

[[email protected]]

Or better yet, a transzorb -- which is electrically roughly the same as a
Zener diode, but specifically designed to generally fail short-circuit,
whereas Zeners typically fail open-circuit.

Wonder if there's any list with available semiconductor types somewhere..
(makes me wonder what more types I might have missed

 

[Mochuelo]

On 18 Jul 2006 07:32:07 GMT,

Use some zener and normal diodes to protect.

That does not protect against connecting the 1-wire line to ground.
You would blow up the PMOS. Same for a transzorb.

 

[Adrian Jansen]

See Figure 4 on the datasheet? Right where it says "1-Wire Bus" insert
a 120 ohm, 0.5 W resistor. Then, add a 5.1 V, 0.5 W zener with anode
to ground and cathode to the left side of the resistor you just
inserted (so, drain of the PMOS). The connector (processor-cable) is
to the right of your 120 ohm resistor. That will allow you to connect
the 1-wire line to 12 V, without blowing anything (not even the
zener). To tell you whether your PMOS will die when you connect the
1-wire line to ground, you need to tell us which PMOS you have. It
needs to stand Ids=-42 mA and P=0.21 W (in the worst case).

Best,

Hmm,

The short to ground is ok, the processor pin is internally current
limited to about 20 mA when driven into a short. It will withstand that
without any further protection.

The short to 12 volt is much harder.
5 volt zeners are a bit 'soft' for this. Typical for a 500mW zener
would be about 0.5 volt rise above the nominal zener voltage at 50 mA
load ( assuming 7 volts drop across the 120 ohm resistor ). Given the
zener tolerance of 4.8 - 5.4 volts, that will pull the processor pin up
to around 5.6 - 5.9 volts worst case, well beyond its Vcc + 0.3 volts
max rating.
Tranzorbs are good for fast spikes, but even 'softer'.

Putting a schottky diode cathode to power is probably better in
principle, except that it feeds the possible 12 volt fault input onto
the 5 volt rail. I doubt the voltage regulators will like this much.

Its not as simple a problem as it seems. But there must be a reasonable
solution, things like Dallas I-buttons have to work in all sorts of
hostile places with direct connections to processor pins.



--
Regards,

Adrian Jansen adrianjansen at internode dot on dot net
Design Engineer J & K Micro Systems
Microcomputer solutions for industrial control
Note reply address is invalid, convert address above to machine form.

 

[Mochuelo]

Hmm,

The short to ground is ok, the processor pin is internally current
limited to about 20 mA when driven into a short. It will withstand that
without any further protection.

The short to 12 volt is much harder.
5 volt zeners are a bit 'soft' for this. Typical for a 500mW zener
would be about 0.5 volt rise above the nominal zener voltage at 50 mA
load ( assuming 7 volts drop across the 120 ohm resistor ). Given the
zener tolerance of 4.8 - 5.4 volts, that will pull the processor pin up
to around 5.6 - 5.9 volts worst case, well beyond its Vcc + 0.3 volts
max rating.
Tranzorbs are good for fast spikes, but even 'softer'.

Putting a schottky diode cathode to power is probably better in
principle, except that it feeds the possible 12 volt fault input onto
the 5 volt rail. I doubt the voltage regulators will like this much.

Its not as simple a problem as it seems. But there must be a reasonable
solution, things like Dallas I-buttons have to work in all sorts of
hostile places with direct connections to processor pins.

You can use a programmable zener (such as a TL431). It will give you a
much more abrupt transition (it includes an opamp) and also the chance
to fine-tune the trip-point.

Regarding short to ground, I thought you were using parasitic supply
(you mentioned a 2-wire cable), in which case it would mainly be the
PMOS who would be at danger.

 

[Adrian Jansen]

You can use a programmable zener (such as a TL431). It will give you a
much more abrupt transition (it includes an opamp) and also the chance
to fine-tune the trip-point.

Regarding short to ground, I thought you were using parasitic supply
(you mentioned a 2-wire cable), in which case it would mainly be the
PMOS who would be at danger.

Thanks, I hadnt thought of an active zener.

Yes I will be using parasitic supply, but from what I see, a 1K passive
pullup will work, and the processor pin has enough pull-down to override
that. So I dont need an active pullup. Correct me if I am wrong.


--
Regards,

Adrian Jansen adrianjansen at internode dot on dot net
Design Engineer J & K Micro Systems
Microcomputer solutions for industrial control
Note reply address is invalid, convert address above to machine form.