AD595 fail ( Thermocouple IC )

[Sajjadkhan]

i havent found any article about my new problem that i found with AD595. i was attempting to make a soldering station and it was a success but sometime values flickered alot, not that often that is. then later i found that this "not that often" time gap is when i use to receive my sms or call or dial a call using my GSM based phone.

1st i thought it was the thermocouple leads which are of a meter long or my analog and digital grounding separation is not that effective. so i made a prototype on the bread board as standalone Celsius meter i.e. no pic attached to it. even then there are pulses going upto Vcc rail of Ad595. As i m using this ic as a standalone Celsius meter so there is no thermocouple attached to it rather a 1 inch jumper wire from pin 14 goes to pin 1 and pin 1 is effectively grounded(without any resistor). then i made a pcb just for ad595 like brakeout board and shorted pin1 and 14 directly under the pcb, noise was dramatically removed but still 0.5v noise remained.

Conclusion: using any sort of filter on pin 1 and 14 is of no use because op-amp on pin 1 and 14 has very high impedance,more than needed so it creates and effective antenna with a small piece of wire. so if you want to use it properly then you may have to shield your box and ground that shield, which i dindnt try yet, nor i m willing to. i will perform same experiment with max6675 and max31855 and will post the results.

 

[FvM]

Conclusion: using any sort of filter on pin 1 and 14 is of no use because op-amp on pin 1 and 14 has very high impedance

My conclusion is quite different. Thermocouples are low impedance sensors, thus thermocouple amplifier inputs can be easily and effectively filtered.

General comment. A popular fault with AD595 and similar differential amplifiers is to forgot the required ground/bias connection for the sensor. RF interfernces shouldn't be an issue. A solid circuit ground, e.g. PCB ground plane and sufficient bypass capacitors may be required however.

See AD595 datasheet:

Common-mode voltages on the thermocouple inputs must remain within the common-mode range of the AD594/AD595, with a return path provided for the bias currents. If the thermocouple is not remotely grounded, then the dotted line connections in Figures 1 and 2 are recommended.

 

[Sajjadkhan]

My conclusion is quite different. Thermocouples are low impedance sensors, thus thermocouple amplifier inputs can be easily and effectively filtered.

General comment. A popular fault with AD595 and similar differential amplifiers is to forgot the required ground/bias connection for the sensor. RF interfernces shouldn't be an issue. A solid circuit ground, e.g. PCB ground plane and sufficient bypass capacitors may be required however.

See AD595 datasheet:

AD595 datasheet 1st page: look into features: High Impedance Differential Input.

I have done what you are saying and whats in datasheet, as i said pin 1 and 14 are grounded and connected together. with thermocouple i always prefer to remotely ground it at the tip.

i also applied filter as in the application notes of AD595 but problem remains, since its in the pins. Have you tired Ad595 as in my conditions?

- - - Updated - - -

I am talking about op-amp impedance not the thermocouple impedance.

 

[FvM]

I am talking about op-amp impedance not the thermocouple impedance.

I know, but it doesn't necessarily involve a high interference succeptibility. The applicable filter impedance is commanded by the sensor impedance.

I believe that the RF interference sensitivity is caused by your grounding and supply bypass scheme.

I'm somewhat sceptic about noise level measurement numbers without a bandwidth specification. My performance criterion would be the stability of DC output value, e.g. measured with a multimeter with a preceeded a low-pass.

 

[ALERTLINKS]

Put a 0.1 uF capacitor between pin#1 and pin#14 and a 330 ohm resistor on pin#1 in series with thermocouple.
Pin#14 is grounded.