I85PREP
⌘ K
UGC NET Electronic Science study materials
UGC NET Electronic Science study resources

Thermocouple front end that saturates and picks up heater noise

UGC NET Electronic Science · Put your knowledge to work in context.

All case studies

Thermocouple front end that saturates and picks up heater noise

A furnace monitoring channel uses a Type-K thermocouple with a Seebeck coefficient of about 41 µV/°C and a cold-junction compensator. The intended range is 0–600 °C, so the full-scale thermocouple voltage is about 24.6 mV. The front end is a three-op-amp instrumentation amplifier followed by a 16-bit, 0–2.5 V ADC; the reference and supply are fixed, and 0.5 V of headroom must be reserved at the ADC input. The prototype uses 1% gain-setting resistors and a gain of 500, but it saturates at roughly 130 °C. The op-amps have 86 dB CMRR at DC, falling to 60 dB at 50 Hz. An ungrounded junction sits near a switched heater that imposes 1.5 V of 50 Hz common-mode voltage on the leads, and the displayed temperature jumps by about ±3 °C whenever the heater turns on. No chopper-stabilized amplifier is available; only shielding, twisted pair and RC filtering can be added.

1. Explain why the gain of 500 saturates and select a gain that covers 0–600 °C with the stated headroom. What input-referred resolution results?

Read the worked answer

Saturation check

  • Full-scale thermocouple voltage: 600 °C × 41 µV/°C = 24.6 mV.
  • At G = 500 the input that produces 2.5 V is 2.5/500 = 5 mV, which corresponds to 5 mV / 41 µV/°C ≈ 122 °C. The observed saturation near 130 °C is consistent with this, so the fault is gain selection, not a faulty amplifier.

Allowable gain

  • Usable output swing = 2.5 V − 0.5 V = 2.0 V.
  • G_max = 2.0 V / 24.6 mV = 81.3, so a standard gain of 80 is appropriate.
  • At 600 °C the output is 24.6 mV × 80 = 1.97 V, which leaves the required headroom.

Resolution

  • 1 LSB at the ADC = 2.5/2^16 = 38.1 µV.
  • Input-referred: 38.1 µV/80 = 0.48 µV, or about 0.012 °C.
  • Resolution is therefore not the limiting factor; CMRR, offset drift, noise and thermocouple nonlinearity dominate. A single 41 µV/°C slope is itself an approximation over 0–600 °C, so a polynomial or table-based linearisation is needed to keep the absolute error within a few °C.

2. Quantify the 50 Hz interference that reaches the amplifier input and specify an input filter that reduces the heater-induced error below 0.2 °C.

Read the worked answer

Interference at the input

  • At 50 Hz the CMRR is 60 dB, i.e. a factor of 1000.
  • 1.5 V of common-mode voltage therefore appears as 1.5 V/1000 = 1.5 mV of differential-mode error at the input.
  • Referred to temperature: 1.5 mV / 41 µV/°C ≈ 36.6 °C. This is far above the observed ±3 °C only because some stray filtering and wiring balance already exist; it is still unacceptable.

Required attenuation

  • Target error 0.2 °C corresponds to 0.2 × 41 µV = 8.2 µV.
  • Required attenuation = 1.5 mV / 8.2 µV ≈ 183, or 45.2 dB at 50 Hz.

Filter choice

  • A two-pole differential RC low-pass with fc = 3 Hz gives (50/3)^2 ≈ 278, or 48.9 dB, at 50 Hz.
  • Residual input error = 1.5 mV/278 = 5.4 µV ≈ 0.13 °C, which meets the target with margin.
  • Place the filter ahead of the high-gain stage so the interference cannot saturate the amplifier, and match resistors and capacitors to preserve CMRR. The 3 Hz corner gives a time constant of about 53 ms, acceptable for a furnace loop.

3. Recommend grounding and shielding practice for the ungrounded junction and explain the effect on CMRR.

Read the worked answer

Bias return

  • An ungrounded thermocouple leaves the amplifier inputs floating with respect to the local ground, so input bias currents charge stray capacitance and the common-mode point drifts.
  • Add high-value resistors, typically 10 MΩ, from each input to ground. This defines the common-mode operating point and gives bias current a return path.

Shielding and wiring

  • Use twisted pair for the thermocouple leads and route it away from heater wiring.
  • Connect the shield at the amplifier end only; grounding both ends creates a loop that couples magnetic interference and can inject noise into the signal pair.
  • Keep the differential filter resistors and capacitors closely matched; unmatched components convert common-mode voltage into differential voltage and defeat the amplifier CMRR.

Resistor tolerance

  • The 1% gain-setting resistors in the difference stage limit practical CMRR to roughly 40 dB regardless of the op-amp specification. Use 0.1% resistors or a trimmed network in the second stage, otherwise the calculated 60 dB at 50 Hz will not be achieved.