LTC2380-24 DC Measurements with DC2289A

We are currently evaluating the LTC2380-24 for precise DC measurements in our Physics lab. For this, we bought the DC2289A in combination with the DC890. Surely, a DC590 or DC2026 with the QuickEval connector would be more suitable for DC measurements, but in theory the DC890 is equally capable to measure DC, or am I mistaken?
For the final lab installation we will not use any of the demo boards because we will need realtime control circuits.

Anyways, I tried to replicate a very basic measurement with the ADC that SHOULD looks like Figure 5 and Figure 6 (see attachment).

I replicated the histograms with a simple python script after capturing a few thousand samples using the PScope software.
However, I did not succeed, the noise looks much worse (in the range of multiple 100mV instead of micro Volts. Can someone explain how the measurement setup for these plots would look like? I am pretty certain that I am doing something fundamentally wrong. Or is the DC890 simply not capable of accuratly measuring DC voltages?

DC Performance

Here the histogram that I captured with PScope and plotted with python:

Parents
  • +1
    •  Analog Employees 
    on Apr 9, 2021 2:04 PM

    Hi,

    The DC890 should be able to capture DC voltages without any problems. Please explain how the inputs of the DC2289A were biased. Enclose a photo if possible. The DC common mode input voltage should be Vref/2. You can't just ground the inputs. The voltage source must be very low noise (Do not use a switching power supply). I suggest passing it through a 1HZ low pass filter. Make sure the R component of the filter is low impedance (<50ohms). The QuikEval plots shown were taken dividing a 25V low noise source (Analogic Model 8200) by ten to further reduce the noise and then passed through a 1Hz filter. The attenuator is 90ohms in series with 10ohms. The capacitor is 15000uF. This 2.5V is then applied to both DC2289A inputs through a shielded cable. It may be necessary to enclose the board in a shielded box if a screen room is not available.

  • Tried it with a 2.5V voltage source and it worked perfectly. My results are by no means as good as the ones in the demo manual as the voltage reference in use is not very stable. I was wondering if there is a different method of measuring the intrinsic noise of the ADC that does not rely on the precision of outside components. Do you have any suggenstions?

    Additionally, I have 2 quick questions.

    1.) When using different common mode voltages (e.g. apply 3.3V at both inputs), the resulting Voltage should be 0V, or am I mistaken? In practice I can measure a voltage that is close to 0V but not as close as it would be with 2.5V

    2.) When measuring the output of the voltage reference on the DC2289A board with a multimeter, I measure a voltage of 5.8V. This is way off from the 5.0V that should be present at that pin. And it's most certainly not a measurement error as this is more than 15% off. Is this normal (e.g. due to other active electronics on the same line) or is something broken on my board?

  • 0
    •  Analog Employees 
    on Apr 20, 2021 8:11 PM in reply to drtlf

    If the input voltage is zero, the noise of the reference should not matter as much.

    1) The common mode voltage of the LTC2380-24 must be held at VREF/2 +/-100mV. With the two inputs tied together, the input voltage must be held at 2.5V +/-100mV when using a 5V reference.

    2) I am guessing when you measure 5.8V, you are floating the inputs. With the inputs floating the driver output voltage can exceed VREF. There is a diode from the inputs to VREF so that the reference voltage will be pulled up if the input voltage exceeds VREF by more than a diode drop.

Reply
  • 0
    •  Analog Employees 
    on Apr 20, 2021 8:11 PM in reply to drtlf

    If the input voltage is zero, the noise of the reference should not matter as much.

    1) The common mode voltage of the LTC2380-24 must be held at VREF/2 +/-100mV. With the two inputs tied together, the input voltage must be held at 2.5V +/-100mV when using a 5V reference.

    2) I am guessing when you measure 5.8V, you are floating the inputs. With the inputs floating the driver output voltage can exceed VREF. There is a diode from the inputs to VREF so that the reference voltage will be pulled up if the input voltage exceeds VREF by more than a diode drop.

Children