1

I am a senior studying electronics, and have happened to make a bandgap reference circuit that gives out an output reference voltage of ~65 mV. I've constructed this after quite a few months of research and simulations alongside my mentor.

I wasn't really paying attention to the output reference voltage I was getting, I was more focused on the circuit's performance, like its Temperature Coefficient or its Line Sensitivity. While all that seems to be decent according to my literature survey, I feel the value of the reference voltage itself isn't useful. It seems a silly question to ask my mentor, but what exactly could I use a 65 mV reference voltage for?

P.S. I'm not willing to share any details of my circuit because I'm still working on it, and would like to keep the work confidential.

Akshar Gandhi
  • 121
  • 4
  • 4
    "~65 mV" doesn't sound very accurate. Maybe give a more detailed specification of initial accuracy and drift including load variations and input voltage change artefacts. Look at this from a standpoint of someone wanting a reference. Draw them in with good specs. 65 mV by itself is useful but, I'm not going to comment on applications because its the ~ in front of the 65 mV that puts me off. – Andy aka Dec 07 '21 at 19:15
  • 1
    I honestly don't know how anyone can comment seriously on this without details. All references can be calibrated for accuracy. But one that can be fabricated using commonly available manufacturing techniques and has a high degree of repeatable initial accuracy guaranteed by solid state physical laws and its design and manufacturing methods and similarly guaranteed limitations to its long term drift and finally able to produce a tight (low noise) signal would be far more valuable than a noisy, difficult to repeatably manufacture, in need of device-by-device expensive calibrations, reference. – jonk Dec 07 '21 at 19:31
  • A "65mV" reference, if characterized and included with a configurable, precision op-amp, could be a 0-10V reference in practice. But accuracy, noise, tempco... will all have to be figured into the resulting spec. – rdtsc Dec 07 '21 at 19:34
  • And what kind of voltage would make it useful in your mind? A bigger number? A rounder number? Did you know regulators step down their output voltage through a resistor divider and compare that to a reference voltage? It doesn't matter what the reference voltage is as long as it is lower than the output and known. – DKNguyen Dec 07 '21 at 19:41

3 Answers3

6

That's a sensible reference voltage for direct measurement of thermocouple voltages (65mV will cover most applications). Usually the signal is amplified before the ADC so there is no need of such a reference, of course so I'm not sure there is much there for you.

Of course you can always amplify the reference to make it whatever you want, you might want to compare the output noise resulting from that compared to commercial offerings. Even a 3:1 amplification brings you up to the 200mV range.

Also, as supply voltages slowly creep downward, a reference that operates from a single alkaline button cell (1.5V, maybe 0.8V at end of life) might benefit from a reference in the 50-500mV range. The (ancient) LM10 (a voltage reference plus op-amp) was not wildly successful in the market, but is still around. It has a 200mV nominal reference voltage but cannot operate from less than about 1.2V. You didn't indicate what the minimum supply voltage is for your circuit, but if it's well in the sub-1V range, that is a good feature.

Spehro Pefhany
  • 397,265
  • 22
  • 337
  • 893
5

If the performance significantly exceeds that of a good integrated amplifier it could be used as an ADC reference voltage for precision sampling signals <100mV, specifically with delta-sigma ADC.

Traditionally a reference of 1.25-5V is used , depending on device this is the reference that is used to compare the input signal, the method depends on the architecture/technology of the ADC. ADC sampling is usually defined in terms of this VREF.

Traditionally, small signals are amplified on die to obtain sufficient bits, typically on a good 24bit ADC with an integrated amplifier the noise performance of the amplifier is 100-120db. Stated in other terms, when amplifying the signal by maximum amount , it adds one bit of noise on a 24bit sample, in ideal conditions for a state of the art delta sigma adc..

I am not certain there is any available ADC that supports a very limited reference voltage to test this application, but it is , theoretically at least, an application that could conceivably be competitive if it reduces cost , die size, or component count .

crasic
  • 5,815
  • 1
  • 21
  • 43
4

Low voltage references are often used in circuits that need to detect low voltages such as across a current shunt (regulators, motor drivers, battery chargers, constant current sources) or a small voltage difference between nodes (BMS cell balancer, ideal diode).

Often it is easier to use a higher voltage reference with a resistor divider to develop lower voltages, but sometimes a separate reference is required. Here's an example from a battery charger IC, where the voltage across a 10 mΩ shunt resistor is compared to a 30 mV reference:-

enter image description here

Bruce Abbott
  • 56,322
  • 1
  • 48
  • 93