Category Archives: Transistors

Gate-Drive and Isolation Transformers

Controlling the current flow between the drain and the source of a MOSFET requires the application of a drive voltage to the gate of the MOSFET. Switching power supplies operate the MOSFET as a current switch by applying a pulsed voltage drive to the gate for turning the drain-source current on and off. Delivering the controlling pulse requires a gate drive transformer to provide isolation between the controlling drive circuit and the MOSFET. Companies like Coilcraft offer off-the-shelf gate drive transformers for the purpose.

Gate drive circuits must provide an isolated or floating bias supply for maintaining the necessary turn-on bias when the MOSFET source rises to the input voltage. While driving the MOSFET gate, not only does the gate drive transformer help in isolating the controlling gate drive circuit from the switch node, it may also scale the output voltage via a suitable turns-ratio between its primary and secondary.

Some applications use optocouplers or digital isolators for driving the MOSFET directly. However, the use of a gate drive transformer is preferable, as it can provide a higher voltage requirement, much lower turn-on and turn-off delay times, and it can scale voltages by the ratio of its turns. These advantages make the simple gate driver transformer the best-performing solution for high-frequency and high-voltage applications that require maintaining accurate and fast signal timing.

Typical low-power applications use a simple single-output, transformer-coupled, high-side gate driver circuit. Additional components like capacitors, resistors, and diodes may be necessary depending on the duty cycle and other circuit conditions. These include preventing the development of a DC voltage across the transformer, as this may cause it to saturate. The additional components also help in the coupling capacitance and magnetizing inductance from resonating with specific duty cycle ratios. For single-ended circuits, the highest duty cycle is preferably 0.5.

Higher power applications may require half-bridge and full-bridge configurations coupled with transformers. Double-ended or DC-coupled bridge configurations may use a theoretical maximum duty cycle of 1.0. Designers use isolation transformers for isolation and voltage scaling in power supply applications. These serve three main purposes.

First, the transformer helps to connect circuits with grounds at different potentials—this prevents ground loop formation. Second, the transformer provides galvanic isolation, thereby preventing any flow of direct current. Lastly, the transformer provides voltage transformation—stepping up or stepping down from one voltage to another.

Isolation transformers may be available as signal transformers, power-supply transformers, communication transformers, data-line transformers, and many others. These are versatile and aptly suited for several industrial and commercial data communications and power supply applications.

Using off-the-shelf gate drive and isolation transformers can simplify the design of the gate drive circuit and significantly reduce the design cycle time. Coilcraft transformers typically use high-permeability ferrite cores for maximizing the inductance and minimizing the magnetizing current.

The designer can determine the required transformer size by the volt-time product of the application. This forms the first selection criterion for a gate drive transformer, as the designer can select the appropriate volt-time or V-µsec rating from the datasheet of the transformer. The rating must be equal to or greater than the highest applicable voltage-time product 

RF Transistors using more Germanium

So far, gallium arsenide was the choice of material for building fast radio frequency transistors. However, that scenario is changing fast. Germanium is fast catching up, as it is less expensive and more compatible with CMOS and silicon. In this connection, the European research institute Imec has presented a pair of papers featuring gate-all-around (GAA) transistors at the 2017 Symposia on VLSI Technology and Circuits in Japan. They claim that GAA transistors can outperform standard CMOS below the 10-nm node, while featuring source/drain contacts with resistances of the order of a billionth-of-an-ohm.

Imec claims their SiGe GAA FETs are super-fast, as they use scaled strained germanium p-channels. With diameters below the 10-nm range, the GAA FETs are integrated on a 300 mm platform, which gives them their superior electrostatic control. They achieve this by using high-pressure annealing (HPA)—the same technology used by Imec for their more traditional FinFET architecture.

According to Imec, they used pulsed-laser annealing and shallow gallium implantation techniques to achieve the very low source/drain resistivity. They claim to have achieved a new world record for resistances of one billionth of an ohm for the source/drain contacts of their p-MOS transistor.

That germanium-on-silicon transistors have the ability to outperform other technologies as radio frequency transceivers is already well known. In practice, this allows using the same CMOS technology throughout the transceiver and avoids using GaAs Pas, which has lattice structures incompatible to silicon.

However, beyond the 10 nm range of advanced nodes, there was no confirmation that SiGe could perform as well for FinFETs or for more advanced architectures such as the GAA FETs, least of all in the state-of-the-art wafers of 300 mm. According to Imec, using HPA, they were able to boost the performance and electrostatic control exceptionally for both p-channel FinFETs and GAAS.

Imec will be sharing their record-breaking billionth of an ohm per square centimeter for P-SiGe source/drain contacts with other CMOS members, namely, TSMC, Sony Semiconductor Solutions, SK Hynix, Samsung, Qualcom, Micron, Intel, Huawei, and GlobalFoundries.

Imec claims to have made it easier to go below the 10-nm range without sacrificing electrostatic control, by making architectural changes they have not yet disclosed. This allowed them to compensate for the smaller bandgap and larger permittivity of SiGe. According to Imec, this has also allowed them to make gate lengths of the order of 40 nm, and nanowires of about 9 nm, the shortest and the thinnest in the world. As a consequence, they claim to have lowered the sub-threshold slope of 79 mV/decade and the drain-induced barrier of 30 mV/V, while retaining the electrostatic control for their GAA-FETs.

By using HPA techniques, Imec is also claiming to have boosted the performance of both their germanium GAAs and FinFETs. Because of HPA at 450°C, Imec researchers were able to improve hole mobility and interface quality to 60 cm2/Vs. By optimizing the HPA technique, they could improve the electrostatic and overall performance of the GAA devices significantly. This allowed them to reach 60 nm lengths and achieve a Q-factor of 15. They were also able to lower the currents from 3 to 10-billionths of an amp per micron.

What Are Super-Junction MOSFETs?

Switching power-conversion systems such as switching power supplies and power factor controllers increasingly demand higher energy efficiencies. For such energy-conscious designers, super-junction MOSFETs are a favored solution, as the technology allows smaller die sizes when considering key parameters such as on-resistance. This leads to an increase in current density while enabling designers to reduce circuit size. With increasing market adoption of this new technology goes up, other challenges are coming to the fore, mainly the requirement for improved noise performance.

High-end power supplies for equipment such as LED lighting, LCD TVs, notebook power adapters, medical power supplies, and tablet power supplies require reduced electromagnetic noise emission. Designers prefer using resonant switching topologies such as the LLC converter with zero-voltage switching, as these have inherently low electromagnetic emissions. Super-junction transistors in the primary side switching in an LLC circuit helps designers achieve a compact and energy-efficient power supply.

Compared to conventional planar silicon MOSFETs, the super-junction MOSFET has significantly lower conduction loss for a give die size. Additionally, architecture of the latter device allows lower gate charges and capacitances, leading to lower switching losses compared to conventional silicon transistors.

Fabricators used a multi-epitaxial process for structuring the early super-junction devices. They doped the N-region richly allowing a much lower on-resistance compared to conventional planar transistors. They adapted the P-type region bounding the N-channel to achieve the desired breakdown voltage.

The multi-epitaxial processes resulted in the N- and P-type structures being dimensionally larger than ideal and led to an associated impact on overall device size. The nature of the multi-epitaxial fabrication also restricted engineering the N-region to minimize on-resistance. Therefore, fabricators now use single-epitaxial fabrication processes such as deep trench filling to optimize the aspect ratios of N- and P-regions to minimize the on-resistance while also reducing the size of the MOSFET.

For instance, the single epitaxial fabrication process allows DTMOS IV family of Toshiba’s fourth-generation super-junction MOSFETs to achieve a 27% reduction in device pitch, while also reducing the on-resistance by 30% for each die area. Similarly, Toshiba’s DTMOS V, based on deep trench process, has further improvements at the cell structure level.

Thanks to the single-epitaxial process, the super-junction MOSFETs can deliver more stable performance when faced with temperature changes. Power converters with conventional MOSFETs are noted for reduced efficiencies at higher operating temperatures, which the super-junction MOSFETs are able to counter. For instance, super-junction MOSFETs show a12% lower on-resistance at 150°C.

Power converters using the fifth-generation super-junction DTMOS V devices can now deliver low-noise performance along with superior switching performance. A modified gate structure and patterning helps to achieve this, resulting in an increase in the reverse transfer capacitance between the gate and drain of the device.

Selecting Universal Motor Controls

Inside the home, one will find a number of gadgets with the universal motor dominating. Mostly, these are used in high speed, low-cost motor applications, such as in power tools, vacuum cleaner, and countertop blenders. However, not all gadgets perform equally. For example, a bargain-basement blender may make a lot of noise when working. Others may be relatively quieter. While some products have a tendency to overheat, others run cool even if you load them over. Actually, the motor itself has little to do with the wide variation in performance. Mostly, the lifespan and performance of the universal motor depends on its drive circuitry.

If speed control is not necessary, gadgets have their universal motors simply connected to the AC mains or to the DC rail, as this is the most cost-effective method for driving the motor and letting it spin. However, the speed of the universal motor depends largely upon the voltage applied, and connecting it directly to the voltage source allows it to spin at its maximum speed at minimum load.

While connecting directly to the voltage source for maximum speed might suit power tools or a vacuum cleaner, other applications may require the speed of the shaft to vary. Designers accomplish these using subtractive measures, mostly by reducing the motor voltage. This helps to reduce the speed to a fraction of the maximum RPM.

One can power universal motors through either alternating voltage or direct voltage, with each approach having its own advantages and disadvantages. While DC control circuits tend to be more expensive than their AC counterparts are, the DC controls have the advantage of prolonging the motor life, and offer noticeably quieter operations, with improved efficiency.

Running a universal motor on alternating voltage and implementing a lowest-cost speed control entails feeding the motor varying amounts of the AC half-cycle. The cheapest open-loop method employs two semiconductor devices, a Triac triggered by a Diac, with a series RC network controlling the phase at which the Diac fires.

Closed-loop controls replace the Diac with a low-cost 8-bit micro-controller. Apart from offering improved control of the Triac, use of the micro-controller results in a closed-loop speed control, more sophisticated user interface, and a proprietary software-based design. By digitally monitoring the motor voltage and current under load, most applications are able to forego the use of a tachometer on the motor shaft for feedback.

Although the above makes for a very economical drive, the downside to using this approach results in a high current ripple, making the operation run fairly noisy. Dissipations in the Triac reduces the efficiency of the approach, with the thermal strain on the brushes ultimately reduces their life.

DC drives, on the other hand, take a different approach. Regulated DC power to the motor is pulse-width modulated, with the rate of rotation of the shaft being directly proportional to the duty cycle of the PWM waveform. When the energy supplied to the motor is low, it spins slower.

The DC drive has the advantage of higher efficiency, reduced noise, highly responsive speed regulation, prolonging the motor life. Overall, the application may use a smaller motor.

Researchers Develop Thermoelectric Organic Transistors

Linkoping University scientists have made possible an organic transistor that is driven by temperature changes instead of by an electrical signal. Made of a thermoelectric material, the transistor brings about an appreciable current modulation for just a single degree rise or fall in temperature.

Professor Xavier Crispin, based at the Laboratory of Organic Electronics of the university, states that heat driven transistor is the first logic circuit to be developed that makes use of thermoelectricity.

Wide Range of Applications

The scientists foresee diverse uses for the new transistor. Since the device can record very small temperature changes, healthcare professionals can use it to fabricate therapeutic dressings that monitor the healing process along with treating the patient.

The scientists say it would be possible to build circuits that would respond to the heat contained in infrared radiations, too. This could be of particular use in developing heat cameras and similar devices.

The organic transistor is highly susceptible to minute heat changes. Compared to conventional thermoelectric devices, it is 100 times more sensitive to a drop or rise in temperature. This high level of heat sensitivity implies that just one electrical connector from a heat sensor electrolyte is adequate for sending a signal to the transistor. The researchers explain that a pair of a thermoelectric transistor and a sensor connector would be sufficient to make up a “smart pixel” for the camera.

A set of these smart pixels could make up a matrix, which may serve as a detector. This could be used in place of the numerous sensors used for detecting infrared rays in existing heat cameras. The researchers hope to add in more developments so that even a device as small as a mobile phone can include a heat camera. Since the materials needed for fabrication are non-toxic, inexpensive, and easily available, the feature could be had at a low cost.

Sunlight Charged Supercapacitor

The researchers built the heat-powered transistor by exploring a technique that allowed charging a supercapacitor by sunlight. The capacitor, developed a year ago, captures the light photons falling on it to convert to electricity, which is stored within it for further use. Crispin explains that it was crucial to establish the working of the heat driven supercapacitor before looking into possible electrolytes and the range of possibilities.

The university team researchers looked through a wide range of conducting polymers to turn out a liquid electrolyte that can produce a potential difference from a temperature gradient a hundred times more than that most electrolytes generate. While the positive ions of the electrolyte are small and move quickly through the liquid, the polymer molecules are negatively charged and massive, and move slowly. When a part of the electrolyte is heated, the lighter positive ions move to the colder regions rapidly. The separation of the positive ions from the negatively charged polymer molecules generates a potential difference or a voltage, which is adequate for transistor applications.

Team members Simone Fabiano, a lecturer, and Dan Zhao, a researcher engineer, have worked extensively with the electrolyte to show that heat signals can be used to make electronics controlled by heat signals.

What is an LDO and How Does it Work?

When you need a voltage regulator for your circuit and do not have much of a voltage head room, the trick is to use an LDO or a low-dropout regulator. Normal regulators need voltage headroom of roughly around 3V to allow good regulation, but LDOs can do with a lot less – of the order of a few 100 millivolts. However, there are other considerations as well.

To regulate and control an output voltage, it is necessary to source it from a higher input voltage supply. For normal regulators, the voltage headroom or the difference between the output regulated voltage and the minimum input unregulated voltage must be more than 3V. For example, if you need a regulated voltage of 5V, it must be sourced from a minimum input voltage of 8V. That ensures the regulated output voltage never dips below 5V. With circuits getting more complex and noise sensitive, new designs must deal with higher currents and lower voltages. Hence, headroom voltages of 3V or more may not be available in all cases, and it is necessary to use LDOs.

Although manufacturers offer datasheet specifications for basic parameters of regulators, they cannot list all parameters for every possible circuit conditions. Therefore, to use the LDO in the best possible manner, designers must necessarily understand the key performance parameters of the LDO and their impact on given loads. A close analysis of the surrounding circuit conditions helps to determine the suitability of a specific LDO.

In applications, LDOs primarily isolate a sensitive load from a noisy power source. The pass transistor or the MOSFET regulating and maintaining the output voltage accurately is always on and dissipates continuous power. This is different from switching regulators, which work as on-off switches. That makes LDOs less efficient and designers must handle the thermal issues related. System power requirements primarily drive the use of LDOs as voltage regulators. Since they are linear devices, they are also used for noise reduction and for fixing problems related to EMI and PCB routing.

As the power dissipation of an LDO is primarily governed by the current through it, LDOs are an obvious choice for very low current loads, bringing with their use simplicity, cost economics, and ease of use. For load currents of more than 500mA, designers must consider other parameters also, such as the dropout voltage, load regulation, and transient performance.

LDOs comprise three basic functional elements – a pass element, a reference voltage, and an error amplifier. Under normal operation, the pass element behaves as a voltage controlled current source. A compensated control signal from the error amplifier drives the pass element. The error amplifier senses the output voltage and compares it with the reference voltage. LDO regulator designs use four different kinds of pass elements – PNP transistor based regulators, NPN transistor based regulators, P-channel MOSFET-based regulators and N-channel MOSFET-based regulators.

While using a specific LDO in their circuits, designers need to consider the performance of the LDO with respect to its dropout voltage, load regulation, line regulation, and the power supply rejection ratio or PSRR.

AT21CS01 from Atmel: This EEPROM Does Not Require External Power Source

AT21CS01 from Atmel is a two pin serial EEPROM. Astonishingly, it does not have a Vcc or power supply pin characteristic of any IC and does not require an external power source to work. This amazing memory IC operates with only a data pin and a ground pin. The memory in the IC is organized as 128×8 bits, that is, a total of 1-kbits.

The single-wire device, AT21CS01, operates with only an SI/O and GND pin. The SI/O signal functions as a combination of data and power line. That means, apart from moving data in and out of the IC, the SI/O pin also provides power to the device. During high time of the protocol sequence, the IC’s parasitic power scheme provides the IC with power.

Each AT21CS01 is factory programmed to include a unique serial number of 64-bits. The SI/O line can be accessed directly from outside the application, because the device complies with the IEC 61000-4-2 ESD tolerance. This memory IC comes in 4-ball WLCSP, 8-lead SOIC and 3-lead SOT23 packages. Market availability is slated for the fourth quarter of 2015.

Possible applications for AT21CS01 include ink and toner print cartridge identification, storing data for analog sensor calibration and management of after-market consumables. There are several advantages in using AT21CS01.

Manufacturers claim AT21CS01 consumes 33% lower power in its active mode when compared to devices offered by the competition. For instance, at 25°C, the typical write current for an AT21CS01/11 is 200 µA, the typical read current measures about 80 µA, and a typical standby current of 700 nA. Each memory location can endure 1 million write cycles.

With such features, the AT21CS01 is eminently suitable as identification markers for cables, batteries, consumables, wearables and IoT. To support different voltage requirements, the AT21CS01 comes in two variants. AT25C501 is suitable for applications operating in the range of 1.7 to 3.6V. However, when operating with Li-Ion or polymer batteries, applications require higher voltage ranges, such as 2.7 to 4.5V, for which, the AT21C511 is suitable.

With its ultra-low active and standby currents, the AT25C501 beats the competition by consuming at least one third less power. The single-wire interface follows the I2C communication protocol. This IEC 61000-4-2 Level 4 ESD compliant device can withstand discharges of +8KV in contact and +15KV in air.

The innovative memory is organized into for zones of 256-bits each, with a security register additional to the 1 Kb memory space. Each EEPROM has a 64-bit serial number programmed at the factory and includes 16-bytes extra for user programmability. That means the user can improve on the uniqueness of the serial number on each device.

The advantages of using AT25C501 are many. The designer needs only one pin from the ASIC/MPU/ASSP/MCU. Because of its smaller footprint, layout is simple and the consumed PCB area reduces. That makes it easy to integrate identification capabilities in cables and or consumables. Its lower energy consumption is a boon for instruments working on batteries. The high-speed mode of AT25C501 even in low power applications results in high performance.

How Does Switching Affect Semiconductors?

Even though ICs rule the world of electronics, the transistor does all the work. Within each IC are millions upon millions of transistors perpetually switching on and off so that the IC can carry out its intended functions. Even if one of the multitudes of transistors were to stop switching, the IC could lose part or all of its functionality.

Circuits handling digital signals most often use transistors to switch from a high state to a low state and vice versa. It is usual to call a circuit point as being in a high state if the voltage at that point is close to the supply voltage. If the circuit point is closer to the ground or zero voltage, we generally call it as being at a low state. The time taken for the transistor to switch from a high to a low state or vice versa is its switching rate. While the transistor does not expend much energy when at either the low or the high state, the same cannot be said for the time when it is actually switching.

Under ideal conditions, a transistor should switch instantaneously. That means the transistor should take zero seconds to change its state. However, ideal conditions do not happen in reality and the transistor takes a finite time, however small, to actually switch over.

Transistors are made of semiconductor material and each junction has a finite capacitance and resistance. Junction capacitances store energy and the combination of resistance and capacitance acts to slow down switching – the capacitance must fill up or empty itself before the transistor can flip. The rate at which the capacitance fills up or empties itself depends on the junction resistance.

The situation gets worse as the switching frequency goes up. As the transistor is driven to toggle faster and faster, the junction capacitance may not get enough time to discharge or charge up fully. That defines the maximum switching rate the transistor can achieve.

Semiconductor manufacturers use various methods to reduce junction capacitances and resistances to induce these special semiconductors switch faster. Although modern semiconductors (transistors and diodes) are capable of switching at MHz or GHz scales, the cumulative effect of the tiny switching losses add up to increase the junction temperature.

Power is the product of voltage and current. When a semiconductor is in a high state, although the voltage is high, the current is negligible and consequently, the power drawn from the supply is negligible. When the semiconductor is a low state, its voltage is close to the ground level and the product of current and voltage is again negligible.

However, during switching, when the voltage is somewhere in-between the supply and ground levels, the current drawn also increases. That makes the product of voltage and current have a significant value and the semiconductor generates heat because of the power consumption. With higher frequencies, this happens more frequently and the heat accumulates to produce higher junction temperature.

If the natural process of heat dissipation can remove the accumulated heat, the semiconductor soon reaches a steady temperature. Else, heatsinks and or forced cooling methods are necessary to remove the heat accumulated.

Phosphorene Challenges Graphene as a Semiconductor

Though silicon has been the basis of semi-conductors for decades, it is facing stiff competition from other materials that promise to deliver several extras to consumers who like to enjoy more flexibility with their gadgets.

For some time, graphene, a one atom thick allotrope of carbon has been under consideration for use in electronic devices because its thin structure allows electrons to travel across it much more rapidly than they would do across silicon. However, graphene has severe limitations, as its conductivity is a little too high to be of much use in electronic devices, which need semi-conductors or materials with medium levels of conductivity. Another newly developed material dubbed phosphorene, which can form identical thin layers and is a semiconductor as well, offers a wider scope in electronics.

Phosphorene particulars

Scientists at the Technical University of Munich (TUM) have prepared a semiconducting material with black phosphorus in which a few phosphorus atoms have been swapped by arsenic atoms. Replacement of the phosphorus atoms with arsenic has caused the band gap to reduce to 0.15eV, which makes the material an effective semiconductor.

Phosphorene or black arsenic phosphorus can form very thin layers like graphene. Unlike silicon, which is hard and brittle, phosphorene is easy to manipulate into different kinds of structures and shapes. This makes possible a great range of electronic devices with considerable mechanical flexibility.

Scientists at TUM have built on technology that allows the fabrication of phosphorene with the application of high pressure. This reduces the production costs considerably. The research workers have been able to fine-tune the band gap exactly according to specific requirements by tweaking the arsenic concentration. According to Tom Nilges, who is heading the research team at TUM this has enabled them to produce a wide range of materials with diverse electronic properties that were not possible earlier.

Field Effect Transistors

American scientists from Yale University and the University of Southern California (USC) have collaborated with the researchers at TUM to build devices like field effect transistors with phosphorene. A group headed by Dr. Liu and Professor Zhou of the Electrical Engineering Department at USC has studied the transistor characteristics.

Infrared Detectors

Further exploration of the material by the scientists revealed that the material when heavily doped with arsenic could be used for infrared detection. For instance, when the arsenic concentration is as high as 83%, the band gap in phosphorene is about 0.15eV. This fact makes it an effective sensor for infrared rays of long wavelengths. Researchers expect that the new substance can be effectively used as Light Detection and Ranging (LIDAR) sensors, which find use in applications for tracing dust particles and pollutants in the atmosphere and as distance sensors in vehicles.

Anisotropic behavior

Another noteworthy feature of phosphorene is its anisotropic nature. Electronic and optical properties of the material were studied using ultra-thin films in two mutually perpendicular, x- and y-axes. It was observed that the properties were different in the two directions.

Phophorene has an edge over other newly discovered thin-layered semiconductors because it is very easy to peel off layers from a parent black phosphorus crystal.

How to Select Voltage References

how to select voltage referencesSensing applications use Analog to Digital Converters and Digital to Analog Converters and the accuracy of their readings depends on the voltage reference used. Most often the voltage reference used are very simple components with only two or three pins. However, the performance of these references depends on several parameters and careful attention is necessary when selecting the proper one. Typically, applications use either a shunt or a series voltage reference.

A series voltage reference is basically a high precision, low-current linear regulator. The load current comes through a series transistor positioned between the input voltage and an internal reference voltage. For the shunt voltage reference, a transistor placed parallel to the load shunts excess current to ground. As the series reference has to supply only the required load current, the shunt reference dissipates more power. The bias current of a shunt reference has to be greater than or equal to the maximum load current plus the minimum operating current of its internal reference.

In general, shunt references offer the user greater flexibility in handling higher input voltages and in creating floating or negative references. They also provide better power supply rejection but consume higher power. On the other hand, series references dissipate lower power and perform better for high-precision applications. The typical way of depicting the use of a shunt reference is by showing the symbol of a Zener diode.

Drift or variation of the reference voltage over temperature is a very important factor and has the units of parts-per-million per degree Celsius or ppm/°C. Most monolithic references use the bandgap reference as their base. Special circuitry is required to maintain drifts lower than 20 ppm/°C with the additional circuitry providing some form of curvature correction. Other types of references use a buried Zener diode voltage combined with the base-to-emitter voltage of a bipolar transistor to provide a stable reference voltage of about 7V. Both types have similar drift characteristics, but buried Zener types offer better noise performance.

All voltage references generate internal noise producing a dynamic error degrading the signal to noise ratio of a data converter. Device datasheets typically provide specifications separately for low and for high frequency noise in addition to the broadband noise in rms microvolts over the 10 Hz to 10 KHz bandwidth. You can reduce the broadband noise by adding a bypass capacitor.

Thermal cycling or change in temperature causes references to show thermal hysteresis. This appears as a shift in the reference voltage. Manufacturers define a thermal cycle as an excursion from room temperature to a minimum and a maximum temperature with a return to the room temperature. This is important as the reference may have to be soldered and this can induce shifts from the desired reference voltage.

Continuous operation may cause long-term stability issues resulting in a typical shift in the reference voltage. Manufacturers usually state the shift after six weeks or 1000 hours of continuous use. Since long-term stability is typically related logarithmically to time, the shift in reference voltage in the first 1000 hours provides a rough idea of the stability of the voltage reference over its life.