Magnetics Overview (03:23)
GMR Overview (02:00)
Today we're talking about magnetism with Daniela
at NVE. So, Danno, what causes magnetism?
In practical applications, magnetic fields are either from permanent magnets or
electric currents. Permanent magnetism is due to the intrinsic magnetic moments
of electron spins.
Here's what the field from a disk magnet looks like:
[diagram showing flux lines from a disk magnet].
What happens to magnetic field strength over distance?
Near the face, the magnet is like a monopole, so the field drops off with the
square of the distance. Farther from the face, the field drops with the cube of
Here's an experimental setup for magnetic field strength vs. distance using a small ceramic magnet and an AAH002 magnetometer chip.
For this demonstration, we powered the sensor IC with a nine-volt battery, and
connected it to a meter to measure the field versus distance.
As the magnet moves closer, the output increases.
[animation of magnetometer meter reading versus distance]
This graph summarizes the data:
[graph of sensor output versus distance]
You can see the inverse square, and inverse cube relationships.
The chart goes out to about an inch and a half. Can you detect a magnet farther away than that?
Maybe with a really big magnet. It depends on a number of things. In this example,
we start to get interference from the earth's magnetic field around here...
Most practical applications use spacings of a fraction of an inch, but sensitive sensors give you more signal to work with.
That's certainly "attractive." What's the actual field from the magnet?
Here's the same graph with the field in oersteds:
[graph of magnetic field strength versus distance]
For reference, the earth's magnetic field is about half an oersted.
[line for the earth's magnetic field]
"Oersteds..." Sometimes I hear tesla or gauss for magnetic fields. Whats
Tesla and gauss are units of magnetic flux density. One tesla is 10,000 gauss. Oersteds and amps per meter are different units of magnetic field strength. Oersteds and gauss have a one-to-one correspondence in air.
Thanks. Where can I get the parts you used in this video?
The sensor, magnet, and a circuit board for the sensor are in the AG-001 Analog
Sensor Evaluation Kit, available from NVE or Digi-Key.
Today we're talking about spintronic "giant magnetoresistance" with Daniela
at NVE. Danno, why does a nanotech company make something "giant?"
refers to a giant change in resistance. GMR is actually a nanoscale phenomenon.
The balls represent electrons. The middle (conduction) layer needs to be thinner
than the mean free-path of conduction electrons, which is only a few nanometers.
The top layer is called the free layer because its electron spins are free to
change. The bottom is the pinned, or reference layer, because its spin orientation
is fixed when the device is made. In this state, the electrons spin in opposite
directions in the top and bottom layers, which causes the electrons in the middle
to scatter, increasing resistance. Then when a field is applied, spins in the
free layer switch, conduction electrons scatter less, and resistance drops. The
magic of GMR is turning the esoteric property of electron spin into resistance,
which can be used by conventional electronics. The large signals can mean smaller
size, more precision, and lower power. We can summarize these advantages with
the "three Bs": boxes, bits, and batteries.
are certainly "giant" advantages. What products use GMR?
come to mind:
* magnetic sensors
* gear-tooth sensors;
* GMR isolators.
And there are videos
on those applications and more on nve.com.
today we're "digging" into spintronic tunneling with Daniela at NVE.
So, Danno, what is tunneling and how does it work?
Tunneling is a nano-scale phenomenon where under the right conditions, electrons
can tunnel through very thin, normally insulating materials, causing lower resistance.
Structures are called Magnetic Tunnel Junctions, or Tunneling Magnetoresistors.
Let's look at one. The section on the right is called the free layer, because
its electron spins are free to change, and the layer on the left is the pinned,
or reference layer, because its spin orientation is fixed when the device is made.
The barrier between the pinned and free layers is thin enough that electrons can
tunnel through, and as the meter shows, the resistance is low, even though this
is a normally insulating material. When a field is applied, the free-layer polarization
switches, reducing tunneling, because of the scattering caused by electrons spinning
in opposite directions. So the tunnel-barrier returns closer to its insulating
state. In this way we can turn the esoteric property of electron spin into something
useful: resistance. The resistance change can be very large. Large signals can
mean smaller size, more precision, and lower power spintronic devices. Here's
an SEM of an actual tunnel junction made here at NVE. This white stripe is an
aluminum-oxide tunnel barriereight atoms thick. This is probably magnified
nearly a million times on your screen, so an eight-inch wafer would appear a hundred
miles in diameter.
Cool. How can our viewers "burrow"
You can go to nve.com for lots more info.
we're learning about "MRAM" with Daniela at NVE. So, Danno, what is
MRAM stands for Magnetoresistive Random
Access Memory. Unlike semiconductor memories that use electron charge to store
data, MRAM uses electron spin.
Why is that better?
charge, electron spin is inherently permanent. MRAM has been called the ideal
memory because of its potential to combine the speed of SRAM, the density of DRAM,
and the nonvolatility of flash memory.
So how does it work?
Let's look at what's called "classic" MRAM. Each bit has a spin-dependent
tunnel junction memory cell, and magnetic row and column write lines. The spin-dependent
tunnel junction produces a large change in resistance, depending on the predominant
electron spin in a storage layer. The tunnel barrier is as thin as a few atomic
layers. It's so thin, that depending on spin polarizations, electrons can "tunnel"
through the normally insulating material, causing a resistance change.
Here's an SEM of an actual tunnel junction made here
at NVE. This white stripe is an aluminum-oxide tunnel barriereight atoms thick.
This is probably magnified nearly a million times on your screen, so an eight-inch
wafer would appear a hundred miles in diameter.
We have a separate video explaining "tunnel junctions."
in MRAM, data is read as the tunnel junction resistance. Data is stored in the
spin-polarization of a tunnel junction magnetic layer. The write lines create
magnetic fields that store data bits by setting magnetic spin polarization.
what's after "classical MRAM"?
MRAM could reduce cell size and power consumption. One next-generation technology
is Spin-Momentum Transfer, which changes the spin of storage electrons directly
with a spin-polarized electrical current, rather than an induced magnetic field.
This has the potential to significantly reduce MRAM write currents and self-heating.
Dannothat was "memorable." Go to nve.com for more info on
MRAM and spintronics.
Sensor Systems (02:00)
we're talking about GMR Switches with Jayne, at NVE. Jayne, what are "GMR
Switches" and how are they better?
GMR Switches are
precision digital magnetic field sensors with integrated GMR sensor elements and
on-board digital signal processing. As this slide shows, GMR Switches are more
accurate than other magnetic sensors, and magnetic operate points are stable over
voltage and temperature. This makes for precision, tight-tolerance magnetic sensing
And what parts are available?
GMR Switches can operate over a wide range of magnetic fields, making them the clear
choice for digital output magnetic sensors. NVE offers three series of GMR Switches:
AD-Series Digital Sensors with a 4.5 volt to 30 volt supply; ADV001-Series bipolar digital sensors;
and AFL-Series digital sensors with supply
voltages as low as 0.9 volts. Parts are available in SOIC-8, MSOP-8, or 2.5 by
2.5 millimeter TDFN-6 packages.
Thanks, Jayne. Where
can I "switch" for more info?
You can click
to go to the digital sensor section of nve.com for info, or to get an evaluation
kit. Or call me, at 800-GMR-7141, extension 2.
Sensor Tips (02:33)
So what's with the bad hat and the music, dude? Is it some kind of geek chic?
Foolproof sensor circuits is today's topic.
Right. The unique NVE DB-002 power switch IC comes in a SOIC-8 and provides:
* short-circuit protection;
* reverse battery protection, and
* automatic thermal shutdown;
as well as signal processing, so sensor systems can be just about foolproof.
So where would you use that?
The IC is designed to work with NVE's AD-1 Series GMR magnetic switch, or any current sourcing, CMOS, or TTL output sensor element.
Some customers even use the IC to ruggedize inductive or photoelectric sensors.
What's a typical circuit?
Let's look at one. The DB-002 provides:
* a sourcing or sinking output with up to 300 milliamps. Integrated transient protection makes the circuit especially good for driving capacitive or inductive loads.
* The LED is driven with three milliamps when the output is on.
* Shutdown delay after a short-circuit is set by the one-nanofarad capacitor. A bigger capacitor makes for a longer delay.
* A 10 nanofarad minimum bypass cap is recommended between Vcc and ground.
* Vreg is a regulated, five-volt output, provided by the IC to power the sensor and other components in the assembly. For noisy environments, Vreg can be bypassed with up to a 100 nanofarad capacitor.
You would be a "fool" not to use that IC!
And you can buy them from the on-line store at nve.com.
vs. Omnipolar Magnetic Sensors (01:40)
today we're "gearing up" to talk about gear-tooth rotation sensors,
with Jayne at NVE. Jayne, tell us about GT Sensors.
GT Sensors are designed for detection of gear teeth and magnetic encoder wheels
in industrial speed sensing. As this slide shows, GT Sensors have four GMR sensing
resistors, which are connected as a Wheatstone bridge. The direction of sensitivity
is parallel to the sensor plane. A biasing magnet provides field, and the flux
lines are deflected into the direction of sensitivity by passing metal gear teeth.
The sensor produces a sinusoidal output with one cycle per tooth. GT sensors use
high-sensitivity, low-hysteresis GMR, to detect even the smallest gear teeth,
and to provide a 50 percent duty cycle output with wide air gap and temperature
How about some application tips for our viewers
to sink their "teeth" into?
Several things come
* About 1.5 millimeters between the back of the sensor and the face
of the bias magnet keeps flux lines flexible and able to follow teeth.
* Locating the sensor and magnet on opposite sides of
a circuit board often provides good spacing.
* Use a thick circuit board with a milled magnet pocket to precisely position the magnet on
a PCB. Most board manufacturers can do that.
* If you don't need zero-speed operation, you can AC-couple the sensor to eliminate offset
* The right magnet is important. Ceramic-8
is cheap and has good field properties. Alnico-8 is good for high-temperature.
We don't recommend rare earth magnets because they can saturate the sensors.
Thanks, Jayne. Where can I "turn" for more info?
can click to go to the GT-Sensor section of nve.com for info, or to buy an
evaluation kit. Or call me at 800-GMR-7141, extension 2.
Angle Sensors (01:59)
Jayne, a question we get here in the NVE Applications Center is: what's the difference
between an "omnipolar sensor" and a "bipolar sensor," and
is that a question for a psychologist or an engineer?
question. Most NVE magnetic sensors are omnipolar, which means they're sensitive
to either polarity. But bipolar sensors are different, with opposite operate and
release points. Let's look first at an omnipolar sensor. Either a north or south
field turns the sensor On. The sensor turns off when the field is removed. An
opposite field also turns on the sensor. This is a bipolar sensor. A south field
above the sensor threshold latches the sensor On; and the sensor remains On after
the field is removed. A north magnetic field turns the bipolar sensor Off.
NVE's ADV001 is a great choice for bipolar applications.
would be handy in certain applications. Where can I get more info?
You can click, to go to the digital sensor section of nve.com. Or call me, at
800-GMR-7141, extension 2.
Meter Sensors (02:44)
we're talking about "TMR" angle sensors with Jayne at NVE. Jayne, tell
us about these remarkable sensors.
NVE's AAT001 sensors use extremely high output spintronic Tunneling Magnetoresistors.
They sense angles around an axis perpendicular to their TDFN package. Typically,
they're used with a split-pole magnet to create flux-lines parallel to the package.
The sensor can determine angle and direction. As the magnet turns, the sensor
has two sinusoidal outputs ninety degrees out of phase. This quadrature lets you
determine direction of rotation. AATs can be used for angle sensing or precision
Why are AATs better than other angle sensors?
We summarize AAT advantages with three "Bs":
* bits, meaning large, precise output signals, without amplification;
* bounds, or wide air-gap tolerance, and;
* batterieshigh sensor resistance means extremely low power consumption.
And what are popular applications?
Several come to mind:
* rotary encoders;
* automotive rotary sensors;
* motor shaft sensors; and
* knob position sensors.
Where can I "turn" for more info?
You can click, to go to the AAT page in NVE's on-line catalog. Or call me, at 800-GMR-7141,
Ultraminiature Sensors (01:20)
Were talking about angle sensors and water meter applications using NVEs
unique AAT-Series TMR sensors. With me is Jaimie, in the NVE Application
AAT-Series sensors use high-output spintronic Tunneling Magnetoresistors, or TMR
[AAT block diagram].
They come in ultraminiature TDFN6 packages [package dimensions].
The sensor measures angles around an axis perpendicular to the package.
The sensor has a wide mechanical air-gap tolerance and a low magnetic field requirement.
The magnet creates flux lines parallel to the package.
As the magnet turns, the sensor provides two sinusoidal outputs ninety degrees
out of phase [AAT sensor animation].
This lets you determine position and direction.
And smart water meters are a popular application?
Absolutely. Heres an example.
A split-pole or bar magnet is attached to the end of the water meter turbine shaft.
The sensor is in a fixed position.
As the turbine spins, the sensor indicates its position [water meter animation].
High sensor resistance is ideal for battery operation. High sensitivity means
more precise output signals without amplification [AAT circuit].
So in summary, AAT advantages in water meters are:
* high output without amplification;
* wide air-gap tolerance;
* high resistance for low power40 kilohms or 1.25 megohms; and
* an ultraminiature package.
Thanks, Jamie. Where can I get more information?
Visit nve.com, or call us at 800-GMR-7141.
for Sensor Systems (03:08)
thinking small today, as we learn about ultraminiature ULLGA sensors, with Jayne
at NVE. Jayne, what's a "ULLGA?"
for Ultra Leadframe Land Grid Array. The ULLGA package is a unique, 1.1 millimeter-square,
surface-mount sensor. These are NVEs smallest packaged sensors. This coffee
cup holds 700,000...
Hey! Don't drink that, Dude!
Anyway, let's take a closer look: 1.1 millimeters square with four 400-micron pads. The
parts are smaller than the head of a pin. In addition to their small size, ULLGA
parts are available in nanopower versions. And all NVE digital sensors have extremely
stable operate points over voltage and temperature.
So how can our viewers reach you for a "little" more info?
E-mail me at firstname.lastname@example.org, or call me at 800-GMR-7141, extension 2.
IsoLoop Isolators Work (02:18)
today we're "sticking" to magnets with Jayne at NVE. Jayne, tell us
which magnets are "attractive" in sensor systems.
The right magnet can make a big difference in system performance, often with not
much cost. Let's start by looking at magnet configurations.
Magnets are often used with simple proximity sensors, such as this GMR Switch.
A magnet might also bias a sensor like this GT gear-tooth sensor, so the field
is deflected by passing metal.
What are some differences in magnets?
main things: material grades, material, and shape.
First, material grades: Ceramic and Alnico magnetic materials are graded between
one and eight. Grade one materials are non-oriented or isotropic. Higher
grades are more fully oriented and have higher field strengths. Cost goes up with
Turning to materials:
* Ceramic, or ferrite, magnets are made of strontium
carbonate and iron oxide.
* Grade-one have the weakest magnetic field strengths, grade five
are inexpensive and popular, while grade eight have good field characteristics
for sensor systems but are still fairly inexpensive.
* Alnico refers to aluminum-nickel-cobalt alloys. Alnico magnets have
working temperatures above 500 degrees Celsius. Alnico grade eight magnets
provide a very stable field over wide temperature ranges, making them the preferred
choice for high temperatures.
* Rare-earth magnets, such as samarium-cobalt or neodymium-iron-boron,
are expensive and have very high field strengths. The high field strengths are
generally not necessary with NVE sensors, and may actually saturate the sensors.
There are almost infinite magnet sizes and shapes. Some of the more common
with our customers are bar magnets around one-half inch long, and disks an eighth
to a quarter-inch in diameter. A couple of specialty-type magnets are
used in sensor systems:
* Split-pole disk magnets are used for angle sensors
such as AAT001s. Split-pole magnets have poles on opposite sides of the diameter,
rather than top and bottom.
* And ring magnets have a number of poles
around the diameter for rotational or course-angle sensing. Ring magnets are
common in ABS brake systems, for example, and can be used with GT Sensors.
So "rare" on earth can our viewers get some of the magnets you talked about?
You can click, to go to the magnet section of our on-line store.
Isolator Overview (03:44)
here with Sandy, in the NVE Applications Center. Sandy, can you explain how NVE's
award-winning "IsoLoop" isolators work?
But first, let's look at optocouplers. As this shows, optocouplers
transmit signals by light through a dielectric that provides galvanic isolation.
The optical elements are slow, have limited life, and limit size.
I hate optos!
Doesn't everyone. But IsoLoop couplers use
spintronics and magnetics, rather than optics, to transmit data. They consist
of spintronic GMR resistors, microscopic integrated coils, and conditioning electronics.
The GMR resistors are spin valves with two stable resistance states. A dielectric
provides isolation. The coil creates a magnetic field proportional to the input
current. The field changes the spin polarization of the spin valves, which changes
their resistance. The output is amplified and conditioned to produce an isolated
replica of the input.
So, IsoLoops transmit signals by
magnetic field and electron spin, rather than light and photons?
And because ground potential variations are common to both sides of the coil,
they don't generate a current. This yields a large common-mode rejection ratio
and true galvanic isolation.
Cool. And you said they were
smaller and faster?
Darn right. IsoLoops are available in MSOPs, QSOPs, narrow and wide SOICs, plus
PDIPs, with up to five channels per device. They run up to 150 megabits per second,
and up to 125 degrees C.
Way cool. Where can I get data sheets or more info?
You can click, to go to the isolator section of nve.com. Or call us, at 800-GMR-7141,
Isolator Speed Metrics Explained (04:13)
I'm here with Sandy, in the NVE applications center. Sandy, tell us about NVE's industry-leading
IsoLoop "IL700" isolators.
Sure. Here's a single-channel functional diagram. IL700s are digital-in, digital-out
galvanic isolators. They use spintronic GMR rather than optical technology. Key
features include 5- or 3-volt operation; 2500-volt isolation, which complies with
UL1577 and IEC61010. Parts also feature low EMC footprint and best-in-class speed
and distortion, with many I/O configurations, packages, and variants.
What I/O configurations are there?
Just about all possible one, two, four, and five channel configurations are available.
There are also several variants for particularly demanding needs:
* IL200 series have five channels per package.
* Dash-one Series are MSOPs and QSOPs, the world's smallest isolators.
* "S" stands for speed, at 150 megabits per second and lowest
pulse-width distortion of 300 picoseconds.
* Finally, T-Series IsoLoops have the highest operating
temperature of 125C.
What parameters set IL700s apart?
Here are a few:
* highest data rate;
* lowest distortion as measured by PWD, prop delay, skew, and jitter;
* excellent transient immunity; and
* wide temperature ranges.
We summarize the advantages with the four "Bs":
* Boxespackage options to maximize channel density or creepage
* Bitsup to five channels per device;
* Baudhighest speed and lowest distortion; and
* Bulletproof, including temperature, transients, and barrier
The four "Bs." As you British say, "brilliant."
What packages are available?
Several choices. I mentioned
MSOPs for highest channel density. There are:
* eight-pin SOICs and PDIPs for drop-in replacement;
* 16-pin narrow SOICs, for four, or five channels per package, plus
* wide SOICs or PDIPs for maximum creepage and clearance.
What are popular applications?
A lot of IL700s replace optocouplers with higher speed, less distortion, longer
life, and higher channel density.
I hate optos...
Join the club. And other popular IL700 applications include:
* serial interfaces;
* isolated SPI;
* isolated A-to-D converters; and
* power interfaces.
Where can our viewers get data sheets and samples?
You can click, to go to the IL700 section of our on-line store for samples and data sheets.
Megabaud vs. Megahertz (01:24)
We're talking about isolator speed with Sandy, at NVE. Sandy: PWD, skew, jitter. All the terms give me the "jitters."
Let's cover key isolator speed metrics. Starting with data rate, which is the maximum rate at which data can be transmitted. Simple enough, but some companies specify a data rate faster than the propagation delay. This is a waveform for an actual device. The data is more than five cycles behind. By the time the data shows up on the output, not only are we on the next clock cycle, it's several cycles behind. And other parameters can make claimed data rate unusable. For example, one competitor's datasheet requires zero rise and fall time and exactly 50% duty cycle. Obviously unrealistic.
That's so lame. NVE would never do that...
That's why "Usable Data Rate" is important. UDR is defined as the maximum rate before the data is a clock cycle behind. So with this propagation delay the full data rate is usable.
Gotcha. And how's "Propagation Delay" defined?
Righty-O. "Propagation Delay" is the time for an edge to propagate. The faster the device the shorter the prop delay. There are also some important metrics related to prop delay. Because fast is worthless if it's garbled:
"Pulse-Width Distortion" is the maximum difference in rising and falling
prop delays. As you can see, the asymmetry distorts the pulse width--in this case,
making it wider. This is important in pulse width modulation and delta-sigma ADCs.
"Channel-to-Channel Skew" is the difference in propagation delay between
channels in the same device. It can be either direction. It's critical in clocked
"Propagation Delay Skew" is the difference in propagation delay between devices. Also important for clocked systems. It's larger than "Channel-to-Channel Skew," and can be a trap for the unwary in synchronous systems. This is an area where IsoLoop isolators shine.
Finally, "jitter" is another under-reported parameter. It's the variation
in the pulse edge position and can cause particularly nasty distortion.
So what are some specs for those parameters?
Here are the specs for NVE's IsoLoop IL700-S Series parts:
* "Data Rate";
* and "Usable Data Rate" are the same for NVE isolators;
* "Propagation Delay" is the time for an edge to propagate;
* "Pulse-Width Distortion" is the change in pulse width at the output;
* "Propagation Delay Skew" is the difference in propagation delay between devices;
* "Channel-to-Channel Skew" is the difference in propagation delay between channels in the same device; and
* "Jitter" is the variation in the pulse edge position.
The NVE parts are best in class for speed parameters.
Thanks, Sandy. Where do you "propagate" more "unskewed," "usable" info?
You can go to nve.com, or call us at 800-GMR-7141, extension four.
a question we got here in the NVE Applications Center: the IL700 data sheet specifies
a maximum data rate of at least 100 megabits per second, but the maximum frequency
is 50 megahertz. So what's the difference between "baud" and "hertz"?
question. As you can see, there are two bits of data in every 50 nanosecond cycle
running at the maximum speed. Therefore, the max data rate is twice the max frequency.
So the maximum data rate is 100 megabits per second for 50 megahertz devices like
IL700-Series Isolators. Extracting, or qualifying, that data requires a 100 meg
clock, but the maximum transfer rate is still 100 megabits per second. In addition
to best-in-class speed, IsoLoop Isolators have remarkable pulse-width distortion
and jitter specifications. Because fast is worthless if its garbled.
Thanks, Sandy. Where can I get data sheets or more info?
You can click to go to the isolator section of nve.com. Or call us, at 800-GMR-7141,
Layout Tips (02:02)
Today we're talking about RS-485 design with Sandy in the NVE Applications Center. Sandy,
give us the "four-one-one" on "four-eight-five" to avoid a bus "nine-one-one."
Sure. RS-485 is a bidirectional, half-duplex transmission system, so data can
be transmitted in both directions, but only one direction at a time. RS-485 allows
up to 32 unit-loads, although high-fanout transceivers like the NVE IL3285 allow
up to 256. Transceivers connect the nodes. Pull-up and pull-down resistors can
make the network fail-safe. Several bus configurations are possible. These configurations
have long stubs that can cause reflections. "C" and "D" are
ideal because the nodes are in a continuous line, although not necessarily straight.
Short spurs to intermediate nodes are often necessary. If so, stub lengths should
be less than one-sixth of the electrical signal length. This is the formula for
electrical signal length: using typical values of 10 nanoseconds rise time
and 78 percent propagation velocity, we get an electrical signal length of
2.3 meters. So the maximum stub length is one-sixth of the electrical length,
or 39 centimeters.
And how long can we have the main bus?
Well, the longer the
cable, the slower the data rate. RS-485 can transmit over 1200 meters, or 10 megabits per second, but not both.
This graph shows the typical tradeoff, depending on noise and jitter tolerance.
What about terminating lines?
Good point. Unterminated lines are only suitable for very low data rates, and very
short cables, otherwise reflections cause errors. Both ends of the bus, but not
every node, should be terminated. Finally, for a fail-safe configuration, bias
for at least 200 millivolts with no active drivers. This ensures the bus will
be in a known state.
Don't forget about isolation!
reduces noise, eliminates ground loops, and improves safety. IsoLoop Isolators
are faster, more reliable, and simpler than Optocouplers. Here's an opto isolation
board full of components.
That's a lot of parts...
Right. But IsoLoop isolators provide a single-chip solution, combining isolation
with communication functions. There's a full line of isolated transceivers, including
the workhorse IL485, the high-speed IL3585, the fractional-load IL3285, and the
low-cost IL3185. There are 16-pin wide and narrow-body SOICs and ultra-miniature
QSOPs. So let's sum up with my RS-485 tips:
* keep stubs short;
* use twisted-pair cable;
* shielding for long or fast busses;
* ground one end of the shield;
* terminate bus ends;
* bias for fail-safe; and, of course,
* use isolators.
Thanks Sandy. Where can viewers go for more info?
You can click, to go to the RS-485 Application
Center at nve.com. Or call us at 800-GMR-7141, extension four.
we're going to learn best practices for data converter board layout with Sandy
in the NVE Applications Center. So how do we optimize data conversion?
are two under-appreciated keys to data converter performance. First, PCB layout;
and second, isolation. Here's an example of a not-so-good layout. The customer
expected about 92 decibels signal-to-noise plus distortion. But he got closer
to just 80.
Eighty decibels! Where did he go wrong? He has a ground plane...
Yes, but high-speed digital currents
are still injected into analog paths. So here we've moved the ADC closer to the
supply, and made sure digital grounds don't travel across, under, or parallel
to analog signals. But there's still noise from digital ground currents. And that
can be eliminated with isolation.
Why is the isolated circuit so much better?
Here's the equivalent circuit for the non-isolated board. Asynchronous
noise is injected into the ADC ground. But isolation separates the digital and
analog paths. Current spikes are routed to ground without interfering with the
analog signal. Complete galvanic isolation and best performance is achieved by
using separate analog and digital power supplies, which is possible with isolation.
Where can I learn more about getting on "board?"
You can click to go to the data converter
application center at nve.com. Or call us at 800-GMR-7141, extension four.
What's with the set and music?
We're talking about "spy."
No, this spy is "S-P-I," which stands for Serial Peripheral Interfacea
common bus for ADCs and DACs. Isolating SPI eliminates ground loops and reduces
noise, which improves accuracy. Here's a single-channel isolated SPI Delta-Sigma
A-to-D Converter using an IL717. The ADC is located on the bridge with no signal
conditioning electronics between the bridge sensor and the ADC. The IL717 isolates
the SPI control bus from the microcontroller, and the system clock is on the isolated
side of the system.
What if I have multichannel A-to-Ds?
Then you need an IL200-Series five-channel isolator. For example, the IL262 dash-3 has
five channels in a 0.15-inch SOIC-16. This circuit is an example of a multi-channel
sampling system with separate A-to-D cells. The IL262 is used to control the SPI
lines and send ADC-busy commands back to the host, for efficient interrupt-driven
sampling. The busy line can also be used as a frame synchronization signal in
I love video. But I hate optos. How are IsoLoops better?
IsoLoops are ideal for
SPI because they have higher speed, less distortion, unlimited life, and higher
channel density than optos.
Where can a guy "spy" more on SPI?
You can click to go to the SPI application
center at nve.com. Or call us at 800-GMR-7141, extension four.
Audio Isolation (01:26)
with the music?
The Can-Can, because we're talking about CAN bus in the NVE Applications Center.
Well, CAN bus isolation is a good idea and allows higher speed and more reliable
operation by eliminating ground loops and reducing susceptibility to noise and
EMI. And, in high-voltage Battery Management Systems, isolation improves safety.
To complement stand-alone CAN transceivers, NVE has several two-channel, bidirectional
isolators, including unique MSOPs. The new IL721 bidirectional isolator is similar
to the popular IL712, but with a reversed channel configuration to better suit
some layouts. Both the IL721 and IL712 have best-in-class ten nanosecond propagation
delay. This minimizes loop delay and maximizes speed over a given bus length.
How about single-chip isolated transceivers?
Right. IL41050 single-chip isolated transceivers simplify CAN circuits even more.
They fully comply with ISO 11898 and de facto industry standards. Loop
delay is a best-in-class 180 nanoseconds. And advanced features make for reliable
bus operation. Wide-body, narrow-body, and QSOPs are available.
So you're saying you "can" isolate "CAN." So then where "can"
I get data sheets and more info?
You can click, to go to the isolator section of nve.com. Or call us at 800-GMR-seventy one
four one, extension four.
Passive-Input Isolators (01:30)
talking jitter with Sandy in the NVE Isolator Applications Center. Sandy, what
is isolator "jitter" and why is it important?
can be a problem with conventional isolators. As this shows, jitter is variation
in the pulse edge position of a data stream.
just so sad!
Indeed. But NVE's IL700 isolators have a
virtually undetectable 50 picoseconds of jitter, making them ideal for precision
How do isolators help audio circuits?
ways. First, galvanic isolation can eliminate hum and noise. And second, low jitter
ensures PCM data phase.
Can you give an example?
Here's a typical circuit. It's good practice to isolate serial CD, or MP3
data from the analog sound system. This eliminates ground loops, speaker hum,
and high frequency pickup caused by digital currents in analog paths.
that "sounds" good. Where can I get more info?
You can click to go to the iso application centers at nve.com. Or call us at 800-GMR-7141,
True 8 (true 8 mm creepage) Isolator
we're talking about, NVE's award-winning passive-input, drop-in opto-isolator
replacements. Sandy, what's "passive input"?
CMOS input isolators, IL600s have a current-sensitive, resistive coil, like an
LED input, but without the voltage drop. The outputs also mirror LED optos with
CMOS and open-drain options. A variety of one, two, and three channel models are
available to upgrade from optos.
Well, I hate optos. Why
are IL600s better?
We summarize the advantages with, four
* Boxesdrop-in PDIPs or SOICs, plus unique ultra-miniature
* Boundsflexible inputs, CMOS or open-drain options;
Baudhigher speed and less pulse-width distortion; and
* Bulletproof, with no LED to be damaged or degrade with time.
"four Bs." As you British say, "brilliant." Where can viewers
get datasheets and samples?
You can click, to go to the
IL600 section of our on-line store for samples and datasheets.
Transient Immunity (03:30)
We're learning about NVE's exclusive, True 8 millimeter creepage Isolators,with
Tammy in the NVE Application center. Tammy, give us the "truth" about isolator
Creepage is the minimum spacing over insulation. Stringent rules are set by IEC60601.
Sixteen-pin wide-body SOICs are popular for isolators, but most don't meet the
eight-millimeter creepage requirement.
Why so, Tamarino?
With most packages, JEDEC tolerances, mold variability, and surface metal in the
creepage path means full 8 millimeter creepage can't be assured.
The shortest creepage path is usually around the end of the package. Ordinary
JEDEC wide-body packages are nominally 7.4 millimeters wide, with approximately
8.1 millimeters between pins, before subtracting tie bars.
I used to wear tie bars...
Well, these tie bars are tabs used in the molding process. Internally connected or not, the exposed metal reduces creepage. The tie bar subtraction, for an ordinary JEDEC package is typically half a millimeter, bringing creepage to 7.4 to 7.6 millimeters. That's not enough for 250 working volt applications, even before tolerances.
So how are NVE True-8 packages, different?
Instead of a general-purpose package, we custom tooled the True 8 package.
It's within the JEDEC standard, so there's no special board layout or handling.
It has much tighter tolerances for width and pin position. And, rather than two
metal tabs, the True 8 has just one thin tab that secures the lead-frame
during molding. Creepage around the end is calculated as the package edge width,
plus two pin-to-end spacings, less the surface metal. Even with worst-case package
dimensions and pin position, the True 8 ensures 8.03 millimeters minimum
creepage. The path over the top is also specified for 8 millimeters. Eliminating
the outdated pin-one edge chamfer ensures over-the-top creepage.
So, what parts come in the True 8 package?
There are oodles of True 8 parts: high-performance IL700/200 Series Isolators, single-chip transceivers, cost-effective IL500s, and IL600 Passive Input Isolators. Common applications include serial busses, SPI, A-D converters, and power interfaces.
Thanks, Tammy. How can viewers learn more about the True 8 package?
We have links to application bulletins, or call us at 800-GMR-7141,
High-Voltage Standards (03:44)
We are talking about NVE isolators' best-in-class transient immunity with Jamie,
in the NVE Applications Center. Jamie, why should we care about transient immunity?
Transient immunity is important in noisy environments. It is also important in
floating supply applications such as power control gate drivers, where the change
in the common-mode power supply voltage can cause spurious switching. Floating
gate drive supplies are used to drive power MOSFET gates [illustrative circuit].
As the MOSFET turns on, the floating power reference voltage increases with the
voltage on the load. This creates a rapidly changing potential between the isolator
output-side supply voltage and the input reference. This changing potential can
cause parasitic currents through stray capacitance proportional to dV/dt. Isolators
are designed to cancel these common-mode currents, but the cancellation is not
perfect, which is why we have transient immunity, or dV/dt, specifications.
So, what part types are used for gate drivers?
Here is a summary [selector guide]:
* IL700-Series isolators have best-in-class 50 kilovolts per microsecond
* Typical pulse width distortion is one nanosecond, which minimizes
dead time and improves efficiency.
* The IL600-Series is failsafe.
* Parts are available in MSOPs to minimize board area.
* Although IL600's are specified at 20 kilovolts per microsecond,
increasing the input drive to 10 mA from the 5 mA minimum can increase immunity
from the specified typical of 20 kilovolts per microsecond to as much as 70 kilovolts
per microsecond or more.
How do you test transient immunity?
IEC 61000-4-4 is the governing standard. Here's a practical test setup [test circuit].
It determines the maximum sinusoidal amplitude the isolator can tolerate without
The oscilloscope monitors the amplitude of the driving sine wave, as well as the
isolator output. The green trace is the isolator input and the purple trace is
the output. The signal generator amplitude is gradually increased until spurious
isolator outputs [oscilloscope traces].
The applied dV/dt is the slope of this curve [slope line].
Thanks, Jamie. Where can I get details on transient immunity testing?
Visit nve.com for detailed application bulletins. Or call us at 800-GMR-7141,
We are talking high voltage with Jamie, in the new NVE Applications Center. Jamie,
give us the four-one-one on 1577, 61010, and 884.
We should explain that 1577 is a UL high-voltage standard, 61010 is an IEC equipment
standard, and 884 is VDE.
What's the difference?
All three standards have one minute destructive sample sample tests and 100% one-second
production tests. An important distinction: 1577 and 61010 are based on voltage
breakdown, which is the voltage that causes high leakage leakage current. But
884 and 60747 use partial discharge, so theyre becoming more popular..
Right. Partial discharges do not completely bridge the isolation barrier. But
they degrade the barrier.
How so, Jame-arino?
Small bubbles or barrier impurities can cause partial discharge. When a high voltage
is applied, there is a discharge through the defect. That damages the isolation
barrier, which makes the device more susceptible the next time. Eventually, the
barrier can break down.
Holy coulombs, Batgirl! And breakdown tests cant detect these failures until
its too late.
Breakdown-based standards such as 1577 and 61010 are based on one-minute breakdown
voltage, so isolators cannot continuously operate at the rating. And its
not a measure of transient tolerance, either. Like pressure testing a surgical
glove, partial discharge tests uncover tiny imperfections that could grow.
So lay down some numbers...
IsoLoop isolators have best-in-class partial discharge performance, including
600 working volts and 4000 volts transient overvoltage for all package types.
All package types? Even MSOPs and QSOPs?
Absolutely. MSOP and QSOPs are amazing.
What about folks who need wide-body isolators?
So lets cover one more standard: IEC 60601, which sets stringent rules for
Wasnt Creepage a Radiohead song?
Not exactly. But most sixteen-pin wide-body SOICs dont meet the eight millimeter
creepage required under 60601 for 250 working volts. Ordinary JEDEC wide-bodies
have two tie bars.
I used to wear tie bars...
Well, these tie bars are tabs used in molding. Internally connected or not, the
exposed metal reduces creepage. For an ordinary package, creepage is reduced below
And the True 8 package?
Its custom tooled, with tighter tolerances, and just one tie-bar. But its
within JEDEC standards, so theres no special board layout or handling.
Are there other high-voltage metrics of interest?
Other IsoLoop best-in-class high-voltage metrics include:
* Barrier resistance;
* Comparative Tracking Index;
* High-voltage endurance; and
* Barrier life.
Thanks, Jamie. How can viewers learn more?
Visit nve.com for application bulletins. Or call us at 800-GMR-7141, extension 4.
Barrier Life (02:05)
Jamie, here's a question we get frequently, here in the NVE Applications Center.
Working voltage for NVE isolators is 600 volts RMS and transient overvoltage is
4000 volts peak.
Yes, best in class.
So, that's great for line-voltage applications. But what voltage can they withstand continuously?
The maximum voltage indefinitely across an isolator is called endurance voltage.
It's not in most data sheets, and may be the least familiar high-voltage specification.
But it's useful in applications where the isolators aren't subject to line voltage
transients. IsoLoop endurance voltage is 1000 volts RMS and 1500 volts DC.
How do you test endurance?
* Endurance tests the isolation barrier. Our extraordinary
endurance comes from a unique polymer/ceramic composite barrier.
* Pins are tied together on each side.
* The endurance voltage is applied; and
* we leave it like that for years.
Years? How long is indefinitely?
Our reliability data show an MTBF of 44000 years at the endurance voltage and
100 degrees C.
So you've had parts on life test since the Pleistocene Epoch?
Well, since the Clinton administration, with standard FITs extrapolations.
And where is endurance voltage important?
Battery management systems come to mind. In this circuit, the last isolation stage
is 300 volts above controller ground. So the isolator sees 300 volts
between supplies. But with an endurance voltage of 1500 volts DC, it
will easily withstand the voltages in this circuit.
And that endurance spec is for all packages? Even MSOPs?
You bet-cha. The unique IsoLoop MSOPs meet that endurance spec.
Well, thanks, Jamie. Where can I get more info?
Visit nve.com for application bulletins. Or call us at 800-GMR-7141, extension
Enable Next-Generation Switching-Mode Power Supplies (06:35)
I'm here with NVE's Quality Guru. Dee, do IsoLoop isolators really last "forever"?
For practical purposes, forever. Optos use LEDs, which degrade over time. Every photon emitted depletes the LED, so the output decreases and eventually it will fail.
That's one reason engineers hate optos.
Sure. But IsoLoop isolators use magnetics and spintronics rather than optics. Nothing is depleted. Nothing degrades with time. Think of a microwave versus an Easy Bake oven.
Neat. But what about the life of the isolation barrier?
The isolation barrier is a patented glass-polymer structure that doesn't degrade. Compared to conventional semiconductors, its like indestructible fiberglass versus rusting sheet metal.
Do you have data to back that up?
Absolutely. Our reliability data show failures-in-time of 2.6 per billion hours at 100 degrees C, for an MTBF of 44,000 years. Some conventional semiconductor isolators brag about a barrier life of just 13 years.
Thirteen years gives engineers "fits." Heck, even my ties last longer
than that. So you've had parts on life test since the Pleistocene Epoch?
Well, at least since the Reagan administration, with standard FITs extrapolations.
Thanks, Dee. How can viewers get IsoLoop reliability reports?
You can contact us at email@example.com.
We're talking about Switching-Mode Power Supplies, or SMPS, with Jamie, in the
NVE Applications Center.
Switching-Mode Power Supplies are widely used because of their power density,
efficiency, and reliability. Today we'll look at SMPS technology and new 2.5 kV
MSOP Isolators that enable better-than-ever power supplies.
So just like you evolved...
Hey, that's not me!
Sorry. SMPS technology has evolved from non-synchronous buck converters to synchronous
rectification. Synchronous rectification has evolved from self-synchronized to
controller-driven. Synchronous gate drive isolation is the most challenging isolation
element in modern SMPS. It's gone from pulse transformers, to optocouplers, to
digital isolators such as IsoLoop Spintronic Isolators.
Some pros and cons of the design approaches:
Non-synchronous converters rely on Schottky diodes, and forward diode drops mean
inefficiencies. The losses are even more significant with newer, low-voltage supplies.
Self-synchronized rectification uses the AC waveform to switch the MOSFETs. It's
simple, but inefficient because it's not well synchronized, and doesn't drive
the MOSFET ON resistance as low as it could be.
Controller-driven synchronous rectification is the preferred design because it
provides excellent synchronization, which maximizes efficiency. But it requires
We'll get back to isolation, but are SMPS designs complicated?
So simple a caveman can do it.
Here's a typical circuit. Key components are:
* a pulse width modulation controller;
* switching MOSFETs;
* synchronous rectifier MOSFETs;
* MOSFET gate drivers;
* gate driver isolation; and
* an error amplifier and linear isolator (optocouplers remain popular
here because next-generation digital isolators aren't suited for analog).
But you're working on that. So what about digital isolators?
There are three types of non-optical digital isolators: inductive, capacitive,
Inductive isolators have limited pulse position accuracy because carrier frequencies
limit accuracy, and the high-frequency signaling causes EMI emissions. Device
life is better than optos at 50 to 100 years, but of course there's a statistical
chance of any particular device failing much sooner.
Capacitive isolators are also imprecise and generate EMI, and their rated life
is only 13 years.
Spintronic isolators have excellent precision because they're not clocked, low
emissions because they don't have carriers or refresh clocks, are the smallest
with MSOP8s available, and have a remarkable 44000-year life.
Why is precision so important in SMPS?
Gate drivers and isolators need to be many times faster than the switching frequency.
Speed and precision improve efficiency. Here's why:
Propagation delay causes the MOSFET to turn on and off after the zero crossings.
That means the first part of the waveform isn't used, and there's energy loss
from a negative voltage, here.
Channel-to-channel skew also causes inefficiencies. With this skew, neither MOSFET
is ON, so the AC energy is lost. This type of skew is even worse, because both
MOSFETs are ON, which means no net energy transfer and there are ohmic losses
from the MOSFETs fighting each other.
Spintronic isolators have best-in-class speed and precision specifications.
Now I understand precision. Let's get back to reliability. You said Spintronic
isolators have a 44000-year life. So you've had parts on life test since the Pleistocene
Well, since the Clinton administration, with standard FITs extrapolations. Here's
a comparison of isolator life specs.
So where are Spintronic Isolators?
We need to expand the scale quite a bit. Past the industrial age, the pyramids,
great civilizations, the first settlements, farming...
Is that a woolly mammoth?
Yes, we're talking serious MTBF here.
Well, thanks for that "powerful" information, Jamie. Where can I "buck
up" with more?
You can sync up at nve.com for datasheets and reference designs.