AnalogTalk

With the increasing popularity of portable devices, demands on rechargeable batteries have driven the growth of its surrounding electronics, such as battery protection IC  and battery charger IC. Battery capacity below 500 mAh is a popular category for small form factor products.  In such applications, LDO (low drop out) regulator based architecture is often used to design to recharge batteries.

One popular question often received from system designers is that the charge current during fast charge stage is insufficient and results in longer charge times. It may also trigger built-in timers and experiences premature termination. When investigated, two common scenarios were found – improper package selection or lack of thermal design. In order to fit in space limited portable systems, the smallest packages are selected, such as the 5-pin SOT-23 or small DFNs. Using a Lithium-Ion Polymer battery as an example, the fast charge voltage threshold is at 3V (typical) and  5V for the power supply of a portable device. The power dissipation can be calculated as following for LDO based battery charger:

Pdisipation = (Vin – Vbattery) x Iout

Ignoring the internal supply current, the worst case value will be 2V x 500 mA = 1 Watt

Typical thermal resistance of a SOT-23 package can be 230 C/W. For one watt dissipation, it will heat up to be 230C above room temp and so is practically not possible to deliver that much charge current.

If delivering less than 200 mA can’t fulfill a designer’s need, he may go for a thermally enhanced package, such as a 2mm x 3mm DFN package where typical thermal resistance value is about 76 C/W.

It is feasible for such a package to handle the current, but why does it still sometimes fails to deliver that much current?

The answer is the enclosed environment.  So the PCB layout and thermal design of the product becomes even more important. Two tips to improve the thermal performance is to allow enough copper pour and apply copper to the exposed pad with vias under the DFN package. This will help to draw the heat away from package .

Piezoelectric horns are used in applications such as smoke detectors, telephones appliances and a wide variety of other equipment.  An over looked characteristic of a piezoelectric horn is the bidirectional nature of the energy conversion process of the piezoelectric element.  Manufacturers of piezoelectric horns warn against mechanical shock resulting in stress being applied to the piezoelectric element but stress during finished product assembly can also cause problems.

In one case the piezoelectric element was mounted in the top half of a smoke detector case.  The bottom half of the case contained the PCB and the PCB had L shaped brackets which contacted the piezo element when the top and bottom half of the cases were snapped together.  If the halves of the case were not aligned properly, when the case was snapped together, contact to the piezo element was not uniformly made.  The L shaped bracket which contacted the piezo element first stressed the piezo element and generated a large voltage which was applied to the second L shaped bracket when it contacted the piezo element.    The current flow which discharged the large voltage damaged the horn driver circuit.

A piezoelectric horn consists of the piezoelectric element already mounted in a plastic case with various types of electrical connections. Some piezo horns are built with pins that allow them to be soldered into a PCB.  In a case where a piezo horn was wave soldered to the PCB, the horn driver circuit was being damaged.  In previous products this process was not a problem, although it was reported that the horns popped as they went through the solder wave.  In the new product prototype runs the horn driver circuit was being damaged.  In the new product the pins of the piezoelectric horn were aligned differently to the solder wave.  Heat from the process caused expansion of the horn case and piezoelectric element which generated a large voltage.  As the PCB passed through the solder wave, the pins of the horn were connected to the horn driver circuit in a slightly different order or with a slightly different timing which allowed the voltage discharge to damage the horn driver.

A manufacturing yield problem related to a piezoelectric horn not operating was traced to stress in the finished product assembly. When the horn was tested after being removed from the case, it worked properly.  The horn was held snuggly in the case with two spring loaded points formed in the top of the plastic case.  The stress produced by the spring loaded points on the horn’s plastic case affected the piezoelectric horns ability to start up properly.  By adjusting the value of a resistor in the horn driver circuit, the piezoelectric horn was able to start up reliably.

As an electronic component, the piezoelectric horn can be relatively easy to use.  The bidirectional nature of the horn’s energy conversion process can cause difficulties during the product manufacturing process.  Understanding the effects of mechanical stress on the electronic performance of the piezoelectric horn may be necessary.

Dynamic range is the ratio between the  largest and smallest possible values of a changeable quantity of a signal, and is most often presented as a ratio or as a logarithmic  value.  Within power measurement, the dynamic range refers to the current range the active power is being calculated over.  For example, if a system needs accurate measurement of power over 100 mA minimum to a 10A maximum, then the dynamic range would be 100:1.

The drive for better power efficiency, power monitoring and power management has started to push for greater dynamic range at higher accuracy as the cost for electricity rises.  Where once typical electricity meters have been designed to a dynamic range of 500:1  to a power accuracy of 0.5%, it is now common to see dynamic ranges of 2000:1 and even 8000:1 in higher end commercial and industrial meters.  Simple power monitoring, which managed very rough accuracy’s, is also pushing towards <5% active power measurement at lower current levels as seen in the next generation of servers.

To meet these needs, improvements  in sensor technology (resistive shunt sensors, current transformers, rogowski coils) and the analog-to-digital conversion process are now allowing cost-effective solutions for higher accuracy and higher dynamic range power management.

When it comes to monolithic instrumentation amplifiers (INAs), there are a variety of methods that manufacturers use in order to implement the gain setting. These methods include fixed gain, programmable gain and using one or two external resistors to set the gain. Each of these methods has advantages and disadvantages.

Fixed gain INAs are very useful in that these devices can be optimized for a specific gain, improving linearity, offset and gain accuracy for the specific gain that is implemented. These devices can also be placed in smaller packaging, since additional pins for setting gain are not needed. The obvious downside is that the gain is fixed, meaning it cannot be adjusted by the end user.

As the name implies, programmable gain amplifiers implement a register set that allows the user to set the gain and possibly adjust other settings as well. These amplifiers typically use a standard I²C or SPI interface that can easily be controlled by a microcontroller. This programmability allows for a much more flexible device, but requires a microcontroller or other serial interface device and can be a relatively large silicon solution, which adds to the cost.

Using external resistors is a very popular method for setting the gain of a monolithic INA. Some devices allow the gain to be set via one external resistor, while on other devices gain is set via the ratio of two external resistors. At a glance, it may seem that using one external resistor is better – eliminating an external resistor saves a little cost, board space and reduces design complexity.

However, there are disadvantages to this single resistor approach. Whether using one or two external resistors, the gain is set via the ratio of two resistors. In the case of a single external resistor, it is the ratio of an internal resistor constructed within the IC and the external resistor provided by the user. This technique requires the internal resistor value to be accurately known, such that the user can set the gain properly. Most manufactures must trim this internal resistor in order to get this level of accuracy. Another disadvantage is that the internal resistor and external resistor will have different drift characteristics, which will lead to gain errors across time and temperature.

Implementing a monolithic INA with two external gain setting resistors eliminates these concerns. With proper layout, the two external resistors will track very closely over time and temperature, and the accuracy of the ratio of the two external resistors is left entirely up to the user.

So the next time your design requires a monolithic INA, be sure to consider the gain setting methodology, and when it comes to using external resistors to set the gain, remember that less isn’t always better.

Digital-to-Analog Converters (DACs) and digital potentiometers (DigiPots) are similar in many ways that make them both suitable for similar applications. For example, both DACs and DigiPots have a digital control (often SPI or I²C) in that control the wiper position, the wiper increments/decrements over a resistive ladder that creates a voltage output.  However, there are system limits that may make one or the other the only suitable solution.

DACs often include an output buffer that enables them to drive low-impedance loads, while DigiPots are often left with an output that isn’t buffered. There is also the option of current output versus voltage output with a DAC.  DACs also offer higher resolution and speed options that aren’t seen in digital potentiometer; high resolution digital potentiometers have around 1024 taps, the equivalent to a 10-bit DAC, while high-end DACs can have resolutions up to 24-bits.

A DigiPot provides access to both the resistive ladder and wiper terminals. The wipers can be tied to either the high side or low-side terminals to effectively make it a variable resistor (rheostat mode).

DigiPots also have the option of a linear or logarithmic taper; a logarithmic taper is useful in audio equipment since the human ear perceives a logarithmic increase in volume (dB) as a linear increase in volume.  With simpler controls and lower resolutions, DigiPots are often cheaper.

MOSFET devices are among the highest power loss contributors in POL converters. In High step down applications, such as computers, that typically operate with high input voltage (12V – 20V) and low operating voltages (.8V – 3.3V), the high side switch has a low duty cycle and the low side switch has a high duty cycle.

During switching, the low side (LS) and high side (HS) MOSFET switch on and off with some dead-time in between to avoid shoot through current.

To optimize performance the upper MOSFET (Q1) gate charge must be very low in order to minimize switching losses and the low side MOSFET (Q2) RDS(on) must be very low in order to minimize conduction losses.

For example, in High Step down applications where Vin = 12V and Vout = 1V, the high side duty cycle is 8% and the low side duty cycle is 92%.  The MOSFET (Q1 and Q2) Conduction loss and MOSFET Q2 switching losses can be estimated using the equations in the spreadsheet below. Other losses such as inductor loss, capacitor loss, gate drive loss, low side switching loss, and body diode losses are not included.

The losses increase with higher switching frequencies and MOSFET capacitance.

The chart below shows the power loss for high side/low side conduction loss and high side switching loss from 1A to 20A output current.

Op Amps can be used in function generation applications, such as triangle waves generator, as shown in Figure below.  In this example, a triangle wave generator consists of an integrator and one comparator, connected in a positive feedback loop. This approach is based on the simple fact that integration of a constant voltage results in a linear ramp. The op amp is configured as an integrator using R1 and C1 to provide the triangular output and the Schmitt triggers are designed with R2 and R3 to change the state corresponding to the desired peak voltages of the triangular wave output. The reference voltage (VREF) is supplied by a low-impedance source. In single-supply applications, VREF is typically VDD/2.

At times designers will find that a little more accuracy is needed out of their system Analog-to-Digital Converter.  There may be reasons not to change to a more accurate device, such as cost, BOM restrictions and hardware redesign, or it is integrated into an MCU that can’t be changed.  And there may be no need to change.  Oversampling is an allows one to get a little more accuracy out of an ADC.  Of course, this depends on the type

Over-sampling with a fast SAR ADC is common to increase accuracy, including with on-board converters often found on MCUs.  Over-sampling and averaging increases accuracy by 0.5 bit for each doubling of the sample frequency.  As one can see, there are limits.  A first order delta-sigma ADC however decreases in-band noise by 9 dB (due to it’s noise-shaping architecture) or adding 1.5 bits of accuracy for every doubling of the over-sampling ratio, three-times better than simple over-sampling of an SAR ADC.

Car electronics have made great strides in recent years.  Strategy Analytics recently released numbers that sensors within the automobile have grown year over year ~12%.  The major driver’s of this growth, which is expected to continue in the future, is fuel mileage, environmental and safety legislation in addition to many new comfort and convenience applications.  A few examples of newer applications you may be familiar with include Park Assist Systems, interior mood lighting, and rearview backup cameras.

It is estimated close to 1500 children a year are injured due to non-traffic incidents and 25% of those are killed.  The more disturbing fact is that in a lot of cases it’s family members that are driving. In late 2010, the National Highway Traffic Safety Administration proposed a mandatory rear-view mirror camera installation in all passenger vehicles by 2014.  That regulation stalled in March of this year and is still spinning its wheels.  With elections in the near future motivation for increasing safety requirements with resulting costs on manufacturers have gotten political and will likely be put off.

Latest products from automotive suppliers is this Magna 3.5” TFT LCD which incorporates the the rear-view camera picture into the rear-view mirror.

If you look at Wikipedia (http://en.wikipedia.org/wiki/Smart_power) you’ll see a description on the use of ‘hard’ & ‘soft’ power strategies as it relates to diplomatic and military uses of power and influence….  While that is an interesting topic, that wasn’t really my intent, so instead we’ll talk about smart power in the context of electrical power conversion.

As mentioned in a previous blog post “Dumb Power May Be Smarter: When Smart Power Is Not So Smart”, there is an increasingly amount of ‘buzz’ surrounding ‘Smart Power’…  so what is ‘Smart Power’?  What constitutes ‘Smart’…  when is something ‘Smart’?  Or is this just Marketing’s gobbledygook?

The entirely unscientific, statistically insignificant, and yet reasonable answer… I don’t know.

While I am initially leaning towards marketing gobbledygook, perhaps there is some usefulness to the terminology. So let’s take a moment, and think about it.  Let’s start with the definition of intelligence (from dictionary.com):

in·tel·li·gence   [in-tel-i-juhns] – Capacity for learning, reasoning, understanding, and similar forms of mental activity; aptitude in grasping truths, relationships, facts, meanings, etc.

When I see ‘learning’, ‘reasoning’, ‘understanding’… I’m not sure any power supply system fit this description. (Give 1 for marketing gobbledygook).

Competitive Intelligence (don’t get me started)

Taking a few ‘smart/intelligent/etc…’ branded parts, the only commonality I see is they have a digital component.  Perhaps they have a fully digital (as in some MCU’s or DSP’s) controller with some analog sensors or gate drive.  Or perhaps it’s the Analog peripheral in a small MCU….  But all of them have some digital that provides an easy way to: at a minimum handle the higher-level, supervisory, communication, and configuration functions, or at most implement fully digital, custom control algorithms.  They still don’t satisfy the definition of intelligence… they can only do what we tell them to do… which is the case for any design.  (another 1 for marketing gobbledygook)

Taking a different angle, what in, or about, an analog power supply is intelligent?

…. I’m still thinking.

… Part of what makes analog so attractive is its elegant simplicity with high performance.  I know, I know, some of you are thinking “Analog…. Simple?!?!?!  … I once had this design that did XYZ….  And I STILL don’t understand it… 25 years later!’.  There are definitely challenges in analog designs, and extra care must be taken when evaluating performance, including at the corners, but tradeoffs are a part of any design… and the fun… or sense of accomplishment!

Getting back to the question  – ‘Smart Analog’ certainly doesn’t fit the definition of intelligence, but it’s not clear that an analog supply can’t do everything a digital supply could do. .  (another 1 for marketing gobbledygook).

There’s little argument that digital, analog, and hybrid digital+analog power supplies each have their own unique strengths, weaknesses, opportunities, and tradeoffs, and are all valuable tools to address today’s power problems.  Given that it’s looking like “Smart power” has no real use outside of marketing gobbledygook, perhaps the only ‘Smart Power’ play is for the designers to intelligently design power supplies to the needs of their applications.