Author Archives: descoeurope

Is ESD Control a Waste of Time (and Money)?

Today’s post is going to be a bit shorter than usual. BUT: that doesn’t mean it’s going to be any less interesting. Quite the opposite! So, let’s jump right in.

A little while ago we were approached by a customer with the following statement:

Generally speaking, most IC’s these days already have adequate protection on their pins, the notable exception being discrete J-FETs, and MOSFETs, especially for RF applications.
It’s difficult to advise when these might be in use on an assembly without giving everyone in-depth training on circuit design, so to avoid trouble in the 1% of cases that matter, it’s a good idea to play safe and keep applying our procedures for the other 99% of parts too.
I am of the opinion that a PCCU in its housing does not need special treatment though. It has ESD protection, and has passed testing for this, so I am not worried about someone touching its pins without wearing a grounded wristband, etc …

So, is this statement true? Is ESD Control obsolete? Let’s find out!

Types of ESD Damage

Remember that there are two types of ESD damage:
1) catastrophic failure and
2) latent defects.

While catastrophic failures cause an ESD sensitive item to be damaged permanently, latent defects only partially degrade an ESD sensitive item that is exposed to an ESD event. It may continue to perform its intended function and may not be detected by normal inspection. However, intermittent or permanent failures may occur later.

Bottom line: Even if an ESD sensitive component has quite a high withstand voltage and no catastrophic failure has been caused, latent defects may still make your life miserable.

Continued Requirement for ESD Control

Here is a link to the ESD Association’s ESD Technology Roadmap. The document illustrates what future thresholds are expected for ESD sensitive devices and how they impact on ESD Control. The thresholds are determined by current trends in the semiconductor industry and are displayed as “roadmaps”. The aim is to predict future limitations of device protection which are driven by performance requirements and technology scaling.

You should head-over now and read through the document. But in case you haven’t got time, here are the main take-away notes:

  • Finally, these trends point to the need for continued improvements in ESD control procedures and compliance.” [section 1.0 Synopsis]
  • Therefore, the prevailing trend will be circuit performance at the expense of ESD protection levels.” [section 2.1 Overview]
  • Therefore, implementation of advanced HBM controls using the limits and qualifications requirements in ANSI/ESD S20.20, IEC 61340-5-1, or JESD625, would become necessary within the next 3 years.” [section 2.2 Device ESD Threshold Roadmaps]

Bottom line: As electronic technology advances, electronic circuitry gets progressively smaller. As the size of components is reduced, so is the microscopic spacing of insulators and circuits within them, increasing their sensitivity to ESD. Therefore, the need for proper ESD protection increases every day.

ESD control procedures and compliance continue to be requiredESD control procedures and compliance continue to be required

For more information on the ESD damage and the costly effects of ESD, check-out this post.

Protect your sensitive devices from ESD Damage

Every company should document the most ESD sensitive device that they are handling.
A prerequisite of ESD control is the accurate and consistent identification of ESD susceptible items. Some companies assume that all electronic components are ESD susceptible. However, others write their ESD control plan based on the device and item susceptibility or withstand voltage of the most sensitive components used in the facility. A general rule is to treat any device or component that is received in ESD packaging as an ESD susceptible item.
This post provides further information on how to set-up an ESD Control Plan.

So, tell us: are there instances in your company where you forego standard ESD Control practices? If so, let us know in the comments – we’d like to hear from you.

ESD Classifications

We were recently approached by a customer who wanted to know more about the different classifications of ESD products. So, we thought this would be a good opportunity to share more details with you. Be warned – this is a very theoretical post: so, loads of text but not too many pictures. We promise, we’ll have some more images in our next post!

Introduction

Part of every ESD Control plan is to identify items in your company that are sensitive to ESD. At the same time, you need to recognize the level of their sensitivity. As explained by the ESD Association, how susceptible to ESD a product is depends on the item’s ability to either:

  • dissipate the discharge energy or
  • withstand the levels of current.

Whilst some items are easily damaged by discharges arising within automated equipment, others may be more susceptible to damages from personnel when being handled.

There are three main classifications based on three different ESD models. There are detailed standards available from the ESD Association:

  • Human Body Model or HBM [100 pF @ 1.5 kilohms]: ANSI/ESDA-JEDEC JS-001-2010
  • Charge Device Model or CDM [4 pF/30 pF]: ESD DS5.3.1
  • Machine Model or MM [200 pF @ 0 ohms]: ESD STM5.2

The two primary models used for ESD events today are the Human Body Model (HBM) and Charged Device Model (CDM).

The Human Body Model (HBM)

The most common model is the HBM.  This model simulates discharge occurring between a human (hand/finger) and a conductor (metal rail). For this model, a 100 picofarads (100 x 10-12 Farads) capacitor is discharged through a 1,500 ohms resistor to simulate a human body. The typical rise time of the current pulse (ESD) through a shorting wire averages 6 nanoseconds (6 x 10-9 s) and is larger for a higher resistant load.  The peak current through a 500 ohm resistor averages 463 mA for a 1,000 volt pre-charge voltage.

If a device has failed if it does not meet the parameters outlined in the datasheet.

Class Voltage Range
Class 0
Class 1A 250 volts to
Class 1B 500 volts to <1,000 volts
Class 1C 1,000 volts to <2,000 volts
Class 2 2,000 volts to <4,000 volts
Class 3A 4,000 volts to <8,000 volts
Class 3B ≥ 8,000 volts

ESDS Component Sensitivity Classification for the Human Body Model (Per ESD-STM5.1)

Charged Device Model (CDM)

This is the most neglected one of the three models but it can severely compromise your ESD control programme. Here, it is the ESDS device itself that becomes charged (sliding out of a tube/bag/sorter/etc.) and when contacting a grounded conductor (table top/hand/metal tool) it will discharge to that conductor and may result in damaging ESD. The length of the discharge may be very short (less than 1 nanosecond) – however, the peak current can reach a high amperage.

The model uses a 4 pF or 30 pF verification module which can simulate from 2 to 30 Amps peak current for non-socked and up to 18 amps for socketed devices.

Class Voltage Range
Class C1
Class C2 125 volts to
Class C3 250 volts to
Class C4 500 volts to <1,000 volts
Class C5 1,000 volts to <1,500 volts
Class C6 1,500 volts to <2,000 volts
Class C7 ≥ 2,000 volts

ESDS Component Sensitivity Classification for the Charged Device Model (Per ESD-STM5.3.1)

Machine Model (MM)

This model simulates a machine discharging through a device to ground. When checking components to the Machine Model (MM), the test replicates MM failures and tells you the MM ESD sensitivity levels for your devices. The criteria is 200 pF at nominal 0 ohms.

Class Voltage Range
Class M1
Class M2 100 volts to
Class M3 200 volts to
Class M4 ≥ 400 volts

ESDS Component Sensitivity Classification for the Machine Model (Per ESD-STM5.2)

Each component in your company should be fully classified using HBM and CDM. That means an item may have a Class 2 (HBM) and Class C1 (CDM).

Bear in mind that these classifications are guides only and do not represent the real world. However, they can be used to:

  • Develop and measure suitable on-chip protection.
  • Enable comparisons to be made between devices.
  • Provide a system of ESD sensitivity classification to assist in the ESD design and monitoring requirements of the manufacturing and assembly environments.
  • Have documented test procedures to ensure reliable and repeatable results.” [Source]

References:

  • ESD Association, Inc.: Device Sensitivity and Testing
  • ANSI/ESDA-JEDEC JS-001-2010: Electrostatic Discharge Sensitivity Testing — Human Body Model
  • ESD STM5.2-2009: Electrostatic Discharge Sensitivity Testing — Machine Model
  • ESD STM5.3.1-2009: Electrostatic Discharge Sensitivity Testing — Charged Device Model
  • ANSI/ESDA/JEDEC JS-002-2014: Joint Standard for Electrostatic Device Sensitivity Testing – Charged Device Model (CDM) – Device Level
  • IEC 60749-26: Semiconductor devices – Mechanical and climatic test methods – Part 26: Electrostatic discharge (ESD) sensitivity testing – Human body model (HBM)
  • IEC 60749-27: Semiconductor devices – Mechanical and climatic test methods – Part 27: Electrostatic discharge (ESD) sensitivity testing – Machine model (MM)

5 common mistakes in ESD Control & how to avoid them

Many companies implement an ESD Control Programme with the aim of improving their operations. Effective ESD control can be a key to improving:

  • Productivity,
  • Quality and
  • Customer satisfaction.

However, problems arise when an organisation invests in ESD protective products and/or equipment and then misuses them. Not only do these companies waste a lot of money but they could also be causing more harm than good. So, with today’s blog post we want to highlight some of the major issues we have come across and how you can avoid or fix them.

Introduction

Remember that for a successful ESD Control Programme, ESD protection is required throughout the manufacturing process: from goods-in to assembly all the way through to inspection. Anybody who handles electrical or electronic parts, assemblies or equipment that are susceptible to damage by electrostatic discharges should take necessary precautions.

Think of viruses or bacteria that can infect the human body. Just like ESD, they are invisible. Yet, in hospitals the defence against this hidden threat is controlled by extensive contamination control procedures including sterilisation. The same applies to ESD Control: you should never handle, assemble or repair electronic assemblies without taking adequate protective measures against ESD.

Treat ESD like Viruses & Bacteria
Treat ESD like you would Viruses and Bacteria

For an ESD Control Programme to be successful, there is discipline required; basic ESD Control principles should be followed:

  • Ground conductors.
  • Remove, convert or neutralise insulators with ionisers.
  • Shield ESD sensitive items when stored or transported outside the EPA.

For more information on how to get your ESD Control Programme off the ground (no pun intended) and create an EPA, check this post.

Common Mistakes in ESD Control

1. Poorly maintained or out-of-balance Ionisers

If an ioniser is out of balance, instead of neutralising charges, it will produce primarily positive or negative ions. This results in placing an electrostatic charge on items that are not grounded. These could then discharge to nearby sensitive items potentially cause ESD damage.

 Check Remember to clean emitter pins and filters using appropriate tools. Create a regular maintenance schedule which will extend the lifespan of your ionisers tremendously.
Consider using ionisers with “Clean Me” and//or “Balance” alarms. These will alert you when cleaning is required.
 Standard All ionization devices will require periodic maintenance for proper operation. Maintenance intervals for ionizers vary widely depending on the type of ionization equipment and use environment. Critical clean room use will generally require more frequent attention. It is important to set up a routine schedule for ionizer service.”
[CLC TR 61340-5-2 User guide Ionization clause 4.7.6.7 Maintenance and cleaning]

This post covers in detail how ionisers work and what type of ioniser will work best for your application.

2. Ungrounded ESD Garments

We’ve seen it so many times: operators wearing an ESD coat (without appropriate wrist straps and/or footwear/flooring) thinking they are properly grounded. Well, here is some news for you: you are not!

 Check Every ESD garment needs to be electrically bonded to the grounding system of the wearer. Otherwise it just acts as a floating conductor. There are a few options to choose from:
·         Wrist Straps
·         ESD footwear/flooring
·         Hip-to-Cuff grounding
 Standard The ESD risk provided by everyday clothing cannot be easily assessed. The current general view of experts is that the main source of ESD risk may occur where ESDS [ESD sensitive items] can reach high induced voltage due to external fields from the clothing, and subsequently experience a field induced CDM [Charged Device Model] type discharge. So ESD control garments may be of particular benefit where larger ESDS having low CDM withstand voltage are handled, and operators habitually wear everyday clothing that could generate electrostatic high fields.
[CLC TR 61340-5-2 User guide Garments clause 4.7.7.1 Introductory remarks]

Another thing to remember with ESD clothing is that they do lose their ESD properties over time. So make sure you incorporate periodic checks (see #3 below).

If you need more information on ESD coats, we recommend having a look at this post.

3. Not Checking ESD Control Products

A lot of companies waste thousands of pounds by buying and installing ESD Control products but then never check them resulting in ESD equipment that is out of specification. They haven’t got the tools in place to check their ESD items and have no idea if they are actually working correctly. Remember, ESD products (just like any other) are subject to wear and tear, workstations get moved, ground cords get disconnected…. The list goes on.

 Check When investing in ESD Control Products, make sure you also establish a Compliance Verification Plan. This ensures that:
·         ESD equipment is checked periodically and
·         Necessary test equipment is available.
 Standard A compliance verification plan shall be established to ensure the organization’s fulfilment of the requirements of the plan. Process monitoring (measurements) shall be conducted in accordance with a compliance verification plan that identifies the technical requirements to be verified, the measurement limits and the frequency at which those verifications shall occur. The compliance verification plan shall document the test methods used for process monitoring and measurements. If the organization uses different test methods to replace those of this standard, the organization shall be able to show that the results achieved correlate with the referenced standards. Where test methods are devised for testing items not covered in this standard, these shall be adequately documented including corresponding test limits. Compliance verification records shall be established and maintained to provide evidence of conformity to the technical requirements.
The test equipment selected shall be capable of making the measurements defined in the compliance verification plan.

[EN 61340-5-1 clause 5.2.4 Compliance verification plan]

For detailed instructions on how to create a Compliance Verification Plan, have a read through this post.

4. Re-Using Shielding Bags with Holes or Scratches

ESD Shielding Bags are used to store and transport ESD sensitive items. When used properly, they create a Faraday Cage effect which causes charges to be conducted around the outside surface. Since similar charges repel, charges will rest on the exterior and ESD sensitive items on the inside will be ‘safe’. However, if the shielding layer of an ESD Shielding Bag is damaged, ESD sensitive items on the inside will not be protected anymore.

 Check Re-using shielding bags is acceptable as long as there is no damage to the shielding layer. Shielding bags with holes, tears or excessive wrinkles should be discarded.
Use a system of labels to identify when the bag has gone through five (5) handling cycles. When there are five broken labels, the bag is discarded.
 Standard ESD shielding packaging is to be used particularly when transporting or storing ESD sensitive items outside an ESD Protected Area. Per Packaging Standard EN 61340-5-3 clause 5.3 Outside an EPA “Transportation of sensitive products outside of an EPA shall require packaging that provides both:
– dissipative or conductive materials for intimate contact;
– a structure that provides electrostatic discharge shielding.

This post provides further “dos and don’ts” when using ESD Shielding Bags.

5. Using Household Cleaners on ESD Matting

A lot of people use standard household cleaners on their ESD matting not realising how damaging this is to their ESD Programme. Many household cleaners contain silicone which creates that lovely shine you get when wiping surfaces in your home. The problem is that on an ESD working surface mat, that same silicone creates an insulative layer which reduces the grounding performance of the mat.

 Check Don’t spend all this extra money on ESD matting and then coat it with an insulative layer by using household cleaners. There are many specially formulated ESD surface and mat cleaners available on the market. Only clean your ESD working surfaces using those cleaners.
 Standard “Periodic cleaning, following the manufacturer’s recommendations, is required to maintain proper electrical function of all work surfaces. Ensure that the cleaning products used to not leave an electrically insulative residue which is common with some household cleaners that contain silicone.”
[CLC TR 61340-5-2 User guide Work surfaces clause 4.7.1.5 Maintenance]

This post covers everything you need to know about ESD protective working surfaces.

Now – the above list is by no means complete. There are many more issues we see when setting foot into EPAs but we think it’s true to say that these issues are some of the ones we encounter more often.

What issues have you come across before? Leave us a comment below.

The link between ESD and EMI

Introduction to ESD and EMI

Current Electrostatic Discharge (ESD) control practices have substantially minimised the dangers from unwanted electrical overstresses that are known to haunt semiconductors and other micro-electrical devices at all stages of their manufacturing, handling and applications.
The act of grounding an ungrounded ESD Sensitive (ESDS) device can trigger an ESD event, yielding latent or catastrophic damage by means of an energy or voltage failure mechanism in the ESDS device. To minimise this potential problem, the rate of discharge must be controlled during grounding and the work potential at the grounding electrode must be increased [1]. Decreasing the rate of discharge will limit the current density of a potential electrical arc (ESD event). Any combination of an increase in resistance or capacitance in the contacting electrodes (the two materials that sustain a discharge) can decrease the rate of discharge and lessen the effects of an ESD Event.
One of the side effects from an electrostatic discharge (ESD) is an induced EMI (Electromagnetic Interference). An ESD-induced EMI in the near-vicinity of mission-critical equipment can cause data errors, temporary resets or even power-up resets requiring operator intervention [2]. This is caused by the EMI undergoing conversion to a voltage or current, which in turn corrupts the operation of the circuit/logic inputs.

The effects from undesired electromagnetic radiation, EMI, on ungrounded or unshielded conductors is commonly underestimated. An ESD event occurring outside an ESDS protective work area can still pose a risk to unshielded and ungrounded conductors within the critical work or ESDS area.

Case Examples of ESD/EMI Problems

Some examples of ESD/EMI problems reported from the Center for Devices and Radiological Health (CDRH) databases are listed by product recall numbers. Recall numbers M485337, M485338, M562311 (March 1994) state that static from bed sheets when a nurse was making the bed caused infusion pumps to sound a “processor lock-up alarm”. Recall number M249358 (October 1991) states that a discharge from an operator to the timer of a radiation therapy system caused the timer’s display to blank just as treatment began. Recall numbers Z3112, Z3212, Z3132, Z142 (January 1992) state that ESD affected infant radiant warmers, causing the heater to turn on or off, the alarm not to activate, and the display to become blank or corrupted, [9].
Today’s TTL and CMOS logic states have a logic “0” at 0.8 Volts or lower and a logic “1” at 2.0 Volts or higher. This leads to a smaller indeterminate range of 1.2 Volts for most TTL and some CMOS logic circuits and places the logic inputs from these circuit traces or cable connections susceptible to induced EMI voltages exceeding this range. One example of an ESD-induced EMI was characterised from office chairs [3, 4]. Induced voltages over 2 Volts have been measured on a printed circuit assembly (PCA) 90 cm from the furniture ESD [4]. 2 Volts is enough voltage to easily drive a TTL circuit let alone an ECL circuit into a logic error.

Table I lists some logic devices and their potential susceptibility to EM energies. Noise margin is a quantitative measure of a device’s noise immunity. The high-level DC noise margin in Table I is the difference between the minimum device output levels for a logic high VOH of the driving gate and the minimum input level VIH required by the driven gate to recognise a “1” logic state. The Indeterminate Range is the difference between the logic inputs’ low level maximum and high level minimum to differentiate between a logic “0” or “1”.
Some types of common lab stools and office chairs can radiate a series of impulsive fields from metal legs due to internal ESD when a person rises from the chair. As many as 12 pulses have been recorded within a 10 second period after a person rises from a chair [3].
Smith stated that a value of tens of millivolts per inch (~2 V/m) is generally not enough to affect digital logic whereas values over one volt/inch (40 V/m) are potential problems. One example observed induced voltages of over 4 volts/inch (>160 V/m) in cables one foot from one type of office chair [3].

Table1Table I – Table of Logic Families’ Power Transition, Noise Margin, and Indeterminate Range [8]

What often looks like software errors in process equipment may actually be caused by an external static charge (or discharge) problem. An ESD event anywhere in a room can cause an EMI. That EMI can couple into a system through cables or open chassis and induce a noise voltage greater than the logic inputs’ indeterminate range and cause a single event upset. EMI effects to microprocessors or other circuit logic latch-ups in process equipment can manifest itself in several ways, such as random hangs, robotic malfunction, or software errors, all resulting in downtime and reduced throughput.

Theoretical Energy Analysis

1. Mechanism of an ESD Event
There are three well-known methods to simulate an ESD Event: the human body model, the machine model and the charge device model. Each has its place to aid in designing the proper ESD Control Programme depending on the application.
Induced voltage from an EMI energy transfer to a logic input trace with typical area of 40mm2 could be as high as 485mV with an ESD-induced 100 MV/m field at 33cm as depicted in Table II. 485mV is enough voltage to flip a logic state of an ECL device as depicted in Table I. From the same ESD event, a data input cable with a receiving area of 40 cm2 can have an EMI-induced voltage of 4.85 which is enough voltage to drive a logic error in any family or subfamily of logic circuits; TTL, CMOS, & ECL.

Table2Table II – EMI Energy Transfer from an ESD to an Isolated Conductor using antenna theory where: the area of the conductor is A=variable, the distance from the source is R=1/3 meter, ESD has a 1 ns rise time and a 3 ns pulse width.

2. ESD Event
An ESD event can have a fast rise time, especially for low voltage discharges [5]. The waveform for an ESD event includes high-frequency components with a frequency range from DC to over 6GHz, [4]. This electromagnetic radiation (EM) can readily couple to circuit traces (conductors acting as antennae). For ungrounded conductors coupled within a capacitive circuit, this EM wave can induce a static charge, building until a discharge, breakdown, recombination or neutralisation occurs. High-speed circuits, by their nature, tend to be very susceptible to high-frequency signals such as those from a nearby ESD event.
The electrostatic field strength (Eo) just before an ESD is proportional to the charged voltage (V) at gap width δ. The gap width, δ, is defined by Paschen’s Law, but may vary in each discharge condition. The electric field strength Eo = V/δ where V is from 0.5kV to 30kV and δ is from 5μm to10mm, can yield an electric field strength as high as 6 GV/m. This extremely high field strength is attributed to a smaller gap width, δ = 5μm. It is important to note that the arc length of an ESD is of greater influence to its disturbance than its voltage [7].

3. EMI
An Electromagnetic Interference (EMI) is an unwanted electromagnetic energy, (whether intentionally or unintentionally generated), of almost any frequency and energy level. EMI is defined to exist when undesirable voltages or currents are present to adversely influence the performance of an electronic circuit or system. Sources of radiated electromagnetic energy from ESD are very common in today’s factories from furniture ESD, raised flooring ESD, Human Body ESD, hand held toolbox ESD and metal-to-metal ESD [3, 4, 6, 7]. An EMI, or summation of EMIs, can over time induce a charge (static voltage) on an ungrounded conductor coupled in a capacitive circuit, i.e., an isolated capacitor. An even more common occurrence is a single ESD induced EMI that can upset a logic circuit and cause systems errors. The very fast rise time of an ESD may be preserved if it flows through a metal conductor, resulting in radiated EMI.

Solutions

  • Assume that all electronic devices are susceptible to damage or logic error states from ESD and EMI, respectively; and take the proper precautions.
  • Proper grounding of isolated conductors and use of ground-planes near active conductors will minimise some of these effects.
  • Shielding the known emitting devices will help, but it is the unknown emitters that will cause the most problems. Thus, shielding the receptors, sensitive logic devices, will help combat EMI-induced logic errors. Start shielding at the device level, for it is less costly than at the system level.
  • Reduce ground-loop areas between interconnected equipment and systems. Route interconnected cables inside conduit, cable trays or raceways when possible. Do not coil excess cable into a helix, but rather fold back and forth to foil antenna gains.
  • Metal-to-metal discharges will always derive the largest current derivatives (di/dt) and hence generate the strongest EMI fields. Treat isolated conductors as charged devices and ground them with an electrically dissipative material (R > 104 Ohms). This will slow down the energy transfer from the conducted ESD causing the resultant EMI to be negligible to any active near or far field system.

Conclusion

A high energy ESD can drive a substantial EMI energy to couple and charge passive circuits or energise active circuits with significant system problems. EMC practices involving shielding designs typically account for EMI from known sources, but should also consider unplanned sources such as ESD events in the near vicinity of the active or sensitive system(s).
With today’s logic devices having smaller noise margins and indeterminate ranges, susceptibility to ESD-induced EMI should be accounted for in the design and implementation of the systems incorporating logic circuits.

White-Space

References:
R. C. Allen, “Controlling Workstation Discharge Times”, Evaluating Engineering, Jan. 1998, pp. 88-92.
G. Chase, “EMI from ESD – An Insidious Alliance”, NARTE News, Vol. 14, No 1, 1996, p 22.
Smith, “A New Type of Furniture ESD and Its Implications”, EOS/ESD Symposium Proceedings, EOS-15, 1993.
Y. Tonoya, K. Watanabe, and M. Honda, “Impulsive ESD Noise Occurred from an Office Chair”, EOS/ESD Symposium Proceedings, EOS-15, 1993.
S. Podgroski, J. Dunn, & R. Yeo, “Study of Picosecond Rise Time in Human-Generated ESD”, Proc. IEEE Int. Symp. Electromagnetic Compatibility, Cherry Hill, NJ, Aug. 12-16, 1991, pp. 263-264.
Y. Tonoya, K. Watanabe, and M. Honda, “Impulsive EMI Effects from ESD on Raised Floor”, EOS/ESD Proceedings, EOS-16, 1994.
D. Pommerenke, “Transient Fields of ESD”, ESO/ESD Symposium, pp. 150-159, 1994.
R. C. Dorf, The Electrical Engineering Handbook, 2nd Edition, CRC Press, pp. 1773-1777, 1997
J. Silberberg, “What Can/Should We Learn from Reports of Medical Device Electromagnetic Interference?”, FDA, Rockville, EMBC95 paper 10.2.1.3, 1995

%d bloggers like this: