Learning from the tellyPaul Gregg checks out research to get more performance out of printed circuit boards Increasing system clock speeds mean analogue and digital circuit designers must look more closely at the conductor trace length on PCBs. With the pressure on bandwidth, there is a move to try and squeeze more performance out of the PCB laminate.
In computing, telecommunications, military and aerospace sectors, designers working on digital circuits operating above 50MHz and with trace lengths above 6in. for digital, are specifying controlled impedance boards. For analogue signals above 300MHz and trace lengths of more than 3in., the same is happening.
According to board test specialist Andy Burkhardt of Polar Instruments, controlled impedance testing is used in the production of some 90 per cent of all high performance multilayer PCBs, as distinct from consumer PCBs. Market estimates suggest that controlled impedance PCBs could account for 70 per cent by volume of all PCB manufacturing within a few years.
The easiest example of controlled impedance we are most familiar with is the coaxial cable that connects the aerial to the television. This cable consists of an inner conductor separated by an insulator from the outer conductor (the shield) where the dimensions of the cable and insulator are carefully controlled to ensure the cable has the desired impedance. The spacing between the inner conductor and the shield, and the material used as the insulator between the two conductors, determines the high frequency impedance of the cable.
Controlled impedance PCBs are effectively simulated coaxial cable, where the coaxial shield is represented by an earth plane, the insulator is the laminate and the coax inner conductor is the conducting trace.
The function of a cable or a trace on a PCB is to transfer power from one device to another. Theory shows that maximum power is transferred when the impedances match.
The TV aerial has a natural characteristic impedance and the coaxial cable impedance is designed to match the aerial's impedance.
The input impedance of the TV is also designed to match the coaxial cable's and aerial's impedance. Hence there is a matched impedance system (aerial to coaxial cable to the TV).
The situation is the same on PCBs operating at high frequencies. Designers need to control the impedances of some critical high frequency traces to ensure that one circuit couples the maximum energy into the other.
The longer the trace and the higher the operating frequency, the greater the need to control the impedance. The PCB maker controls the impedance by varying dimensions of the particular trace. The specific effects of having the wrong impedance depend on the particular circuit, but problems encountered can include low gain in an amplifier, excessive electrical noise, and random errors in digital systems due to reflections from an impedance mismatch. Once the board is loaded with components it becomes difficult to find these faults.
Components on boards have various tolerances, so one batch of components may tolerate an impedance mismatch more effectively than a different batch. This can mask an incorrect impedance trace.
In addition, a component's parameter may alter with temperature, so it can absorb the mismatch at room temperature but random faults will occur as it warms up. For this reason the board designer will specify a value and tolerance of impedance for specific traces, and will rely on the PCB manufacturer to conform to the specification.
Typical PCB impedances are specified between 40 and 120 Ohm and either consist of a trace sandwiched between two planes or a trace above one plane.
The trace and the plane(s) constitute the controlled impedance. Specialist board maker MEPD first became involved in the production of controlled impedance PCBs about three-and-a-half years ago due to work it was doing for customers in the military sector. Its research now forms the basis of a draft IEC specification.
MEPD's boards were required for high frequency use. While it was possible to produce boards with a specific characteristic impedance by controlling parameters such as trace width and fabrication material, there was no simple way to verify the results. Impedance tended to vary from batch to batch.
MEPD was commissioned by the Defence Research Agency - the main research and technology adviser to the Ministry of Defence - to define a set of design and measurement standards for the electronics industry. During this work the company collaborated with board test company Polar Instruments.
It soon became clear that while time domain reflectometry (TDR) was the best measurement technique for verifying the impedance of bare boards. Software development would be needed to mask the complexity of the control and data interpretation behind a user interface.
Close cooperation between the two companies' engineers let the task be done in less than a year, with MEPD acting as a beta site. As a result, Polar Instruments could introduce its CITS100 controlled impedance tester, followed by the upgraded CITS200.
The CITS200 uses Windows software to give a rapid pass/fail presentation of the controlled impedance. For a pass, the trace must fall between two vertically, hatched areas. TDR techniques have eased measurement of the reflection of fast rise-time pulses to give a graphical presentation based on the trace's characteristic impedance.