Crossing the divide
4 mins read
Mixed signal controllers pose organisational challenge for design engineering teams.
In the early days of electronics, a microprocessor contained no more than a processing core or arithmetic logic unit (ALU), some internal registers and a memory bus. All other elements of the design, including rom/eprom, ram, uarts and other digital and analogue peripherals, were connected to the microprocessor via the data bus or I/O bus (so called 'memory mapped' or 'I/O mapped' peripherals).
As design and manufacturing processes improved, memory and peripherals began to be integrated onto the same die, leading to the development of the microcontroller, which combined the core and ALU with memory and digital peripherals (counters, timers, uarts and other serial links). Simple analogue peripherals (a/d converters, d/a converters and comparators) followed soon after.
In effect, this integration process simplified product design to the point where circuit design for simple embedded systems can now be accomplished by little more than ticking boxes in a microcontroller selector guide to find the right memory sizes, performance and peripherals. The inclusion of simple analogue peripherals in a digital device has become accepted as the norm for microcontrollers, and the resolution, speed and accuracy of these peripherals has also improved with advances in manufacturing technologies.
This integration process has now gone a big step further, with the introduction by Cypress of the PSoC 3 mixed signal controller. PSoC 3 gives the design engineer access to operational amplifiers, programmable gain amplifiers and transimpedance amplifiers, as well as digital filters, multiplexers and even mixers on the same silicon as the processor.
While this is an apparently natural evolution towards even greater integration, it has raised organisational, management and technical problems that potential users must overcome.
A mixed signal controller requires the coordinated practice of many design disciplines – digital, analogue, software and programmable logic – across one, shared hardware platform. However, today's product development teams are normally demarcated rigidly along hardware and software lines – and often along analogue and digital lines as well. While this is a good fit for conventional microcontrollers, the approach will require some adjustment in order to make the best of a device such as PSoC 3.
So the decision to adopt an integrated mixed signal controller can lead to confusion or conflict about responsibilities. This can all too easily become politicised, as one side worries that the other will design them out of a job. Empire building, or empire protection, can damage a company's efforts to implement a new architecture, such as that of PSoC 3.
There may also be some opposition to this new technology from adherents to existing platforms who just prefer the status quo. And some might be concerned that they will be locked into a single source, non standard microcontroller family – although this concern can equally be applied to nearly all the best selling microcontrollers in use today.
In fact, organisations resist change unless strong leaders push it through, and the implementation of PSoC and mixed signal technologies in general requires active sponsorship at a management level. Management must create a single, united design team that coordinates the digital and analogue application development on a shared hardware resource.
This can put strain on the technical capabilities of the team. In a mixed signal environment, one engineer may be required to develop the complete system from external signals or sensor hardware, through analogue and digital manipulation, to the output as an analogue signal. This means they will need to map the system design on to a single device and to ensure that accuracy and resolution are maintained throughout.
Product managers should be alert to the need to pull in external technical assistance from third parties, such as Future Electronics, which can give design teams the benefit of tried and tested processes and techniques deployed at other customers.
Company managers must also be prepared to commit time to training the design engineers who are to use PSoC. There is a perception that the PSoC Creator design environment must be complex because it has so many functions to implement. In fact, it has a similar look and feel to many other integrated development environments. But, like any new tool, it needs to be learned, and a design engineer who is new to it will not be immediately productive.
A great deal of support is available free of charge from Cypress, including application notes, webinars and other materials. But there is no alternative to design engineers taking the time to immerse themselves in the new tool.
From a purely technical perspective, a design team can encounter tensions in the transition to a mixed signal controller platform. In particular, engineers who are used to designing with discrete analogue ics will find the analogue peripherals in PSoC 3 more difficult to debug because they are embedded in mixed signal silicon. When using a PSoC device, you cannot 'single step' through analogue circuitry. It is, however, possible to bring out internal signals to a pin to view them on an oscilloscope.
On the other hand, the quality of the analogue functions in PSoC 3 is comparable to that of widely used discrete analogue parts. It is true that earlier implementations of mixed signal processors did make some compromises in the analogue components, which were most noticeable in the amplifier and comparator specifications. PSoC 3 devices, however, offer up to four op amps that compare favourably with the industry standard LF356 op amp. Similarly, typical discrete designs specify an external LM4040 as a 0.1% reference. PSoC 3 includes an internal 0.1% reference (temperature range -40 to 85°C), saving board space and around $2 (in low volume) on the bill of materials.
Another common concern among analogue engineers is that the combination of analogue and digital processing on the same silicon will cause excessive noise and crosstalk. As the result of careful routing design, the estimated pin to pin crosstalk in PSoC 3 is estimated to be better than -80dB at 100kHz; the crosstalk between individual amplifiers will probably be too low to measure. The amplifier noise figure is estimated at 100nV/vHz at 10kHz.
Adopting PSoC 3 as a replacement for a conventional 8bit microcontroller is not a straightforward technical decision to be made with the help of datasheets and evaluation boards alone. A company must be prepared for some managerial and organisational disruption when adopting a programmable mixed signal controller.
But when a PSoC 3 adoption programme is planned and thought through properly, the advantages are considerable – higher integration, savings in the bill of materials, board space and component count, and greater flexibility to change product specifications throughout and beyond the initial design cycle.
Author profile:
David Clark is technical solutions manager, Future Electronics (UK)