Hector Garcia, Power Specialist, Avnet Abacus
It is often assumed that in any product design, if you can use a standard ‘off-the-shelf’ power supply it will be the lowest cost option compared with a full custom or configurable part. It’s true that with today’s fast-paced product launches and upgrades, if a custom part is commissioned, the Non-Recurring Engineering (NRE) costs may not be recoverable, risks are high and the time for development could be way too long. Even in military development programs, which can be measured in decades, Commercial off-the-shelf (COTS) parts are often preferred. Configurable parts should not be dismissed though.
Power supply manufacturers do concentrate their marketing on particular standard models and try to specify them to capture the largest share of applications. When a manufacturer can sell most of its product as a single configurable style though, the cost on the market can be surprisingly close to a ‘standard’.
An argument often put forward is that a ‘standard’ part is shipped to many customers and therefore manufacturers benefit from economy of scale. If you believe that the savings are always passed on to customers then this is surely a good thing. But there’s a fallacy here: each power supply manufacturer needs to have hundreds if not thousands of ‘standards’ to meet market demands. Just look at the power product listings boasting a multitude of variants available from typical suppliers. For them, this means huge approved vendor lists and mountains of documentation. If they could sell fewer types that fit more applications, real savings could be made and passed on to the end user. This is where ‘configurable’ products can score for both manufacturers and their customers.
Does the power supply have to fit in the ‘space that’s left’ anymore?
With energy and cost saving constraints, system designers are now acutely aware that their power system cannot be an ‘afterthought’ as in the bad old days. This has the unfortunate effect though that the power supply specification must be set early on in development with guesses at power rating, size, voltage rails and even safety compliance levels.
Then there’s the monitoring and control features to anticipate. The result can be that product designers have to work around the pre-determined power specification, chosen from standard offerings, severely limiting the scope of their designs.
No responsible designer would compromise too far, so there is the very real risk that the power supply and its specification would need to change mid-program with consequential disruption to mechanical arrangements. Even if the designer works around his power specification constraints, could marketing change their requirements during development? Surely not. Have circuit designers never been asked to ‘just add an extra interface’ or ‘just add in a medical version’? That decision to fix the power supply as a standard gets thrown out of the window and the search starts again with all the costs and uncertainties involved.
The alternative solution is to specify a configurable or programmable power supply from the outset. These typically are a base unit with mains filtering and a power factor correction stage followed by plug-in modules for individual output isolation and regulation. Now, any specification changes can be accommodated with, at worst, a module swap and sometimes no more than an adjustment of a potentiometer to a different voltage. Many configurable supplies for example can adjust a single output from 5V to 3.3V with no effect on other outputs.
Module power ratings can be chosen up to a maximum for the base unit and outputs can be put in parallel or series to increase current and voltage levels respectively. Usually effort is made by the manufacturer to make the configurable supply meet worst case specifications. As such you can expect them to meet the most stringent EMC performance and medical safety specifications, even patient-connect type (2 x MOPP).
You might say that the argument for specifying a configurable supply at some extra unit cost relies on the assumption that a standard power supply choice will always change during program development with its associated disruption, necessitating choice of another standard part. No one wants to anticipate failure so the argument would not be a popular one, especially with the financial guys. However, that’s not the whole story.
There are positive advantages to going the route of a configurable supply from the start. As long as the right footprint is allocated in the end product, power supply selection can be left to the later stages of product development when actual voltages, power levels and signalling requirements are known. The product designer is then freed up to optimise their part of the design without power constraints. And more, the inherent flexibility can be used to tailor the end product for its application. Those marketing guys might also want a bare-bones version – just remove or downsize power modules. They might want a fully featured option – just add or upgrade power modules. Time to market for variants is minimised by having that configurability.
Configurable power supply designers know that for their product to be attractive, flexibility is the key so they will often feature all-isolated outputs, for example.
Typical standard power supplies will have a common ground for multiple outputs constraining grounding systems from being optimum for EMC and functionality. Standard multioutput supplies will also often require a minimum load on a specified ‘main’ output, below which other outputs fall out of specification. Configurable supplies, with their individually regulated outputs don’t suffer from this restriction. The typical way individual outputs are controlled is shown in Figure 1, an example from the Excelsys CoolX range.
Programmability gives flexibility and ‘future proofing’
Perhaps the biggest potential advantage of configurable and programmable supplies is that because of their modular nature with individual output regulation, they will often incorporate comprehensive signalling and control. Recent programmable power supply products incorporate multiple microcontrollers for full digital loop control. Opening up the internal control bus to users via a Graphic User Interface (GUI) gives the possibility of adding factory and field programmability and extra functionality to the end product. Figure 2 shows an example of the monitoring features of the free software supplied for the Artesyn iHP configurable product.
Click image to enlarge
Figure 2: Artesyn iHP series monitoring software GUI
Now, power supplies can be set in the factory to a particular configuration for an end-product variant, field service is facilitated by hooking up to portable diagnostics and end users can fine-tune power supply performance to match their load characteristics.
Using the manufacturer-supplied GUI, some programmable power supplies can even be dynamically and automatically adjusted – for example in battery charging where voltage might need to change depending on cell temperature and mode change from constant current to constant voltage might be required.
Your choice of configuration and suppliers
So, what are the choices on the market today? Leaders in configurable supplies are Artesyn, Excelsys and MEAN WELL. Artesyn has two ranges, µMP configurable and iMP/iVS, fully configurable and programmable. All are compliant to EN60601-1 medical specifications, the µMP can have up to 12 outputs while the iMP can have 21 and the high-power iVS 24. A free downloadable GUI is available for the µMP and iMP and the iVS can be controlled by standard PMBus commands over an i2C interface. The high power iHP series can be expanded with modules up to 24kW in 3kW increments with up to eight outputs. The system holds industrial safety approvals with additional compliance to SEMI F47 for semiconductor processing equipment.
Excelsys has gone down the route of convection cooling, with leading efficiency designs that still give high power in small form factors at high temperatures. Its CX06M, for example, provides up to eight outputs at 600W combined up to 40⁰C with derating to 85⁰C. The package size is 1U x 4.5 x 8 inches (1U x 114.3mm x 203.2mm). The parts are particularly suitable for medical applications where the lack of fan noise is often appreciated.
MEAN WELL has recently introduced its NMP series of configurable 2 x MOPP medical grade supplies at 650 and 1200W. The fan cooled NMP650 is just 1U x 6.3 x 3.5 inches (1U x 160mm x 88.9mm) and the NMP1K2 1U x 6.3 x 5 inches (1U x 160mm x 127mm). Like the Excelsys offering they include a 5-year warranty.
All the products offer monitoring and programmability to some degree, typically able to report on voltage levels, output currents and temperature by PMBus signals. Fault threshold monitoring can be set and protection is comprehensive. Even mode of operation can be set in some models, changing from constant voltage to constant current with foldback as required. Figure 3 shows some examples.
Click image to enlarge
Figure 3: Example configurable power supplies from Artesyn, Excelsys, and MEAN WELL
Flexibility from stock
With the extreme flexibility of configurable supplies available at keen prices, they can realistically compete with the rigidly specified ‘standard’ parts and enable system designers to pass on that flexibility to end users. Specifications can be tailored for their exact application along with ‘future-proofing’. There is also the possibility of easy upgrades and output specification changes through exchangeable modules and remote programming through friendly GUIs. Programmable power supplies from Artesyn, Excelsys and MEAN WELL are available through Avnet Abacus from stock.