The Fascinating World of Bespoke Engineering
Do you like the "grey zone"?
I am a bespoke engineer with Kikusui’s System Technology Group. While earlier in my career I was involved with our standard (off-the-shelf) range, I have now actually spent longer working with bespoke devices. In this series, I discuss various aspects of bespoke engineering.
What do you think when you hear the term "bespoke"?
The electronic equivalent of avant-garde cuisine
Off-the-shelf devices are highly refined in their design: these devices are the product of repeated prototyping and design tests (in which the necessary tests are performed to obtain the necessary data), requiring a significant outlay of time and labor. A given level of performance is guaranteed, and quality remains uniform, no matter how many iterations roll off the production line. When building a device to order, on the other hand, you start off with the technical design for a standard model, and mix in the client’s requests (the specifications) and your own bespoke engineering expertise. It’s much like a one-off culinary creation─of avant-garde cuisine at that.
Avant-garde cuisine is all well and good provided your guests like it, but it has its drawbacks─namely a lack of reproducibility. While there might be a basic design behind the idea for a dish, because you can’t guarantee that the prevailing conditions (materials, equipment, budget and production time) will be the same, reproducing it is actually very challenging. (Obviously, repeat orders are nice to receive.) However, I believe that the ability to reproduce the "taste" of the original is a mark of an engineer’s skill, and what makes bespoke engineering so interesting.
This story is from early in my career, when I had yet to get the knack of creating the same "taste" twice. The device in question was a fixed-voltage bipolar power supply, a bipolar supply being one that is able to not only source, but so also sink, current. As you would expect, I got together with the client to discuss specifications, including the required current and voltage ranges, frequency response, other characteristics including ripple and dynamic response, protective features, and controls.
Once the design was ready, I obtained the necessary components, and assembled and wired up the device. I then cautiously turned it on and proceeded to test its performance. I tested a range of performance parameters ranging from performance under a given load (referred to as "static characteristics" because the load is constant), to performance under variable loads (referred to as "dynamic characteristics" and "transient characteristics"). It goes without saying that I sent the power supply off to the client confident that it would work. Indeed, it was delivered to the client and operated without issue.
Several years later, the client, who was happy with their first power supply, placed a repeat order. As the first power supply had performed without issue, I procured the same components, and performed the same assembly, adjustment, and testing when building the second one. With the test results all normal, I delivered the device to the client.
Mark 2 malfunctions
A few months after the second power supply had been delivered, I received an enquiry from the client. The client said that a particular test subject that worked fine on the first power supply was causing the output voltage of the second power supply to drop. Based on the client’s description of the issue, I assumed that they were using the power supplies under conditions that caused the maximum current rating to be exceeded, thereby causing a voltage drop. The client was concerned that the second power supply was registering a voltage drop when the first one did not. This, despite the fact the second power supply had been checked and compared against the voltage and current waveforms observed when the first power supply was tested.
First, I had to work out what was happening. I visited the client with a range of testing equipment, and took waveform measurements of voltage and current. (I can’t reveal what the test subject was.) My findings were as expected: the client was using the power supply under conditions that exceeded its maximum current rating.
Notably, maximum current was only exceeded for a matter of milliseconds. As you have probably already guessed, the difference between the first power supply and the second was the response time of the overcurrent protectors, this being a characteristic that was not included in the specification (or at least was only roughly specified as "x milliseconds or less").
The "grey zone"
When designing the bipolar power supply, the default circuit constant of the overcurrent protector had not been changed. The design of the circuit was such that despite both power supplies using identical error amplifiers and other components, the properties of the circuits varied within the tolerances of the components used. For the purposes of clarity, I should note that it would have been possible to specify the time constant of the overcurrent protection used, although this would have meant imposing conditions on the circuit’s dynamic characteristics (on the duration and value of power variations).
A pedant might say that I had failed to verify all the specifications. In reality, however, it is extremely difficult (theoretically impossible, in fact) to include every single item in the specification for a device. Rather, the primary considerations when drawing up specifications are to "decide what to decide", and "articulate and quantify the items that both parties believe are necessary", recognizing that it you cannot cover everything. No matter how detailed your specifications are, something will always be left out, or fall into a "grey zone". (A "grey zone", these days referred to as "off-white", is a variable that is either not mandatory, or is difficult to include in the specifications). The existence of a set of specifications does not guarantee its completeness, something that a lot of people don’t understand.
If you take the wrong approach to properties and characteristics not included in the specifications, difference in assumptions between parties about what is "normal" or the status of items that are not explicitly specified can cause misunderstandings, and ultimately major problems. For this reason, you should make an effort to make the specifications as explicit as possible, although there will always be some items that are not explicitly included in the specifications or are not discussed. Now, later in my career and with more experience behind me, I have a better feeling, based on the list of specifications prescribed by the client (i.e. the description of the device being built), for what specifications need to be verified (i.e. the traps), so far fewer potential problems arise. I have not eliminated all problems, however.
Why bespoke engineering is fun
Happily, when I explained the cause of the above problem (the difference in circuit properties) to my client, they were understanding, and the issue was resolved by simply adjusting the properties of the second power supply so that they matched the first. I also asked the client to provide us with data on overcurrent characteristics when placing repeat orders.
These types of "grey zone" are part and parcel of bespoke engineering, so your suitability for bespoke projects comes down to how you regard these fuzzy areas. To use a baseball analogy, are you the kind of person who would field a ground ball that lands in the "hole" between the shortstop and third base? If you’re the kind of person who says, "that’s not my job", you will find bespoke engineering very stressful. For me, picking up those ground balls and working out how I am going to deal with them is about applying my skills, and the fun part of bespoke projects. In any case, real life is full of such "ground balls"─for instance, TV detectives often say they have to go back to the scene of a crime 100 times to look for clues. The same is true of bespoke engineering. If you’re the kind of person who likes "grey zones" and getting your hands dirty, you will be a good match for bespoke engineering.
In conclusion, if you get flustered whenever you encounter an issue that has "grey" cause, it means you are still a newbie. Conversely, if your eyes light up when you encounter one of these thorny problems, it is a mark of experience. Myself? I get told that my eyes sometimes look like those of a dead fish, but enough about me. Onward and upward!
Solutions Development Division
[Major achievements in product development]
PAL, PAK-T series regulated DC power supplies
Custom built regulated AC power supplies; made to order charge/discharge power supplies
Custom built electronic load devices; custom built bipolar power supplies, etc.