My troubled experiences with isolation amplifiers and external noise
During my several decades at Kikusui working with custom products, I have gained experience with a range of different products, but I have also had my fair share of setbacks. My tendency to lecture the younger members of my team is the result of lessons I learned the hard way when I was younger. In fact, my experience with overcoming obstacles is an important part of who I am today.
While we tend to take a negative view of obstacles, the experience of overcoming challenges is a valuable one (albeit stressful at the time). Malfunctions arising from negligence or carelessness really should not happen in the first place, but what about those times when you thought you did everything right, but something still goes wrong? I like to think of these experiences as “blessings in disguise”. (Also, if I don’t, the alternative is too depressing!)
On that note I would like to share a story of such a “blessing” early in my career.
My first major assignment.
This story is about the first time I was assigned to install a large-scale system (a power supply system with over a dozen channels) on a client site. Rather than merely distributing a single power supply over multiple channels, the system comprised over a dozen DC, bipolar, and polarity switching*1 power supplies with a range of different voltage and power specifications, which could be drawn on both discretely and in combination in various permutations. In total, the Client’s system comprised around 100 channels, with a portion of those being supplied by Kikusui.
The client’s principal requirements were that the system supply constant current when used in constant current mode, and that the power, control and monitoring modules be isolated from the output.
I was charged with handling the overall design of the control and interface module and formulating the overall specifications of the system. I allocated the design work duties amongst the other members of my team. On the basis of my prior experience, I assumed that all we needed to do to meet the specification was combine a constant voltage reference with an error amplifier and a current sensor. The components were procured, and the system assembled and wired up. Just as we were getting ready to “fire it up”, however, disaster struck.
*1 A power supply whose output polarity can be switched using a relay or similar device.
A careless choice of amplifier.
I was contacted by a colleague on site, who said that the system was not meeting specification in terms of its ability to supply constant current. Upon measuring the voltages across various parts of the system, we detected fluctuation of the voltage reference, which was supposed to be constant. The cause of the problem was the isolation amplifier used in the interface module located between the client-side control unit and the power supply: having had no prior experience with isolating amplifiers, I had simply selected a model I had seen used by other more experienced members of my team, assuming it would work in the same way as an operational amplifier. However, when I looked at the specifications of the isolating amplifier later, I realized that it had a larger offset and slightly poorer thermal characteristics than its op-amp counterpart. What is more, because, in an effort to reduce the number of amplifiers in the system I had configured the isolating amplifier to produce gain, the offset gain of that amplifier was also increased, thereby further worsening current stability. Simply assuming that the device my colleague used would be fine, I had not bothered to check its specifications. I had made a real beginner’s mistake.
Without delay, I said about fixing the problem, which I did by (1) replacing the original isolating amplifier with a more stable one, (2) reducing the gain of the isolating amplifier to 100% and (3) moving the operational amp used to create gain upstream of the isolating amplifier (thereby boosting the input/output voltage of the isolating amplifier and in turn reducing offset effects).
With the support of my team, I was able to modify all of the several dozen channels affected just in time to deliver the system on schedule. I had really learned my lesson!
My next “blessing” involved unexplained “hunting”.
When my team works on a major installation, we need to prepare the worksite, check that the system works properly, and have it inspected by the client. In the case of the system I will write about next, several days’ worth of work was split into a number of sessions in which we delivered the equipment to the site, checked the operation of the system, and had it inspected by the client. They say that work is 80% preparation: having made all the necessary preparations to ensure we would achieve project specification and test requirements, we completed the job with no significant problems.
However, just as I was congratulating myself of a job well done, I received yet another “blessing in disguise”. Several months after the system had been delivered, the client contacted me complaining of an inconsistent output current and a noisy power supply.
I hurried to the site to verify the symptoms and attempt to determine the cause. The power supply in question was a series-regulated DC power supply, and it was exhibiting symptoms of “hunting”*2. “Hunting” was observed irrespective of output current, but was not readily reproducible, presenting differently each time.
After adjusting the phase-controlled error amplifier and observing the waveforms at various parts of the circuit, we concluded that the “hunting” was the result of noise on the AC input line. The client said that it could not find any source of such a noise anywhere in its own system and asked me to fix the problem as soon as possible as only the Kikusui power supply was faulty.
*2 A series-regulated direct current power supply is comprised of a phase-regulated pre-regulator in series with a resistor. Hunting is said to occurs when the operation of the phase regulated pre-regulator becomes unstable (oscillates). Intermittent oscillation results in noise from chokes and transformers. (Also see Knowledge Plaza: Basic principles of power supplies on the Kikusui website).
The culprit was yet to be found.
After discussing the matter with my supervisor, I tried adding a filter to the AC input line and even changing the number of filters but was unable to resolve the problem. (At the time, while filters were for low-power applications were commercially available, high-power filters were not, so I had to build these myself). Despite further, repeated attempts to fix the problem, each time with different components, after three visits I was no closer to a solution, and a full six months after the system had been delivered, I was still in the dark about what to do. Sometimes “blessings in disguise” are punishing.
In the end, the mystery solved itself. What preparing for yet another visit, I received a call from the client who told me that that the problem had been rectified. It turned out that the source of the noise was a high-capacity power supply from another manufacturer that had been connected to the client’s system to drive a pulse generator. The client discovered the cause of the trouble after noticing that a control device on the client’s side was also malfunctioning: the malfunctioning device shared AC input and earth lines with the power supply that had been causing the “hunting”. The extensive branching of the earth line also appeared to be partly responsible for the problem. The client responded by arranging an independent AC supply and earth line for the power supply at fault before reconnected the device. With the faulty earth fixed, the “hunting” stopped.
Because this problem was ultimately the fault of a third-party manufacturer, the months I spent working on a solution were in vain. However, it was a good lesson in the importance of earth line and common line design when designing devices.
While Santa Claus doesn’t deliver gifts to grown-ups, the same does not apply to these “blessings in disguise”. I say that because I receive one every year…
I really don’t want any more of these “blessings”. My inbox is full of them!
