Follow @kepco_careers
Advanced Search
Kepco Power Solutions


Power supplies exist to provide five separate and distinct functions:
  1. Change the output to a different voltage level than the input (step up or step down)
  2. Rectify a-c into d-c
  3. Isolate the output from the input
  4. Stabilize the output against the influence of:
    • Source changes (source effect)
    • Load changes (load effect)
    • Temperature changes (temperature effect)
    • The passage of time (time effect or drift)
  5. Provide the user with a means to vary the output
Not all power supplies provide all of these functions. Instrumentation power supplies are distinct from the power supplies used to provide fixed voltages to power devices in that they are optimized to provide the user with precise means for exercising control over the output. There are two fundamental control methods: ANALOG control and DIGITAL control. Kepco makes both analog-controlled instrumentation power supplies and models that are fully digital.

Kepco also makes digital interfaces for the analog-controlled models.


By analog control, we refer to a means for controlling a power supply's output by smoothly varying a signal. Common signals include variable resistances, variable voltage and variable current. When a variable resistance is chosen,the ratio of resistance to output voltage (ohms/volt) corresponds to a current. With few exceptions, Kepco has standardized on a control current of 1.0 mA which corresponds to a control ratio of 1000 ohms per volt. For control by a voltage signal, most Kepco analog power supplies may be characterized as three-terminal amplifiers in which the power supply's (+) error sensing terminal serves as the common between the input and the output. The control voltage must be positive 0 to +10V with respect to the (+) sense terminal. The output is negative as the power supply will function as an inverting amplifier.

Because it is likely that an available signal will not match the polarity and amplitude requirements, Kepco builds one or more "preamplifiers" or "uncommitted amplifiers" into many of its analog-programmable power supplies. These can serve to invert and scale the available control signal into the necessary 0 to +10V. Additionally, the uncommitted amplifiers can be used to do operational functions such as integration or summing. To characterize these extra operational amplifiers which are incorporated into Kepco's analog-programmable power supplies, the specifications contain information about the influence of source voltage changes, temperature and drift on the input offset voltage and offset current. The fixed part of the offsets are zeroable. The tabulated specifications are for the changes induced by the principal influence quantities.

The basic series-pass linear technique for stabilization offers the possibility of amplifier-like control techniques. Kepco calls this "Operational Programming" because the control arithmetic closely resembles the equations used to characterize analog operational amplifiers. Like op-amps, operationally programmable power supplies may be used for modeling and simulation of real-world situations. You can use them for amplifying, scaling, inverting, integrating and combining various input signals to produce powerful outputs that can drive motors, light lamps, run heaters, charge and discharge batteries and control machinery of all sorts.


In all power supplies, the design aspect that governs the rate at which the output can follow a fast changing input signal is the output capacitor. Specifically, it is the rate that the capacitor can be charged and discharged. This rate is proportional to the ratio of the output current setting (the current limit) to the value of the capacitor in microfarads. To enable users to calculate the speed, Kepco lists the value of this capacitance in the model table of its operationally programmable models. Those units that have removable capacitors or no capacitors at all (BOP) have the "equivalent capacitance" listed. This is the capacitive value computed by working backward from the observed response. It is not a physical capacitor.

A special technique that uses adjustable lag networks allows certain of our operationally programmable models to function without a conventional output capacitor. This, in turn, allows a relatively wide signal bandwidth for modulation and rapid output slewing. In high speed automatic test applications, this allows for rapid level shifts. Such high speed power supplies (ATE, BOP, BHK-MG) function better than their conventionally filtered counterparts when in current stabilization mode. Their response to step load transients is up to 1000 times faster.

FIGURE 1 - The effect of a changing load on a current stabilizer. The output capacitor's charge and discharge time controls the recovery time


To understand current stabilization one must first stop and consider just what is meant by the idea of both voltage stabilization and current stabilization.

In one sense, these are descriptions of a power supply's expected response to changes in the load. A voltage stabilizer will respond by varying its current along a straight line locus that defines a fixed voltage (Figure 3a) while a current stabilizer does just the opposite (Figure 3b). A plot of the successive operating points created as the load is allowed to vary creates a straight line at a fixed current for the current stabilizer.

Figure 3a and 3b - The concept of "voltage stabilization" or "current stabilization" relates to the locus of points that a varying load will trace if you observe the changing output voltage and current of the power supply being loaded

We understand resistance or impedance to be the ratio of the change in voltage to the change in current. In the plot of operating points from the voltage stabilizer, we observe a very small change in voltage for a large change in current corresponding to a low source impedance. If the stabilizer were ideal, this impedance would approach zero. The corresponding plot for the current stabilizer suggests a very high impedance and, indeed, the ideal current source would exhibit nearly an infinite source impedance.

The schematic symbol for a voltage source is a battery and the idle condition is an open circuit. The symbol for a current source is an encircled arrow and its idle condition is a short circuit. See Figure 2.

FIGURE 2 - Schematic


In this catalog we tabulate the actual source impedance of Kepco's precision analog stabilizers. Impedance values are shown in both voltage and current modes. This is in addition to simply stating the rated degree of stabilization or regulation that the power supply achieves with its high gain feedback amplifier. The figures for resistive inductance and capacitance are in the model tables. One reason for doing this is that output impedance is a widely misunderstood power supply specification that is actually defined at d-c when load effect stabilization is specified. Its significance is apparent at frequencies other than d-c. The output impedance of a power supply is characterized both in terms of a d-c resistance and a reactive element. At high load frequencies, the reactive component dominates the specification.

FIGURE 4a and 4b - Plot of output impedance vs. frequency for a voltage stabilizer and for a current stabilizer

When stabilizing voltage, the reactance is the effective series inductance which, at high frequencies, introduces a significant non-zero component. The impedance rises at 6db/octave as the frequency increases.

When stabilizing current, the reactance is an effective shunt capacitance. This prevents the impedance from being infinite. Indeed,the impedance falls at 6db/octave as the frequency increases.

The impedance table will permit you to plot the reactive impedance of inductors and capacitors. It is a log-log plot. See Figure 5.

FIGURE 5 - Impedance table


A conventionally filtered power supply has a quite large output capacitor for energy storage as a voltage stabilizer and for dynamic stabilization. This capacitor shunts the output with its low impedance. This is fine for voltage mode but not-so-good when current is the thing to be stabilized.

A problem arises with power supplies that are dual-mode, attempting to be both voltage and current stabilizers. They are called automatic crossover designs. The problem is that they can't be very good current stabilizers with a big low impedance output capacitor stuck across the output. The idea of current stabilization mode is that while the current remains fixed as the load changes, the voltage must remain free to vary in proportion to the load resistance. A capacitor inhibits any voltage change across its terminals and thus is incompatible with the voltage agility that must characterize current stabilization.

Kepco's ATE and BHK-MG allow users to disconnect the output capacitor when they wish to optimize performance in current mode. The BOP models have no output capacitors to begin with.

A capacitorless power supply is dynamically less stable than a conventionally filtered design. It is much less tolerant of reactive loads, oscillating if the load reactance is not compensated. Such units and operating modes should not be chosen for conventional use, especially if the load is reactive.


The straight lines in Figures 3a and 3b would tend toward infinite voltage or infinite current if not bounded in some way. For the voltage stabilizer,the prospect of infinite current has, traditionally, a host of solutions: fuses, circuit breakers and, in more modern designs, current limiters. For the current stabilizer, the corresponding prospect of infinite voltage is far less familiar, and therefore less intimidating. It is no less a real problem.

Kepco's ATE, BHK-MG, MAT and MST power supplies are "automatic crossover" designs. Common to all of these is the idea that the gain, precision and performance of the current control channel is as close to the performance of the voltage control channel as we can get it. Also, these power supplies have the ability to operate over fully 0-100 percent of the voltage and current rating. The BOP power supplies are 4-quadrant designs. They are fully bounded by a voltage limit when stabilizing current and a current limit when stabilizing voltage. A selector is used to determine whether voltage or current is the parameter being stabilized. It is not automatic. BOP are, therefore, not considered automatic crossover.

An automatic crossover power supply uses the complementary nature of voltage and current modes to form boundaries for each other. See Figure 6. Here the voltage locus combines with the current locus to produce a closed,fully bounded rectangle whose voltage and current lines can be positioned as the user wishes. The vertical current stabilizing locus provides a maximum current limit for the load current produced by the voltage stabilizer,while the horizontal voltage stabilizing locus provides a maximum voltage limit for the load voltage produced by the current stabilizer.

FIGURE 6 - The rectangular locus of an automatic crossover design in which the voltage mode serves to protect the current from overload and vice versa

The opposite of overload is idle. If we understand the difference between what constitutes overload to a voltage stabilizer and current stabilizer, then we can appreciate what it means for them to be idle. For the voltage mode, it is intuitive. Zero current is idle. This can be achieved by opening the circuit to its load with a switch or relay. In Kepco's MAT and MST,the "output-enable" relays open the connection to the load when idling the units in voltage mode. For current mode, it is less intuitive. A current stabilizer is idle when it is producing no voltage. This can be achieved by shorting the output with a switch or relay. The MAT and MST output enable relays do just this when idling the units in current mode.

Note: For a power supply to be a real Automatic Crossover design, it MUST be capable of operating indefinitely into a dead short circuit, the current mode's "idle." While the mode is at idle, the power supply, by virtue of its series-pass topology, is working very hard. All Kepco power supplies have sufficient series-pass "horsepower" to do this effortlessly. Many competitive units do not. See the section on HEAT DISSIPATION for an explanation of this. ABC is of course, a switch- mode design.


The value of a power sinking capability in power supplies lies in two applications: When a BOP is driving reactive loads, particularly in the frequency domain, there will be a part of each cycle when the voltage and current are out of phase and one may be positive where the other is negative. In this situation, the load is supplying power to the power supply for a portion of each cycle and the power supply is a sink.

Another situation is when a power supply is used to exercise batteries, perhaps simulating the light-dark cycle of orbiting solar cells. During the dark, or discharge portion of the cycle, the sink capability draws energy from the batteries. In a sense,such application is like an electronic load.


Power supplies that produce a single voltage polarity and single current polarity are, naturally, single quadrant units. The voltage-current rectangle of an automatic crossover power supply lies in a single quadrant. However, there are actually four quadrants. Voltage can be plus or minus and current can be plus or minus. Kepco's MAT and MST models with their polarity reversal relays can be thought of as two quadrant instruments. See Figure 7.

FIGURE 7 - Two quadrant operation results when relays are used to provide polarity reversal

When the current (in the conventional sense) flows OUT of the +voltage terminal... s in a battery...we say that it is a SOURCE. When the current flows INTO the +terminal in a resistor...we say that it is a SINK. Power supplies are usually thought of as SOURCES. Loads, electronic or resistive, are thought of as SINKS. It is possible to combine load and power supply in ways that make it behave as both source and sink. See Figure 8.

FIGURE 8 - Source and sink operation is achieved by preloading the current source in series with the preload resistor, RPL. This makes the preload current, IPL independent of the voltage, EO. If EO is a fixed voltage, a simple resistor may be used to create the preload. The sum of +IL and -IL is the current rating of the voltage source EO

Kepco's BOP series, bipolar power supplies, operate in four quadrants. They can produce both positive and negative voltage and current and operate as both source and sink. BOP are true ide-band d-c amplifiers that can reproduce a complex aveform going smoothly and linearly through zero. They are, nevertheless, solid d-c power supplies capable of producing a fixed output for unlimited time which is quite beyond conventional amplifiers. Hence,the phrase "Bipolar Operational Power Supply"(BOP).

FIGURE 9 - Four quadrant operation from a Kepco BOP power supply

BOP are the ultimate analog power supply responding to both amplitude and polarity signals with high speed. When they are harnessed to digital controllers, either their built-in BIT card or an external SN card, their output is, of course,digitized. No longer do they slew smoothly through zero to any voltage or current position in the four quadrants. Digitally controlled by a bus, they stutter step in rigid increments, with a resolution controlled by the digital system.

While the polarity reversal afforded by the relays in MAT and MST power supplies is mechanical, rather than electronic, as in BOP, when the speed limits and resolution constraints of digital control are considered, the net effect is similar. Unless the SINK capability of BOP is required, relay reversal of polarity is a way of obtaining bipolar (two-quadrant only) output.


The key to being able to function in various quadrants, at high speed, without de-rating, with no limits on duration or load, is an efficient heat dissipating arrangement. The concept of the linear series-pass topology is that the power transistors (or MOSFETs in the high-voltage models) are able to convert unwanted power into HEAT. By so doing, they can meter the flow of power to your load very accurately. Since the power can be quickly switched between dissipator and load, such LINEAR power supplies react very quickly to changing load demands.

The BOP High Power models are four quadrant source-sink power supplies that use switch-mode technology for improved efficiency. Their bandwidth is therefore necessarily smaller than for linear design. To avoid dissipating sinked energy, the BOP High Power models use energy recuperation, passing the load sink back to the a-c power mains.


If a 4-wire Kelvin connection is made between power supply and load (Figure 10), it is possible to arrange the wires so that one pair carries the load current, and the second pair is made to sense the output voltage without the voltage drop error induced by the flow of current through the resistance of the connecting wires. By this means, power supplies can be designed to compensate for the resistance of the load cables. All of Kepco's linear power supplies have been designed with one extra volt available (a 0-6V supply is really capable of 0-7V); this extra volt allows you to drop as much as 0.5V per load wire and still get full rated output at the load. High current, switch-mode models (BOP High Power) have an extra 0.5V which allows up to 0.25V drop per wire.

FIGURE 10 - Circuit illustrating the use of remote error sensing for voltage stabilization

Some care must be taken when using remote sensing. It is easy to introduce noise through pickup. Wires should be shielded in a noisy environment, and may need to be twisted together, or with the respective load wire to minimize the inductance present in a long run. Most important,care should be taken when connecting, disconnecting, or switching 4-wire load circuits. The load wires must always mate before the sense wires, and the sense wires must break before the load wires. To do otherwise risks running heavy current down light gauge sense leads.


Kepco stabilized power supplies have a special diode connected across each error-sensing link (to the respective output terminal) whose function is to conduct should the connections be inadvertently omitted. These diodes prevent an uncontrolled response when the links are open and a remote connection is missing. See Figure 11.

FIGURE 11 - Protection diodes

The diodes present a possible hazard to users who wish to switch their loads on and off. If remote error sensing is used,and if the power supply is heavily loaded, a switch that opens only the load wires throws the burden onto the sensing leads. Not only will this likely damage the error sense diodes, but may produce dangerous overheating in light gauge sense wires used between the power supply and its load.

When switching the load on a power supply that uses a 4-wire connection to its load, the switch must interrupt both the load circuit and the sense circuit. Preferably the sense circuit should be interrupted first and reestablished last.


ProductsSupportLiteratureContact UsCareersAbout

TEL (718) 461-7000 • FAX (718) 767-1102 • email: