ENIAC                             U. S. Army

A REPORT ON THE ENIAC -- Part I, Chapter 2

II. INITIATING UNIT

The initiating unit of the ENIAC is the device which contains controls for turning the power on and off, for initiating a computation, for initial clearing, and for selective clearing a group of accumulators, as well as program controls for the reader and printer. Certain devices for testing the ENIAC are also located on the initiating unit.

The following topics are discussed in this chapter: Section 2.1, starting and stopping the ENIAC power and initial clearing; Section 2.2, reader and printer program controls on the initiating unit; Section 2.3, initiating a computation; Section 2.4, selective clear program controls; and Section 2.5, testing features. The following drawings are referred to in this section:

Initiating Unit - Front View	PX-9-305               
Initiating Unit - Front Panel  	PX-9-302
Cycling Unit and Initiating Unit
 Block Diagram	PX-9-307
Power System Block Diagram	PX-1-303
A-C Power Distribution Rack	PX-1-304

2.1 Starting, Stopping, and Initial Clearing

Nearly all the characteristic functions of the ENIAC depend on d-c power. This, however, is derived from 240 volt, 3 phase, a-c. The latter has some immediate uses in addition to furnishing the d-c. There are in all five principal uses for the a-c power. These are as follows:

  1. 1. for the heaters of the numerous tubes of the ENIAC units.
  2. 2. for the heaters of the rectifier tubes in the ENIAC's power supplies which convert a-c into the different d-c voltages.
  3. 3. for the plates of the rectifier tubes.
  4. 4. for the fans which dispel the great amount of heat generated by the preceding
  5. 5. for the control circuits needed in starting and stopping the ENIAC power, in furnishing protection to various circuits, and in initial clearing.

    The first four items referred to above are identified by the corresponding numbers on PX-1-303. The last item is noted there as control circuits and is more explicitly dealt with on PX-9-307. The control circuits govern the connection of the other items to the a-c lines, cause d-c to be supplied to the units of the ENIAC, and control the initial clearing of these units.

    Program controls for these circuits are found on the initiating unit. Other auxiliary program controls and elements of the control circuits are found on the power distribution rack, the condenser cabinets, and the units of the ENIAC themselves. In this section we shall discuss the events involved in starting and stopping the ENIAC (Section 2.1.1.) and in initial clearing (Section 2.1.2.) 2.1.1. Starting and Stopping the ENIAC

    In this discussion it is assumed that the main a-c safety switch is closed. By a "safety switch" is meant one whose opening not merely cuts off power, but actually opens all lines of the circuit controlled by the switch. We also assume here that the 2 safety switches for the ENIAC heaters and those for the fans and for the heaters and plates of the power supplies are all on. With the last 2 switches off, only the a-c circuits can operate; with any of the others off, neither a-c nor d-c can.

    When the start button on the initiating unit (see PX-9--302) is depressed, the amber pilot light goes on immediately and the following sequence of events takes place: the ENIAC heaters and the power supply heaters are connected to the a-c and the ventilating system is turned on. One minute later, after the heaters have had an opportunity to warm up, the plates of the power supply tubes are connected to the a-c. Simultaneously, initial clearing, which lasts for 10 seconds, begins. After the ENIAC has been initially cleared, the green pilot light on the initiating unit goes on and the ENIAC is ready to operate.

    The heaters clock on the front of the initiating unit, which keeps count of the number of hours that the power supply heaters are on, starts to record as soon as the start button is pushed. On each of the remaining 39 panels on the ENIAC, there is also a heaters clock and an on-off switch for the heaters. When the a-c is turned on, the heaters in a panel go on only if the switch for that panel is in the 'on" position. The associated heaters clock records the number of hours that the heaters of the panel are turned on.

    Before a more detailed discussion of the starting sequence is give, the elements involved in various phases of starting will be pointed out on the schematic diagram of the a-c control circuits shown on PX-9-307. The elements enclosed within the heavy lines are not in the initiating unit. The 28 under-voltage release relays and their 14 associated pick-up relays (designated by M) are located in the condenser cabinets. In the Moore School installation the power supply heater fuse relays and the d-c fuse relays are in a cabinet beside the d-c fuse cabinet and relays A, B, and K are located in the machinery laboratory. The remaining items, except for the door switches and thermostats which are in the ENIAC panels, are on the power distribution rack in the ENIAC room (see PX-1-304).

    Relays A and B connect the heaters of the ENIAC units to the 3 phase a-c power. Relay D is the power supply heaters contactor. F, an adjustable timer which has been set for 1 minute, provides for the delay between the turning on of the power supply heaters and plates. When timer F has counted the specified period of time, relay G is activated. This relay connects the plates of the power supplies to the a-c so that the d-c is turned on when relay G is activated. Timer J which has been set for 10 seconds and relay H, the main initial clear relay, are activated after the d-c is turned on. Relays 3 and 4, auxiliary initial clear relays, are each responsible for the emission of one of the signals involved in initial clearing (see Section 2.1.2.). Ten seconds after timer J starts to count, relay K is activated and the initial clear period is terminated, thus bringing the starting sequence to an end.

    It can be seen on PX-9-307 that in addition to the start and stop buttons on the initiating unit which operate both the a-c and d-c circuits, separate d-c start and stop buttons have been provided. Through the use of the d-c stop button, only the d-c circuits (controlled by relay G), can be turned off, leaving the a-c circuits unaffected. With the a-c power on, pushing the d-c start button connects in the d-c circuits and causes initial clearing to take place. Isolation of the d-c from the a-c circuits has been provided in order to make possible leaving the heaters turned on even when the ENIAC is not to be operated or when there is a failure (see the discussion of protective circuits below) in the d-c circuits. This has been done because it is hoped that, by cutting down the number of times that the heaters are turned on and off, tube life will be lengthened.

    It is to be noted that the operation selector switch on the cycling unit must be set at continuous when the power is turned on. In Section 2.1.2. where initial clearing is discussed, it is pointed out that when the power is first turned on, a number of flip-flops may come up in the abnormal state and it is also remarked that the resetting of these often depends on the pulses and gates emitted by the cycling unit. These pulses are not given out immediately unless the ENIAC is in continuous operation. The danger of having these flip-flops remain in the abnormal state is that, as a result, a number of tubes that should be off most of the time and on only a short period of time (i.e. tubes in circuits that have been designed for a low duty cycle) remain on for a long time and thus cause damage to themselves and other elements.

    Certain protective devices included in the control circuits are also shown on PX-9-307. Of these the most important are relays C, Q, N, and L. The action of these will be discussed in the following paragraphs. Their distinguishing characteristics are as follows: under proper operating conditions C and N are on; L and Q are off. C may be turned off by a thermostat or a door switch. Since it is believed undesirable to turn off the heaters unless it is absolutely necessary, C acts through a timer P which may be set between 5 and 15 minutes. When this time has elapsed and the trouble has not been remedied, both a-c and d-c circuits are turned off. The other three relays act without any delay but affect only the d-c. Relay Q is turned on by the blowing of any heater fuse. This cuts off the d-c power supply including its heaters. Relay N is turned off by phase in the plate supply or under-voltage in the output of a d-c power supply. The effect is to turn on L. This is also accomplished by the d-c stop button or the failure of a d-c fuse. When L is turned on or when there is any phase failure in the heaters, the plate supply to the rectifiers is cut off, but the heaters are left on. The distinction between N and L is that there is a provision for inhibiting the action of N during starting. These actions will now be discussed in more detail.

    Relay C is a master relay which controls both a-c and d-c circuits. This relay, which is activated when the a-c safety switch is closed, operates in conjunction with the door switches (see below), thermostats, and timer P. Found at the back of each ENIAC panel and at the front of the power supply and condenser cabinets, is a door switch. When the cover of a panel or cabinet is removed, the door switch on the panel opens,* causing relay C to be deactivated. If, however, the door switch shunt button on the initiating unit (see O/x-9-302) is held down while the cover is off, relay C is not deactivated. Relay C is also deactivated when a thermostat opens as a result of the overheating of a unit. When relay C is not activated, contact C[1] closes and timer P, which is set for 5 minutes, starts to operate. First its clutch (CL) is thrown in, and next the motor (M) is connected into the circuit through contact CL[1]. A warning lamp above the power distribution rack (see PX-1-304) also lights. Necessary repairs can be made on the machine during this 5 minute period (which may be adjusted to as much as 15 minutes if more repair time is required). If, at the end of 5 (or 15) minutes, the condition which caused relay C to be deactivated has not been corrected, then contact P[1] opens and relay A is deactivated. This turns off both the a-c and d-c circuits. The start button on the initiating unit is used to turn the power on again after the fault has been corrected.**

    * At the present time, there is a permanent shunt for the door switches so that removing a cover does not cause relay C to be deactivated. The description in the text above applies to the intended method of operation of the door switches.

    ** If both the amber and green pilot lights are off, the start button on the initiating unit must be used. If only the green pilot light is off, the power may be turned on through the use of the d-c start button.

    The door switches have been provided as a safety measure for both personnel and the machine since the opening of a panel exposes dangerous voltages (as much as 1500 volts in the case of the d-c) and also, by drawing air from the ventilating system to the open panel, may cause another unit to overheat.

    Relay Q protects the d-c circuits and the power supplies. When Q is activated, contact Q[1] opens so that relay D is de-energized. This turns off the power supply heaters and causes contact D[1] to open. With contact D[1] open, F is de-energized so that contact F[1] opens and relay G, the d-c contactor is deactivated. Relay Q is activated when a contact on one of the power supply heater fuse relays closes. This latter event takes place if a power supply heater fuse blows. If the d-c is turned off because Q has been activated, the d-c start button on the power distribution rack must be used to turn the power on again.

    The remaining protective devices shown on PX-9-307, relays L and N with their associated devices, control only the d-c circuits, leaving all heaters turned on in case of a failure. If one of these circuits detects a failure and turns the machine off, the power can be turned on again through the use of the d-c start button. The main and power supply heater phase failure relays connected in series with timer F detect faults in the three phase which goes to the heaters of the ENIAC and of the power supplies. These phase failure relays are activated so that the contacts shown on PX-9-307 are closed under proper operating conditions. In the event of a phase failure, F is de-energized so that contact F[1] opens and relay G drops out. As soon as the fault is repaired, timer F is again activated and, one minute later, contact F[1] closes.

    Relay L is the d-c cut-off relay. When this relay is activated, contact L[1] opens so that relay G is de-energized. This results in cutting off the d-c power. With the a-c on (so that contact A[4] is closed), relay L can be picked up through the closing of the d-c stop button, the activation of the d-c fuse relays when a d-c fuse blows, or the non-activation of relay N (see the discussion of relay N in the next paragraph).

    Relay N operates in conjunction with the power supply phase failure relays and the under-voltage release relays. The power supply phase failure relays in this circuit detect faults in the three phase a-c which goes to the plates of the power supply tubes. These relays are activated and their contacts closed under proper operating conditions. There is an under-voltage release relay for each power supply. During the starting sequence while initial clearing takes place, relays M are activated. These relays provide the high voltage required to pick up the under-voltage release relays. After the starting sequence is completed, the under-voltage release relays remain activated and their contact are closed unless the voltage emitted by a d-c power supply drops below a specified level. During the initial clear period while the under-voltage release relays are being picked up, contact K[2] of relay K provides a circuit which shunts the under-voltage release relays and the power supplies phase failure relays.* Thus, relay N is activated and contact N[1] is open at all times unless a fault is selected.

    * Timer J should not be set for less than 10 seconds since this delay is required when turning the d-c on to permit the under-voltage release relays to pick up before the shunt across them is removed.

    The starting sequence which takes place when the start button in the initiating unit is pushed ins described chronologically in Table 2-1. In some cases, a contact is classified as both a pick up and hold contact for a circuit, since the contact must close for the circuit to operate and since the circuit continues to operate only so long as the contact remains closed. In other cases, the pick up and holding functions are performed by separate contacts.

    When the stop button on the initiating unit pushed, the ENIAC is completely turned off. Relay A, then B, E, D, G, H, and K are de-energized.

    When only the a-c circuits are on, the d-c start button is pushed, the following events take place: Relay L is deactivated, and through contact F[1] (closed provided that the a-c is on and there is no phase failure in the power for the ENIAC and power supply heaters) and L[1] (closed when L is deactivated), relay G is picked up. This turns the d-c on and then initial clearing follows as indicated on Table 2-1.

    When the d-c stop button is pushed, relay L is activated. Since contact L[1] then opens, relay G drops out and the d-c is disconnected. Contact G[1] also opens, causing relay K to drop out.

    With regard to the matter of interrupting a computation, it might be pointed out that it is not necessary to push the stop button on the initiating unit or the d-c stop button for this purpose. Even though the power is turned on, a computation can be stopped in a number of different ways. If a program cable which delivers a program output pulse to a program tray is removed, the pomputation in progress ceases with the program whose projgram output pulse is eliminated in this way. If the card reader exhausts the cards in its magazine (see Section 8.3.), the computation is terminated with the program just before the one in which reading would take place. A computation ceases, similarly, when the cards in the magazine of the card punch are exhausted (see Section 9.1.).

    2.1.2. Initial Clearing

    When the ENIAC is turned on, it is a matter of chance as to which flip-flops in the various counters, both numerical and program ring, or which program flip-flops (in receivers, transceivers and common programming circuits) will come up in the abnormal state. It is obvious that a computation must start with the numerical and program rings in the clear position and with program flip-flops in the normal state in order that the correct answer may be obtained. Furthermore, if a flip-flop in a transceiver or a program control flip-flop such as the printer start flip-flop (see Section 9.1.) comes up in the abnormal state, not only is the associated program commenced, but also, upon the completion of the program, an output pulse is transmitted which, in turn, may stimulate another program control, etc. This it is also necessary before starting a computation to break program chains or sequences which are accidentally begun when the ENIAC is turned on. Furthermore, it is convenient to be able to stop a computation at a certain point (without turning the ENIAC power off), erase all data stored in accumulators and the master programmer, and then start afresh.

    The initial clear circuits in the ENIAC provide for the contingencies mentioned above. The initial clear circuits consist of the initial clear push button on the initiating unit, relays H and K, which were referred to in Section 2.1.1., and initial clear relays 3 and 4 (see PX-9-307). When the ENIAC's power is turned on, initial clearing takes place automatically immediately after the d-c goes on (see Section 2.1.1.). The initial clear push button is pushed when, with the power already on, it is desired to clear the accumulators and the master programmer. It is to be noted, that the operation selector switch on the cycling unit must be set at continuous for initial clearing to take place. Relay H is the main initial clear relay. When activated, this relay causes initial clearing to take place. Relay K terminates the initial clear period. Initial clear relay 4 is responsible for emitting the initial clear gate (ICG) which, in general, clears the counters used for either numerical or programming purposes. Initial clear relay 3 causes the master programmer clear gate (MPC) to be emitted. The MPC is used in the master programmer to break program sequence (see the discussion in the latter part of this section.)

    When the start button on the initiating unit or the d-c start button is pushed, relay K is not activated so that relay H and the ten second timer J are picked up through contacts G[1] and K[4]. At the end of 10 seconds, contact J[1] on the timer closes. Through J[1], relay K is picked up. From then on, relay K holds through contact K[1] and the initial clear switch which is normally closed.

    When the power has been on and the initial clear button is pushed, relay K is de-energized so that K[4] closes. Since G[1] remains closed as long as the d-c is on, relay H and timer J are then picked up through G[1] and K[4].

    When relay H pick up, contact H[1] closes, thus activating relay 3. Contact 3-1 then closes and the MPC is emitted. As a result of the activation of relay 3, contact 3-3, which is normally closed, opens. Now with 3-3 closed, there is a circuit which allows a small amount of current to flow through the coil of relay 4 by not enough to pick this relay up, and very little passes through the large resistor to the condenser. While 3-3 is open, however, the condenser is charged.

    Ten seconds after relay H is activated, K is activated. Contact K[4] opens and H is, thus, deactivated. This causes contact H[1] to open and relay 3 to drop out. At this time, contact 3-3 closes. This allows the condenser to discharge through the coil of relay 4. In this way, relay 4 is activated and contact 4-1 is closed. With contact 4-1 closed, the initial clear gate is emitted. Initial clear relay 4 is restored to the normal state with contact 4-1 again open in about 1/2 a second when the condenser had discharged.As can be seen from the discussion above, the 10 second period (when the green light is off and when timer J is operating) designated by the phrase initial clear period, is actually devoted to the master programmer clear signal. The initial clear gate comes on after the MPC goes off and lasts for about 1/2 a second. Both the MPC and ICG are carried to the other units of the ENIAC in the d-c voltage cable.

    At the time of writing of this report, the MPC is taken only to the master programmer's stepper output gates (see Section 10.3.1.). The MPC, a negative signal closes down these gates so that no program output pulse can be emitted by the master programmer while the MPC is on. Although a program sequence may be initiated because the flip-flop of some transceiver comes up in the abnormal state, it is impossible for a program sequence lasting 10 seconds (of continuous operation) not to go, at some time in that period, to the master programmer. Since the master programmer, however, cannot transmit a program out pulse while the MPC is on, program sequences which have started accidentally are broken here.

    The way in which the initial clear gate is used in the units of the ENIAC to prepare them for computation is shown on Table 2-2. The reader will probably find it convenient to refer to this table in connection with Chapters IV-X. The circuit elements referred to in Table 2-2 can be identified on the block diagrams for the various units. The reader will notice that in many cases clearing depends on the carry clear gate and the central programming pulse emitted by the cycling unit. It is for this reason, that the cycling unit must be in continuous operation for initial clearing to be accomplished.

    On Table 2-2, two difficulties inherent in the present method of initially clearing the divider and square rooter are noted. One of these difficulties, that the flip-flops in the transceivers may not be reset by the end of the initial clearing period, arises from the fact that in the divider and square rooter, as in the other units of the ENIAC, no special provision has been made for directly resetting the transceivers. In other units of the ENIAC, this causes no difficulty. For, suppose that a transceiver in the high-speed multiplier comes up in the abnormal state when the power is turned on. The multiplier then proceeds, during the time that the MPC is on, to carry out the program set-up on the switches associated with that transceiver. In a maximum of 14 addition times, the program is completed and the transceiver is reset.

    In the divider and square rooter, however, there is no upper limit on the length of time required for a division program (division by zero, for example, requires an infinite length of time). Therefore, if a division program is started because a transceiver comes up in the abnormal state when the ENIAC is turned on or because an accidentally begun program sequence stimulates it, there is no certainty that the program will be completed and the transceiver be reset by the end of the initial clear period.

    Plans have been made to revise this initial clearing difficulty by causing the clear flip-flop in the divider and square rooter to be set during the initial clear period. Since the clear flip-flop in the abnormal state causes the CL and CL[1] signals to be emitted, any flip-flops now reset by CL and CL[1] will also be reset by the modified method of initial clearing. The CL signal also resets the clear flip-flop. The normally negative output of the clear flip-flop provides a reset signal for the divider and square rooter's transceiver.

    Until the initial clearing process for the divider and square rooter is modified, the operator can circumvent this first difficulty by setting the operation switches on this unit at square root instead of divide and the interlock switches at NI (no interlock). Since the maximum time for a square rooting program is 400 addition times (less than a tenth of a second), an accidentally begun square rooting program is certain to be completed by the end of the initial clear period. The reason for setting the interlock switches on the program controls at NI is that, even though a program were completed, a program output pulse would not be emitted and the transceivers would not be reset unless the interlock flip-flop also came up in the abnormal state or unless some program sequence, accidentally started, provided for an interlock pulse.

    The second difficulty, that no provision has been made for resetting the interlock coincidence flip-flop, is also to be remedied. Plans have been made for making a small modification in the divider and square rooter's common programming circuits which will eliminate the need for this flip-flop. Until this modification is made, the operator must pay particular attention to the interlock coincidence flip-flop neon (see PX-10-302) before starting a computation. When the interlock coincidence flip-flop is in the normal state, this neon is off. If this flip-flop comes up in the abnormal state at the end of initial clearing, initial clearing should be repeated until this flip-flop does come up in the normal state.

    2.2. READER AND PRINTER PROGRAM CONTROLS ON THE INITIATING UNIT

    2.2.1. Reader Program Controls

    Certain reader program controls are found on the initiating unit (see PX-9-302 and 9-307). These include the reader start flip-flop and program pulse input terminal (Ri), the reader interlock flip-flop and interlock pulse input terminal (R1), the reader finish flip-flop, the reader synchronizing flip-flop and program pulse output terminal (Ro), and associated gates, buffers, and inverters. The reader start button is also on the initiating unit.

    The reader start flip-flop is flipped into the abnormal state either when Ri is pulsed or when, at the beginning of a computation (see Section 2.3.), the reader start button is pushed. When the start flip-flop is in the abnormal state, a start relay in the constant transmitter is activated so that the reader is stimulated to read a card and cause information read from the card to be stored in the constant transmitter. A little less than half way through the card reading cycle (see Chapter VIII), a reset signal from the reader resets the start flip-flop, so that, even though reading is not yet completed, the start flip-flop is capable of again being flipped into the abnormal state (by the reception of a pulse at Ri) to remember that reading is to take place again.

    When reading is completed, the reader emits a finish signal which causes the reader finish flip-flop to be flipped into the abnormal state. The interlock flip-flop is flipped into the abnormal state when an interlock pulse arrives at R1 or, at the start of a computation, when the reader is stimulated to read by the reader start button. The reader interlock flip-flop makes it possible to carry on a sequence of programs in parallel with reading and then to stimulate the next program sequence when both reading and the parallel sequence have been completed since no program output pulse is emitted from terminal Ro unless the interlock flip-flop is flipped into the abnormal state (see below). If a computation does not call for a sequence in parallel with reading, the operator can provide an interlock pulse by sending the pulse which goes to Ri also to R1.

    The coincidence of signals from the interlock and finish flip-flops causes gate 69 to emit a signal. The output of gate 69 gates a CPP through gate 62 which then sets the reader synchronizing flip-flop. The CPP gated through 68 by the normally negative output of the synchronizing flip-flop gates a CPP through 68 and, thus, provides a reader program output pulse which is emitted from terminal Ro. The reason that the synchronizing flip-flop and gate 68 are used after gate 62 is to ensure a program output pulse of the proper shape and in synchronism with other program pulses.

    Neons correlated with the flip-flops mentioned above are shown on PX-9-305. Program controls for the reader in addition to those on the initiating unit are discussed in Chapter VIII.

    2.2.2. Printer Program Controls

    The printer program controls on the initiating unit include the printer start flip-flop and program pulse input terminal, the printer finish flip-flop, the printer synchronizing flip-flop and program pulse output terminal, and associated gates, buffers, and inverters. Neons correlated with the flip-flops appear on PX-9-305.

    A program input pulse received at Pi flips the printer start flip-flop into the abnormal state. This causes a start relay in the punch to be activated so that the tubes in the printer are set up for the data to be printed and so that a card punching cycle is initiated (see Chapter IX). About 1/4 way through the card punching cycle, the punch emits a finish signal which resets the start flip-flop and sets the printer finish flip-flop. The output of the finish flip-flop in the abnormal state gates a CPP through gate 66. The output of 66 sets the printer synchronizing flip-flop- whose output gates a CPP through gate 69. The output of gate 69 is transmitted from PO as a program output pulse.

    The printer program controls are discussed in greater detail in Chapter IX.

    2.3. INITIATING PULSE FOR A COMPUTATION: Reader Start Button and Initiating Pulse Button.

    Once the starting sequence is completed (amber and green pilot lights are on), the ENIAC is ready to begin computing. To stimulate the computation to begin, however, a program pulse must be delivered to the input terminals of the program controls on which are set up the programs that begin in the first addition of time of the computation. Two alternative methods exist for stimulating the beginning of a computation.

    If the first event of a computation consists of the reading of a card, the computation can be started by pushing the reader start button on the initiating unit (see Section 2.2.1.). When reading is completed, then a program output pulse is emitted from terminal Ro. This pulse can be used to stimulate the programs of the computation which immediately follow reading. As was noted in Section 2.2.1., pushing the reader start button also results in setting the reader interlock flip-flop so that no interlock pulse need be provided for a reading initiated by the reader start button.

    The terminal marked R[s] on PX-9-302 parallels the reader start switch and is used for remote control (see Section 2.2.1.).

    The second procedure for initiating a computation is to connect the termial marked Io (see PX-9-302) to the same program line as the input terminals of the program controls used for the first programs of the computation. When the initiating pulse button is pushed, the initiating pulse input flip-flop (see PX-9-307) is set. Its output allows a CPP to pass through gate 66 and set the synchronizing flip-flop. The output of the synchronizing flip-flop gates a CPP through gate 69 which resets the input and synchronizing flip-flops and causes a program pulse to be emitted from terminal I[o]. Neons correlated with the flip-flops mentioned above are shown on PX-9-305.

    The initiating pulse button has a second important use in connection with testing the ENIAC. One of the chief techniques for localizing errors in either the machine or the set-up of the machine is to operate the ENIAC in the one addition time mode or in the one pulse time mode. Here, the pulses for one addition time or 1 pulse time at a time respectively are given out in sequence every time the 1 pulse - 1 addition time button on the cycling unit is pushed (see Chapter III). In this way, there is an opportunity to observe the numerical and programming neons. Frequently, it is more convenient to proceed through a portion of the computation with the ENIAC operating in its normal or continuous mode and then switch to 1 addition time or 1 pulse time operation than it is to progress through the entire computation non-continuously. This may be arranged by disconnecting the program cable which delivers the pulse used to initiate the programs which are to be examined non-continuously. We call this point where the program cable is removed a break point. When the initiating pulse button is pushed, the computation begins and progresses to the break point. With the necessary switch made in the cycling unit (see Chapter III), computation in the non-continuous mode can be stimulated by delivering the initiating pulse from terminal I[o] to the program line from which the program cable was removed. The reader will notice that after the initiating pulse button is pushed, two addition time cycles, one in which a CPP passes through gate 66 and one in which a CPP passes through gate 69, are required before the initiating pulse is delivered.

    The emission of the initiating pulse may also be stimulated by remote control. The terminal marked I[s] on PX-9-302 is used to parallel the initiating pulse switch with a switch which may be carried anywhere around the ENIAC room and which is connected to I[s] via a program line which has no load box.*

    * Also see the discussion of the portable control box in Section 11.6.

    2.4. SELECTIVE CLEAR CONTROLS

    There are 6 selective clear program controls on the initiating unit. Each control consists of a transceiver with a program pulse input (Ci) and output (Co) terminal on the front panel. The six selective clear transceiver outputs are connected in parallel to a line of the synchronizing trunk. When a selective clear transceiver is stimulated, its flip-flop emits a signal called the selective clear gate (SCG). One addition time later, the transceiver is reset by a CPP and a program output pulse is emitted. Neons associated with the selective clear program controls are shown on PX-9-305.

    The selective clear gate is delivered by the synchronizing trunk to the 20 accumulators. When the SCG is give out, any accumulator whose selective clear switch is set at SC clears in accordance with the setting of its significant figures switch (see Section 4.2.3.). Notice that selective clearing lasts but one addition time and clears only the decade and PM counters of accumulators. The selective clear feature provides a convenient means of clearing the group of accumulators which store data for the printer (see Chapter IX) after printing takes place (see the illustrative problem discussed in Sections 8.7. and 9.5.).2.5. DEVICES FOR TESTING THE ENIAC

    Located on the initiating unit (see PX-9-302) are the following devices for testing the ENIAC: d-c voltage meter and associated voltage selector switches, d-c voltage hum oscilloscope, and a-c voltage meter and voltage selector switch.

    The d-c voltage meter together with the two d-c voltage selector switches provide a means of examining any of the ENIAC's 78 d-c voltages. The d-c voltage chart below the selector switches indicates which voltage is measured as a result of the combination of settings on the switches.

    The a-c voltage meter and switch are used to measure the three phases of one of the two bus systems supplying 110 volt a-c to the filament transformers of the various units. Further details concerning the use of the testing devices mentioned above as well as others not located at the initiating unit are to be found in the ENIAC MAINTENANCE MANUAL.



    UP | PREV | NEXT