Text preview for : 660A(Model660A).pdf part of Keithley 660A(Model660A) Keithley 660A 660A(Model660A).pdf

Back to : 660A(Model660A).pdf | Home


We warrant each of our products to be free
from defects in material and workmanship. Our
obligation under this warranty is to repair or
replace any instrument or part thereof (except
tubes and batteries) which, within a year after
shipment, proves defective upon examination.
We will pay domestic surface freight costs.
To exercise this warranty, call your local
field representative or the factory, DDD 216.
795-2666. You will be given assistance and
shipping instructions.

Keithley Instruments maintains a complete re-
pair service and standards laboratory in Cleve-
land, and has an authorized field repair
facility in Los Angeles.
To insure prompt repair or recalibration serv-
ice, please contact your local field representa-
tkve or the plant directly before returning the
Estimates for repairs, normal recalibrations,
and calibrations traceable to the National Bureau
of Standards are available upon request.


Section Page Section Page

1. GENERAL DESCRIPTION ........ 1 5-5. Adjusting Hum Bal Control . 19
5-6. Zener Diode Replacement . . 19
1-1. Description ......... 1 5-7. Meter Adjustment . . . . . 19
1-2. Operating Modes ....... 1
l-3. Applications ........ 1 6. CALIBRATION . . . . . . . . . 21
l-4. Specifications ....... 2
1-5. Accessories ......... 3 6-l. General . . . . . . . . . . 21
1-6. Equipment Shipped ... ... 3 6-2. Calibration Schedule . . . 22
6-3. Kelvin-Varley Divider
2. OPERATION. ............ 5 Verification . . . . . . . 22
6-4. Range Calibration , . . . . 23
2-l. Front Panel Controls 6-5. Reference Voltage Supply
and Terminals . . . . . . . 5 Stability Test . . . . . . 25
2-2. Rear Controls and 6-6. Oscillator Adjustment . . . 27
Terminals . . . . . . . . . 5
2-3. Preliminary Procedures . . 6 7. ACCESSORIES , . . . . . , . . . . 35
2-4. Operating Procedures . . . . 6
2-5. Recorder Output . . . . . . . 8 7-L. Model 660LA High Voltage
2-6. Effects Due to Kelvin- Divider . , . . . . , , . 35
Varley Divider Out- 7-2. Rack Mounting . . . . . . . 35
put Resistance . . . . . . . 9 7-3. Placing in Rack . . . . . 36
2-7. Loading and Off-Null
Resistance . . . . . . . . . 9 8. REPLACEABLE PARTS , . . . . . . : 37
2-8. Thermal EMF Precau-
tions . . . . . , . . . . .lO 8-l. Replaceable Parts List . . 37
2-9. AC Effects on Measurement . .lO 8-2. How to Order Parts . . . . 37
Model 660A Replaceable
3. APPLICATIONS . . . . . . . . . .lL Parts List . . . . . . . . 38
Model 6601A Replaceable
3-1. Procedures for Measuring Parts List . . . . . . . . 44
Resistances . , . . . . . .lL Model 66OA Schematic Diagram
3-2. Theory of Measuring Resis- 182693. . . . . . . . . . 47
tances . . . . . I . . . . .ll Model 6601A Schematic Diagram
1632LB . . . . . . . . . 49
4. CIRCUIT DESCRIPTION . . . . . . . .13
$0 Change Notice
4-l. General . . . . , . . . . .13
4-2. Reference Voltage Supply . .13
4-3. Kelvin-Varley Divider . . . .14
4-4. Null Detector . . . . . . . .14
4-5. Guarding . . . . . , . .15

5. SERVICING . . . . . . . . . . . .17 -'- Yellow Change Notice sheet is in-
cluded only for instrument modifi-
5-l. General . , . . . . . .17 cations affecting the Instruction
5-2. Servicing Schedule . . , . .17 Manual.
5-3. Parts Replacement . . . , . .17
5-4. Troubleshooting . . . , . .L7




a. The Keithley Model 660A Guarded dc Differential Voltmeter is a convenient, self-
contained potentiometer. It measures from 100 millivolts to 500 volts with 0.02% limit
of error, and below 100 millivolts within 20 microvolts. Since the limit of error includes
all stability considerations, the unit need not be periodically calibrated or manually
restandardized for a full year. The 500-volt reference supply permits infinite impedance
at null for measurements from 0 to 500 volts.

b. Features for convenient use include: five in-line readout dials with automatically
lighted decimal points; 10 to 25-millivolt recorder output; input polarity switch; floating
operation up to 500 volts off chassis ground; line frequency rejection greater than 45 db.
Also, full guarding minimizes leakage problems.

L-2. OPERATING MODES, The Model 660A can be used as a potentiometer or as a conventional
vacuum tube voltmeter. As a potentiometer, it can measure from 100 millivolts full scale
to 500 volts with i-0.02% limit of error and from 100 microvolts full scale to LOO milli-
volts within 20 microvolts. As a VTVM, the Model 660A measures from 1 millivolt full
scale to 500 volts with an accuracy of -f3% of full scale. It can also measure resis-
tances from 10 megohms to 100,000 megohms within i-5%.


a. The Model 660A is used for measurements over a wide range. Typical applications in
the microvolt region are the matching of semiconductors and the monitoring of noise, tran-
sients and drift. High level voltage applications include power supply and meter calibra-
tion, and tube potential measurements.

FIGURE I.. Keithley Instruments Model 660A Guarded dc Differential Voltmeter.

1.065R 1

b. Its Long-term stability makes the Model 660A useful in measurements of extended
duration. Measurements of Long-term power supply stability, Long-term drift runs, and
monitoring during environmental and reliability tests are possible uses.

c. The null-detector output permits use with potentiometric recorders and digital volt-
meters equipped with automatic print-out. The Model 660A is useful in quality control,
product development, inspection and production.

d. Floating operation to 500 volts is provided for measurements such as plate potential
differences of balanced amplifiers.



LIMIT OF ERROR: +0.02% of reading or 20 microvolts, whichever is greater, after 30-minute

LONG-TERM STABILITY: Will operate within stated Limit of error for one year.

TFXPERATURE COEFFICIENT: Does not exceed 0.002% per OC.

REPEATABILITY: Within 0.005%.

MAXIMUM NULL SENSITIVITY: LOO microvoltb full scale with 2-microvolt resolution.

INPUT RESISTANCE: Infinite at null, from 0 to 500,voLts.

FLOATING OPERATION: 500 volts maximum off chassis ground.

INPUT ISOLATION: Circuit ground to chassis ground: LOS ohms shunted by 0.05 microfarad.


Maximum Usable
Input Voltage Maximum Dial Full-Scale Null Meter
Range > Resolution, Sensitivity, Resolution,
volts millivolts millivolts I microvoLts

50 - 500 10 LOO
5 - 50 1 LO
0.5- 5 0.1 2
i 0 - 0.5 0.1 2


VOLTAGE RANGES: 0.5 volt full scale to 500 volts in four decade ranges.

NULL RANGES: 100 microvolts full scale to 100 volts in seven decade ranges.

VTVM ACCURACY: t3% of full scale on all ranges, exclusive of noise and drift.

ZERO DRIFT: Less than 10 microvolts per 24 hours, non-cumulative, after 30-minute warm-up.

2 0665R

INPUT RESISTANCE: 50 megohms, 0.5 to 500-volt ranges;
10 megohms, O.l-volt range;
1 megohm, 0.1 to lo-millivolt ranges.



LINE STABILITY: Better than 5 ppm for 10% change in Line voltage.

output: Adjustable 10 to 25 millivolts dc for full-scale meter deflection.
Output Resistance: 300 ohms maximum.
Noise: 2 microvolts peak-to-peak referred to input up to 1 cps.
Note: Recorder used must have fully isolated input, 1010 ohms minimum to ground.

POLARITY: Positive or negative, selectable by switch.

CONNECTORS: Input: Binding posts. Output: Banana jacks

POWER: 105-125 or 210-250 volts (switch selected), 50-400 cps, 45 watts.

DIMENSIONS, WEIGHT: 5-l/2 inches high x 17-l/2 inches wide x 13-l/2 inches deep; net
weight, 24 pounds.


a. Model 6601A High Voltage Divider is a LOO:L divider which extends the range of the
Model 660A to 5000 volts. The divider accuracy is ?O.Ol% and its input resistance is
10 megohms. The overall limit of error of the Model 660A with the Model 660lA is ?0.03%.
Section 7 gives operating instructions and Section 8 contains the Replaceable Parts List
and the Schematic Diagram for the Divider.

b. Model 4000 Rack Mounting Kit, containing two brackets and a top cover, converts the
Model 660A to fit standard 19-inch racks, Rack mounted, the Model 660A is 5-l/4 inches
high x 19 inches wide x 13-l/2 inches deep. Section 7 has assembly instructions.

1-6. EQUIPMENT SHIPPED. The Model 660A Guarded dc Differential Voltmeter is factory-
calibrated and is shipped with all components in place. All units are shipped for bench
use. Model 4000 Kit may be ordered for rack mounting; refer to Section 7 for assembly
instructions. The shipping carton also contains the Instruction Manual.

1066R 3

PIGURE 2. Model 660A Front Panel Controls and Terminals. Circuit Designations refer
to Replaceable Parts List and the Schematic Diagram.

FIGURE 3. Model 660A Rear Controls and Terminals.

4 1065R



a. Power Switch. A toggle switch turns the instrument on when it is set to the ON

b. POLARITY Switch. The POLARITY Switch selects the input polarity. The Switch rever-
ses the polarity of the internal reference voltage supply so both positive and negative
voltages may be measured; it does not reverse the meter polarity.

c. NULL Switch. The NULL Switch sets the null detector sensitivity for seven decade
ranges from 0.1 millivolt full scale to 100 volts. When the Switch is in the VTVM position,
the Model 660A operates as a conventional vacuum tube voltmeter for the four ranges of the
RANGE Switch.

d. RANGE Switch. The RANGE Switch adjusts the sensitivity of the VTVM in four steps:
0.5, 5, 50 and 500 volts full scale. It also determines the voltage across the Kelvin-
Varley divider and the position of the decimal point light - which also serves as a pilot
light - between the five Reference Voltage Dials.

e. Reference Voltage Dials. Five in-line dials at the top of the front panel set the
reference voltage when the Model 660A is used as a potentiometer. The last dial is con-
tinuously variable.

f. METER ZERO Control. The METER ZERO Control adjusts the meter needle to zero. The
Control is needed on only the 0.1 and l.O-millivolt null ranges; on the other ranges, the
needle will normally be on zero without adjustment. The Control has a range of approx-
imately i30 microvolts.

25. Input Terminals. At the lower left front panel, the black LO Post is for connections
to the low impedance terminal of the unknown voltage and the red HI Post is for connections
to the high impedance terminal. A second set of binding posts marked LO and GND is provided
for grounding the LO input terminal to the chassis when desired. The LO terminals are
connected together internally.


a. Fuse. For 105-125 volt operation, the Model 660A uses a l-ampere 3 AG fuse. FOK
210-250 volt operation, the Model 660A uses a 0.5-ampere 3 AG fuse.

b. Power Cord. The Model 66OR is designed for a 105-125 volt, SO-&O0 cps line source,
unless otherwise specified on the rear panel. The 3-wire power cord with the NEMA approved
3-prong plug provides a ground connection for the cabinet. An adapter for operation from
2-terminal outlets is provided.

c. NULL DETECTOR OUTPUT. Two terminals, marked + and -, supply a dc signal from the
null detector.

d. OUTPUT ADJUST. +4 screwdriver control next to the OUTPUT terminals adjusts the null
detector output between 10 and 25 millivolts full scale.

1066R 5

e. 117-234 Switch. The screwdriver-operated slide switch sets the Model 660A for 117
or 234-volt ac power lines.


a. Check the 117-234 Switch and the Fuse for the proper ac line voltage. Connect the
power cord.

b. Set the Model 660A as follows:

Power Switch ON
RANGE Switch 500
Reference Voltage Dials zero

The decimal light between the third and fourth Dials will light. Allow the instrument to
warm up for 30 minutes to meet the specified accuracy on all ranges.

c. With the input terminals open, set the NULL Switch to 0.1 MV and zero the meter with
the METER ZERO Control. Then return the NULL Switch to VTVM. The stability of the Model
660A is such that no adjustment should be required in eight hours after a 30-minute warm-up.


a. The Model 66OA is used first as a VTVM to determine the approximate value of the un-
known voltage. It is then used in the potentiometric mode to determine the voltage to iO.OZ"/,.

0000 ,
Voltage -

Potential -

I m
FIGURE 4. Input Connections to Model 660A. The two diagrams show the input circuit for
measuring at ground and ,for floating.
In A, the unknown voltage has one terminal at ground. The shorting link is between the
LO and GND Posts of the Model 660A.
In 13, the unknown voltage has both terminals off ground potential. Note this floating
or off-ground potential must be less than 500 volts. Also note the shorting link is - not

6 1066R


The Model 6601A High Voltage Divider extends the Model 660A range to 5000 volts.
Refer to Section 7 for operation instructions.

b. VTVM Operating Procedures.

1. Eleven full-scale ranges are available for VTVM operation. When the NULL Switch
is at VTVM, the RANGE Switch determines one of four full-scale ranges. BY putting the
five Reference Voltage Dials at zero, the Model 660A can then operate as a VTVM on the
seven null ranges.

2. Connect the unknown voltage to the input terminals, using the LO Post for the low
impedance terminal of the unknown. Refer to Figure 4.

3. Switch the RANGE Switch to the most sensitive range for an on-scale meter deflection

c. Potentiometric Operating Prbcedures.


Avoid large overload voltages on the null detector. No permanent damage will
occur even with 500-volt overloads, but some open circuit offset will be caused
in the null detector. The offset, due to the polarization of the input filter
capacitors, will disappear after about 5 minutes.

1. Leave the RANGE Switch at the last setting used in the VTVM operation. If the
VTVM reading is negative, reverse the POLARITY Switch position.

2. Set the first two Reference Voltage Initial Most Sensitive
Dials to the first two digits of the un- NULL Switch NULL Switch
known voltage found in the VTVM operation. Setting

3. Set the NULL Switch to the initial
null setting shown in Table 1. Adjust
the Voltage Reference Dials progressively
for zero meter deflection while increasing
the null detector's sensitivity with the
NULL Switch. Deflections to the right TABLE 1. Recommended Null Sensitivities
indicate the voltage being measured is and Settings.
more positive than the Reference Voltage
Dial setting.


The most accurate resistors in the Kelvin-Varley divider are in the first two
Reference Voltage Dials. Therefore, to obtain the most accurate readings, use
the first two dials as much as possible.

4. The value of the unknwon voltage is read directly from the Reference Voltage Dials.

a) The Dial reading will be within the specified limit of error if the NULL Switch
is at the most sensitive setting (Table 1) for the range used and if the meter indi-
cates as close to null as possible. Null does not have to be reached.

1066R 7

b) When the first Reference Voltage Dial is used, only the five Dials need be read
to be within specifications (fO.OZ% of reading or 20 microvolts). HOWeVer, the meter
may be read as an approximation of a sixth digit.

c) When the first Reference Voltage Dial is not used, read the voltage directly
from the remaining four Dials.

d) Use the meter as a null indicator when balancing voltages. When the first
Reference Voltage Dial is not used, the meter approximates a fifth dial reading.
However, the loading effect of Kelvin-Varley divider on the meter causes some quanti-
tative inaccuracies when the meter is off null. (See paragraph Z-6.)


a. Recommended recorders for use with the Model 660A are the F. L. Moseley Autograf
680 series recorder and the Minneapolis Honeywell recorder (lOmv-O-10mv scale, 50 kilohms
input resistance). Any recorder used must be able to float 500 volts off ground and its
input must be fully isolated (lOlo ohm minimum leakage resistance to ground).

b. Before attaching the recorder, set all Reference Voltage Dials to zero. Disconnect
the unknown voltage and short both Model 660A input terminals. Set the NULL Switch to
10 MV. Connect the recorder to the OUTPUT terminals on the Model 660A rear panel.

c. Set the Reference Voltage Dials to 10 millivolts to apply an accurate lo-millivolt
potential to the null detector on the LO-millivolt null range. This will provide a full-
scale recorder output which can be matched to the recorder's range between 10 and 25 milli-
volts by adjusting the OUTPUT ADJUST Control.

d. To obtain accurate results and/or to prevent damage to the instruments, the recorder
must be ablt to float off-ground with the
Model 660A. Leakage and pickup between the
two instruments should also be minimized.

1. Make sure neither recorder terminal 4
is grounded. Use a 3-wire grounded power 1
line for the recorder. If a Z-wire line
is used, connect the recorder chassis and
the Model 660A chassis with a separate lead.
10 pf 1

2. Minimize all sources of leakage be- output ' 1 9
tween the output terminals, the recorder TBlack

and ground. Use polystyrene or Teflon-
insulated wire where possible. If the
connecting wires are shielded, connect
the shield to the LO Post.
FIGURE 5. Recorder Filter. A filter between
3. Avoid long leads between the Model the Model 660A and the recorder may be neces-
660A and the recorder. sary when using the O.l-millivolt null de-
tector range.

4. If difficulty is encountered in off-ground measurements, such as unstable readings,
cauwct a lo-microfarad capacitor between the LO and GND terminals on the Model 660A
front panel.


Do not short either Model 660A output terminal to the case; this may damage the
Kelvin-Varley divider.

e. If there is substantial recorder jitter on the O.l-millivolt null range, place a
filter between the Model 660A and the recorder. Refer to Figure 5 for this connection.
Note the filter must also be insulated from ground.


a. When the Model 660A is used for nulling on the O.l-millivolt range, the last Refer-
ence Voltage Dial may appear to be inaccurate. The apparent error is due to a voltage drop
across the Kelvin-Varley divider. This effect involves only the null detector sensitivity
and not the accuracy of the Dial setting. When the Model 66OA is as near to null as pos-
sible, the Reference Voltage Dial setting is correct within the instrument's specifications.
There is no effect present at null.

b. The effect is most apparent on the 0.1, 1 and lo-millivolt null ranges. For example,
on the O.l-millivolt range a O.l-millivolt off-null setting of the reference voltage will
not produce a full-scale meter deflection. This is because the Kelvin-Varley divider out-
put resistance is significant compared to the shunt resistance across the null detector
meter. The LR drop across the divider will cause the meter to be off up to 6%, depending
upon the Reference Voltage Dial settings. On the lOO-millivolt range the maximum error
is 1%. The effect cannot be observed on the other null ranges of the Model 660A.

c. The amount of deflection on the meter is equal to the ratio

where R, is the shunt resistance across the meter (50 megohms for the 100 to l-volt null
ranges, 10 megohms for the 100~millivolt range, and 1 megohm for the 10 to O.l-
millivolt ranges);

Rkv is rhe output resistance of the Kelvin-Varley divider, which is a maximum of
62.4 kilohms at Reference Voltage Dial settings of 2 4 5 4 5 and 2 5 4 5 5 and
a minimum of 100 ohms at settings of 4 9 9 9 8 and 0 0 0 0 2.


a. The input resistance of the voltmeter for the seven null ranges varies from 50 to
1 megohm as given in Table 2. This resistance, however, is not the effective input resis-
tance of the Model 660A. Its input resistance is considerably higher due to the potentio-
metric principle of operation. The value is given by

Rin = Ed Rn Equation 1

Where Rin is the effective input resistance of the Model 660A;
Ed is the setting of the Reference Voltage Dials in volts;
R, is the shunt or input resistance of the null detector meter in ohms;
V is the null detector meter reading in volts.


b. To find the loading effect the Model Voltage Input
660A will have on a circuit, "se Equation 1 Range Resistance
to compute the effective input resistance.
At null, the input resistance is infinite.
Off null, the input resistance is usually 500 volts 50 megohms
high compared to the internal resistance 50 volts 50 megohms
of the unknown voltage, and loading will 5 volts 50 megohms
not be enough to affect the measurement 0.5 volts 50 megohms
.XC"iZXY. For example, if the Reference
Voltage Dials are set at 1.0000 volt on the Input
l-millivolt null range for a reading off Null Resistance
null by 10% of full scale, the Model 660A Range (Slewing)
input resistance is lolo ohms.
100 volts 50 megohms
2-S. THERMAL EMF PRECAUTIONS. Observe 10 volts 50 megohms
standard thermocouple techniques to reduce 1 volt 50 megohms
thermal emf errors for measurements using 100 millivolts 10 megohms
the most sensitive null ranges. Since the 10 millivolts 1 megohm
Model 660A can read to 2 microvolts, thermal 1 millivolt 1 megohm
emf's can introduce considerable errors 0.1 millivolt 1 megohm
into the measurements. In general, use
pure copper leads throughout the system TABLE 2. Model 660A VTVM Input Resistanct
when measuring in the microvolt range. FO?? by Ranges. Resistance for Null Ranges is
extensive measurements in the microvolt re- when the Reference Voltage Dials are set
gion, request the article, DC Microvolt to zero.
Measurements, from Keithley Instruments,
Inc., or its representative.

2-9. AC EFFECTS ON MEASUREMENTS. To minimize errors from ac signals present in thq un-
known voltage, the Model 660A employs a chopper-stabilized null detector operating at a
42-cps chopping rate with a 3-section R-C filter at the input. Very large ac components
on the measuring lines, however, may reduce off-null sensitivity. Also, heavy 60-cps
pick-up will be observed as needle quiver. If ac components affect measurements by the
Model 660A, additional filtering is required. For a single-frequency ac signal, a twin-T
filter is effective. For a variable frequency signal, "se an ordinary low-pass filter.

10 106611


3-1. PROCEDURES FOR MEASURING RESISTANCES. The Model 660A can be used to rapidly measure
resistances from 1 megohm to 100.000 megohms with an accuracy of is%. To measure resis-
tance, connect the unknown resistor across the Model 660A HI and LO terminals. Use a short
isolated lead to the HI terminal to prevent measuring leakage between the leads. Set the
RANGE Switch to 500. Then determine the value of the resistor as follows:

a. For resistances between 1 megohm and 1000 megohms, set the NULL Switch to 10 volts;
adjust the Reference Voltage Dials to obtain a full-scale meter deflection * Subtract
10.000 from the Dial setting and multiply the difference by 5 to obtain the value of the
resistor in megohms.

b. For resistances between 1000 megohms and 10,000 megohms, set the NULL Switch to 1
volt; adjust the Reference Voltage Dials to obtain a full-scale meter deflection. Sub-
tract 1.0000 from the dial setting and multiply the difference by 50 to obtain the value
of the resistor in megohms.

c. For resistances between 10,000 megohms and 100,000 megohms, set the NULL Switch to 1
volt; adjust the Reference Voltage Dials to obtain a convenient deflection on the meter.
Calculate the value of the resistor using,

Rx = 50 Ed megohms Equation 2

Where & is the unknown resistance;
Ed is the Reference Voltage Dial setting in volts;
V is the meter reading in volts.

above method for determining the value of an
unknown resistor is based upon the equation
for the circuit.(See Figure 6). If an un-
known resistance is across the Model 660A
input terminals, then

Ed = i (Rx + Rn) Equation 3

Where Ed is the Reference Voltage Dial set-
ting in volts; Ed
i is the current in the circuit;
Rx is the unknown resistance;
R, is the i'nput resistance of the null
detector meter in ohms.

The current can be written i = V/R,, where
V is the null detector meter reading in Measuring Resistances. Rx is the unknown
volts. Equation 3 now becomes resistance. R, is the input resistance of
the null detector; V is the null detector;
Ed is the buckout voltage.
Equation 4

1065R 11

If measurements are made on the 1 to loo-volt null ranges, the input resistance, R,,, is
50 megohms. Equation 4 becomes

Rx = 5~10~ ( 7 -1) Equation 5

This is thebasis for simplified calculations in paragraph 3-l.

12 1065R


4-1. GENERAL. The Model 660A Differential Voltmeter measures voltage by the potentiw
metric (null) method. The variable known voltage is an ultra-stable 500-volt reference
supply used in conjunction with a precision multi-dial Kelvin-Varley divider. Electronic
referencing of the 500volt output to a zener diode standard maintains the reference sup-
ply's stability and accuracy. This method eliminates repeated manual standardization.
The difference between the divider output and the unknown voltage is indicated by the
null detector, a chopper-stabilized vacuum-tube voltmeter. At null the unknown voltage
can be read directly from the in-line dials of the Kelvin-Varley divider. The input and
null detector are fully guarded to avoid leakage.


Refer to Schematic Diagram 18269E for cir&it designations.


FIGURE 7. Simplified Model 660A Circuit Diagram.


a. Unregulated voltage from transformer T3001 is rectified by a silicon half-wave
rectifier, D3001, and is filtered by capacitors C3001 and C3002. The voltage then is
applied to the regulator series pass tube, V3004. Regulator tube V3005 is used to keep
the screen of V3004 at a constant potential.

b. To obtain a stable, accurate voltage, the 500-volt output of V3004 is sampled by a
divider network of wirewound resistors, R3023 to R3028. The divider network ratio is ad-
justed with potentiometer R3025 to better than 0.01%. Light modulator E3002 compares the
sample voltage from the divider network to the voltage across zener diode, D3003. Any dif-
ference between the two voltages is chopped by E3002 and amplified by a 2-stage ac-coupled

1065R 13

amplifier, V3001. The amplified output of V3001 is converted to a dc signal by light
modulator E3001 and then is amplified by the two-stage differential dc amplifier, V3002
and V3003. The amplifier output is applied to the grid of the series tube, V3004, to
nullify input variations. Capacitor C3004 is used in the ac feedback circuit.

C. The temperature-compensated zener diode, D3003, is used as the basic reference since
typical variations are limited to less than 20 ppm par year and 5 ppm per OC. Thus, a
highly stable reference which eliminates manual standardization is provided with respect
to both time and temperature, The zener diode will also withstand shock and vibration.

d. The regulated 500-volt output of V3004 is either applied directly to the Kelvin-
Varley divider or it is divided to 50, 5 or 0.5 volts by very stable wirewound resistor
networks. The WGE Switch, 53008, determines which network is used. The 50-volt range
divider consists of resistors R3030, R3032 and R3033; the 'j-volt, of R3030, R3035 and
R3036; and the 0.5-volt of R3030, R3038 and R3039. Using potentiometers R3032, R3035
and R3038 accurately sets the voltage division on each range.


a. The Kelvin-Varley divider precisely divides the reference voltage for nulling an
unknown voltage. It is, in effect, a constant input impedance decade potentiometer, con-
sisting of resistors R3040 through R3088. The resistors within each decade are matched;
the decades are matched for each instrument.

b. Each decade of the Kelvin-Varley divider, except the first, R3040 through R3045,
parallels two resistors of the preceeding string. Between the two contacts of the first
Reference Voltage Switch, S3003, the total resistance is 40 kilohms (80 kilohms in para-
llel with the 80 kilohms total resistance of the four remaining strings). With the
RANGE Switch set at 500, 100 volts dc will appear across the contacts of Reference Voltage
Switch S3004, 10 volts across 53005, 1 volt across 53006, and 0.1 volt across 53007.

4-4. NULL DETECTOR. The Model 660A uses a null detector with a chopper stabilized,
feedback amplifier. The input signal is attenuated, if necessary, and sent through a
three-stage R-C filter. The signal is then amplified and applied to the meter.

a. The null detector has three full-scale sensitivities, 0.1, 1 and 10 millivolts.
Above the lo-millivolt range, the input is divided by resistors R1004 through R1009 to
the 10 millivolts. The full-scale sensitivities are determined by one of three feed-
back resistors, R1040 to R1042, in the circuit. A three-stage R-C filter, consisting of
RlOlO, RlOll, 61003, R1012, C1004, Rl018 and C1005, is used to decrease the ac input

b. The light modulators El001 and El002 convert the difference between the filtered
input voltage and the output of the Kelvin-Varley divider into an ac voltage, which is
fed to a four-stage ac coupled amplifier, VlOOl and V1002. The amplifier output is then
demodulated by light modulator El003 and filtered by capacitor C1019. The null detector
meter, MlOOl, indicates the value of the filtered signal. A 42-cps oscillator drives
the light modulators. Using a drive source harmonically unrelated to the standard line
frequency minimizes 60-cps pickup effects.

c. One arm of the feedback network is formed by resistors R1043 to R1046, and one of
the feedback range resistors, R1040 to R1042. Resistor R1039 forms the second shunt arm.
The feedback is applied to light modulator E1002.

14 0665

d. A zero-control network is used to buck out thermal emf's at the input on the two
most sensitive ranges. The network consists of a 1.34-volt mercury battery, BTlOOl, and
resistors R1013 through R1016. The zero control on the front panel, R1016, has approxi-
mately a 60-microvolt span.

a. The null detector output is obtained across resistors R1045 and R1046, which are
in the feedback network. Potentiometer R1045 adjusts the output from 10 to 25 millivolts
at full scale. The output voltage is proportional to the full-scale meter reading.

4-5. GUARDING. Guarding is accomplished by floating the null detector and the input
circuitry at a voltage equal to the input voltage from a low impedance source. This full
guarding eliminates leakage between the input terminal and ground. Such leakage in an
unguarded circuit is difficult to avoid, under laboratory conditions, and can result
in sizeable errors. For example, in an unguarded circuit with a 1-megohm source, leakage
of 108 ohms will introduce 1% error. A guarded circuit eliminates this element of error.
The effectiveness of guarding in the Model 660A is demonstrated by setting the null det-
ector on the O.l-millivolt range, with the input circuit open, and Reference Voltage
Dials set to 500 volts. Even at this extreme condition there is no deflection on the meter,
demonstrating there is no leakage.

0665 15


5-l. GENERAL. Section 5 contains the maintenance and troubleshooting procedures for the
Model 660A. Follow these procedures as closely as possible to maintain the accuracy and
stability of the instrument.

5-2. SERVICING SCHEDULE. The Model 660A needs no periodic maintenance beyond the normal
care required of high-quality electronic equipment. No part should need replacement un-
der ordinary use except a pilot lamp, fuse or, occasionally, a vacuum tube.


a. The Replaceable Parts List in Section 8 describes the electrical components in the
Model 660A. Replace components only as necessary, and use only reliable replacements
which meet the specifications.

b. Replace resistors within any one of the first three Kelvin-Varley divider decades
only as an assembly. Refer to the Replaceable Parts List for the part number for resistors
R3040 to R3067. Resistors R3023, R3024, R3026, R3033, R3036 and R3039 are also parts of
assemblies. Reorder using the Keithley part number (see Section 8) and replace all resis-
tors in the assembly. Because of its importance in maintaining the power supply stability,
order aener diode D3003 only from Keithley Instruments, Inc., or its representative. Re-
fer to paragraph 5-6 for instructions to replace the zener.


a. The following procedures are for repairing troubles which might occur in the Model
660A. Use these procedures to troubleshoot and use only specified replacement parts.
Table 3 lists equipment recommended for troubleshooting. If the trouble cannot be readi-
ly located or repaired, contact Keithley Instruments, Inc., or its representative.

b. Paragraph 7-2 describes how to remove the Model 660A cover. Before proceeding with
the troubleshooting, check the vacuum tubes. Normally, replacing tubes will clear up the
difficulty. All tubes can be readily tested on a grid-modulated tube tester. If repla-
cing a tube does not correct the trouble, continue the procedures. Replacing tubes does
not necessitate recalibration of the instrument.

c. Table 4 contains the more common troubles which might occur. If the repairs indica-
ted in the Table do not clear up the trouble, the difficulty will have to be found through

Instrument USI2

Tektronix Type 503 Oscilloscope Check wave forms

dc voltmeter, 10% accuracy, minimum lOO- Circuit checking
megohm input resistance, 1 volt to 500 volts

Grid-modulated tube tester Test vacuum tubes

TABLE 3. Equipment Recommended for Model 660A Troubleshooting. Use these instruments or
their equivalents.

0665 17

Difficulty Probable Cause Solution

eedle quiver R1048 is out of adjustment Adjust per paragraph 5-5

eter rattle, drift or er- Tube VlOOl, V1002 or V2001 Check tubes; replace if faul-
or is observed on all null is faulty tY

ine voltage variations Tube V3005 is defective Check tube; replace if faul-
ause measurements to be tY
ut of specifications

,eference voltage supply V3001, V3002, V3003 or Check tubes; replace if fual-
rifts after 30-minute V3007 is defective tY
arm-up, requiring frequent
djusting of the range con- One of the divider resis- Return the instrument for
rols, R3025, R3032, R3035, tars, R3023 to R3028, is factory check-out
,3038 changing value rapidly dur-
ing warm-up.

Zener diode, D3003, is un- Return the instrument for
stable factory check-out

Light modulator E3002 is Check light modulator by re-
defective placing

easurements out of toler- Out of calibration Refer to paragraph 6-4 for
nce on all ranges recalibration

One of the divider resistors, Return the Instrument for
R3023 to R3028, is faulty factory check-out

easurements are out of Resistor in the range di- Paragraph 4-2 d points out
pecifications on one vider network is faulty the resistors used for each
ange other than 500- range. Try to bring the ran-
olt range ges within specifications by
calibrating with the potenti-
ometers in the network. See
paragraph 6-4. If this does
not work, check individual

easurements are out of Resistor R3030 is faulty Replace the faulty resistor
pecifications on all ran-
es other than 500-volt

TABLE 4 (Sheet 1). Model 660A Troubleshooting.

18 1065R

Difficulty Probable Cause Solution

Measurements are out of One of the Kelvin-Varley di- See paragraph 6-3
specifications on any vider resistors is faulty
range when the Reference
Voltage Dials are at any
setting other than

Instrument is out of spec- Resistor R1003 is faulty Check resistor; replace i
ifications on all null faulty
ranges above 10 millivolts

TABLE 4 (Sheet 2). Model 660A Troubleshooting.

a point-by-point check of the circuits. Refer to the circuit description in Section 4 to
find the more crucial components and to determine their function in the circuit. The
complete circuit schematic, 182693, is found in Section 8.


a. Potentiometer ~1048 (Figure 18) min-
imizes 60-cps pickup in the null detector.
Misadjustment will reduce sensitivity and
cause needle quiver.

b. To adjust the potentiometer, set the
Reference Voltage Dials to zero and short
the input terminals. Attach an oscillo-
scope to the junction of capacitor Cl016
and resistor R1035 (from pin 6 of V1002).
Figure 8 shows the wave form if potentio-
meter R1048 is adjusted for minimum output.

D3003, is a reference for the voltage divide] I, FIGURE 8. Wave Form with R1048 Adjusted.
resistors R3023 to R3028. The values of re- A Type 503 Oscilloscope was used; horizon-
sistors R3027 and R3028 (Figure 20) are de- tal sweep was 20 msec/cm; vertical, 500
termined by the reference voltage across di- mv/cm.
ode D3003. When the zener is replaced, the
value of these two resistors may have to be


a. Potentiometer R1043 (Figure 18) is the internal meter sensitivity adjustment. It
sets the current through the meter to indicate a full-scale deflection for a full-scale
applied voltage.

b. Warm up the Model 660A for 30 minutes. Set the NULL Switch on the 0.1 millivolt
range, and set the Reference Voltage Dials to .OOOlO. Adjust potentiometer R1043 until
the meter reads full scale.

1066R 19



a. The following procedures are recommended for calibrating the Model 660A. Use the
equipment recommended in Table 5. If proper facilities - especially important for cal-
ibrating an +0.02% instrument - are not available or if difficulty is encountered, con-
tact Keithley Instruments or its representative to arrange for factory calibration.

b. Four procedures are covered: Kelvin-Varley divider verification, voltage range
calibration, reference voltage supply stability test and oscillator adjustment.

c. If the Model 660A is not within specifications after the calibration, follow the
troubleshooting procedures or contact Keithley Instruments, Inc., or its representative,

Instrument Use

Electra Scientific Industries Model SV194B Range voltage calibration
Joltage Calibrator, *0.005% accuracy with
corrections on certificate

Electra Scientific Industries Model RV7'22 Voltage divider for range calibration
)ecade Voltage Divider; terminal linearity,
hl ppm; certificate corrected to *0.2 ppm

Slectro Scientific Industries Model LC875B Range voltage calibration
Lead Compensator

Hewlett-Packard Model 200CD Oscillator Monitor oscillator frequency

(eithley Instruments Model 150A Micro- Null detector for range calibration

(eithley Instruments Model 241 Regulated Voltage supply for range calibration
$igh Voltage Supply

(eithley Instruments Model 662 Guarded dc Check voltages in Kelvin-Varley divider
Xfferential Voltmeter

losley Instruments Model 680 Direct Reading Recorder for reference voltage supply sta.
&corder bility

Cektronix Type 503 Oscilloscope Check wave forms

jeston Instruments Model 3 Type 7 Saturated Range calibration and reference voltage
;tandard Cell supply stability

Jeston Instruments Model 66 Oil Bath Range calibration and reference voltage
supply stability
CABLE 5. Equipment Recommended for Model 660A Calibration. Use these instruments or tht r

1065R 21

I Control
Refer to