Download Discrete OTA Measurement Report #1

Measured results from commercial OTA ICs and a discrete OTA
Tim Davis
October 2006 to March 2007
Below is measured data from a discrete OTA, an LM13700 purchased from Digi-Key, a
CA3080E purchased from a seller out East and a CA3280E sent to me by a friend. The
LM13700 is the only part in production as both the CA3080E and CA3280E are obsolete.
Figure 1 shows the test setup used to measure the data in Figure 2 and Figure 3. Figure 2
compares the measured results of the current mirrors in the OTA. Figure 3 gives an
indication of how well the +/- (p/n) mirrors match. Figure 5 shows the effect of the
output impedance of the CA3280E and a discrete OTA.
+13V
Iota
0.1uF
+
DC_IN
+/- 200mV
Iout
OTA
+
Vout~0V
-
+
0.1uF
I_Meter
-
-13V
Figure 1: Test setup used for measuring data in this report.
The discrete OTA performs very well versus the commercial OTA ICs. The LM13700 is
the only IC still in production. The discrete OTA could be built in production as well.
All of the commercial OTA ICs start to deviate in performance at currents above 200uA.
This is largely due to thermals at the higher currents. The discrete OTA was tested out to
3mA and could be tested to even higher currents. There are significant problems starting
to occur in the commercial ICs at 1mA and above even. (Spec gives 2mA max)
The next step is to measure the offset voltage of the various OTAs versus Iota. Figure 5
gives one source of error that will generate an input referred offset error. The input pair
will also be another source of Vos.
LM13700, CA3080E, CA3280E, Discrete, Iout/Iota vs. Iota, Vin =+/-200mV, Vout =0V,
Vsupply = +/-13V
1.10
1.09
1.08
1.07
1.06
CA3080E
1.05
Iout/Iota, 1+13700
Iout/Iota, 1-13700
Iout/Iota, 2+13700
Iout/Iota, 2-13700
Iout/Iota, +CA3080E
Iout/Iota, -CA3080E
Iout/Iota, +CA3280E
Iout/Iota, -CA3280E
Iout/Iota, 2+CA3280E
Iout/Iota, 2-CA3280E
Iout/Iin, +discrete
Iout/Iin, -discrete
1.04
1.03
1.02
Discrete
Iout/Iota
1.01
1.00
0.99
0.98
LM13700
0.97
CA3280E
0.96
0.95
0.94
0.93
0.92
0.91
0.90
0.89
0.88
0.001
0.01
0.1
1
10
100
1000
10000
Iota, uA
Figure 2: Effect of OTA’s bias current through its current mirrors – various OTAs.
Ratio of +/- currents in mirrors vs Iota, Various OTAs, Vout=0V, Vsupply=+/-13V
1.05
Ratio of +/- current in mirrors
1.04
1.03
I-/I+, discrete
I+/I-, 1_CA3280
I+/I-, 2_CA3280
I-/I+, CA3080E
I-/I+, 1_13700
I-/I+, 2_13700
1.02
1.01
1.00
0.99
0.001
0.01
0.1
1
10
100
1000
10000
Iota, uA
Figure 3: Ratio of mirror currents at the output. This is a good indication of how well the +/- (p/n)
mirror’s match and this mismatch will introduce a source of an input referred offset voltage.
Figure 3 has a couple of devices that don’t fair well at all. The second device or B device
of the CA3280 and the CA3080 doesn’t do well either. Both of these devices will
measure higher input referred offset voltages and will give greater variation versus Iota.
+13V
Iota=
100uA
0.1uF
+
DC_IN
+/- 200mV
Iout
Vout=
-10V to +10V
Source
+
Measurement
Unit (SMU)
OTA
+
-
0.1uF
Force V
Measure I
-13V
Figure 4: Measurement setup to measure output impedance. Data is in Figure 5.
Iout vs. Vout, CA3280E and discrete OTA, Vin = +/- 200mV, Iota =100uA, Vsupply = +/- 13V
1.02E-04
n-side
p-side
1.00E-04
Iout, Amps
9.80E-05
Iout, 3280_+
Iout, 3280_Iout, 3280_2+
Iout, 3280_2Iout, discreteIout, discrete+
9.60E-05
9.40E-05
p-side
9.20E-05
n-side
p-side
n-side
9.00E-05
-10
-9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
9
10
Vout, Volts
Figure 5: Effect of output impedance of CA3280E and discrete OTA. This wasn’t measured yet on
the other commercial OTA ICs. Discrete’s data on upper two lines. The flatter the line, the better.
Figure 5 shows that the discrete OTA stacks up well against the CA3280 in comparison
of output impedance. This isn’t a big surprise as the discrete OTA can use transistors for
both pnp and npn that have better VA (early voltage). This can be tougher to achieve for
both bipolar polarities in an IC process.
+13V
0.1uF
Iota
+
DC_IN
uV source
Vos
Iout
OTA
+
-
Iout~0A
Vout~0V
+
-
0.1uF
I_Meter
-
-13V
Figure 6: Circuit setup used to measure the data shown in Figure 7 to Figure 9.
Vos vs Iota, Discrete OTA, CA3280
0.0004
No 3mA pt for 3280
0.0003
0.0002
0.0001
Vos, Volts
0.0000
Discrete OTA
CA3280
-0.0001
-0.0002
-0.0003
-0.0004
-0.0005
-0.0006
-0.0007
1.0E-08
1.0E-07
1.0E-06
1.0E-05
1.0E-04
1.0E-03
1.0E-02
Iota, Amps
Figure 8: Offset voltage vs. Iota. Comparison of the best CA3280 I have measured to date versus the
discrete OTA.
The CA3280 measured for this Vos data is the better of the two CA3280s that have been
measured thus far. At this point, only one discrete OTA has been built and tested. At the
time of this writing, a second one is almost complete.
Figure 8 shows that the offset voltage of the CA3280 is definitely better than the discrete
OTA’s at Iota currents less than 1uA. However, the offsets begin to become comparable
above 1uA, and at currents of 100uA and greater, the discrete OTA performs better.
Further circuit design has already occurred to make improvements at the lower Iota
currents and the new OTA design will be measured and reported on soon. The output
current of the discrete OTA is also more stable than that of the CA3280 under many
operating conditions and that fact is attempted to be shown in Figure 9 and Figure 10.
If one has a good low offset voltage CA3280 and intends to design the tail current in the
range around 10nA to 1uA, then a CA3280 is a better choice than this discrete OTA –
given this data. Under the remaining biasing conditions however, the discrete OTA
performs better. The next revision of the discrete OTA is making significant
improvements at the lower Iota currents.
Minimum Iout including noise and drift while trimming offset at each Iota vs Iota,
discrete OTA and CA3280
1.0E-07
Iout Minimized, Amps
1.0E-08
1.0E-09
Iout, minimized, dis
Iout, minimized, ca
1.0E-10
Smaller number is better
1.0E-11
1.0E-12
1.0E-08
1.0E-07
1.0E-06
1.0E-05
1.0E-04
1.0E-03
1.0E-02
Iota, Amps
Figure 9: Minimum value of current including noise and drift while trimming Vos at each individual
Iota. Discrete is blue, CA3280 is pink.
Figure 9 shows how low the output current can be trimmed by using Vos at each Iota
current. Except for the 100uA current value, the discrete OTA is equal in performance or
better at EVERY Iota current value. This is interesting because the discrete design seems
to drift less and be less noisy than the CA3280.
Ipicoammeter, minimum includes noise and drift. Part trimmed at Iota=1mA and then varying
Iota keeping Vos the same, Discrete OTA and CA3280
1.0E-05
No 3mA pt for 3280
I minimized during trim
1.0E-06
1.0E-07
Ipico, minimized, dis
Ipico, minimized, ca
1.0E-08
smaller number is better
1.0E-09
Trimmed at 1mA
1.0E-10
1.0E-11
1.0E-08
1.0E-07
1.0E-06
1.0E-05
1.0E-04
1.0E-03
1.0E-02
Iota, Amps
Figure 10: Trim Vos at 1mA Iota and then measure the output current when varying Iota and
keeping Vos constant. Discrete is blue, CA3280 is pink.
In Figure 10, the discrete OTA performs equally well or better than the CA3280 at Iota
tail currents of 200nA and above. The CA3280 is thermally unstable at Iota currents
above 2mA. The discrete OTA was measured to Iota currents of 3mA and should/could
have been measured to higher current values yet. This graph also shows, and it was
witnessed during test, that the discrete OTA doesn’t drift and has more stable readings
than the CA3280. The current shown on the y-axis in Figure 10 is taken to be the
absolute value.