Copyright © 2011 by Wayne Stegall
Updated November 4, 2011. See Document History at end for
details.
Phono Preamp Input Protection
Dilemma
On first impulse it seems advisable to protect the front end of a phono
preamp with a circuit such as in
figure
1. The clamping diodes are able to protect from the real
possibility that a line output could be connected to a phono
input. However, reservations about the distortion produced by
the diodes may deter their actual use. I expect you too have
similar concerns. In view of their usefulness however, it would
be better to analyze the
circuit before making a hasty decision.
Figure
1:
Diodes
protecting
phono
preamp
input


A Typical Example
A typical illustration of the need of input protection for a phono
preamp is illustrated in
figure 2.
Here
a
typical
lownoise
bipolar
operational
amplifier
has
internal
protection
diodes
limited
to
10mA
to
protect. Without protection,
Z
_{G} may allow excessive current to flow through internal
diodes damaging an expensive operational amplifier. Discrete
circuits could present similar difficulties, i.e. an overloaded JFET
could see reverse gate current exceeding specifications in spite of
some
protection by a source resistor.
Figure
2:
Example
of
diodes
protecting
bipolar
operational
amplifier


Z_{F} and Z_{G}
here simplify a more complicated RIAA equalization circuit.
Internal protection diodes are taken out of the signal path by the
feedback loop in normal operation.

Here the voltage across Z
_{G} causes the external diodes to
shunt the majority of current that would pass through the internal
diodes if the input is overloaded. Because diodes have an
exponential transconductance characteristic, addition of voltages
multiplies currents and subtraction of voltage divides currents by
commonly known exponential and logarithmic relations.
Exponential Axioms


Corresponding Logarithmic Axioms

(1)

e^{x+y}

=

e^{x} × e^{y} 
(2) 
e^{xy} 
=

e^{x}
e^{y} 


(3)

ln(a) + ln(b)

= ln

(

a × b 
)

(4) 
ln(a)  ln(b) 
= ln


a
b 


Where V
_{e} is the limiting voltage across the external diodes,
V
_{i} that across the internal ones, and V
_{zg} that
across Z
_{G}, the following voltage subtraction creates a
related current bypass ratio:
(5)

V_{i} = V_{e}  V_{zg}

Solve for the current bypass ratio presuming external protection diodes
have the same saturation current I
_{O}
as the internal ones:
(6)

I_{ext}
I_{int}

=

I_{O}(e^{40×Ve} 
1)
I_{O}(e^{40×Vi}
 1) 
= 
e^{40×Ve}  1
e^{40×Vi}  1 
Under the overload conditions of interest, e
^{40×V} >> 1,
therefore:
(7)

I_{ext}
I_{int}

=

e^{40×Ve}
e^{40×Vi} 
= 
e^{40×(VeVi)}

=

e^{40×Vzg} 
In the limiting condition where a Z
_{G} of 10Ω passes 10mA and
drops 100mV:
(8)

I_{ext}
I_{int}

=

e^{40×(100mV)} 
= 
e^{4}

=

54.5982 
This means the external protection diodes will bypass 545.982mA before
the opamp input diodes exceed their specifications, conditions easily
met by
reasonable overload conditions. This is well below the current
output limit of most operational amplifiers and the output of circuits
with
limiting resistors as well. Only the improbable connection of
speaker leads to the protected input could damage the operational
amplifier. The sacrifice and replacement of a 1N914 or a 1N4148
small signal diode is well worth protecting more expensive circuitry,
although a medium diode like a 1N4004 might absorb all current from any
line output without destruction.
More Exact Calculations
My presumption in the above calculations that the saturation currents
are equal served the purpose of demonstration, because I knew it to be
a worst case example. Actual saturation current, however, is
roughly
proportional to the size and current capacity of the diode. I
expected the operational amplifiers input protection diodes to be
smaller than any external ones used for protection and to result in a
greater current bypass ratio.
If the presumption of equal I
_{O} were not met
equation 7 would have been:
(9)

I_{ext}
I_{int}

=

I_{Oext}
I_{Oint} 
× e^{40×Vzg} 
Consider some actual saturation currents:
Diode


I_{O} 



AD797 input protection


1.0fA

1N914


64.7335pA

1N4001


31.9824nA

Still in the limiting condition where a Z
_{G} of 10Ω passes
10mA and
drops 100mV,
now an AD797 protected by 1N914's gives:
(10)

I_{ext}
I_{int}

=

64.7335pA
1.0fA 
× e^{40×(100mV)} 
= 
64.7335k × e^{4}

=

3.53433M(A/A) 
In this case the 1N914 would have to pass an impossible 35kA before the
internal diodes reached their limit of 10mA.
and an AD797 protected by 1N4001's gives:
(11)

I_{ext}
I_{int}

=

31.9824nA
1.0fA 
× e^{40×(100mV)} 
= 
31.9824M × e^{4}

=

1.74618G(A/A) 
In this case the 1N4001 would have to pass an absurd 17MA before the
internal diodes reached their limit of 10mA.
Either of these real results leave the operational amplifer to bear
negligible overload current.
SPICE Analysis
The following is the distortion analysis of a typical MM phono input
loop.
Figure
2:
SPICE
schematic.


SPICE model
Fourier analysis of
v_{out} for
v_{in}
of 5mV
_{RMS} protected by 1N914s:
No. Harmonics: 16, THD: 6.20501e06 %, Gridsize: 200,
Interpolation Degree: 1
Harmonic 
Frequency 
Magnitude 

Norm.Mag 

Percent 

Decibels 









1 
1000 
0.00696354 

1 

100 

0 
3 
3000 
3.5818e10 

5.14364e08 

5.14364e06 

145.775 
5 
5000 
6.19818e11 

8.9009e09 

8.9009e07 

161.011 
7 
7000 
7.47346e11 

1.07323e08 

1.07323e06 

159.386 
9 
9000 
8.88345e11 

1.27571e08 

1.27571e06 

157.885 
11 
11000 
1.03415e10 

1.48509e08 

1.48509e06 

156.565 
13 
13000 
1.17256e10 

1.68386e08 

1.68386e06 

155.474 
15 
15000 
1.29017e10 

1.85275e08 

1.85275e06 

154.644 
Even harmonics were omitted because the minuscule values
calculated
represent only calculation noise inconsistent with circuit symmetry
expecting their values to be zero.
Distortion here is vanishingly low!
I initially preferred the use of small signal diodes to medium ones
because I expected the distortion to increase with the saturation
current. So here I add a distortion analysis with a medium diode
as well.
Fourier analysis of
v_{out} for
v_{in}
of 5mV
_{RMS} protected by 1N4001s:
No. Harmonics: 16, THD: 0.000811766 %, Gridsize: 200,
Interpolation Degree: 1
Harmonic 
Frequency 
Magnitude 

Norm.Mag 

Percent 

Decibels 









1 
1000 
0.00696751 

1 

100 

0 
3 
3000 
5.65599e08 

8.11766e06 

0.000811766 

101.811 
5 
5000 
2.24354e11 

3.22001e09 

3.22001e07 

169.843 
7 
7000 
1.34994e14 

1.93748e12 

1.93748e10 

234.255 
9 
9000 
2.00863e14 

2.88285e12 

2.88285e10 

230.804 
11 
11000 
4.37608e14 

6.2807e12 

6.2807e10 

224.040 
13 
13000 
4.30256e14 

6.17517e12 

6.17517e10 

224.187 
15 
15000 
1.82514e14 

2.6195e12 

2.6195e10 

231.636 
Even harmonics were omitted because the minuscule values
calculated
represent only calculation noise inconsistent with circuit symmetry
expecting their values to be zero.
The third harmonic here has increased as expected until it is low but
no
longer vanishing. It is odd that the remaining harmonics have
decreased.
Last Word
Because diode distortion is
presumed more objectionable than that of FETs and tubes, It is still a
subjective matter whether an input protector's distortion is acceptable
in your circuit.
Document History
November 3, 2011 Created.
November 3, 2011 Made improvements. Added much new
material. Updated SPICE results based on better diode models.
November 4, 2011 Corrected some grammar.