Hi all,

I've designed a circuit which drives and protects my photodiode. I'd
now like to interface this circuit to my codec, the TI AIC23EVM. To do
this I've included a voltage limiter to my origional circuit -

http://www.geocities.com/bg_ie/circuit.htm

under the heading - "Limiter Design" on that page.

The final circuit I've presented on that page is the circuit I'm
planning to interface to my codec. First of all, have I designed this
final circuit correctly, and will it protect my codec? Sould I connect
the opamp output directly to the ADC, or should there be a resistor
between them? What would its value be? And should I include a cap also
to high-pass filter any high frequency noise? The AIC23 has a
selectable gain stage giving a gain of between 12 and -34.5dB. If I
was to select a gain of 12 and input a 1.5*cos(2*pi*f) signal, would
my codec be safe?

Thanks for your help,

Barry.

Barry wrote:

> Hi all,
> 
> I've designed a circuit which drives and protects my photodiode. I'd
> now like to interface this circuit to my codec, the TI AIC23EVM. To do
> this I've included a voltage limiter to my origional circuit -
> 
> http://www.geocities.com/bg_ie/circuit.htm
> 
> under the heading - "Limiter Design" on that page.
> 
> The final circuit I've presented on that page is the circuit I'm
> planning to interface to my codec. First of all, have I designed this
> final circuit correctly, and will it protect my codec? Sould I connect
> the opamp output directly to the ADC, or should there be a resistor
> between them? What would its value be? And should I include a cap also
> to high-pass filter any high frequency noise? The AIC23 has a
> selectable gain stage giving a gain of between 12 and -34.5dB. If I
> was to select a gain of 12 and input a 1.5*cos(2*pi*f) signal, would
> my codec be safe?
> 
> Thanks for your help,
> 
> Barry.

Barry,

I have lots of questions. Here are some:

What are 1N5817s? Surely, they're not simple diodes to prevent the rails
from reversing polarity. Save me the trouble of finding the data sheet.

Is the 1N5711 there to provide temperature-compensation leakage?

How much voltage can the codec safely withstand? (For many circuits, it
doesn't hurt to swing inputs to or slightly over the rails. Clamping
the input to the rails with Schottky diodes might me enough.

Power transients seem to be the motivation for the bound circuit. Do you
know how it will behave when the power is switched?

Bounding components can be built into the existing op-amp. Is is
simplicity that motivated as separate one? Will the extra noise prove
bothersome?

Is the bias voltage on the photodiode critical enough to warrant the
2N222 voltage regulator? With most photodiodes, probably not.

The photodiode impedance (when dark) is probably much higher than the
10K load/feedback resistor. Is there a reason not to include the cheap
temperature compensation that a series 10K resistor from the op-amp's
non-inverting terminal to ground would provide?

You mention replacing the photodiode with a second 22K resistor. Where
is the first one?

Jerry
-- 
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Hi Jerry,
 
> What are 1N5817s? Surely, they're not simple diodes to prevent the rails
> from reversing polarity. Save me the trouble of finding the data sheet.
> 
> Is the 1N5711 there to provide temperature-compensation leakage?

I had a huge amount of trouble when biasing my photodiode. Basically,
when I switched on and off the circuit, the rails would swing
resulting in the damage of my first photodiode. I therefore got help
of the guys at electonics.design as my analog skills are nothing to
write home about. The 5817s hold the lines, the 5711 ensures that the
bias applied to the photodiode is quickly removed when switched - that
it doesn't go positive, and the 2222 is overkill more than anything
else. I've been using this circuit, the first circuit on -

http://www.geocities.com/bg_ie/circuit.htm

for months now, and I've had no trouble with it - it works perfectly
as far as I'm concerned. Up until now I'v been viewing the photodiode
output on a digital scope, but now I need to interface it to my
AIC23EVM.

> How much voltage can the codec safely withstand? (For many circuits, it
> doesn't hurt to swing inputs to or slightly over the rails. Clamping
> the input to the rails with Schottky diodes might me enough.

I cant see anything in the AIC23 specs regarding the max voltage
applied to the input -

http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  
http://www.lami.pucpr.br/~afonso/Graduacao/Outros/Audio/tlv320aic23-manual.pdf

other than this -
                                             MIN     NOM     MAX    
UNIT
Analog input voltage, full scale – 0dB                1             
VRMS

> Power transients seem to be the motivation for the bound circuit. Do you
> know how it will behave when the power is switched?
>

Well the way I look at it is that the the opamp is capable of
outputing 15 Volts.
 
> Bounding components can be built into the existing op-amp. Is is
> simplicity that motivated as separate one? Will the extra noise prove
> bothersome?
> 
> Is the bias voltage on the photodiode critical enough to warrant the
> 2N222 voltage regulator? With most photodiodes, probably not.
>

I agree.
 
> The photodiode impedance (when dark) is probably much higher than the
> 10K load/feedback resistor. Is there a reason not to include the cheap
> temperature compensation that a series 10K resistor from the op-amp's
> non-inverting terminal to ground would provide?

Yeah, I'm going to add that this time also.

> 
> You mention replacing the photodiode with a second 22K resistor. Where
> is the first one?

I've changed the load resistance a few times - the first 22K was the
load across the photodiode at that point in time. I'm actually using a
33K at the moment. I'll have a variable resistance in there
eventually. Sorry about the confusion.

In your opinion is the inclusion of the limiter enough to protect my
AIC23? Do I need a resistor between the op-amp output and the AIC23
input also? Thanks again for your help,

Barry.

This is the electonics.design thread where i got help -

http://groups.google.ie/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&threadm=731cea69.0310181320.164b8d18%40postin
g.google.com&rnum=1&prev=/groups%3Fq%3Dauthor:bg_ie%2540yahoo.com%2B5711%26hl%3Den%26lr%3D%26ie%3DUTF-8%26oe%3DU
TF-8%26selm%3D731cea69.0310181320.164b8d18%2540posting.google.com%26rnum%3D1

Barry wrote:

> Hi Jerry,
>  
> 
>>What are 1N5817s? Surely, they're not simple diodes to prevent the rails
>>from reversing polarity. Save me the trouble of finding the data sheet.
>>
>>Is the 1N5711 there to provide temperature-compensation leakage?
> 
> 
> I had a huge amount of trouble when biasing my photodiode. Basically,
> when I switched on and off the circuit, the rails would swing
> resulting in the damage of my first photodiode. I therefore got help
> of the guys at electonics.design as my analog skills are nothing to
> write home about. The 5817s hold the lines, the 5711 ensures that the
> bias applied to the photodiode is quickly removed when switched - that
> it doesn't go positive, and the 2222 is overkill more than anything
> else. I've been using this circuit, the first circuit on -

If it works, go with it.

> http://www.geocities.com/bg_ie/circuit.htm
> 
> for months now, and I've had no trouble with it - it works perfectly
> as far as I'm concerned. Up until now I'v been viewing the photodiode
> output on a digital scope, but now I need to interface it to my
> AIC23EVM.
> 
> 
>>How much voltage can the codec safely withstand? (For many circuits, it
>>doesn't hurt to swing inputs to or slightly over the rails. Clamping
>>the input to the rails with Schottky diodes might me enough.
> 
> 
> I cant see anything in the AIC23 specs regarding the max voltage
> applied to the input -
> 
> http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  

Section 2.2 of the link above, "Recommended Operating Conditions", shows
1 V RMS for the input. That's more than 1 V peak, of course, and the key
word is "operating". Just above, under 2.1, "Absolute Maximum Ratings
...", the allowed analog input range is -0.3V to AV_DD+0.3 V.

Using a bounded op-amp as a limiter gives a very sharp clipping knee
(which you don't need), but no assurance that the output won't go ape
when the supply is shut off, particularly if the positive and negative
rails have different time constants. The spec sheet shows that back-
biased Schottky diodes from xLINEIN to AGND and AVDD should be adequate
protection.* The xLINEINs are internally that's OK for you, fine.
Otherwise, you may need to experiment to see what works. Of course, the
protective diodes need to be on the codec side of the coupling
capacitor.

That's enough if's and maybe's. If after you look it over you have more
questions, ask them here or by email.

   ...

Jerry
_________________________________
* Schottky diodes can go to 0.4 V, but they will be OK for this brief
transient. If that makes you queasy, try germanium diodes (2.2 V).
-- 
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

OK, I'm ready to tackle this problem again. So where were we? Well I'm
trying to interface my photodiode protection circuit with my codec
evaluation module, the TI AIC23EVM, in a fashion which protects the
codec board. According to the specs, the AIC23 chip accepts input
voltages in the range -

-0.3 to 3.6Volts,

but this isn't strictly true. It is my understanding that this is the
acceptable voltage range after the input signal is biased by an
internal signal Vmid, which is equal to +1.15 Volts. I'm now using the
line inputs on the codec. I other words, the actually chip inputs
expect AC signals at the input. There is the added complication also
of an internal amplifier with a programable gain of between 12 and
-34.5dB. So what might happens if I input a 1 Volt sine wave when the
gain is set to 12dB? Could I damage the chip?

The limiter circuit I origionally proposed will provide no protection
when the photodiode circuit is being switched on or off, as the
positve and negative rails have different time constants. So what
might I do instead considering the fact that I'd much prefer to leave
the EVM as it is, and have any protection circuitry I need external to
it?

My photodiode circuit -

http:\\baz.perlmonk.org/circuit.html

AIC23EVM -

http://www.lami.pucpr.br/~afonso/Graduacao/Outros/Audio/tlv320aic23-manual.pdf

AIC23 chip -

http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  

Any suggestions as to how I might approch this,

Thanks in advance,

Barry.




Jerry Avins <j...@ieee.org> wrote in message news:<4021108f$0$8372$6...@news.rcn.com>...
> Barry wrote:
> 
> > Hi Jerry,
> >  
> > 
> >>What are 1N5817s? Surely, they're not simple diodes to prevent the rails
> >>from reversing polarity. Save me the trouble of finding the data sheet.
> >>
> >>Is the 1N5711 there to provide temperature-compensation leakage?
> > 
> > 
> > I had a huge amount of trouble when biasing my photodiode. Basically,
> > when I switched on and off the circuit, the rails would swing
> > resulting in the damage of my first photodiode. I therefore got help
> > of the guys at electonics.design as my analog skills are nothing to
> > write home about. The 5817s hold the lines, the 5711 ensures that the
> > bias applied to the photodiode is quickly removed when switched - that
> > it doesn't go positive, and the 2222 is overkill more than anything
> > else. I've been using this circuit, the first circuit on -
> 
> If it works, go with it.
> 
> > http://www.geocities.com/bg_ie/circuit.htm
> > 
> > for months now, and I've had no trouble with it - it works perfectly
> > as far as I'm concerned. Up until now I'v been viewing the photodiode
> > output on a digital scope, but now I need to interface it to my
> > AIC23EVM.
> > 
> > 
> >>How much voltage can the codec safely withstand? (For many circuits, it
> >>doesn't hurt to swing inputs to or slightly over the rails. Clamping
> >>the input to the rails with Schottky diodes might me enough.
> > 
> > 
> > I cant see anything in the AIC23 specs regarding the max voltage
> > applied to the input -
> > 
> > http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  
> 
> Section 2.2 of the link above, "Recommended Operating Conditions", shows
> 1 V RMS for the input. That's more than 1 V peak, of course, and the key
> word is "operating". Just above, under 2.1, "Absolute Maximum Ratings
> ...", the allowed analog input range is -0.3V to AV_DD+0.3 V.
> 
> Using a bounded op-amp as a limiter gives a very sharp clipping knee
> (which you don't need), but no assurance that the output won't go ape
> when the supply is shut off, particularly if the positive and negative
> rails have different time constants. The spec sheet shows that back-
> biased Schottky diodes from xLINEIN to AGND and AVDD should be adequate
> protection.* The xLINEINs are internally that's OK for you, fine.
> Otherwise, you may need to experiment to see what works. Of course, the
> protective diodes need to be on the codec side of the coupling
> capacitor.
> 
> That's enough if's and maybe's. If after you look it over you have more
> questions, ask them here or by email.
> 
>    ...
> 
> Jerry
> _________________________________
> * Schottky diodes can go to 0.4 V, but they will be OK for this brief
> transient. If that makes you queasy, try germanium diodes (2.2 V).

Barry wrote:

> OK, I'm ready to tackle this problem again. So where were we? Well I'm
> trying to interface my photodiode protection circuit with my codec
> evaluation module, the TI AIC23EVM, in a fashion which protects the
> codec board. According to the specs, the AIC23 chip accepts input
> voltages in the range -
> 
> -0.3 to 3.6Volts,
> 
> but this isn't strictly true. It is my understanding that this is the
> acceptable voltage range after the input signal is biased by an
> internal signal Vmid, which is equal to +1.15 Volts. 

No. According to the spec sheet, which I'm not looking at, but believe I 
remember, that is the "absolute maximum" stress that can be applied to 
the package pin.

> I'm now using the
> line inputs on the codec. I other words, the actually chip inputs
> expect AC signals at the input. There is the added complication also
> of an internal amplifier with a programable gain of between 12 and
> -34.5dB. So what might happens if I input a 1 Volt sine wave when the
> gain is set to 12dB? Could I damage the chip?

Not if the chip is competently designed, which I assume it is.

> The limiter circuit I origionally proposed will provide no protection
> when the photodiode circuit is being switched on or off, as the
> positve and negative rails have different time constants. So what
> might I do instead considering the fact that I'd much prefer to leave
> the EVM as it is, and have any protection circuitry I need external to
> it?

Use Schottky or fast germanium diodes as I described earlier.
To protect my reputation as an engineer at the risk of tarnishing my 
reputation as a communicator, nothing is snipped.

> My photodiode circuit -
> 
> http:\\baz.perlmonk.org/circuit.html
> 
> AIC23EVM -
> 
> http://www.lami.pucpr.br/~afonso/Graduacao/Outros/Audio/tlv320aic23-manual.pdf
> 
> AIC23 chip -
> 
> http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  
> 
> Any suggestions as to how I might approch this,
> 
> Thanks in advance,
> 
> Barry.
> 
> 
> 
> 
> Jerry Avins <j...@ieee.org> wrote in message news:<4021108f$0$8372$6...@news.rcn.com>...
> 
>>Barry wrote:
>>
>>
>>>Hi Jerry,
>>> 
>>>
>>>
>>>>What are 1N5817s? Surely, they're not simple diodes to prevent the rails
>>>
>>>>from reversing polarity. Save me the trouble of finding the data sheet.
>>>
>>>>Is the 1N5711 there to provide temperature-compensation leakage?
>>>
>>>
>>>I had a huge amount of trouble when biasing my photodiode. Basically,
>>>when I switched on and off the circuit, the rails would swing
>>>resulting in the damage of my first photodiode. I therefore got help
>>>of the guys at electonics.design as my analog skills are nothing to
>>>write home about. The 5817s hold the lines, the 5711 ensures that the
>>>bias applied to the photodiode is quickly removed when switched - that
>>>it doesn't go positive, and the 2222 is overkill more than anything
>>>else. I've been using this circuit, the first circuit on -
>>
>>If it works, go with it.
>>
>>
>>>http://www.geocities.com/bg_ie/circuit.htm
>>>
>>>for months now, and I've had no trouble with it - it works perfectly
>>>as far as I'm concerned. Up until now I'v been viewing the photodiode
>>>output on a digital scope, but now I need to interface it to my
>>>AIC23EVM.
>>>
>>>
>>>
>>>>How much voltage can the codec safely withstand? (For many circuits, it
>>>>doesn't hurt to swing inputs to or slightly over the rails. Clamping
>>>>the input to the rails with Schottky diodes might me enough.
>>>
>>>
>>>I cant see anything in the AIC23 specs regarding the max voltage
>>>applied to the input -
>>>
>>>http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  
>>
>>Section 2.2 of the link above, "Recommended Operating Conditions", shows
>>1 V RMS for the input. That's more than 1 V peak, of course, and the key
>>word is "operating". Just above, under 2.1, "Absolute Maximum Ratings
>>...", the allowed analog input range is -0.3V to AV_DD+0.3 V.
>>
>>Using a bounded op-amp as a limiter gives a very sharp clipping knee
>>(which you don't need), but no assurance that the output won't go ape
>>when the supply is shut off, particularly if the positive and negative
>>rails have different time constants. The spec sheet shows that back-
>>biased Schottky diodes from xLINEIN to AGND and AVDD should be adequate
>>protection.* The xLINEINs are internally that's OK for you, fine.
>>Otherwise, you may need to experiment to see what works. Of course, the
>>protective diodes need to be on the codec side of the coupling
>>capacitor.
>>
>>That's enough if's and maybe's. If after you look it over you have more
>>questions, ask them here or by email.
>>
>>   ...
>>
>>Jerry
>>_________________________________
>>* Schottky diodes can go to 0.4 V, but they will be OK for this brief
>>transient. If that makes you queasy, try germanium diodes (2.2 V).
 >>--
 >>Engineering is the art of making what you want from things you can get.
 >>¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Thanks again.

OK, heres the left line-in circuitry concerned with the AIC23EVM -

    |--------------------------------------------------------
    |
    |                    AIC23EVM daughterboard
    |
    |
    |    4.9K                        0.47uF   |------------ 
   ----/\/\/\/\----|------------|------)|-----|  LINEIN
RCA input          |            |             |
   -----|          |            |             |   AIC23 
    |   |          \            |             |   
    |   |          /           --- 47pF       |   CHIP
    |   |          \ 4.9K      ---            | 
    |   |          /            |
    |   |          |            |
    |   |          |            |
    |   |----------|------------|
    |                           |
    |                          \/ AGND 
    |


I've measured the voltage at the 0.47uF Cap and yes there is a DC
voltage present here, of around 1.5 Volts, even when no voltage is
applied to the RCA input. This allows the condition -> -0.3 to +3.6
Volts to be obeyed. Therefore the schottkys would make sense between
ground and LINEIN and LINEIN and the 3.3Volts analog rail, but as you
say, only after the 0.47uF Cap. And thats my problem. The above
circuitry is implemented on PCB. I'm also obtaining the ground and
3.3Volts analog rail from my 6711DSK, through both daughtercard I/Fs.
And thats why I was hoping I could protect the board prior to the RCA
input, otherwise I would have to interfere with the PCB daughterboard.

Any ideas?

Thanks,

Barry.




                 




>>The spec sheet shows that back-biased Schottky diodes from xLINEIN
to AGND and AVDD should be adequate protection.* The xLINEINs are
internally that's OK for you, fine. Otherwise, you may need to
experiment to see what works. Of course, the protective diodes need to
be on the codec side of the coupling
capacitor.








Jerry Avins <j...@ieee.org> wrote in message news:<4032c507$0$3072$6...@news.rcn.com>...
> Barry wrote:
> 
> > OK, I'm ready to tackle this problem again. So where were we? Well I'm
> > trying to interface my photodiode protection circuit with my codec
> > evaluation module, the TI AIC23EVM, in a fashion which protects the
> > codec board. According to the specs, the AIC23 chip accepts input
> > voltages in the range -
> > 
> > -0.3 to 3.6Volts,
> > 
> > but this isn't strictly true. It is my understanding that this is the
> > acceptable voltage range after the input signal is biased by an
> > internal signal Vmid, which is equal to +1.15 Volts. 
> 
> No. According to the spec sheet, which I'm not looking at, but believe I 
> remember, that is the "absolute maximum" stress that can be applied to 
> the package pin.
> 
> > I'm now using the
> > line inputs on the codec. I other words, the actually chip inputs
> > expect AC signals at the input. There is the added complication also
> > of an internal amplifier with a programable gain of between 12 and
> > -34.5dB. So what might happens if I input a 1 Volt sine wave when the
> > gain is set to 12dB? Could I damage the chip?
> 
> Not if the chip is competently designed, which I assume it is.
> 
> > The limiter circuit I origionally proposed will provide no protection
> > when the photodiode circuit is being switched on or off, as the
> > positve and negative rails have different time constants. So what
> > might I do instead considering the fact that I'd much prefer to leave
> > the EVM as it is, and have any protection circuitry I need external to
> > it?
> 
> Use Schottky or fast germanium diodes as I described earlier.
> To protect my reputation as an engineer at the risk of tarnishing my 
> reputation as a communicator, nothing is snipped.
> 
> > My photodiode circuit -
> > 
> > http:\\baz.perlmonk.org/circuit.html
> > 
> > AIC23EVM -
> > 
> > http://www.lami.pucpr.br/~afonso/Graduacao/Outros/Audio/tlv320aic23-manual.pdf
> > 
> > AIC23 chip -
> > 
> > http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  
> > 
> > Any suggestions as to how I might approch this,
> > 
> > Thanks in advance,
> > 
> > Barry.
> > 
> > 
> > 
> > 
> > Jerry Avins <j...@ieee.org> wrote in message news:<4021108f$0$8372$6...@news.rcn.com>...
> > 
> >>Barry wrote:
> >>
> >>
> >>>Hi Jerry,
> >>> 
> >>>
> >>>
> >>>>What are 1N5817s? Surely, they're not simple diodes to prevent the rails
>  
> >>>>from reversing polarity. Save me the trouble of finding the data sheet.
>  
> >>>>Is the 1N5711 there to provide temperature-compensation leakage?
> >>>
> >>>
> >>>I had a huge amount of trouble when biasing my photodiode. Basically,
> >>>when I switched on and off the circuit, the rails would swing
> >>>resulting in the damage of my first photodiode. I therefore got help
> >>>of the guys at electonics.design as my analog skills are nothing to
> >>>write home about. The 5817s hold the lines, the 5711 ensures that the
> >>>bias applied to the photodiode is quickly removed when switched - that
> >>>it doesn't go positive, and the 2222 is overkill more than anything
> >>>else. I've been using this circuit, the first circuit on -
> >>
> >>If it works, go with it.
> >>
> >>
> >>>http://www.geocities.com/bg_ie/circuit.htm
> >>>
> >>>for months now, and I've had no trouble with it - it works perfectly
> >>>as far as I'm concerned. Up until now I'v been viewing the photodiode
> >>>output on a digital scope, but now I need to interface it to my
> >>>AIC23EVM.
> >>>
> >>>
> >>>
> >>>>How much voltage can the codec safely withstand? (For many circuits, it
> >>>>doesn't hurt to swing inputs to or slightly over the rails. Clamping
> >>>>the input to the rails with Schottky diodes might me enough.
> >>>
> >>>
> >>>I cant see anything in the AIC23 specs regarding the max voltage
> >>>applied to the input -
> >>>
> >>>http://www.tij.co.jp/jsc/docs/msp/da/products/pdf/tlv320aic23.pdf  
> >>
> >>Section 2.2 of the link above, "Recommended Operating Conditions", shows
> >>1 V RMS for the input. That's more than 1 V peak, of course, and the key
> >>word is "operating". Just above, under 2.1, "Absolute Maximum Ratings
> >>...", the allowed analog input range is -0.3V to AV_DD+0.3 V.
> >>
> >>Using a bounded op-amp as a limiter gives a very sharp clipping knee
> >>(which you don't need), but no assurance that the output won't go ape
> >>when the supply is shut off, particularly if the positive and negative
> >>rails have different time constants. The spec sheet shows that back-
> >>biased Schottky diodes from xLINEIN to AGND and AVDD should be adequate
> >>protection.* The xLINEINs are internally that's OK for you, fine.
> >>Otherwise, you may need to experiment to see what works. Of course, the
> >>protective diodes need to be on the codec side of the coupling
> >>capacitor.
> >>
> >>That's enough if's and maybe's. If after you look it over you have more
> >>questions, ask them here or by email.
> >>
> >>   ...
> >>
> >>Jerry
> >>_________________________________
> >>* Schottky diodes can go to 0.4 V, but they will be OK for this brief
> >>transient. If that makes you queasy, try germanium diodes (2.2 V).
>  >>--
>  >>Engineering is the art of making what you want from things you can get.
>  >>¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Barry wrote:

> Thanks again.
> 
> OK, heres the left line-in circuitry concerned with the AIC23EVM -
> 
>     |--------------------------------------------------------
>     |
>     |                    AIC23EVM daughterboard
>     |
>     |
>     |    4.9K                        0.47uF   |------------ 
>    ----/\/\/\/\----|------------|------)|-----|  LINEIN
> RCA input          |            |             |
>    -----|          |            |             |   AIC23 
>     |   |          \            |             |   
>     |   |          /           --- 47pF       |   CHIP
>     |   |          \ 4.9K      ---            | 
>     |   |          /            |
>     |   |          |            |
>     |   |          |            |
>     |   |----------|------------|
>     |                           |
>     |                          \/ AGND 
>     |
> 
> 
> I've measured the voltage at the 0.47uF Cap and yes there is a DC
> voltage present here, of around 1.5 Volts, even when no voltage is
> applied to the RCA input. This allows the condition -> -0.3 to +3.6
> Volts to be obeyed. Therefore the schottkys would make sense between
> ground and LINEIN and LINEIN and the 3.3Volts analog rail, but as you
> say, only after the 0.47uF Cap. And thats my problem. The above
> circuitry is implemented on PCB. I'm also obtaining the ground and
> 3.3Volts analog rail from my 6711DSK, through both daughtercard I/Fs.
> And thats why I was hoping I could protect the board prior to the RCA
> input, otherwise I would have to interfere with the PCB daughterboard.
> 
> Any ideas?
> 
> Thanks,
> 
> Barry.
   ...

No. There certain unavoidable truths in Electrical engineering, just as
there are in life.
1) The most that a single flip-flop can remember is one bit.
2) To refer a signal to the voltage at a particular pin, you need access
    to that pin or its surrogate.
3) Lumped constants aren't. Example: every capacitor has inductance.
Experience, mostly painful, will equip you to add to this list.

If the terminals of the 0.47uF capacitor are accessible, you don't need
direct access to the LINEIN pin. The diodes can stay on the board when
you're done with it. A reasonable case can be made that they should have
been there all along.

I imagine that there are input protection diodes in the chip itself, and
all that is needed for brief transients is a sufficiently large resistor
in series with the input signal. I saw nothing to assure me of that in
the spec sheet, though. Maybe a TI application engineer can address that
issue.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Jerry Avins wrote:

> Barry wrote:
> 
>> Thanks again.
>>
>> OK, heres the left line-in circuitry concerned with the AIC23EVM -
>>
>>     |--------------------------------------------------------
>>     |
>>     |                    AIC23EVM daughterboard
>>     |
>>     |
>>     |    4.9K                        0.47uF   |------------    
>> ----/\/\/\/\----|------------|------)|-----|  LINEIN
>> RCA input          |            |             |
>>    -----|          |            |             |   AIC23     |   
>> |          \            |             |       |   |          
>> /           --- 47pF       |   CHIP
>>     |   |          \ 4.9K      ---            |     |   |          
>> /            |
>>     |   |          |            |
>>     |   |          |            |
>>     |   |----------|------------|
>>     |                           |
>>     |                          \/ AGND     |
>>
>>
>> I've measured the voltage at the 0.47uF Cap and yes there is a DC
>> voltage present here, of around 1.5 Volts, even when no voltage is
>> applied to the RCA input. This allows the condition -> -0.3 to +3.6
>> Volts to be obeyed. Therefore the schottkys would make sense between
>> ground and LINEIN and LINEIN and the 3.3Volts analog rail, but as you
>> say, only after the 0.47uF Cap. And thats my problem. The above
>> circuitry is implemented on PCB. I'm also obtaining the ground and
>> 3.3Volts analog rail from my 6711DSK, through both daughtercard I/Fs.
>> And thats why I was hoping I could protect the board prior to the RCA
>> input, otherwise I would have to interfere with the PCB daughterboard.
>>
>> Any ideas?
>>
>> Thanks,
>>
>> Barry.
> 
>   ...
> 
> No. There certain unavoidable truths in electrical engineering, just as
> there are in life.
> 1) The most that a single flip-flop can remember is one bit.
> 2) To refer a signal to the voltage at a particular pin, you need access
>    to that pin or its surrogate.
> 3) Lumped constants aren't. Example: every capacitor has inductance.
> Experience, mostly painful, will equip you to add to this list.
> 
> If the terminals of the 0.47uF capacitor are accessible, you don't need
> direct access to the LINEIN pin. The diodes can stay on the board when
> you're done with it. A reasonable case can be made that they should have
> been there all along.
> 
> I imagine that there are input protection diodes in the chip itself, and
> all that is needed for brief transients is a sufficiently large resistor
> in series with the input signal. I saw nothing to assure me of that in
> the spec sheet, though. Maybe a TI application engineer can address that
> issue.
> 
> Jerry

P.S. "Sufficiently large" could just be 4.9 K. It seems plausible.
--
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯

Thanks Jerry, I emailed TI. Would the OA90 be suitable in your opinion
- forward voltage drop of 0.2V. Im not at all familiar with Germanium
diodes to be honest.

Cheers,

Barry.

OA90 have a lower forward voltage drop of 0.2V 


Jerry Avins <j...@ieee.org> wrote in message news:<403a72ac$0$3094$6...@news.rcn.com>...
> Jerry Avins wrote:
> 
> > Barry wrote:
> > 
> >> Thanks again.
> >>
> >> OK, heres the left line-in circuitry concerned with the AIC23EVM -
> >>
> >>     |--------------------------------------------------------
> >>     |
> >>     |                    AIC23EVM daughterboard
> >>     |
> >>     |
> >>     |    4.9K                        0.47uF   |------------    
> >> ----/\/\/\/\----|------------|------)|-----|  LINEIN
> >> RCA input          |            |             |
> >>    -----|          |            |             |   AIC23     |   
> >> |          \            |             |       |   |          
> >> /           --- 47pF       |   CHIP
> >>     |   |          \ 4.9K      ---            |     |   |          
> >> /            |
> >>     |   |          |            |
> >>     |   |          |            |
> >>     |   |----------|------------|
> >>     |                           |
> >>     |                          \/ AGND     |
> >>
> >>
> >> I've measured the voltage at the 0.47uF Cap and yes there is a DC
> >> voltage present here, of around 1.5 Volts, even when no voltage is
> >> applied to the RCA input. This allows the condition -> -0.3 to +3.6
> >> Volts to be obeyed. Therefore the schottkys would make sense between
> >> ground and LINEIN and LINEIN and the 3.3Volts analog rail, but as you
> >> say, only after the 0.47uF Cap. And thats my problem. The above
> >> circuitry is implemented on PCB. I'm also obtaining the ground and
> >> 3.3Volts analog rail from my 6711DSK, through both daughtercard I/Fs.
> >> And thats why I was hoping I could protect the board prior to the RCA
> >> input, otherwise I would have to interfere with the PCB daughterboard.
> >>
> >> Any ideas?
> >>
> >> Thanks,
> >>
> >> Barry.
> > 
> >   ...
> > 
> > No. There certain unavoidable truths in electrical engineering, just as
> > there are in life.
> > 1) The most that a single flip-flop can remember is one bit.
> > 2) To refer a signal to the voltage at a particular pin, you need access
> >    to that pin or its surrogate.
> > 3) Lumped constants aren't. Example: every capacitor has inductance.
> > Experience, mostly painful, will equip you to add to this list.
> > 
> > If the terminals of the 0.47uF capacitor are accessible, you don't need
> > direct access to the LINEIN pin. The diodes can stay on the board when
> > you're done with it. A reasonable case can be made that they should have
> > been there all along.
> > 
> > I imagine that there are input protection diodes in the chip itself, and
> > all that is needed for brief transients is a sufficiently large resistor
> > in series with the input signal. I saw nothing to assure me of that in
> > the spec sheet, though. Maybe a TI application engineer can address that
> > issue.
> > 
> > Jerry
> 
> P.S. "Sufficiently large" could just be 4.9 K. It seems plausible.