Simon, I went back to recheck the specs on the mic input. It is 0.2--not 2.0V! I started with 4.
The card is now blown. This is ok since it will prevent future disasters with better cards. Now the proper zeners may be selected for protection in case something goes haywire with my preamp circuit.
Thank you very much for your guidance.
Oh dear!
I think there's been some discussion on the subject of protecting soundcards on here recently, with suggestions ranging from zeners and series resistors (to protect the zeners and driving source, they're not bulletproof either) through to the more sophisticated Pete Millett's soundcard interface, complete with built in RMS meter.
Interestingly, he also got p***ed off with calibrating soundcards hence the RMS meter and his design looks good to me. (I should also mention that he puts quite extraordinary effort into scanning and putting on his site technical publications for free.)
Fwiw, I'm thinking about simpler/cheaper protection schemes myself, so I'd be happy to look over any ideas you might have - I tend to tie myself in knots with what-ifs so you'd probably be helping me keep a little focus.
Well, Millet uses two fast acting diodes on an opamp. The positive power terminal is connected to the non-inverting input; vice-versa for the negative. I do not understand how this works, or if it would work with only a pos supply scheme.
I liked the idea of back to back zeners because it is cheap and easy. I am only interested in protecting the card in case of malfunction or inadvertent high SPLs while the gain is high. The opamp I may be using can withstand a dead short on the output. I also found the clip voltage on the pair will be less than the zener voltage by about 0.7V due to the voltage drop of the other zener in series.
There is another method that does not flat top, but instead rounds it. I cannot find it right now. However, here is a good general discussion: Overvoltage Protection (OVP) for Sensitive Amplifier Applications - Maxim
Elliot's Clipping Indicator Power Amp Clipping Indicator, his speaker protection Project 111 - PIC Based Speaker Protection.
The rounding clipper may be buried in Elliot, I think.....
I liked the idea of back to back zeners because it is cheap and easy. I am only interested in protecting the card in case of malfunction or inadvertent high SPLs while the gain is high. The opamp I may be using can withstand a dead short on the output. I also found the clip voltage on the pair will be less than the zener voltage by about 0.7V due to the voltage drop of the other zener in series.
There is another method that does not flat top, but instead rounds it. I cannot find it right now. However, here is a good general discussion: Overvoltage Protection (OVP) for Sensitive Amplifier Applications - Maxim
Elliot's Clipping Indicator Power Amp Clipping Indicator, his speaker protection Project 111 - PIC Based Speaker Protection.
The rounding clipper may be buried in Elliot, I think.....
Diodes to the power rails works by stopping the signal from going higher than the +ve power rail (+ one diode drop) or lower than the -ve rail (- one diode drop)
As the maxim paper you link to points out, this means that the power supply concerned has to be able to absorb the current being supplied to it, ie in the opposite direction to what it expects, if the circuitry connected to it doesn't draw that much. The current that can be supplied to it depends on the series resistor preceding the diodes clamp - 49.9k 2W in PM's design.
The maxim paper also very usefully points out that the varying-with-voltage capacitance of reverse biased zeners is much bigger than that of ordinary diodes. So rather than simply wiring the zeners in series with one another, wire them each in series with a normal diode, then parallel those. I've attached the relevant figure.
As the maxim paper you link to points out, this means that the power supply concerned has to be able to absorb the current being supplied to it, ie in the opposite direction to what it expects, if the circuitry connected to it doesn't draw that much. The current that can be supplied to it depends on the series resistor preceding the diodes clamp - 49.9k 2W in PM's design.
The maxim paper also very usefully points out that the varying-with-voltage capacitance of reverse biased zeners is much bigger than that of ordinary diodes. So rather than simply wiring the zeners in series with one another, wire them each in series with a normal diode, then parallel those. I've attached the relevant figure.
Attachments
The zeners seem to be the simplest. With prototype development where the circuit may change everyday, this quad setup is simply moved wherever needed. In my app for instance, only the mic output needs to be clipped at around 0.2V.
I am concerned with the influence of this extra capacitance may do (oscillation), and any extra noise it may add, before it is actually activated near zener voltage, and while it is clipping as well.
Unfortunately for me, the noise level in my mic circuit needs to be kept low. At high gains the threshhold is inconsistently reached in the recognizer, and sets a start signal (with the cheap mics, and probably the cheap sound cards as well). This is most undersireable, since it would cause the post processor, the NLP part of the system (what is it trying to say), to do extra work and probably increase errors.
So, I need to come up with a near-Neumann-quality mic array with just a pocket full of shillings. Carelessness like the preceding debacle are prohibited!
I am concerned with the influence of this extra capacitance may do (oscillation), and any extra noise it may add, before it is actually activated near zener voltage, and while it is clipping as well.
Unfortunately for me, the noise level in my mic circuit needs to be kept low. At high gains the threshhold is inconsistently reached in the recognizer, and sets a start signal (with the cheap mics, and probably the cheap sound cards as well). This is most undersireable, since it would cause the post processor, the NLP part of the system (what is it trying to say), to do extra work and probably increase errors.
So, I need to come up with a near-Neumann-quality mic array with just a pocket full of shillings. Carelessness like the preceding debacle are prohibited!
Update
The calibration source has been modified to .707 times 200mV. 200mV is indicated in an old Creative data sheet as "full scale" mic/line input. This is for the 3.5mm jack connection, not the RCAs which is 2.0V full scale.
The signal amplitude was decreased again to compensate for the percent error of my voltmeter, in case it indicated a lower number than it actually is.
Now there is no clipping and the Visual Analyzer should be working properly as would any PC based oscilliscope.
The Creative Titanium Fatal1ty was repaired! Scrutinizing the area from the input jack to the ADC, two flat and rough-looking solder pads were seen where a resistor was indicated. Narrowing it down, the path between the pad farthest away from the ADC and the blocking cap indicated infinite resistance where about 10k should have been. A SMD resistor of around that value was surgically removed from a scrap board and "installed" by soldering one side to the back side of the cap through-hole, and connecting a wire to the pad closest to the ADC.
Apparently, the series, current limiting resistor had blown off. There are other pads on the board where components are indicated, but these either are completely clean or have clean solder balls on them.
As far as overvoltage/current protection for any device to be plugged in, I could not find any zeners that even come close to the millivolt range.
So, it appears the easiest way to prevent card burnout would be to size an under wattage series resistor, or a glass tube circut breaker as the last component in the mic output circuit. This is for inadvertent overcurrents, and not intended to be activated normally. An led could be placed parallel to this device to indicate the trip. Obvoiusly, the circuit should be designed to limit the output voltage at any gain.
Here is one circuit I found which may serve two purposes: Automatic Gain Control Pre-Amplifier Circuit Diagram
An an authoritative engineering article: http://www.eecg.toronto.edu/~kphang/papers/2001/martin_AGC.pdf
The calibration source has been modified to .707 times 200mV. 200mV is indicated in an old Creative data sheet as "full scale" mic/line input. This is for the 3.5mm jack connection, not the RCAs which is 2.0V full scale.
The signal amplitude was decreased again to compensate for the percent error of my voltmeter, in case it indicated a lower number than it actually is.
Now there is no clipping and the Visual Analyzer should be working properly as would any PC based oscilliscope.
The Creative Titanium Fatal1ty was repaired! Scrutinizing the area from the input jack to the ADC, two flat and rough-looking solder pads were seen where a resistor was indicated. Narrowing it down, the path between the pad farthest away from the ADC and the blocking cap indicated infinite resistance where about 10k should have been. A SMD resistor of around that value was surgically removed from a scrap board and "installed" by soldering one side to the back side of the cap through-hole, and connecting a wire to the pad closest to the ADC.
Apparently, the series, current limiting resistor had blown off. There are other pads on the board where components are indicated, but these either are completely clean or have clean solder balls on them.
As far as overvoltage/current protection for any device to be plugged in, I could not find any zeners that even come close to the millivolt range.
So, it appears the easiest way to prevent card burnout would be to size an under wattage series resistor, or a glass tube circut breaker as the last component in the mic output circuit. This is for inadvertent overcurrents, and not intended to be activated normally. An led could be placed parallel to this device to indicate the trip. Obvoiusly, the circuit should be designed to limit the output voltage at any gain.
Here is one circuit I found which may serve two purposes: Automatic Gain Control Pre-Amplifier Circuit Diagram
An an authoritative engineering article: http://www.eecg.toronto.edu/~kphang/papers/2001/martin_AGC.pdf
- Status
- Not open for further replies.