Sound Card Input Circuit Protection

[AndrewT]

Originally Posted by Osvaldo de Banfield

............. use two zeners in reversal series...........

Originally Posted by ITPhoenix

............the zeners are back to back, in series, .......................
Would not the introduction of standard diodes defeat clamping operation?

I was imagining paralleled Zeners. I did not note Osval said series connected.

In parallel the diode does not defeat the clamping action.

But the series connected uses fewer components and the effective capacitance is quartered. Go for series connected.

[SoNic_real_one]

Still, the capacity is very voltage-dependent. So, for an input that has high impedance it might be a problem. Plus, the zenner might be to slow anyway and let-trough som edangerous pulses.
Usually the CMOS IC have protecting diodes inside on every input, so doing this outside is redundant and unnecessary.
If the input is associated with an OpAmp, the better option is two fast-acting diodes, one connected to negative rail, the other to positive rail.

[SY]

Take a look at the schematic for Pete Millett's soundcard interface to see how protection should be done. In fact, if you want to use a sound card for measurement, just build that interface and be done with it.

[ITPhoenix]

Originally Posted by AndrewT

I was imagining paralleled Zeners. I did not note Osval said series connected.

In parallel the diode does not defeat the clamping action.

But the series connected uses fewer components and the effective capacitance is quartered. Go for series connected.

That is quite alright. Sometimes deep insights heretofore unseen are derived from misunderstandings.

I submitted a post with an image of the basic circuit yesterday, along with some details of the larger project, but it failed to publish thus far.

[ITPhoenix]

Originally Posted by SoNic_real_one

Still, the capacity is very voltage-dependent. So, for an input that has high impedance it might be a problem. Plus, the zenner might be to slow anyway and let-trough som edangerous pulses.
Usually the CMOS IC have protecting diodes inside on every input, so doing this outside is redundant and unnecessary.
If the input is associated with an OpAmp, the better option is two fast-acting diodes, one connected to negative rail, the other to positive rail.

Interesting. I need to be at the tentative design stage of the complete circuit to consider this impementation.

I do not trust soundcard makers to provide overcurrent protection. I already blew one card simply by using a boosted zoom mic; however, I am unsure if that is what did it or if it was the cutting edge, beta driver furnished by ALSA to adddress complaints of low mic level inputs. We know aggressive software drivers can blow hardware. I will disect the mic I used when it arrives next week, to see what is in it and how the component are wired.

[ITPhoenix]

Originally Posted by SY

Take a look at the schematic for Pete Millett's soundcard interface to see how protection should be done. In fact, if you want to use a sound card for measurement, just build that interface and be done with it.

Yes, Sonic's post basically shows the method. It appears the diodes conduct when input exceeds supply voltage and sinks any overvoltage thereto. Most interesting.

Millet's device is beautiful; however, my app uses the mono mic input.

[Simon B]

Originally Posted by ITPhoenix

..........I already blew one card simply by using a boosted zoom mic; however, I am unsure if that is what did it or if it was the cutting edge, beta driver furnished by ALSA to adddress complaints of low mic level inputs.........

ITPhoenix, this is the first real information that you've given about what it is you're trying to protect against. Apart from that we knew only that you wanted to keep a 600R input impedance.

I had dismissed Tesla coil performance monitoring as unlikely for this forum and was dithering between lightning strikes and accidental connection to the mains supply, slightly favouring the latter. I appreciate that actually specifying things tends to curtail the discussion but occasionally a bit of focus is nice.

Microphone input usually implies low impedance, low signal level, needing good low noise amplification etc etc.

Will the 'boosted' mic be the the only one you're using? In which case won't it work on the line level input?

Might I suggest that the best way for you to stop blowing soundcards is to stop using the mic inputs? They're never as good as a proper external mic preamp anyway.

[Osvaldo de Banfield]

Originally Posted by SoNic_real_one

If the input is associated with an OpAmp, the better option is two fast-acting diodes, one connected to negative rail, the other to positive rail.

It is very true, but you must have access to internal wiring, which isn´t recommendable por people that don´t know exactly they are doing. In place, a simple 5.6V 1/4w back to back may be accomodated inside the plug and don´t need to remove nor a screw from the computer. And nonlinearity, buaaaaa, it may be significant when diodes start acting. Also can be tested a MOV, I don´t know how MOV´s can perform in audio, but a 6V+ varistor may work properly, at the price of (perhaps) increased self capacitance

[ITPhoenix]

Originally Posted by Simon B

ITPhoenix, this is the first real information that you've given about what it is you're trying to protect against. Apart from that we knew only that you wanted to keep a 600R input impedance.

I had dismissed Tesla coil performance monitoring as unlikely for this forum and was dithering between lightning strikes and accidental connection to the mains supply, slightly favouring the latter. I appreciate that actually specifying things tends to curtail the discussion but occasionally a bit of focus is nice.

Microphone input usually implies low impedance, low signal level, needing good low noise amplification etc etc.

Will the 'boosted' mic be the the only one you're using? In which case won't it work on the line level input?

Might I suggest that the best way for you to stop blowing soundcards is to stop using the mic inputs? They're never as good as a proper external mic preamp anyway.

Agreed, that is why I submitted three paragraphs explaining the application with an image of the basic AC zener clamp. It was never posted. It appears some users are experiencing posting problems at this time.

The mic is a mono, boosted zoom that is normally used in camcorders. This is fine at SPLs originating at a distance. It is also attached to an autonomous robot so there is no way of knowing when a close, loud noise source, such as someone screaming at it may occur, not to mention possible mechanical shock from a fall or a punch in the face.

Whether or not overvoltage/current caused the destruction has not been determined, but we know this is possible if enough energy is available. It still may have been caused by poorly written software. The app needs to pick up speech at a max distance of about 15 meters.

I do not see why I cannot use the line inputs, come to think of it, as long as the input signal is compatible. The only problem might be is the recognizer may only read mic input. It would work for training the recognizer as we can record in mono, but then a different channel would be used in operation. It is critical the same channel hardware be used in training and in the recognition mode.

Gas tubes are the best protection against HV discharges, from my limited research. A friend experienced a direct hit to the pole in front of his house. He was forced to go electronics shopping for almost all he had.....If he only had one on the service panel............

[Simon B]

Given the very variable signal levels you're getting, and the application, it would probably be worth having the mic feed an automatic gain control circuit. As the mic has an inbuilt pre-amp, I don't think you need worry about input impedance of the agc too much.

To go into the mic input, a simple resistive voltage divider, after the agc, should do the trick and provide definite protection to the soundcard input.

Hopefully your robot won't be suffering too many lightning strikes, or wandering up too many pylons. My thankfully limited experience of nearby lightning strikes is that gdts etc are of limited value - anything conductive takes a massive common mode hit. Further back, gdts will save stuff that otherwise would have fried. In my case the strike was to the telephone line, entering the house downstairs, where stuff died. Upstairs, ie out of the direct path to ground, stuff survived.

I've added a note about the posting problem to a relevant thread in Forum Problems