diyAudio (
-   Everything Else (
-   -   How to protect sound card during amp measurements? (

okapi 10th August 2008 04:51 PM

How to protect sound card during amp measurements?
I am attempting to make my first measurements (frequency response etc.) with my laptop sound card from some amps i've built.

i want to reduce the amp output (high voltage) to line level so i can run it back to my sound card's line-in for evaluation.

the amps output is also attached to a resistor load (8 or 4 ohm)

i have a 100X oscilloscope probe and i was wondering if i could just use the probe to scale down the amps signal (amp max output voltage peak to peak is about 40 volts) or should i take some other measures to protect the sound card?

for example, will turn on surges be a problem and should i therefore be concerned about the order in which i hook things up/turn things on?

additionally, is the DC offset (25-75mV) of the amp potentially harmful to the sound card? Could i run the amp output through my ac coupled oscilloscope remove any DC?


bwaslo 11th August 2008 03:46 AM

If you know the input impedance of your soundcard (usually around 10k ohms), you can just use a big series resistor from the amp output to the soundcard's line input center conductor. Something like 100k ohms in series (1/4Watt or more) should work fine for measuring 40V types signals. For most soundcards, the inputs can handle about 2V pk linearly (damage level would be several times that). If the signal is too weak to drive the card, drop the resistor value. Ive done this a LOT.

You'll also need to connect grounds between amp and soundcard. But be careful NOT to connect the soundcard's ground to the power amp's [-] terminal unless you are SURE that it is ground -- with a bridged amp both output terminals are hot. Best ground to use is probably the amp's input connector ground shield.

The scope may be a way to do it, too, but you'd be putting more circuitry in line -- it would make distortion measurements pretty meaningless. Also be careful, some scopes can output enough voltage to blow a soundcard input.

The big series resistor is probably a safer way to do it. That's actually what a scope does anyway (the probe is just a resistor driving low capacitance cable with a capacitive trim shunt to flatten response).

okapi 11th August 2008 04:21 AM

thanks for the help.

i think i've got some appropriate resistors on hand i can try. i just need to get it in the ballpark as i can adjust the soundcard input gain to optimize the amplitude.

i am using a LM4780 chip in parallel configuration so the speaker (-) terminal should be ground but i will use the amp's rca input as ground (signal ground/line output ground) as you suggest.

i won't use the scope but this bring up another question. if i did run it through the scope couldn't i separately assess the scope's contribution to the distortion and then subtract that out later. i was planning on using this approach for the sound card itself by measuring distortion when i ran the line output to the line input, as well as measuring the distortion of the line output when driving the amp - a possibly more difficult load than the line input of the sound card.

thanks so much for you help. frying my work laptop sound card, or perhaps worse, would not have been good.

Ron E 11th August 2008 03:01 PM

I "blew" the inputs on my IBM T21 Laptop with just the headphone out. Some soundcard outputs are more powerful than others. I kept upping the output and the S/N kept getting better and better and the Speaker Workshop traces cleaner and cleaner until something broke and now the inputs actually still work, but are so noisy as to be unusable. Not that you need a cautionary tale....

bwaslo 12th August 2008 12:07 PM


You can't subtract out distortion like that, it is not an additive thing, but happens from a cascade of stages (one stage is input to the next, rather than all the outputs getting combined together). Even if there weren't "distortions of distortions" effects like that, subtraction could only work if what you were measuring had absolutely flat response and no group delay -- subtract one signal from another that is phase shifted or of wrong amplitude and they don't cancel anymore.

It is a good idea to be wise to what RonE said -- start with large resistors and work your way down if needed. But with a 100k series resistor, you can only get half a milliamp of current from a 40V signal, that shouldn't be able to hurt any soundcard input. Don't go probing inside tube amps (hundreds of volts) with that resistor, though!

You can put a cap in series with that resistor to block DC voltages if needed. Something like 1uF (at 50V or more, film type is best) should do it. Even with the cap though, the resistor has to be scaled to handle any DC level + signal peak level, since at the instant you connect it, the cap will pass current until it charges up.

BTW, be sure and put the series resistor at the SENSING end of the cable (not at the soundcard end). A shielded cable is a capacitor, particularly if it is long, and you don't want it directly across the power amplifier (some amps can get upset from that).

Conrad Hoffman 12th August 2008 01:26 PM

I've had good success with test systems at work using a large resistor (depends on the possible voltage you might apply) and back-to-back zeners. Size the resistor so the maximum possible voltage won't exceed the zener current rating. Obviously the zener voltage should be low enough to protect the sound card. I also couple it with a large film cap as suggested above. No doubt this could increase the distortion a bit if you don't get it sized right, but it's way better than a blown sound card.

okapi 13th August 2008 01:52 PM

thanks for all the feedback.

it seems like i have all the info required to move forward.

i will report my slow progress here.

progress will be slow because i need to order a few parts, i am coding the procedures to run the data acquisition and analysis (using igor pro from wavemetrics) myself, and the math behind the signal generation and analysis is also new to me.

thanks again for your help.

jackinnj 13th August 2008 03:04 PM

2 Attachment(s)
This looks a bit like the input to my Boonton THD% Analyzer (you have the luxury of a bipolar supply in the analyzer, however). If the sound card looks like a 10K load the 4.7uF cap should give you flat response down to a few Hz.

If there is any DC offset on the amp it will show up on the other side of the capacitor -- the two diodes will limit any DC -- the input to your sound card is probably capacitatively coupled

okapi 26th August 2008 03:46 AM

this weekend i put together a sound card protection circuit.

thanks for help!

i left out the capacitor because i would like to use the device during phase measurements and because jackinnj said: 1. that DC will still get across the cap and 2. my sound card is probably capacitatively coupled anyhow.

however, i did not implement the circuit that jackinnj suggested because it was over my head. i did not understand how that circuit would protect against DC or the function of the 5Vcc ( i assume they are related).

i incorporated a shunt resistor and used a different value series resistor. i hope this change is inconsequential?

i tested my circuit, and the diodes clip at the expected voltages. my only concern is that the protection won't last very long as the diodes get hot (and ring) relatively quickly. i guess the solution would be to use diodes with higher current ratings.

thanks again. ca... schematic.png

here is a pic of the guts: ca...ction guts.JPG

here is a pic of the exterior: Ca...n Exterior.JPG

Jan Dupont 26th August 2008 04:46 AM

I'm working on an active version (using Opamps) at the moment, and see some problems with your circuit....

Your scope would have an input impedance at app. 1 MOhm.
Depending on the switch settings of your box, the input impedance could be as low as just a little above 1K.

Measuring 40V with 1K input impedance means a current load of 40mA from the circuit you have under measurement. This is ok for Power Amplifier output stages, however if you try to connect your probe to a more sensitive circuit like the VAS stage you may end up with some blown low power transistors....

All times are GMT. The time now is 04:43 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2