Making soundcards out appear as 150ohm load to preamp

[Category 5]

I need to build an attenuator that will reduce the output of my soundcard by about 50dB, and at the same time make the output appear as a 150 ohm load to the preamp.

My thinking is that the only way to get accurate frequency, noise and distortion measurements is by way of such a device.

I can crank my soundcards output to -43dB and get the level I need, but it is very close to the noise floor at this point so my noise measurements are useless. Also, since the souncards output isn't 150ohms I believe the preamp isn't giving frequency and distortion measurements that would correlate with it's performance while using a 150ohm mic.

Any simple way to build such an attenuator and load dummy into a balanced 1/4" to XLR cable so I can use the cable to get reasonably accurate measurements?

Thanks as always!
Shane

 

[gyraf]

Use a standard attenuator

50 dB is some 300x

You want ~150R output impedance, so use this value resistor to ground.

Use 300x this value in series with the signal ~47K Ohm to attenuate 300x

So an attenuator 47K/150 Ohms should work..

Jakob E.

 

[Category 5]

So is this right...



Pin1--------------------------------------------------Pin1

47k
Pin2-------------------/\/\/\/\-----------------------Pin2 <-----Source
\
/
\ 150ohm
/
Pin3-------------------/\/\/\/\-----------------------Pin3
47k


Or does the 150 go from pin2 to pin1 and pin3 gets no resistor at all?

Also, is there a need to put a 5 or 6k resistor between pins 2 and 3 on the source side to simulate a higher impedance to the soundcard? I am ultimately tring to get sine sweeps and test tones from my sound card into my DIY preamps, but I need the preamps to perform as though the signals were actual microphone sources.

Thanks Gyraf!

Shane

 

[gyraf]

Ah - I wasn't thinking balanced.

For balanced, you will need two 22KOhm resistors in series with + and -, and 150R between XLR in pins 2 and 3.

Pin1 - ground - goes directly.

This gives 44K load to the output of the soundcard, and some 150 Ohms source impedance to the mic input.

Jakob E.

 

[Category 5]

Thanks again! That's what I thought, but I didn't know the values. I remember seeing a site somewhere that had the formula (X ohms required for Y dB reduction).

I appreciate all the help!

Shane

 

[PRR]

This will adapt a high level unbalanced line source, like hi-fi or sound card, to a balanced mike input.

It isn't really balanced. This should not matter to any proper bal-mike input. If you must be balanced, you need to add a 150:150 transformer, and accept the response and distortion issues that go with transformers.

Loss is quoted for UN-loaded condition. Obviously if you hang a true 150Ω load on it, the voltage will drop another 6dB, or -54.4dB. With typical 1.5K or higher "low Z mike input" the added loss is 1dB or less.

 

[alk509]

Umm... Assuming this computer is at a studio, why not just use a DI box? Isn't that what they're for? :?

Peace,
Al.

 

[NewYorkDave]

This or this would work very well. But PRR's resistive U-pad is by far the most affordable solution.

 

[tk@halmi]

Wouldn't the resistive U-pad add noise to the setup? The 22k resistors are in series with the signal.

 

[Category 5]

[quote author="tk@halmi"]Wouldn't the resistive U-pad add noise to the setup? The 22k resistors are in series with the signal.[/quote]

If I used another transformer wouldn't that almost guarantee a change in distortion and frequency character? These are what I am trying to measure at the preamp.

Also, PRRs example looks good for unbalanced to balanced connections, but my DA has balanced outs so I think it is more cut and dry, like Gyrafs example. Am I wrong?

What's this about resistors adding noise? One of the reasons I am doing it this way is to get accurate noise specs too. My DA will attenuate the signal 50dB by itself, but the noise floor remains constant. If I attenuate after the DA stage I will be attenuating the noise too, so the SN and Dynamic range won't be skewed so much in my measurements.

Shane

 

[PRR]

The 10K:150Ω transformers are helpful but not enough to get 50dB loss. They are 8:1 or 18.2dB voltage loss; you need another 30dB+ of loss somewhere. I might split this as 15dB before and after, to isolate transformer flaws from source and load.

Yes, resistors have noise. Everything does. Get over it. In this case, for artificial test reasons, we MUST reduce the available signal, and work closer to the universal noise. But the alternative, using the resistors in the sound card to reduce level and then amplify it again, is worse. Assuming zero (or small) source noise, and a good resistor (not something toasty pulled from a tired Fender Champ), the output noise of this adaptor will be very-very close to the theoretical noise of a 149.5Ω resistance.

A good DA is probably good. A bad DA (even one that works fine for guitar recording) will confuse things. Be sure it will handle the signal level, and has enough loss: signal generators and sound cards can be a lot hotter than guitars.

If you really have a balanced source (Pro-type sound card), omit the ground strap on the source side.

 

[NewYorkDave]

[quote author="Category 5"]What's this about resistors adding noise? [/quote]

Very short answer: All resistors possess an irreducible degree of noise related to the value of resistance, independent of the composition of the resistor. This is called "Johnson" or thermal noise.

The expression for Johnson noise is:

(Vn)2 = 4KTRB V2/Hz

where

Vn is the noise voltage (V)
K is Boltzmann's constant
(1.38 X 1023 J/°K)
T is the temperature in degrees Kelvin (°K)
R is the resistance in ohms (W)
B is the bandwidth in hertz (Hz)

Then there is 1/f or "flicker" or "excess" noise, which is dependent on the composition of the resistor. For example, metal film resistors give less 1/f noise than carbon comp resistors. The level of 1/f noise is proportional to the voltage across the resistor. This can be expressed as microvolts of noise per frequency decade, per volt across the resistor. Here are some values, according to the book The Art of Electronics, for various resistor types:

Carbon comp .1 to 3 uV
carbon film .05 to .3
metal film .02 to .2
wirewound .01 to .2

A web search should yield hours of reading material on the subject.

In practical terms, though, it's usually not something we need to worry about except when very high values of resistance are involved.

 

[tk@halmi]

[quote author="Category 5"] If I attenuate after the DA stage I will be attenuating the noise too, so the SN and Dynamic range won't be skewed so much in my measurements.
Shane[/quote]

You are right. The attenuator attenuates everything including noise. The 22k resistor adds about 2uV of noise at 10KHz and gets added to the output noise of the sound card. Then it all gets attenuated.
As Dave hinted: the output noise of the sound card is most likely higher than what the pad adds.

 

[Category 5]

[quote author="PRR"]
If you really have a balanced source (Pro-type sound card), omit the ground strap on the source side.[/quote]

It is an RME multiface. It is balanced on the source.

So basically 22k in line with pins 2 and 3. 150R shunt across the 2 and 3, and leave pin 1 floating? (why float pin1?)

The noise floor of the output is about -98dB so I am not worried about a tiny amoutn from the resistors. I imagine the preamps are more around -80dB themselves.

Shane

 

[alk509]

No need to float it, just connect the grounds (pin 1) together, or just on one end for shielding. If your feeding a truely balanced input, it will amplify the difference between pins 2 & 3, not the signal referenced to ground.

I think :?

Peace,
Al.

 

[Bryson]

And turn off the 48v phantom pwr.