Tee attenuator - Which input impedance?

For the measurement of a phono peramp, I intend to feed an inverse RIAA signal via soundcard to the moving coil input.
Therefore, I would like to use a simple T-attenuator delivering -60dB.
The output impedance of the soundcard is 100 Ohms, Picoscope 600 Ohms. My moving coil cartridge 10 Ohms, the preamp input e.g. 100 Ohms

I can design the attenuator for e.g. 600Ohms at the input - and this is my question - Shall the input impedance be 600 Ohms or far above, for example 10k or 100k.
Similar question related to the output impedance of the attenuator - 100Ohms or 10Ohms (factor 10 below input impedance of the preamp)

Many thanks & best wishes
M
 
Others may offer helpful and more specific critique, but here are my thoughts on attenuators in general.

The "T" configuration is not necessary, since you want a very low output impedance of the attenuator (10 ohms). I would use a simple voltage divider. The series resistance (called R1) goes from the output of the sound card to the output of the attenuator. The shunt resistance (called R2) goes from the output of the attenuator to ground. The output voltage of the attenuator is across R2. For an approximate gain of -60dB (which is 0.001 in V/V) you would make R1 equal to 10K ohms and R2 equal to 10 ohms. In general, the loss of an attenuator is not simply the ratio of R2/R1, but is R2/(R1+R2). In this case, since R1>>R2, we have R1+R2 approximately equal to R1, hence, the gain is approximately R2/R1. The impedance looking back into the attenuator output is given by R2//(R1+RS) where RS is the impedance driving the attenuator input and the symbol // means "in parallel with"). Because the output resistance of your soundcard is 100 ohms, the attenuator would have an output resistance of 10 // (10K + 100) or about 10 ohms.

By the way, R1 // R2 is equal to 1/((1/R1)+(1/R2)).

Tom
 
That's a good point using a L-pad rather than a T because of the low output impedance.

Regarding the value of the impedance, there are 2 approaches:
1) setting the input impedance to the same value of the output impedance of the source in order to reduce reflection
2) setting the input impedance at least 10 times higher

I am still puzzled ....
 
To reduce reflection the source and load impedances need to match the cable. Audio is rarely setup this way as reflection is not really an issue (RF devices can be fried by reflected energy, and you have to treat a cable as a transmission line at high frequencies to get any range at all).
 
Member
Joined 2004
Paid Member
For your application model it as a DC voltage divider. The Voltage source (your sound card) has a series 100 Ohm resistor on its output. You are checking a phono preamp which has a 40 dB difference in gain between 20 Hz and 20 KHz. The source impedance going to the phono preamp should be very low. I would like 1 Ohm. That may be too much attenuation in practice. You could get 60 dB of attenuation with a 1 Ohm resistor and an 901 Ohm in series with the 100 Ohm internal resistor. If you can source 1V unloaded with the attenuator you have 1 mV which is a reasonable level for testing a phono preamp (at 1 KHz.) However the RIAA correction from the source will need to go to 10V at 20 KHz which may not be possible. Change the terminating resistor to 10 Ohms and you have approx. 40 dB. (Pretty small error). You can have errors from parasitic and stray capacitance but the impedances are so low it would be very small.
What software will you use? Has it been validated for this? You don't want to be lost because of a bug but most likely it will just work, its a pretty simple problem for a computer.
 
I was thinking that the sound card and the attenuator are intended to model the moving coil cartridge. So why not make the output impedance of the attenuator match the output impedance of the moving coil cartridge?

If I'm designing an attenuator for best performance I like to get as low an output resistance as possible. For example, if the attenuator can deliver the required signal level at a lower impedance the amplifier's signal-to-noise ratio can improve (for some amplifiers and within limits). So I agree with that principle.

But in this case, the goal is to characterize the amplifier when driven by a source with a 10-ohm impedance. Honestly I don't think 10 ohms or 1 ohm will make much difference but I'm curious if I'm missing something. Of course, the situation becomes more complex (pun intended) if we include the effects of capacitance and inductance.

Tom
 
Member
Joined 2004
Paid Member
Moving coil cartridges range from about 3 Ohms (Ortofon) to around 40 Ohms with the "high output" cartridges. There are questions around the effects of loading etc. on the transducer but I would submit you want to start with the preamp having the least interaction with the transducer.
10 Ohms would be a noise source of .4 nV/rtHz or 56 nV across the audio band. The only way to get noise that low on a preamp is with a transformer.
The MC cartridges also have an inductance but those I have measured were really small. Emulating a moving magnet cartridge is a different story however.
 
I have now designed the output impedance to 100 Ohms, equal to the input impedance of the phono preamp with an attenuation of 80dB
Going lower to approximately 10 Ohms would be only achievable with low parallel resistance adding noise and channel inbalance.
There is still the option to increase the input impedance of the phono preamp. However, subsonic corner frequency will change with the impedance.
My MOTU M4 has a high output up to 16dB balanced, or 9dB on RCA. With 24bits, this offers a wide range for measurement.
Audacity brings an inverse RIAA already. Some appreciate REW. I could also import my iRIAA excel spreadsheet to ARTA. Usually I try ARTA first, because I am more used to.

For reliable results, I believe that an iterative procedure is necessary. Measurement with digital iRIAA, simulation, passive iRIAA simulation, test record, simulation with cartridge and cable...

Best
M
 
Member
Joined 2004
Paid Member
Going lower to approximately 10 Ohms would be only achievable with low parallel resistance adding noise and channel inbalance.
I think there is some confusion here. 10 Ohms source will be lower noise than 100 Ohms. I do not understand the channel balance issue. If you are using a 1meg resistor in series with the 100 Ohm resistor to get 80 dB of attenuation the stray capacitance in parallel with the 1 Meg will be significant and alter the attenuation at higher frequencies.
 
You are right. However, when building the attenuator with parts from stock, my collection of resistors around 4Ohms is rather limited, and any deviation of values will result in higher channel inbalance than with higher resistance - a non-technical reason :)

The T- attenuator works quite well. 50dB für MM, 80dB for MC. Linearity up to 10kHz, with -0.7dB at 20kHz.
Input impedance is set to 100 Ohms for MC and 47k for MM. Using the Pico with 600 Ohms or 100Ohms from the MOTU does not affect the frequency response with the MC input..
 
Member
Joined 2004
Paid Member
"Input impedance is set to 100 Ohms for MC and 47k for MM" Input impedance or source impedance? Moving magnet cartridges are large inductors that usually resonate with the input C of the preamp. 47K source will just be really noisy and not realistic. If you have access to an old MM cartridge just use that between your attenuator and the preamp input.