Attenuators for measurement

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I know many on this forum use sound-card analyzers that have a very limited input voltage range. The obvious solution for measuring higher voltages is to use an attenuator at the analyzer input. However, many measurements are done in two parts (e.g. SNR) and the noise from the attenuator may throw off the results. This impacted some users of AverLAB as well. I wrote an article about it and after a long delay, audioXpress published the article in their March 2019 edition. audioXpress March 2019 Is Now Available. Time for Audio Test and Measurements! | audioXpress

In fact, the entire issue is a good read for anyone on this forum.

Also, I put together an Excel calculator sheet for designing a high voltage attenuator that one can download from the Avermetrics website. Avermetrics | Download

Enjoy
 
Perhaps not much use if the article itself isn't linked.
Unfortunately, the audioXpress owns the copyright and keeps the articles behind their registration page. I would think that many here already subscribe to this magazine. It is one of the few magazines that deliver any technical information to the audio industry (unless you consider articles about expensive uni-directional cables to be technical). Their sister publication, VoiceCoil, is even more targeted. It is just for speaker and driver designers.
 
Might want to include a cell for source impedance, and sometimes folks use T-networks.
I had briefly thought about source impedance but there are not many high voltage audio devices that aren't power amplifiers and modern ones tend to have relatively low source impedances (typically <1 ohm). That is why I assume the Thevinin equivalent impedance to be just the parallel combination of the two resistors in the L-pad.

T and Pi attenuators are pretty common in RF designs where you need to match impedances. In my 15 years of selling audio analyzers, I never came across a situation where such a topology was needed. That doesn't mean there aren't such situations though. Do you find you need to use them frequently?
 
I used to use vector and scalar network analyzers in my RF design days. In fact, I had briefly worked at the Network Measurements Division of HP that made those analyzers. When I first got into audio, I would think about using something like a 3577 to measure the s-parameters of audio devices. I wanted to look at the S22 of an amplifier and the S11 of drivers with the idea of making the crossover also serve as a matching network. However, once I was in audio, I no longer had access to RF test gear and I was traveling too much to entertain any serious bench work.
 
I used to use vector and scalar network analyzers in my RF design days. In fact, I had briefly worked at the Network Measurements Division of HP that made those analyzers. When I first got into audio, I would think about using something like a 3577 to measure the s-parameters of audio devices. I wanted to look at the S22 of an amplifier and the S11 of drivers with the idea of making the crossover also serve as a matching network. However, once I was in audio, I no longer had access to RF test gear and I was traveling too much to entertain any serious bench work.

The S-parameter test set of the HP3577a has a lower limit of 100kHz -- have been toying with the idea of making one suitable for audio from the November 1984 issue of HP Journal: http://www.hpl.hp.com/hpjournal/pdfs/IssuePDFs/1984-11.pdf
 
Member
Joined 2004
Paid Member
The directional bridges are not simple. I had a 10 KHz directional coupler from a long time ago and may still have the manual for it. If I can find it I'll share it. You may be more successful using active circuits than transformers at audio frequencies. What do you hope to see?
 
Unfortunately, the audioXpress owns the copyright and keeps the articles behind their registration page. I would think that many here already subscribe to this magazine. It is one of the few magazines that deliver any technical information to the audio industry (unless you consider articles about expensive uni-directional cables to be technical). Their sister publication, VoiceCoil, is even more targeted. It is just for speaker and driver designers.
Pity you teased us but can´t provide the article :)
I am a qualified Voice Coil subscriber which is from the same Publishing House but even that does not open the doors for me, even less for somebody less related.
Agree that it´s a serious Magazine.

Maybe you can link to or upload previous work which led to your article without being it in its present form.

That said, congratulations on your publishing Tech work in a recognized Trade Magazine :)
 
JMFahey, the good news is that there is nothing special about the article. It is just practical advice on building L-pad or U-pad attenuators. If measuring a power amp, your best bet is to design the load as your attenuator. However, if you need to test the unloaded output you need a high-impedance attenuator. The things you need to worry about include:

* The noise produced by a high impedance attenuator will impact measurement accuracy (e.g. SNR and THD+N)
* Different heating on the series and shunt elements can produce changes in attenuation that can also throw off measurements.
* Resistor choices factoring in noise, temp stability, cost and availability

The bottom line is that 0.6 watt thin film resistors are easily available and relatively cheap. They can be arranged in series-parallel networks to achieve just about any desired value while keeping the heat dissipation per element relatively low.

The spreadsheet (link in the first post) has all the formulas for designing an attenuator that meets your requirements for attenuation, measurement accuracy and heat dissipation. The spreadsheet is protected which means you cannot see the formula. The article itself contains the specific formula used.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.