Measuring impedances at audio frequencies

and

have

9kHz

through

voltage

this

apparently

for

met

Maybe you can find some obsoleted US Navy sonar equipment.

I wonder if you could shoot a short current pulse into the speaker and digitize the voltage waveform. Then, pretending the current is a unit impulse, fourier the voltage and voila you have impedance.

You'd likely have to signal average to get enough resolution and to reduce noise.

Somebody has done the software that turns a sound card into an audio-range VNA; you might google for that.

John

Sounds to me like you're trying to make this too difficult.

I don't know how accurate you need it to be, but for example, I've used a 1kohm resistor in series with an oscillator output to provide a near constant current source. Measure the voltage across the transducer and scale it. That'll give you the magnitude of the impedance.

Phase angle can be derived by comparing the relative phase of the voltage across the transducer and the sig gen.

Graham

Hi Graham,

Nope, what you're describing is what I had in mind... just that, realistically, I'm probably not going to have the patience to measure more than a few dozen frequency points over the span of some minutes -- which is accurate enough for my purposes anywhere -- whereas a VNA can measure 10,000 frequency points in about a second. :-)

"Joel Kolstad">

** Better stick with your original idea as posted.

By using separate voltage and current values for each test frequency - errors due to variations in amplifier gain ( response ) and generator output level are eliminated. Bound to be quite significant ones over the 10 Hz to

50 kHz range otherwise.

Also - you will be working with large enough values to overcome errors due to electrical and ambient acoustic noise ( speakers are microphones too).

It is also possible to determine phase fairly accurately angle by using a scope in X-Y mode.

....... Phil

Given that the impedance curve likely won't be that complicated a couple of dozen points is probably fine.

Graham

Joel Kolstad wrote: I'm probably not going to have the patience to measure

The procedure is pretty straight forward, you just need a PC with appropriate software. I-e.

and soundcard plus an audio amp and a 10R/5W resistor. The software is free for 30 days but continues working also after that period. It has a few bugs, so always check the probability of the charts.

--
ciao Ban
Apricale, Italy

There are dynamic signal analyzers, that do from zero to 100kHz or 10MHz.

Rene

*** Free account sponsored by SecureIX.com *** *** Encrypt your Internet usage with a free VPN account from ***

Is money an issue?

GPIB control of basic instruments using Labview software? Labview is not all that cheap however.

Steve

The HP 3577A network analyzer goes down to 5 Hz (up to 200 Mhz) and has the math capabilities that let you display impedance magnitude and phase directly. I've used it for just this application. I noticed you can get them on eBay for around 3K. Plus about $200.00 shipping. Those suckers are heavy!

Regards, Glenn Dixon

Very nice, thanks for the pointer!

For this particular application, yeah, but I've been Ridley's offering before -- I attended one of his mini-courses some years ago at a power supply design seminar put on by On Semiconductor -- and it does look pretty nice if you're doing commercial work.

Nice, I'll keep that reference in mind. Sounds like a great VNA to have around for most HF and some VHF RF work as well? Hmmm... at least with the

35677A S parameter test set...

John Larkin skrev:

Using a maximum length sequence as signal and correlation to get the impulse response will give you both low crestfactor and averaging at the same time

we had one at uni called MLSSA, Maximum Length Sequence Sound Analyser or something like that.

should be simple to to do, play a maximum length sequence through an amplifier and a series resistor, record before and after the resistor using left and right channel. process recorded wav file in matlab or somthing like that.

-Lasse

I loved the '3577I I had when I was still employed. Be advised however, that there isn't enough memory onboard to do full 8-term error correction with the S-parameter test set. HP has software to accomplish this, but it wasn't a stand alone solution. You can do full one-port reflection calibration and/or a through transmission cal, however.

I had the S-parameter box and seldom used it, of course on the other side of the shielded room an HP8510C resided. [g]

If you are not opposed to a bit of experimenting and some construction, take a look at the N2PK VNA. Since I lost "my" '3577 when I retired, it's the next best thing. See:

It only works down to 50 KHz as designed, but I believe that this is simply a limitation of the broadband transformers used. Some single-ended to differential amps driving the detector inputs should take it lower in frequency.

Dunno if this will do all you need, but from the summary and some reports it seems pretty good. Look up the Wallin test jig on the site.

- YD.

--
Remove HAT if replying by mail.

Thanks, that's a nice project too; thanks for the link.

The HP 3577A is definitely going on the "post-winning-the-Lottery" list!

and

have

9kHz Analog Devices recently came with a chip ( AD5933 ) that may do the trick, a DDS from 0-100 kHz and a complex FFT of the incoming signal.

Wim