D-104 Crazy Curve

[ki4nr]

Was playing around with my audacity program. Check radios and whatever I could test. Always wanted to see how the D-104 looked. So I got one of my polk audio floor speaker and wired it all up for a little testing.

Stuck the mic on a stand in the middle of the living room with the polk speaker about 2 feet away and ran some white noise thru the system. Did an analyse run on Audacity and this is what it looked like. I played around with my other stereo speakers and the curve was always the same.

Those Astatic engineers back in the day knew what they were doing. The curve is just what a transmitter needs to sound good. Now my D-104 is running into a MPF102 FET buffer at 10 meg of load. The bass is great right to 90 hz and rolls off like DSP , same on the top big peak and rolls off like a SSB Crystal filter. Well no duh , it a crystal audio transducer.

The response is so crazy and sharp you think you were looking at a Crystal lattice filter in a SSB transmitter. I aways wondered why the D-104 aways drive my radio like crazy. One thing all the charts show the treble peak around 3k,  My mic has it's peak at 2650hz and no matter what I did in testing it was always the same. The big dip is at 1950hz  Very interesting to say the least.

[KD6VXI]

Interesting.

I wonder.  Was the mic facing the speaker or turned 90 degrees from the speaker.

I'd be very interested in the response both ways.

--Shane
WP2ASS / ex KD6VXI

[MikeKE0ZUinkcmo]

Pretty standard, Heres Astatic's published response curve,



and the element's response curves with different terminating load resistances.

[ki4nr]

It's was directly facing the speaker.

[N1BCG]

You'd need a laboratory transducer and careful sound baffling placement for any meaningful microphone response measurements. Speaker placement in a room (ie, center vs corner) will wildly affect the perceived low frequency response.

The graph posted by KE0ZU is pretty much it but that assumes you have a like-new crystal element and that the padding within the microphone enclosure is clean and in good condition.

[MikeKE0ZUinkcmo]

Actually the data from my test setup, shown HERE, is typically pretty close to that shown in Astatic's 5Meg curve, since I just used the meter's 10 Meg input resistance as the load.   Do note however my graph is in 5dB steps not ten as per Astatic's.







I've done several elements, and 90% or better followed their curve pretty closely.   Obviously if one wanted to be with in a tenth of a dB, a much better "chamber" or open air would be required.      

[MikeKE0ZUinkcmo]

OOPS

[W3SLK]

Mike, I was looking at your set-up that you have for repair. I have several that have low or now output. How do you determine if the crystal is good or bad? I did a crap shoot on one, I figured nothing ventured, nothing gained. When I reassembled it, it was no better than when I tore it down.

[K9MB]

The D104 is a cool looking mic. I have several of them.
The thing about the classic curve that is being discussed is that crystals that are good are getting more rare.
Also, even though they exhibit desirable characteristics, they are very sensitive to load variations, which limits their advantages IMO.
I am attaching an article published in QEX by Jim Tonne for a much more flexible and adjustable speech spectral response.
Jim likes electret elements that are flat and much less load dependent and the response is produced by using a simple active filter with a bump in the 2-3kHz area to provide the enhanced readability to phone modulation and the curve can be varied in center and the amount of bump.
In the old days, the D104 response was connected to speech amps that had no other methods of enhancement.
Jim gives an analog solution that works well.
Of course, modern numeric digital filters can also provide a brick wall above the desirable high frequency and represents the ultimate solution.
The simplicity of Jim’s designs and ease of construction are still a great way to mimic this response curve and his later circuits provide a complete speech chain to provide a complete solution.
73, Mike K9MB

http://tonnesoftware.com/downloads/Tonne-QEX-Sep-Oct-2018.pdf

[MikeKE0ZUinkcmo]

I've found that typically their output level is in the -35 to -40 dBV range at the peak.   

It seems to be more or less the luck of the draw so far as what you might get when done.   I've repaired 5 so far and out of those I had two with lower than what was called normal output levels.   One not too bad and one with very little output.   

To verify the output levels you need an SPL meter because the standard audio test Signal Pressure Level  is 94dB SPL, AT the plane of the microphone's diaphragm.