More on D-104 interfacing to boat anchor transmitters

[AB2EZ]

Hi

Following up on my earlier post... I verified the following... using on-air tests with several well known AMers:

Using an un-amplified D-104 with my KW-1 (1 Megohm grid leak resistor / 12AX7 microphone pre-amplifier... similar to many 1950's and 1960's boat anchor transmitters)... I tested the resulting audio with and without adding a capacitor across the microphone output.

The D-104 I was using for this experiment is one that has a ~1000pF series capacitor in its equivalent circuit.

Without the added capacitor across the microphone output, the low frequency response was poor (~3dB rolloff at ~159 Hz).

With an added 4700 pF capacitor, across the microphone output,  the low frequency rolloff was greatly improved (~3dB rolloff at 28 Hz). The parallel capacitor reduces the output by approximately 15 dB, because it forms a 1 : 5.7 voltage divider in conjunction with the 1000 pF series capacitance of the microphone. The KW-1 has more than enough available audio gain to bring the level back up when I switch in the capacitor.

If I do this same experiment with a different flavor of D-104... one that has a larger series capacitance (5000 pF for one that I have)... there is, as expected, no noticable improvement in the low end response... because that flavor of D-104 will already have a good low-end response when looking into an ampllifier with a 1 Megohm grid leak resistor.

Why would you use the approach of putting a 4700 pF capacitor across the output of your D-104, rather than increasing the value of the pre-amplifier grid leak resistor to 4.7 Megohm?

Well, in the case of the KW-1, changing the grid leak resistor would be very difficult.

In addition, constructing a stand-alone pre-amplfier for a non-amplified D-104 may be easier, for some types of pre-amplifier designs, if one does not require a pre-amplifier low-frequency input impedance that is higher than 1 Megohm

As a further note: if I place  a 4700 pF capacitor in series with a 10k ohm  resistor (instead of just a capacitor by itself) across the output of the microphone... I get the additional benefit of a high frequency boost of 6 dB per octave that starts at around 3400 Hz (3 dB point). If you don't want that boost... then leave out the resistor.

Best regards
Stu

[Steve - WB3HUZ]

Hmm. The D-104 already has a HF boost. Sounds like you are removing it with the capacitor and then adding it back in with the resistor. Just work the D-104 into the load recommended by Astatic for the desired LF response and skip all the rest. KISS.

I'll post freq response curves for the D-104 with difference loads later today.

[AB2EZ]

Steve

Hmmm...

I modeled the D104 analytically... I measured it, to verify the model... I performed an experiment to verify my predictions... I asked people to do on-air, A/B comparisons ... I took the time to carefully write my post... several very technically competent people have agreed with me, regarding my conclusions...

You were able to respond to my post, within minutes of the posting, with a negative critique.

Wow... either you must be a really fast thinker or you didn't take the time to read and understand what I was saying

Stu

[Steve - WB3HUZ]

It wasn't intended as a negative critique, just an alternative. Unless you are claiming yours is the ONLY correct way, I see room for differences. As to the time parameter, I'm quoting Astatic data. I'm willing to bet they have more time involved with the D-104 than you or I. Just a thought. 

From the D104 spec sheet.

[W2XR]

Hi Stu,

Out of curiousity, what kind of a test set-up and procedure do you employ for measuring the acoustic response of the microphone (acoustic input) vs. the measured electrical response (microphone electrical output)  into the various resistive and capacitive loads you have described? As you are aware, the frequency and amplitude of the acoustic input along with the test environment in which the microphone is being tested must be carefully controlled to get meaningful frequency response data of the microphone under test.

It would be especially difficult to get a source of acoustic energy below say 35 hz, as most reproducers (i.e. loudspeakers) really roll off in terms of usable acoustic output around that point.

Again, just curious; not tryingto challenge you at all on this!!! The improvement you have posted looks very impressive and those using D-104s would surely benefit from your findings.

73,

Bruce

[Steve - WB3HUZ]

It would be especially difficult to get a source of acoustic energy below say 35 hz

And even harder to get a truly anechoic environment below about 100 Hz, unless you have a pretty big room. Measurements in a nonanechoic environment are subject to variations and error.

[N8LGU]

An easy way to remove (reduce) room acoustics (echo) is to simply move up on the  D-104 whilst reducing the mic gain. Far too many lean back and bump up the gain and transmitt all of the room echo as well as the kids fighting and the tv playing. As an added bonus you will note that the bass response improves, also. Just be sure you don't "pop" the diaphragm with your breath sounds.

[AB2EZ]

Bruce

Hi! Good questions!!!

I performed two types of acoustic measurements: one objective, and one subjective

1. Objective measurement:

a. I used a white noise generator to drive a 15 band graphic equalizer, which drives an audio amplifier, which, in turn, drives a Bose bookshelf speaker in parallel with a Yamaha subwoofer... to produce an acoustic field in the room...which is fixed (I do not assume that it is white, either spectrally or spatially). The white noise generator is flat (white) between ~10 Hz and ~50 kHz (3dB down points)

b. I used a D-104 (the one with about 1000 pF of series capacitance) to drive a 10 megohm input impedance instrument amplifier. The output of the instrument amplifier was measured with an audio spectrum analyzer (computer sound card + Spectrum Laboratory application).

c. I held the D-104 in a fixed position... which was ~ 2 meters from the bookshelf speaker, and also approximately 2 meters from the subwoofer

d. I observed the low frequency spectrum (0-1000 Hz) of the output of the instrument amplifier, and adjusted the graphic equalizer to make the observed spectrum approximately flat in the range 0-1000 Hz.

Note that the spectrum analyzer sampling rate is 96kHz, and the number of FFT samples per tranform is 16384. Thus the resolution of the FFT is about 6 Hz. However, the white noise source I am using is only flat at frequencies above ~ 10 Hz

e. I switched the .0047 uF capacitor across the microphone output

f. I observed an ~10 dB reduction in the output of the instrument amplifier, but no change in the shape of the output spectrum (which remained ~flat from 0-1000 Hz)

g. I placed a 1 megohm resistor across the input of the instrument amplifier (to simulate the input resistance of a typical boat anchor's transmitter microphone preamplifier)

h. I, again, measured the spectrum of the output of the instrument amplifier... with and without the .0047 uF capacitor across the output of the D-104... this time adjusting the graphic equalizer to produce an approximately flat spectrum when the .0047 uF capacitor was in parallel with the microphone

i. When I switched out the .0047 uF capacitor, the spectrum above 200 Hz rose by about 10 dB, but the new spectrum showed a rolloff (approximately 6 dB/octave) that was about 10 dB down from the flat portion of the spectrum at 50Hz. This is consistent with a 3dB rolloff frequency of about 150 Hz.

2. Subjective measurement

I asked well-known AMers to listen to my signal with and without the .0047 uF capacitor in parallel with the D-104 output (driving the 1 meg-ohm input impedance preamplifier in my KW-1). They said that there was noticibly more low frequency content with the capacitor switched in.

[Bacon, WA3WDR]

A piezoelectric mike like a D-104 acts just like a moderate impedance resistive source in series with about 1000pF.  With a load of 1 meg to ground, it rolls of at 159 Hz as claimed.

A 4700 pF to ground makes a capacitive divider.  With a 1000 pF source and a 4700 pF shunt, the level is dropped about 15dB as claimed.

The very high frequency response is rolled off slightly, but the source resistance (assuming 2K) sees the capacitors in series (1000pF and 4700 pF in series looking like about 820pF), so the corner frequency for that is up around 97 KHz (no biggie).

The 1 meg sees the capacitors in parallel - with the 2K in series with the 1000pF, but that doesn't matter much.  4700pF in parallel with 1000 pF is 5700 pF, and then with the 1 meg load, that has a corner frequency of about 28 Hz as claimed.

Hence the parallel capacitor reduces the level and lowers the corner frequency.  Not all rigs have 15dB of extra available gain, so it might be easier to substitute a 10 meg resistor for the 1 meg and have the -3dB point down at 16 Hz.  I used a D-104 and two big resistors in series for about 15 megs of grid leak back in 1971, and I had 9 Hz seismic vibrations modulating my screen-modulated 6146.  I couldn't believe that!

Yes the D-104 has a very useful midrange peak, but I always added at least one stage of R-C boost starting around 750 Hz back in those days.  I had the audio unit set up to switch in as many as three of these stages of boost as desired.  I guess a resistor of about 47K in series with the 4700pF in this circuit would provide one stage of this boost.  Interesting.

[W2XR]

Hi Stu,

Thank you for the detailed overview as to the set-up and test procedure that you utilized for this testing. Nicely done and certainly well thought out!!

73,

Bruce

[Rob K2CU]

For those experimenting with Spice simulators, the following is a close model of the D104 element:

[AB2EZ]

Rob

Interesting.

It seems simple enough at low frequencies (below 500 Hz). Where did the higher frequecy (>500 Hz) model come from? I.e., was it based on the physical model of the element, including its packaging... or was it based on empirical measurements. Do you have a published reference I can obtain for where this circuit model came from.

As you know, below 500 Hz, this circuit boils down to a voltage source in series with a 1 nF capacitor. I found that, among my three D-104 microphone elements, one behaves as having a 5nF series capacitance. Do you have any information about what the differences are among D-104 microphone elements. I don't know if all three of my microphone elements are crystal types... so perhaps the larger capacitance element is a ceramic type.

[Jim, W5JO]

I found that, among my three D-104 microphone elements, one behaves as having a 5nF series capacitance.

Wonder if you know the history of the three microphones?  If the crystal has been jostled or subjected to undue heat-moisture, then the characteristics may have changed.  Wonder if the output of all three are equal or is there a difference?

[Rob K2CU]

I do a lot of work with spice simulations, especially of vintage equipment, and had decided it would be convenient to have a model of the D-104 as a signal source. I had seen the curves that Steve put in his post in the past,  and had made up a circuit to mimic the response and loading characterisitcs as closely as possible.

From work I had doen in electro-acoustics years ago, I know that somewhere there is an electro-mechanica mathmatical model of the mic.

I just received two of the Kobitone crystal mic elements that seem to be pretty close to the one in the D-104, from Mouser P/N 25LM024 @ $5.26 each. They are about the same diameter, but have a protective cover on the front with multiple large holes in it.  Check out the data sheet:

http://www.mouser.com/catalog/specsheets/KT-400025.pdf

I plan on setting up an A/B type compare of the elements, both on air (AM and SSB) and off.

[Todd, KA1KAQ]

Wonder if you know the history of the three microphones?  If the crystal has been jostled or subjected to undue heat-moisture, then the characteristics may have changed.  Wonder if the output of all three are equal or is there a difference?

I was wondering the same thing, Jim. I use a stock crystal D-104 on my KW-1, but the grid resistor has been bumped up to a 4.7 meg. It doesn't take a lot to make those crystal elements sound a bit off.

Why would you use the approach of putting a 4700 pF capacitor across the output of your D-104, rather than increasing the value of the pre-amplifier grid leak resistor to 4.7 Megohm?

Well, in the case of the KW-1, changing the grid leak resistor would be very difficult.

Nah, it's really not. A bit time-consuming removing the harness from the back of the audio deck and getting the deck in/out, but not terribly hard. Loosen the terminal screws, remove front door, remove the 4 screws holding the deck in plae and out she comes. 2 screws holding the small cover over the bottom and you're there.

I'm curious though, Stu - was your grid resistor still a 1 meg? I was under the impression from speaking with Bob that when he rebuilds these things, he changes some of the components to improve the audio. I sent him the list of changes I made to mine back when he got his first one, and I think he added a few more tricks along the way. So I would've guessed that resistor would be 4-5 megs?

There are a number of simple things that will improve the audio. It tends to make the hard core collectors cringe, but since Bob doesn't do actual 'accurate restorations' and instead rebuilds them to work, he's not burdened by such concerns.

Your approach of modifying the mic makes good sense for someone who wants to keep the transmitter stock, though.

[Rob K2CU]

HEy, before goign to the trouble of pulling out the audio deck of the KW-1, just pull the tube and measure the resistance to gnd.

[W1UJR]

HEy, before goign to the trouble of pulling out the audio deck of the KW-1, just pull the tube and measure the resistance to gnd.

Now you're thinking Rob, that is the easy way, measuring right at the tube socket!
Never thought of that, indeed a great idea, beats turning over the chassis.

Todd, your comments are right on, aside from a 1950s vintage broadcast unit, I can’t think of a simpler transmitter than the Collins KW-1 to service. Like its little brother the 30K, all the decks have terminal strips, making electrical and physical removal a snap. You could pull the deck, do the simple mod, and have it back in the cabinet before you finished a cup of coffee.

Stu has an interesting experiment, but I’m not certain that it is one which is repeatable in the real world.
With the D-104, you're not working with a new "off the shelf" mic like the RE-20.
Most D-104s have been around for a longggggggggg time, have suffered various forms of use and abuse, and are often not kept in the best of temp/humidity controlled spaces.
Someone pointed out about the difference in D-104s, each of the three D-104s which I use with my rigs sound different, some have more highs, some are "dead sounding", others very "peaky". I’ve matched each D-104 to the rig it sounds the best with, and keep it that way.

Frankly, my feeling is that the best audio chain is the simplest.
I know this is not a universal belief, but each stage of gain and processing adds it own level, however minute, of distortion.
I use my STOCK D104, into my STOCK Collins 30K-1, and consistently receive excellent audio reports.
Of course that might be due in part to my rigid use of Radio Shack oxygen free cable and gold connectors. 

75th Modulators Hard at Work at 1UJR


Has anyone found a replacement for the D-104 elements?
Now that Astatic has stopped production, I wonder if anyone has stepped in to fill the gap?

73 Bruce W1UJR
www.w1ujr.net

[Jim, W5JO]

[Has anyone found a replacement for the D-104 elements?
Now that Astatic has stopped production, I wonder if anyone has stepped in to fill the gap?

73 Bruce W1UJR
www.w1ujr.net

Omintronics owns Astatic, and not to many months ago I send the head of one of my D104s to them after dropping it on the concete follor in the shack.  The replaced the element and sent it back for about 45 dollars.  Go to the astatic website and send them an email.  They don't mention the service.

That is if you want a stock D 104.

[KF1Z]

Has anyone found a replacement for the D-104 elements?
Now that Astatic has stopped production, I wonder if anyone has stepped in to fill the gap?

73 Bruce W1UJR
www.w1ujr.net

I understand Mouser sells an element very similar for about $5.00.....

They have a ceramic element,  response  30-15khz  ,   8k impedance  1" diameter   for $4.43....

Really don't know how it compares to the D104, but heard some guys on the air discussing it....

Mouser part #25LM037

They have a crystal element with about the same specs, except 12k impedance.....


-------------------------------
Just noticed K2CU's post about the Kobitone elements...... those may have been the ones I had heard about.................

[Todd, KA1KAQ]

Todd, your comments are right on, aside from a 1950s vintage broadcast unit, I can’t think of a simpler transmitter than the Collins KW-1 to service. Like its little brother the 30K, all the decks have terminal strips, making electrical and physical removal a snap. You could pull the deck, do the simple mod, and have it back in the cabinet before you finished a cup of coffee.

You got a good look through the KW-1 in late April when we were tracking down the HV zorching issue. It's no empty box by any means but it's a dream to work on compared to, say...a 32V!    Nowthere's a nightmare waiting to happen. I've been avoiding the repairs to the V-2 because of the way everything is layered and buried in the LV and HV sections. Interesting that a company's lower end transmitter would be ten times harder to service that its top of the line model. Incentive to upgrade, perhaps?

I've been thinking that some winter weekend morning would be a good time to get on the air locally with someone who can record, and make comparison tests between D-104 heads. As well as showing the differences, it would be handy for weeding out the crapped out heads for later rebuilding or experimenting. I must have 8-10 D-104s, all crystal except for one ceramic. Also a crystal T-3 head. I'm thinking along the lines of numbering each one, making a brief audio test for recording each, then tracking the results to discover how similar the good heads sound and which ones are dogs. By using the known good head as a reference and a plain jane unamplified stand, it puts things on a more level playing field for (hopefully) clearer results.

We could even make it an 'experimenter's event' of sorts. Have a few guys with good recording capabilities like Bill 'IFR and Brent record the tests and have stations come on to test their Astatic mic heads for on-air performance. No techincal critiquing required, merely let your ears be the judge. It's certainly no trans-atlantic test, but it would liven up an otherwise quiet Saturday or Sunday morning.

Any takers?

[WU2D]

This is an old topic which has been buried for some time. I was talking with Mark, WA1QHQ, who is redoing yet another Ranger..

He brought up another D-104 question.

According to the all of the published stuff, the 12AX7 input resistor (grid to ground) which is normally 470K is pulled and should be replaced with something between 4.7 - 10Meg in order to get good low end response.

Several circuits show a series resistor from the element to the grid. This is usually 4.7K to 22 K Ohms.

Most of the circuits also include a 47pf to 220 pF capacitor either directly on the input connector hot to ground or directly across the 10M (after the 4.7K). I have assumed that this is an RF bypass capacitor.

Even this small value is going to cause quite a rolloff against 10M, perhaps starting between 300 Hz and 1000 Hz.

Here is the question.

Is this a bypass capacitor blunder  - or de-emphasis designed to work against the built-in pre-emphasis of the D-104's response and flatten it?

Mike WU2D
   

[AB2EZ]

Mike

What confuses a lot of people who use a D-104 is the fact that the equivalent circuit of the D-104 is (to very good approximation) an ideal voltage source in series with a capacitor whose value is around 1000 pF. Most people are used to thinking about signal sources that look like an ideal voltage source in series with a resistor... and this leads to a lot of bad intuition about how the preamplifier works in conjunction with the D-104.

The issue is simply whether the impedance looking into the grid of the preamplfier is forming a voltage divider in combination with the series capacitor of the D-104, or not.

First, the preamplifier series resistor does not impact on the frequency response, because its impedance (for a 4700 ohm resistor) is negligible compared to the impedance of the 1000 pF series capacitor in the microphone's equivalent circuit for frequencies that are below 10 kHz. [The impedance of a 4700 ohm resistor equals the impedance of the 1000 pF series capacitor at a frequency of 33,862 Hz]. The purpose of this resistor is to provide a finite (non-zero) grid leak resistance for the tube in the event that the preamplifier is used with a microphone that includes a transformer with a very low secondary resistance, or in the event that the preamplifier is used with a dynamic microphone whose coil has a very low reisistance.

Now, with respect to the r.f. bypass capacitor. Since its value is typically around 47 pF to 300 pF, it forms a capacitive voltage divider with the 1000 pF series capacitance of the D-104. Since the impedance of the bypass capacitor is significantly higher than the impedance of the 1000 pF series capacitor there is a small frequency-independent reduction in the voltage at the grid of the tube v. the voltage source in the microphone.

With respect to the grid leak resistor, the objective is to pick the grid leak resistor to be a high enough value so that its resistance will be larger than the impedance of the D-104 series capacitor for all (low) frequencies of interest. If we wish to meet this objective down to 50 Hz, then the grid leak resistor must have a value equal to the impedance of the 1000 pF series capacitor at 50 Hz. This implies a grid leak resistor whose value is 3.183 megohms, or greater.

Best regards
Stu

[WU2D]

Fantastic Stu,

This is what Mark was looking for.

Mike

[W4EWH]

OM,

IIRC, the D-104 was available with two elements. I have one, but there's no indication of which element is in it. Is there a way to tell without opening it up?

TIA.

73, Bill W1AC