D-104 Amplified MODS

[WBear2GCR]

Ok, so the stock D-104 amplfiied base is probably best bulldozed to use Tim's term and replaced with a simple FET follower... but I had one, and it was functional, and I really did not want to rip it out, and I was curious if it could be made to sound halfway decent. Keep in mind that the follower is unity gain and the FET follower, and the stock amp has beaucoup gain, like >20dB no problem!

The first image is the stock circuit.
It sounds like hell.
The transistors are biased the crappy way, but that probably keeps the batt draw low.

In a QSO with Stu, AB2EZ he mentioned that he had dropped a cap to ground in parallel with his mic at the input, and that this serves to tailor the response by effectively boosting the bass... so the 470pf in the stock schematic might be doing some of that, although it is in series with a 150k looking to ground...

The modified schematic shown has the response shown in the plot below. The yellow line is ouput B and the blue line is output C. The main diff is that C is lower in level, because of the resistor divider and has the greater roll-off at 1mHz., which might be good in some situations. The cap at C does that extra roll-off out of the passband.

The overall response is down a little at 10kHz. which is generally an advantage, plus the mic's response is NOT modelled here, meaning that there is a presence peak anyhow that goes out to ~5khz. The LF response is where the greatest diff is between this and the stock amplifier. As you can see this is flat to 100Hz. and rolls gently below that.

The key here is still the increased value of the first base resistor and the increased value of the input cap... I adjusted the values of the other two coupling caps, C2 and C3 to achieve the best looking response below 100Hz. C3 in the original is quite large, even though the input end of the circuit doesn't pass any bass! If you scale up C2 and C3 just a little bit, like double the value, the LF rolloff point drops a bunch, and by 1ufd you're looking at flat response down to like 10Hz. A bit much, imho.

C5 has been added to give a HF rolloff just starting at ~10khz or so... then that second cap on the "C" output (not in the original) gets rid of ultrasonic, RF even more.  It doesn't take much reduction in C5 to make the rolloff point 20kHz and not 10kHz.

It looks like C6 should be increased. Maybe more than the listed value. That input cap is in series with the capacitance of the crystal element, so it wants to reduce the capacitance, making the lows roll off more - the bigger that cap the less that effect happens. I did not find a good way to model the capacitance of the source in the SPICE simulation program... so it ain't there. The value I show is 10x the stock value.

C4 is there for possible RF control - it might have an effect in the audio band with the real capacitive crystal element, but I couldn't get the model to reflect that. Without that it has nil effect on the audio.

The pot in the stock unit is shown as fixed resistors. The gain is high in the stock unit, so you'd have to set the levels low anyhow. The output at "C" seems more reasonable in level, being only 8dB hotter than the input level.

I didn't get around to increasing the emitter resistor on the second transistor, but that seems like it might be a good idea - increases the linearity and drops the gain.

Bottom line is that I think you can use the stock amp with these mods and get decent audio.

_-_-WBear2GCR

[AB2EZ]

Bear

Hi!

A few suggestions:

1. If you want to model the series capacitance of the microphone element, just change the value of C6. The series capacitance of the microphone element is around 5000 pF on my D104 element. I think that some of the older D104's may have a series capacitance that is significantly lower (around 1000 pF). Since the existing 0.1uF capacitor, C6, in your schematic will be in series with the series capacitance of the microphone, just change the value of C6 to somewhere between 1000 pF and 5000 pf for the purpose of the simulation (or add another capacitor in series with C6).

2. In the original D104 circuit, the current flowing through the 470 pF capacitor bypasses the gain of the first transistor. For frequencies below 1000 Hz, it behaves as if it were going directly to ground... because it's impedance is much larger than 150k ohms. The 150k ohm resistor will boost the frequency response above around 2000 Hz, because it sets a lower bound on the impedance that is effectively connected between the base and ground. I.e., the impedance of the capacitor equals 150k ohms at 2257 Hz.

3. I think you will find that once you change C6 to reflect the 1000pf - 5000pf capacitor that is in series with the voltage source in the equivalent circuit of the microphone element, you will find it desirable to add some capacitance to ground (~ .005 uF) at the input of the amplifier... if you want to get down to very low frequencies in the overall response of the microphone-amplifier combination

4. With 150k ohms in the emitter of the first transistor, and assuming a current gain of 50, the resistive part of the impedance looking into the base of the first transistor is ~ 7.5 megohms (i.e., 50 x 150 k ohms). From that perspective, the large biasing resistor (R2) makes sense. However, I think you will find that if/when you place some capacitance across the input (as described above) it will not be necessary to have such a high impedance first stage design. The design you have now has such a small base bias current, that I wonder if it has good linearity. Reducing both the value of the 150k ohm emitter resistor and the value of the biasing resistor by a factor of 10 (after adding capacitance from base to ground) might improve the linearity

Best regards
Stu

[WBear2GCR]

Stu,

Re: 2) no voltage gain, it is a follower. Adding or removing the cap in the simulation did not materially change the response - I think it was for RF in the main, unless it works as you surmise - but that did not show up in the simulation so far (there may be a few more things to try in that regard).

Re: 3) changing the input cap or placing an additional series cap of 2000pf did not materially change the overall response in the simulation.

re: 4) seems surprisingly ok... in the simulation. In practice your idea may or may not be an improvement.

Thanks,

      _-_-bear

[w3jn]

Beware the THWONK coming from the coupling cap charging up; the PTT circuit switches the preamp on an off as I recall.

This transient can be a mod xformer killer.  Better to leave the amp powered all the time.

As an aside, I found another transient in my FT1000MP that I was concerned was gonna kill my linear.  I noticed a full power spike on the scope when I keyed the mike; I traced it down to the PTT switch clicking when it was keyed, this sound spike was then sent thru the audio chain and appeared as a full-power spike no matter what the power setting.  Solution was to use a different mike and a footswitch 

[WBear2GCR]

Yeah it does! Bwwwwammm!

I think one could use a transistor to switch the power to the circuit without any trouble.
Probably that would reduce the transient... especially if the transient is being caused by switch "bounce" or straight arcing....

Leaving the power on all the time will reduce the batt life... 

   _-_-bear

[Rob K2CU]

There are several things to consider her to model the circuit use correctly. You have to include the source impedance. I have read that the D-104 crystal cartridge should be modled with a 150K resistor in series with a 4700 pF capacitor.  The input stage base to emitter capacitor is not effectively connected to Gnd. On the contrary, the emitter follws the voltage (almost) onthe base, so that at frequencies well below Ft of the transistor, there will be very little current in the capacitor.  If your simulator privides current probe function, compare AC current in the cap as is and then with it connected to Gnd. Compare it with something like a 5K signal.You will see a difference by a factor of around 16x.  A minor point is to better stabilize the DC operatring point of the intput stage by adding a 33K resistor between the power supply and the junction of the collector and 1M resistor of the first stage. As expected, most of the 2% distortion comes from the input stage too. Try connecting the simulated mike input directly to the second stage and eliminate the first stage all together. You will want to up the 180 Ohm emitter resistor to say, 1K and the collector resistor to 10K. The distortion drops to negligible. The big power on thump comes from the fact the at the instant of turn on, the output transistor is not conducting and the collector will try the shoot up to 9V, but be limited by the output RC times constant to more like a 6V spike and then dro-p down as the transistor comes on. You can eliminate the thump by adding a 1K resistor in series with the power and a 100 uF cap to gnd on the circuit side of the resistor. 

[WBear2GCR]

There are several things to consider her to model the circuit use correctly. You have to include the source impedance. I have read that the D-104 crystal cartridge should be modled with a 150K resistor in series with a 4700 pF capacitor.

ya mon. I was thinking about adding some series resistance.

  The input stage base to emitter capacitor is not effectively connected to Gnd.

Don't think I said it was. I said I modeled it connected to ground to see what it looked like that way.

On the contrary, the emitter follws the voltage (almost) onthe base, so that at frequencies well below Ft of the transistor, there will be very little current in the capacitor.

So, what do you suggest its purpose may be?

  If your simulator privides current probe function, compare AC current in the cap as is and then with it connected to Gnd. Compare it with something like a 5K signal.You will see a difference by a factor of around 16x.

I would expect so, but it has a very minor effect on the overall response

  A minor point is to better stabilize the DC operatring point of the intput stage by adding a 33K resistor between the power supply and the junction of the collector and 1M resistor of the first stage. As expected, most of the 2% distortion comes from the input stage too. Try connecting the simulated mike input directly to the second stage and eliminate the first stage all together. You will want to up the 180 Ohm emitter resistor to say, 1K and the collector resistor to 10K. The distortion drops to negligible.

Of course, but this is/was an exploration of the stock circuitry - I'd just go with a JFET follower if I was doing all new.

The big power on thump comes from the fact the at the instant of turn on, the output transistor is not conducting and the collector will try the shoot up to 9V, but be limited by the output RC times constant to more like a 6V spike and then dro-p down as the transistor comes on. You can eliminate the thump by adding a 1K resistor in series with the power and a 100 uF cap to gnd on the circuit side of the resistor. 

Excellent idea! My comment on the switching tranistor had the idea of a time constant on the base in my mind, although I didn't say so... 

    _-_-WBear2GCR

[Rob K2CU]

Hi Bear,

As to the possible purpose of the 470pF cap....It could have been connected in error and was supposed to go to ground. But, usually, to be effective as an RF bypass, if that were the intent, you want some series R, even if only a 100 Ohms.  As connected, any hifh frequency energy will simply couple directly to the .1 uF coupling cap to the output stage base anyway. If anything, it is a potential detriment to the overall application of the circuit.  One final possibility is that they found some units input stage oscillating at some high frequency and needed to degerate the base. There apparently is no power supply by pass cap, and a 9 volt battery, like an electrolytic, probably looks more like an inductor at RF than a big cap. 

A single FET could do the job nicely, and you could use a circuit that supplies DC power from the rig, over the audio output line. For example:

[k4kyv]

I have a pre-WW2  UTC LS-series input transformer that is designed to match a crystal mic to a 500-ohm balanced line.  Never have tried it with the D-104, since it is physically larger and a lot heavier than the mic itself.  I think the stated impedance ratio is 100k to 500 ohms.  It would be interesting to see what the D-104 response would be, since a 100k grid leak severely limits the low frequency response.  Transformer coupling might not affect it the same way.  UTC gives the frequency response of the transformer as 30~ to 15,000~.

[Steve - WB3HUZ]

Make the gate resistor something like 4.7 or 10 Meg. Below is a neat circuit that also includes some tone controls.



http://www.hamelectronics.com/ham/k1deu/pages/ham/audio/mic_preamp.jpg

[WBear2GCR]

Rob,

What is the "tline" doing there??
As in, what is its function/purpose/design??

        _-_-WBear2GCR

[Rob K2CU]

it represented the mic cord to distinguish between what's in mic vs what's in radio.

[WBear2GCR]

I presume then that you think there actually is something in my radio??

 

        _-_-WBear2GCR

[Rob K2CU]

Merklejammers