Are Tantalum Capacitors Really Needed?

Thread Starter

SamR

Joined Mar 19, 2019
5,031
There is no need to switch the battery negative connection. Also, a DPST switch on the back of a pot is very rare. S1 can be SPST.
Yep, planning to simply switch the Pos lead from a battery holder. Even finding pots with any switch is becoming difficult and already gave up on that one. Now it's all encoders (with push switch) and digipot chips.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Consider replacing R1 with a fixed resistor and trimpot in series. This lets you adjust the static current through the output transistors to prevent excessive power dissipation.
Interesting idea. What values would you suggest for the components?
 

AnalogKid

Joined Aug 1, 2013
10,987
Interesting idea. What values would you suggest for the components?
Depends on what part values you have or can get. Since someone thinks 180 ohms is a good value, a starting point could be 150 ohms fixed plus a 50 ohm pot. Or 120 or 130 ohms fixed and a 100 ohm pot.

Start with the pot adjusted to its minimum value. This probably will decrease the transistor bias voltages so low that the output has noticeable crossover distortion. This is the point of minimum (zero) static current through the output transistor pair. As you increase the trimpot resistance, the voltage between the two output transistor bases will increase until both transistors begin to conduct at the same time. At this point, the crossover distortion should be gone. Increasing the resistance further will increase the static current going through the output pair at all times. More current means less distortion, but a lot more heat.

To increase the stability of the operating point, mount the two diodes so they are in contact with the output transistor cases (physical contact, but electrically isolated). The diode Vf will track the transistor Vbe as they change with temperature.

ak
 
Last edited:

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Thanks AK! I'll spend a bit of time experimenting with it. Started breadboarding and will do some measurements and experiment a bit before proto boarding. Had to order some 50Ω trimmers and got some 20Ω while I was at it. Didn't know they even came in less than 100Ω. I can play with the 100Ω ones I have until they arrive.
 

AnalogKid

Joined Aug 1, 2013
10,987
Basically, you have two parallel strings. One has two diodes and a resistor, and one has two transistor base-emitter junctions. The voltage across both strings is identical. The idea is that the two diode Vf's are approximately equal to the two transistor Vbe's, and sorta-kinda track the transistor Vbe changes with temperature. Even though all junctions are silicon, a signal diode Vf tends to be a bit less than a transistor Vbe, Thus, without R1, there is an operating region where neither transistor is conducting. The result is crossover distortion. This is the reason for the R1, to increase the voltage across the diode string until it starts to try to push the two bases far enough apart to bring on conduction at all points in the output voltage range. It is a balancing act.

Why go to all of this work? Because the temperature coefficient of Vbe is negative. The relationship between base voltage and collector current is a curve on a plot, not a sharp corner at exactly 0.6 V. As a transistor heats up, the Vbe necessary for any given collector current decreases. If the base is biased through a fixed resistor string, the base current will increase as Vbe decreases, further increasing the collector current and heating, decreasing Vbe even more, etc. This loop is called thermal runaway, and was the cause of many output transistor deaths in the early days of transistor power amplifier design. The solution was (and still is) making the transistor bias point track the temperature changes; hence, the diodes.

ak
 

Audioguru again

Joined Oct 21, 2019
6,673
I have not used tantalum capacitors since 1975 (47 years ago) when most blew up.

Most speakers are 8 ohms. Those weak little output transistors have a maximum allowed output current of only 100mA but when playing loudly in that circuit they are zapped with 4V/8 ohms= 500mA. Even 32 ohm headphones will zap the output transistors.

The tiny 47uF output capacitor is also much too small. It cuts mid-frequencies and all bass frequencies.
1000uF (21 times more) is usually used to feed an 8 ohms speaker. 125uF can feed frequencies above 40Hz to a 32 ohms earphone.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
You do know that this is to amplify the output from a crystal radio instead of using a piezoelectric headset? The acronym GIGO comes to mind... Yes, it's an old design from an old book "Radio and Electronics Cookbook" from the Radio Society of Great Britian circa 2001 and I'm sure the designs are even older than that. I'll be comparing it against an LM386 which is a much simpler circuit. But, hey I'm retired and need something to do...
 

Audioguru again

Joined Oct 21, 2019
6,673
A "crystal radio" needs an audio amplifier with a very high input resistance that the original amplifier circuit, my fixed circuit and an LM386 do not have.

A crystal radio sounds very muffled because the AM radio station frequencies are so close together that to avoid adjacent channel interference, radios cut frequencies above 3kHz which is 3 octaves. Frequencies above 3kHz contain the "consonant" sounds (intelligence) of speech.

A crystal radio was used 95 or 100 years ago when there was only one or two radio stations in your area. Today a crystal radio will probably pickup a few or more local stations at The Same Time!
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
OK, not too bad. Built as shown except substituting electrolytics for tantalums. First swept from 10Hz to 2kHz then lowered the input voltage from the 4V default to 10mV and swept again adjusting the input pot for loudness. Shown are screen shots from @100Hz and 1.5kHz (about my range of hearing sans hearing aids. Don't see any clipping or waveform distortion except at low freq and that could be a noise component. I've found that using my scope it helps to run the input signal through an op-amp buffer to remove the noise in my environment. Which I didn't do this time.
IMG_1047[1].JPGSDS2104X Plus_PNG_108.pngSDS2104X Plus_PNG_107.png
 
Last edited:

Thread Starter

SamR

Joined Mar 19, 2019
5,031
A crystal radio was used 95 or 100 years ago when there was only one or two radio stations in your area. Today a crystal radio will probably pickup a few or more local stations at The Same Time!
Well I have a dead zone to the east (Atlantic ocean) closest AM radio station to the south is ~30 miles, ~60 miles to the north and west. I live in a bit of a dead zone as most have gone to FM.
 

Audioguru again

Joined Oct 21, 2019
6,673
Sam, you show the amplifier built on a breadboard that has rows of contacts, long wires all over the place and unshielded wires for the input that are picking up all the interference around it.
Your scope shows a 384kHz RF signal drown in noise, maybe because the input level is extremely low.
The other scope photo shows a low level of 18.6kHz pulses added to very low level sinewave of 154Hz.

Since the speaker is part of the circuit then it should be fed through a shielded audio cable.
 

Thread Starter

SamR

Joined Mar 19, 2019
5,031
Yep, I've become a believer in not using breadboards but they are still useful for quick builds. Laying it all out on a breadboard helps me with planning spatial arrangements for protoboarding. The actual speaker is a rear speaker from an old Panasonic early surround sound "stereo" tower. Surpisingly it was designed for twinlead zip cord and not shielded cable although, now that you've mentioned it, I should fix that. There was no noticeable audio distortion (other than my wife hollering "are you having fun" while it was sweeping like a siren) and I haven't opened the bookcase sized enclosure to see just what is inside. I left it running for a while and did a quick "touch test" of the components and could barely detect any warmth from Q3 & D2. I should probably hook up a thermocouple and see just what temp they are...
 
Top