How to make a one-transistor bjt amplifer work over 3.3 to 4.2 V Vout=1-V

[Salome2]

Thanks again for everyone's help!

 What I do need to solve, for Granny's sake, is how to recognize the signature waveform(attached) picked up by a mic near the intercom/Front Door Opening Sw/Doorbell while not being triggered by noise![/size] 

The scope waveform shows the bursts of  doorbell sounds as seen at the output of the 1-transistor mic preamp.

Each press of the doorbell at street level gives two "dongs" per quick push(or repeats if the street sw is held down.)

Open for ideas..what would be your way to do this(without physically modifying the intercom inside her apt, or mounting a wireless doorbell switch on the street door ring panel.

The idea here is to mount a little battery-powered black box(with a mic and a wireless doorchime push switch dongle inside that detects the weak existing doorbell sound and so activates the loud battery powered chime boxes placed in her bedroom and kitchen).

A most important part of a solution proposed would be to achieve the longest battery life and not respond to other noises..and to chime three times over 10-secs so Granny will get the idea someone's at the door!

Should I  post this call for project help as a new topic?
----------------------------------------------------------------------------------------------------------
I No Longer Need Help with this mic amp  which was needed for the mic pickup part of my design. The last version of my posted two-transistor temperature compensated voltage-divider CE circuit posted below works fine.!

----------------------------------------------------------------------------------------------------------------------------------------------------------------------

Need help with my doorbell extender design circuit:   (see: microcontrollers/help-grandma-hear-her-doorbell)

I can easily get a single transistor CE NPN amplifier to work perfect at 5V without distortion and ~25 uA operating current, but I need this BJT amp to work over the voltage range of  Li-ion 3.3 to 4.2V battery power.

Design Goals:
BJT 1-transistor design only. Please, not interested in using any micropower Op-Amps
Vout >= 1 V PP output undistorted (no clipping). Output to drive only ADC input of MCU operating over same battery voltage range.
Freq of operation 2KHz
Needs to work over a room temperature range (18 to 30 C).
Quiescent  current <= 25uA
Input Impedance ~100k
AC coupled input from 12K output impedance electlet capacitor mic.
Operational Voltage Power Supply Range 3.4 to 4.2V
Voltage Gain constant as set in a range of =10 to 25
Allow possibility of design to have up  to 4 transistors total to make circuit work.

Right now my 5V one-transistor design(works perfect at 25 Deg C room temperature):
(1) Transistor BC547
Collector resistor 100K
Emitter resistor    10k
Voltage Divider biasing of base:  1-Meg to 5V, 130K(Std 5% value selected for V-collector=2.38V quiescent) to Gnd.

Thanks in advance for help!

[moffy]

Because of the low supply and varying voltages it is best to compensate the BJT biasing network by taking into account the varying Vbe drop. The following circuit is an example of that, unfortunately it needs a second BJT in the biasing network. The input signal is 0.2Vp-p and the output almost 2Vp-p.

[Salome2]

Moffy, thanks!  How does your circuit work to keep  Vcoll ~Vbat/2 with Vbat  3.4V to 4.2V Vdd?

Operation over a wide temperature range is an issue, and Quiescent collector voltage kept to ~ 1/2 of battery voltage is most important.
The collector voltage can vary so long as a max 1-V PP undistorted output is possible.
Circuit is only required to operate over room temperature range to 18 to 30 Deg C.

I now have now added the base current mirror compensation transistor(diode) to the circuit.

[PCB.Wiz]

The following circuit is an example of that, unfortunately it needs a second BJT in the biasing network.

That's fine, many vendors now offer single package SOT363 etc matched pair transistors, for exactly this sort of task.  eg BCM857B BCM847B

[moffy]

The following circuit is an example of that, unfortunately it needs a second BJT in the biasing network.

That's fine, many vendors now offer single package SOT363 etc matched pair transistors, for exactly this sort of task.  eg BCM857B

Probably the BCM847 since it's NPN, older times when one wanted only cheap matching, and everything was through hole you would select two transistors for matching then super glue the flat faces of the TO-92 packages together for reasonable thermal contact.

[magic]

Your circuit already consumes more than 25μA, because you have a few μA in the base voltage divider

Isn't it good enough as it is? Tweak it to get 1.5V across Rc and 0.15V across Re at 3.3V supply. Vce is 1.65V. Seems close enough?

If supply increases by 0.9V, base (and emitter) voltage increases ~0.1V, collector voltage decreases ~1V (but also increases 0.9V). Vce drops to maybe 1.45V. Still good enough for 1Vpp, I guess?

The reason this works is because your base divider ratio is similar to the amplifier's DC gain (Rc/Re).

edit

It will no longer be so when the divider is adjusted to work at 3.3V supply. But maybe things could still be made to work?

If not, you could add a PNP emitter follower in front of the NPN. This will eliminate most DC offset between the input and Re, so whatever bias voltage you set on the PNP base will also be the voltage at Re. Set it to VCC/21 and you're done.

[Salome2]

Thanks PCB.Wiz, but rather than order the matched pair,  I will assume(with some testing using a soldering iron tip to heat the transistor pair to very warm) a superglued transistor pair for temperature comp is good enough for my simple requirement to  keep output capability of 1-V PP undistorted collector swing over room temperature operating range.

Thanks Magic, can you do a simulation on LTSPICE (or show me how) to verify this?

The very real problem is to have B-E bias voltage correct at Vbat=3.4V to 4.2V over temp range.

I am confused about your idea to add a PNP emitter follower (without temp compensation) is supposed to work? This idea also adds more current drain and will ace the ~25 uA goal?

Vcc/21 

[Salome2]

I realize that I can add a Microchip MCP1640 low-power boost convertor(see attached datasheet), but this adds  ~400 uA min to battery drain, where without this chip total current drain incl. extra Low Power PIC16F18456 MCU <1mA(yet untested).

In PWM mode the 1640 efficiency for 5V out with 1mA load and 3.4V battery Vin is only 60 to 70 percent(see attached), so another ~200uA of battery drain must be added to this total? So total battery drain at 3.4V totals ~800 to 900uA?

I only have the 1640D so PWM mode is always the only mode of operation.

The (yet to be verified by testing) total battery current without the boost chip adds up to about 240uA.

But with the whole applecart rolling at 5V, everything is so easy to design!


Is there a single transistor solution yet to be found to get the mic preamp to work over an unregulated battery voltage range?

[PGPG]

Is there a single transistor solution yet to be found to get the mic preamp to work over an unregulated battery voltage range?

Experiment with negative feedback from collector to base.
So emiter directly to GND.
Input signal through 100k to base (you will have 100k input impedance).
1M from collector to base - to have gain around 10x.
At first try with 10k in collector.

[Salome2]

PGPG.. I searched fairly thoroughly(googled) for design info on this CE biasing with neg. feedback but couldn't find anything!

I don't quite get how things will add up to functionality with adding a 100K series resistor to base and 47K C-B.

I don't think your solutions will solve the problem of temp compensation while keeping Vcol at BatV/2 over battery V range.

Does anyone know how to LTSpice this?

[PGPG]

I searched fairly thoroughly(googled) for design info on this CE biasing with neg. feedback but couldn't find anything!

I didn't suggested to search but to experiment (simulating) .

and 47K C-B.

How did you translated my 1M into your 47K?
And I'm not saying it is ready solution, but something that you can consider.
Negative feedback have a chance of compensating partially supply voltage changes and temperature changes.
You can have different feedback for DC (polarization) and different for AC.
If you want it be more DC stable you can add resistor in emiter that then is shorted for AC with capacitor (or 'partially' shorted by adding a smaller resistor in serie with this capacitor).

The most variable value is transistor hfe. It would be big problem if you want to design circuit that will be mass-produced with no individual calibrations but as you are doing only one piece you can just select resistors to exact transistor you use.

You can also use negative feedback only for polarization. So having as in your circuit 10k in emitter and 100k in collector and not shorting emitter resistor with capacitor you than can put input signal directly at base. To block negative feedback working at AC you can divide collector-base resistor into two and block midpoint with capacitor to GND.
Just experiment then think then experiment and so on.

[Zero999]

Design Goals:
BJT 1-transistor design only. Please, not interested in using any micropower Op-Amps

Why not?

Vout >= 1 V PP output undistorted (no clipping). Output to drive only ADC input of MCU operating over same battery voltage range.

Can't you configure your ADC so it doesn't need any amplification?

Freq of operation 2KHz

Why? Seems a bit low for audio.

Needs to work over a room temperature range (18 to 30 C).
Quiescent  current <= 25uA
Input Impedance ~100k
AC coupled input from 12K output impedance electlet capacitor mic.

Why <25µA? You do know the electret mic will draw an order of magnitude more current than that.

[PCB.Wiz]

...
I don't think your solutions will solve the problem of temp compensation while keeping Vcol at BatV/2 over battery V range.
Does anyone know how to LTSpice this?

Why the obsession with BatV/2 ?
This is an AC signal into an ADC, you can use the average to remove any relative DC shifts, plus your ADC may be using an internal reference anyway.

Here is a classic jellybean NPN collector feedback amp that seems to meet your signal specs. Distortion is not high spec Audio, but fine for a doorbell.
The natural feedback keeps the output  collector voltage relatively constant over battery range.  Easily manages 1vp-p, gain is ~ 22x  Zin is ~ 150k
If you are driving an ADC, you might want to add a parallel low pass capacitor as this is a relatively high output impedance  220pF~470pF C-E seems ok

[Salome2]

Thanks for  your help, Zero999
I choose to not use a micropower op-amp because I would need to wait for ordered parts and the op-amp would likely need a PCB instead of through-hole which I can make do with immediately. Gramma can't wait!

The simulated doorbell tone signal to be picked up by the mic is, <60mV P-P at the mic output but varies, of course, with the distance of the mic to the intercom box, I haven't considered that a 12-bit ADC could use the 1.024V FVR  and have enough resolution to easily identify the bell sound's signature waveform points.

The freq. of operation is underestimated. and the BW is better described as the audio spectrum, and so would have to have a much wider bandwidth than 2KHz to display the waveform without distortion: The belltone heard by the mic is a 700-mS period burst of sound that my scope shows is made up of a 2-mSec period waveform that repeats every 2-mSec and the bell waveform has distinctive and consistent peak amplitudes every 2-mSec.
Please see the attached amplified mic output scope snapshot of the bell ring. Unamplified, my scope cannot display clearly the waveform to work with. My 200-MHz scope is basically useless to see clearly<50mV/div signals because of unavoidable HF noise below 100mV/div swamping out the signal of interest.

Obsession:  BatV/2 gives the collector quiescent point the largest swing in voltage output to the A2D without clipping. The ADC is using  BatV as the reference.

The mic only uses 40uA of current, and while the MCU is in sleep mode the A2D uses only about 60uA continuous running, but my choice of ~25uA is based the goal of using the least amount of battery current, but, you are correct in assuming my choice is somewhat arbitrary, but was chosen to be comparable to the current drain of a micropower op-amp.

[Salome2]

PCB.Wiz, your circuit uses meg-ohms values that makes this circuit very sensitive to noise and doesn't seem to be temperature compensated, so would it work over the Li-ion voltage range and temperature range of interest?
Using such high meg resistors also makes probe loading errors to be reckoned with when testing this circuit.
I would want the collector voltage to vary over the battery range so as to achieve the widest voltage swing without clipping.
Vcol=BatV/2 is therefor my goal.

Thanks for reminding me to add a filter(tank) capacitor at the A2D input.

[PCB.Wiz]

PCB.Wiz, your circuit uses meg-ohms values that makes this circuit very sensitive to noise and doesn't seem to be temperature compensated, so would it work over the Li-ion voltage range and temperature range of interest?

If you want uA, you will have high component values, just keep the design compact.
It changes about 5% over your temp spec.

Using such high meg resistors also makes probe loading errors to be reckoned with when testing this circuit.

Yes, you would need a 10M scope probe.

I would want the collector voltage to vary over the battery range so as to achieve the widest voltage swing without clipping.
Vcol=BatV/2 is therefor my goal.

You asked for 1V pp, which this does fine, and such simple circuits are not going to clip symmetrically, or even manage rail-rail.
Usually, a designer forgets any BatV/2 and instead look for roughly symmetric clipping. 

This is a MIC amp for a doorbell, feeding a MCU, you are not making a Medical Imaging system  here 

[Salome2]

PCB.Wiz..Well said..thanks.
My circuit now achieves near Vbat/2 for Vcol over 3.4V - 4.2 V battery  range.

Even a 10x scope probe or DVM probe will certainly require some math to reveal the actual voltages of your amp circuit.

My circuit (attached) achieves Av=10(this seems to be optimal) and Vcol=1.8 at V=3.4V and Vcol=2.38V at 4.2V for nearly optimal voltage swing.
Scope or DVM probing without math.
Current use by preamp ~26uA
Current use by Mic  ~40uA
Quite stable Vcol over desired temp range.

[magic]

You could try replacing the dummy transistor with 1N4148 for ultimate space saving.

[Salome2]

Unless you are one of Cinderella's ugly sisters, the current idea of using a mirror should be to have a match  of what's to be mirrored.

[PCB.Wiz]

You could try replacing the dummy transistor with 1N4148 for ultimate space saving.

Ultimate space saving is to use a dual NPN matched pair.

[PCB.Wiz]

PCB.Wiz..Well said..thanks.
My circuit now achieves near Vbat/2 for Vcol over 3.4V - 4.2 V battery  range.

If you really want Vbat/2, and symmetric clipping, here is a symmetric NPN/PNP design

The Nexperia BC847BPN SOT363 45 V, 100 mA NPN/PNP general-purpose transistor pair data says
• Closely matched current gain
• Reduces number of components and board space

but does not define how closely matched. The curves suggest appx HFE of 305 and 290

[PGPG]

Even a 10x scope probe or DVM probe will certainly require some math to reveal the actual voltages of your amp circuit.

Are you sure of this 'certainly' in your sentence?
Thanks to feedback output resistance is much smaller than 220k resistor used in collector and you are interested only in voltage at collector and not at base.
I don't expect any real problems with seeing it with 10M probe.

[Salome2]

PGPG. Sorry..My apology.
I confused your circuit with PCB.Wiz's circuit. Your circuit would not have any problem with probing.

[PGPG]

I confused your circuit with PCB.Wiz's circuit. Your circuit would not have any problem with probing.

But I was speaking about PCB.Wiz's circuit. He used 220k in collector.

[Salome2]

PGPG, right again, measuring VBE voltage would be a problem, not VCE and that scared me enough to not like this circuit. Besides, it is not compensated for temp and I have a  buncha BC547 but no BC847.  Also, on hand, I have very few meg res values to play with.

Nonetheless, the probe problem was my bad invention and thanks for pointing this out to me.