Dynamic mic pre-amp into sound card

[systemloc]

Hi, I've been playing around with the microphone input on my Creative SoundBlaster Live 24-bit card and a low impedence, unbalanced dynamic microphone designed for a radio tranciever. The input level is extremely low, and in my readings I have found that cheap condenser mics require bias voltage, and put out a much larger voltage signal. I also found that this particular soundcard outputs +5V for bias on a separate line from the input of the speaker, using all three pins of a standard stereo jack. I found one person who cleverly designed the circuit below to use the bias voltage to power a simple Class A amplifier to boost the signal of a dynamic mic to something the sound card will pick up:

**broken link removed**
(Picture Source)

I'm trying to understand exactly how this amp works, and I'm trying to learn the concept of impedance matching. I'm also trying to build and use this circuit in practical application.

My first problem is that the BC547B isn't readily available at the RadioShack down the street. Rather than order it, I'd like to substitute it and learn more about the parameters that govern transistor selection. Here's the **broken link removed**, as well as the substitute I selected: the 2N4401. This substitution yielded a circuit that doesn't really provide sufficient gain. I'm guessing by looking at the datasheets that the 2N4401 was designed to function at higher currents and doesn't have a high enough hFE at low base current.

Other possible substitutions are the 2N3903, and the 2N2222. I'm guessing by the spec that the 2N222 would be the best choice as it is designed to work at roughly the same currents as the BC546, correct?

For the sake of completeness, here is the data sheet for the MPS2222A, which is Radio Shack's equivelent to the BC546, but not available to me currently.

My next question is in relation to how this circuit works. My understanding is that the signal first enters R1. R1 is present because the base impedance of the transistor is low. R1 increases the impedance presented to the mic and prevents it from being loaded down. C1 prevents any DC current from passing back to the microphone. R2 is the bias resistor which adds voltage to the incoming signal so that the signal is no longer alternating but always above ground voltage. This is characteristic of a Class A amp. I don't understand how to calculate the proper amount of bias here, and I read that this affects the amount of amplification. I'm guessing that with my higher power 2N4401, I would have to decrease the value of R2 to increase bias voltage to get higher gain out of the transistor, but this would also waste current from the +5V source. As for the output, C2 prevents DC from flowing out the output. Since this is electrolytic, though, I would think that it would be reverse biased during the negative part of the audio wave and not work correctly. Wouldn't this be better as a thin film cap? C3 is just a filter to provide cleaner voltage from Vin, and R3 is to knock down the voltage to the appropriate level to power the transistor. I'm not sure how the value for R3 is chosen, either. Using ohm's law, if the transistor is putting up no resistance, 1mA is flowing through R3. So, I'm guessing that 1mA is desired on the emitter, so the R3 value was chosen accordingly. Would this need to be modified for use with the 2N4401?

Finally, how would you determine the output impedance? I'm guessing that this would be a characteristic of the transistor. I know that the soundcard input is high impedance. My understanding is that a low impedance output into a high impedance input offers optimum voltage transfer, however high-impedance in and out offers optimum current transfer. Which provides more gain to the soundcard? I read that with modern equipment, low-Z into a high-Z input is normal and desireable.

Thank you very much for any input!!!
TJ Harrell

 

[Dr.EM]

I expect a 2N3904 would work quite well, and naturally 549 etc. As far as I know the 2N2222 is noisy and probably isn't desirable in this application.

In this case, you are after voltage/signal transfer, so a low output impedance is desired.

 

[kchriste]

I don't understand how to calculate the proper amount of bias here, and I read that this affects the amount of amplification. I'm guessing that with my higher power 2N4401, I would have to decrease the value of R2 to increase bias voltage to get higher gain out of the transistor, but this would also waste current from the +5V source.

What you want is for the collector of Q1 to be at 1/2 the supply voltage or 2.5Vdc. Choosing the value of R2 requires you to know the Beta (gain) of the Q1 and the value of R3. We know what R3 is, but the transistors Beta will vary between different ones of the same part number. The easiest way is just to sub different resistors until you see 2.5V at the collector of Q1.
An simple formula if you know the Beta of Q1:
R2 = Beta * R3

Finally, how would you determine the output impedance?

On a commom emitter amplifier such as this, it is aprox equal to R3

C2 prevents DC from flowing out the output. Since this is electrolytic, though, I would think that it would be reverse biased during the negative part of the audio wave and not work correctly.

No, because the input to the audio card should already go to ground through a pulldown resistor etc.... There should be 2.5V collector of Q1 so an electrolytic is OK.

 

[audioguru]

I simulated it with a "typical" 2N3904 that has spec's about the same as a BC547B. A 2N4401 has spec's that are very broad, all over the map.

Single transistor circuits with high gain like this one produce horrible distortion at high levels. This circuit clips off the bottom because the base current is too high:

 

[audioguru]

Here is the same circuit but with the value of the base bias resistor increased to 1M so the transistor doesn't have too much base current. The distortion compresses the top so much that the voltage gain can't be counted.

 

[hawk2eye]

mike amplifier

< I'm trying to understand exactly how this amp works,
< and I'm trying to learn the concept of impedance
< matching. I'm also trying to build and use this circuit in < practical application.
.....................................................................................
IMPEDANCE MATCHING
The math indicates that if R_signal_source = R load
then the maximum Power, (NOT voltage ) is transferred to the load.

An important consideration for a signal amplifier S/N
……………………………………………………….
Using R_signal_source = R load in a power supply design would be wrong , WRONG.
In a power supply R_out should be close to R_out =0 ohm to improve efficiency.
………………………………………………………………………………………..
Operation of the circuit shown:
Q1 is connected in the same configuration as an IC op-amp.

C1 blocks +Vbe from biasing the mike

R1 acts as the input resistor of th e “op-amp”

R2 acts as the feedback resistor of the “op-amp”

R2/R1 gives gain of x40 as mentioned in < Picture Source >
note that gain varies with the transistor chosen and
distortion is much worse than any op amp

C2 blocks +Vce from the load connected to the output
…………………………………………………………………………………………
In fact Q1 could be replaced by an op-amp IF < RING , Bias from SB > can provide suitable +Vcc.
- for op -amp:
R2 value changed to 39K , to maintain gain of R2/R1 = x40

-C3 can be used to insure low ripple on the +Vcc connection

-R3 is not used by the IC op-amp,

hawk2eye

 

[Nigel Goodwin]

< I'm trying to understand exactly how this amp works,
< and I'm trying to learn the concept of impedance
< matching. I'm also trying to build and use this circuit in < practical application.
.....................................................................................
IMPEDANCE MATCHING
The math indicates that if R_signal_source = R load
then the maximum Power, (NOT voltage ) is transferred to the load.

Except you don't want impedance matching, it's a BAD idea for audio (or for the majority of connections) - by definition impedance matching is very inefficient, only 50% at the very best.

You want to feed a low impedance source into a higher impedance input, for maximum VOLTAGE transfer, you should be looking for 5 to 10 times higher impedance on the input.

 

[systemloc]

After reading some replies and thinking about this, I've come up with a few more thoughts/questions.

As Dr. EM and Nigel Goodwin both point out, the hi-Z input of the sound card should be fed with a low-Z output from this circuit to achieve optimum voltage (signal) transfer. The goal of this circuit is to increase the gain of the signal going into the soundcard. Since the soundcard has a hi-Z input, we aren't really trying to feed more current, but higher voltage peaks than the mic outputs. A transistor is a current device, though. My understanding is that the gain it is providing would be current, and it would not increase the voltage peaks. Thus, this circuit really isn't amplifying voltage, which is what I think I need.

So, can someone clarify whether I'm looking for voltage or current gain? In the case of this circuit, is it providing current gain as I guessed, and what specifications should I look at on the data sheet to determine the gain? If I am indeed looking for voltage gain, shouldn't I be using a FET?

Another thought is that kchriste says that R3 sets output impedance. R3 is 4.7K, not what I would call low impedance! In order to provide the low-Z output, I would need a circuit with an R3 of 600 ohms or so, right? I'm not sure of how the selection of R3 relates to Q1. Can I choose any R3 I like as long as I adjust R2 according to the formula:
R2 = Beta * R3

On the subject of impedance, is impedance similar in concept as internal resistance? For example, a mic outputting a 5V peak-to-peak sine continuous sine wave with a low impedance might be equivalent to 5VAC with a 50 ohm resistor in series with it's output, while a high impedance mic with the same output might be equivalent to 5VAC with a 50k ohm resistor in series with it's output. This explains why it is very bad to connect a hi-Z mic to a low-Z input. The low-Z input would have low resistance drawing near maximum current from the hi-Z mic, therefore it would load it down because the mic would have such a high load and the mic would not be able to produce a well formed wave.

In response to hawk2eye, would an op-amp be a better choice for this circuit, and if so, for any reason besides producing a cleaner signal? Also, what requirements does a typical op-amp have? Can it run on 5VDC, and how much current would it require? I did test this circuit with the 2N4401, which was mentioned to be very noisy and it didn't introduce a terrible amount of noise, however it didn't provide much gain, as previously mentioned.

TJ

 

[Nigel Goodwin]

As Dr. EM and Nigel Goodwin both point out, the hi-Z input of the sound card should be fed with a low-Z output from this circuit to achieve optimum voltage (signal) transfer. The goal of this circuit is to increase the gain of the signal going into the soundcard. Since the soundcard has a hi-Z input, we aren't really trying to feed more current, but higher voltage peaks than the mic outputs. A transistor is a current device, though. My understanding is that the gain it is providing would be current, and it would not increase the voltage peaks. Thus, this circuit really isn't amplifying voltage, which is what I think I need.

PC mike inputs are really strange?, I've never seen one yet that works by plugging a mike in it?. I have read in the past that the original Sound Blaster card (which most are based on) was designed to use a carbon mike (as in antique telephone handsets).

If you're looking for a decent quality mike input, it's probably best to use an opamp preamp and feed the line input instead of the mike input.

So, can someone clarify whether I'm looking for voltage or current gain? In the case of this circuit, is it providing current gain as I guessed, and what specifications should I look at on the data sheet to determine the gain? If I am indeed looking for voltage gain, shouldn't I be using a FET?

It provides both current gain and voltage gain, in fact I thought it was a really nifty little circuit - should do all that's needed, use a high gain, low noise audio transistor - personally I'd probably use a BC109, the original metal canned version of the BC159.

Another thought is that kchriste says that R3 sets output impedance. R3 is 4.7K, not what I would call low impedance! In order to provide the low-Z output, I would need a circuit with an R3 of 600 ohms or so, right? I'm not sure of how the selection of R3 relates to Q1. Can I choose any R3 I like as long as I adjust R2 according to the formula:
R2 = Beta * R3

The output impedance will be considerably lower than R3, you would probably need to measure it to find out what it was.

On the subject of impedance, is impedance similar in concept as internal resistance? For example, a mic outputting a 5V peak-to-peak sine continuous sine wave with a low impedance might be equivalent to 5VAC with a 50 ohm resistor in series with it's output, while a high impedance mic with the same output might be equivalent to 5VAC with a 50k ohm resistor in series with it's output. This explains why it is very bad to connect a hi-Z mic to a low-Z input. The low-Z input would have low resistance drawing near maximum current from the hi-Z mic, therefore it would load it down because the mic would have such a high load and the mic would not be able to produce a well formed wave.

Yes, that explains why it's bad, but the reason isn't so much a 'well formed wave', it's mostly that the internal impedance of the mike and the input impedance of the preamp form a potential divider - reducing the already tiny output from the mike.

In response to hawk2eye, would an op-amp be a better choice for this circuit, and if so, for any reason besides producing a cleaner signal? Also, what requirements does a typical op-amp have? Can it run on 5VDC, and how much current would it require? I did test this circuit with the 2N4401, which was mentioned to be very noisy and it didn't introduce a terrible amount of noise, however it didn't provide much gain, as previously mentioned.

I built one (very quickly) using an opamp fed from a 9V battery (for an on-line training course with Sony).

 

[audioguru]

The voltage gain of the original single transistor circuit is 80, and it has severe distortion when its input voltage exceeds only 20mV p-p which is only 7.1mV RMS. If its bias current is reduced to make it more symmetrical then its distortion is much worse. An opamp is much better.

A transistor's collector is a constant current source or sink so the output impedance is its collector resistor.

 

[systemloc]

Ack! I totally hosed my implementation of the circuit. I reversed the emitter and collector. Not sure what that even does. I'm thinking it functioned, but with no gain. I'm going to fix the problem with the 2N4401 I have in hand and try it out.

I'm still curious to know what would happen if I used a 2N3903 instead. I'm also still unclear as to what criteria to use to select a transistor in the first place. I'm guessing hFE and noise?

 

[Nigel Goodwin]

I'm still curious to know what would happen if I used a 2N3903 instead. I'm also still unclear as to what criteria to use to select a transistor in the first place. I'm guessing hFE and noise?

Basically noise really, as it has negative feedback the gain isn't as critical as it might be.

Try different transistors and see what happens?.

 

[audioguru]

A transistor with its collector and emitter reversed will have a very low AC and DC gain.

These simple high gain transistor amplifiers hardly have any negative feedback. Therefore their DC operating point varies considerably due to the wide range of a transistor's DC gain. Also there is hardly any negative feedback to reduce distortion.

The DC gain of a 2N3903 is from 40 to 120 at the current it is used at in this circuit, a 2N4401 has a wide DC gain range that is about the same, although has a much higher DC gain at higher currents. The range of gain is so wide that you won't know which transistor has the correct amount of DC gain.

Try a 2N3903 and it works well. Try another and it works lousy. The same for a 2N4401.

 

[Dr.EM]

Undoubteble the suggestion to use an op-amp preamp system powered exteranally by a 9v battery (or 2) into the line input will yield the best results. I have microphones made with electret cardtridges, such as the ones used in PC mics, and connected via a mixer into the line in. The sound quality is massively better with much lower noise and higher headroom than using the onboard mic preamp, and it's not a rubbish sound card either.

It may be possible to use the 5v bias into a DC-DC convertor to give a high enough supply to power the preamp system, though i'm not sure about available current (most soundblaster models have the bias supplied via 2.2k).

 

[audioguru]

The simple transistor preamp circuit goesn't have a gain control. So loud sounds will be terribly distorted. An opamp preamp circuit can have a gain control as its negative feedback resistor so it won't overload when it is turned down.

 

[systemloc]

These simple high gain transistor amplifiers hardly have any negative feedback. Therefore their DC operating point varies considerably due to the wide range of a transistor's DC gain. Also there is hardly any negative feedback to reduce distortion.

I'm unclear on a few concepts here..

Negative feedback in this circuit is provided by R2, correct? The output is phase inverted, so feeding it into the base through R2 subtracts from the input signal to some degree. What's the point of doing this? It would seem to attenuate the amplification, but why? This prevents distortion during signals that peak the transistor?

DC gain is the Hfe value, correct? So the transistors are widely variable in construction, so one may give much more gain than another out of the same box. That much I understand. What is DC operating point, then? I'm guessing this has something to do with the range of currents that the transistor behaves in a linear fashion? Assuming I know the measured Hfe of a particular transistor, could I calculate the expected voltage gain for a given signal?


I want to stop here for a second and say thank you to everyone who has taken the time to post here. This thread has already taught me quite a bit that I didn't previously understand. I'd also like to acknowledge that I understand the Op amp solution is obviously better, and I'll definitely build that as well. At this point, I'm simply playing with this circuit as a means to learn more and understand it may not give me a very usuable product. I did fix my wiring mistake, and I still don't get enough gain. Cranking up the soundcard's software gain control adds a lot of static that may well be from my circuit. Another thing I noticed is that I get some AM reception when I touch the mic casing. Sounded like talk radio. I'd imagine that's a grounding problem easy enough to chase down.

TJ

 

[audioguru]

Negative feedback in this circuit is provided by R2, correct?

Yes.

The output is phase inverted, so feeding it into the base through R2 subtracts from the input signal to some degree. What's the point of doing this? It would seem to attenuate the amplification, but why?

If the transistor's DC gain is low, then it won't have enough base current so its collector current will be too low, causing its collector voltage to be too high and the top of its signal will be clipped. Since R2 connects to the collector then when its voltage rises then the current in R2 will increase and provide more base current to compensate.
The opposite occurs if the transistor's DC gain is too high.
The signal in R2 cancels some of the input signal which reduces the circuit's AC gain and also cancels some of the non-linear distortion.

DC gain is the Hfe value, correct? So the transistors are widely variable in construction, so one may give much more gain than another out of the same box.

No. Hfe is AC current gain. hFE is DC current gain.

What is DC operating point, then? I'm guessing this has something to do with the range of currents that the transistor behaves in a linear fashion?

In theory, if the collector operating point DC voltage is halfway between the supply voltage and ground, then the collector signal voltage can swing the max amount possible. But as I showed, with high AC gain the top of the waveform is compressed with severe distortion. I tried it with lots of negative feedback and the AC gain was only 1, output voltage was nearly max and the distortion was barely visible.

Assuming I know the measured Hfe of a particular transistor, could I calculate the expected voltage gain for a given signal?

No. Hfe hardly affects low level signal voltage gain. hFE also hardly affects low level voltage gain. The low level AC gain without negative feedback is the ratio of the collector resistor (and any load in parallel with it) to the internal resistance of the transistor's emitter to ground. A transistor's internal emitter resistance is about the same for all silicon transistors and is figured as 0.026/emitter current in mA. If the transistor has a resistor to ground in series with the emitter and the resistor doesn't have a capacitor bypassing its signal, then it enters into the AC gain calculation since it causes another type of negative feedback which reduces gain.

 

[Nigel Goodwin]

I'm unclear on a few concepts here..

Negative feedback in this circuit is provided by R2, correct? The output is phase inverted, so feeding it into the base through R2 subtracts from the input signal to some degree. What's the point of doing this? It would seem to attenuate the amplification, but why? This prevents distortion during signals that peak the transistor?

It's as I suggested earlier, it means you don't have to select the resistor values for every single one you build - the negative feedback is at DC as well as AC, so it stabilises the stage (to some extent). If you feed the base resistor from HT rather than the collector you don't lose the gain, but you do have to manually select the bias resistor for every single one you build.

 

[hawk2eye]

low level or or high level?

Except you don't want impedance matching, it's a BAD idea for audio (or for the majority of connections) - by definition impedance matching is very inefficient, only 50% at the very best.

You want to feed a low impedance source into a higher impedance input, for maximum VOLTAGE transfer, you should be looking for 5 to 10 times higher impedance on the input.

+++++++++++++++++++++++++++++++++++++
In order to get best S/N performance from a signal source operating at its the lowest level, impedance matching is the ONLY way to go.

I agree with your comments ....for a "majority of connections" which can be defined for high S/N levels, then maximum voltage transfer gives an advantage over
impedance matching.

BUT: Using your technique, a mike pickup would produce poorer S/N for low level signals.

hawk2eye

 

[Nigel Goodwin]

+++++++++++++++++++++++++++++++++++++
In order to get best S/N performance from a signal source operating at its the lowest level, impedance matching is the ONLY way to go.

I agree with your comments ....for a "majority of connections" which can be defined for high S/N levels, then maximum voltage transfer gives an advantage over
impedance matching.

BUT: Using your technique, a mike pickup would produce poorer S/N for low level signals.

Why?, and what are you defining as 'low level'.

And it's not 'my' technique, it's EVERYONE'S technique - can you name a single audio input where impedance matching is used?. If you're thinking of suggesting microphone transformers?, they are simply used to change the impedance, which is then fed into a much higher impedance as normal.