on instrument amplifiers again.......

[bachevelle52]

Okay so,
Initially i wanted to design a buffer for the front end of an INA 217 based micpre built around the circuit provided in the datasheet. I believe it was PRR who so eloquently asked why would i want to stick "a gitar in a mic hole?" Thats a great question that ultimately led me to dismiss that idea altogether. So the new idea would be this: a box i can plug said "gitar" into that can amplify my signal to drive a line input.

I own a copy of "the Art of Electronics" and i've been reading and re-reading this book for years. I still don't really understand, perhaps my friends here at prodigy-pro can help. With that said, please understand that i'm not looking to get this designed for me, i desperately want to do this on my own, i just REALLY need to be nudged in the right directions.

Here are my thoughts:
1. the treatment of emitter followers in TAOE states that input and output impedances are relatively easy to adjust. Based on the given example, biased from +20V there is a voltage divider setting the base bias. The apparent input impedance is dictated by the bias divider in parallel with re -the intrinsic emitter R. They seemingly arbitrarily choose 1mA quiesent current and that is the basis for the value of Re. Then there is the issue of Rc which is stated to set the gain.
So my questions start here.
1) how do you choose the quiesent current? What does this affect?
2) Rc was determined based on what?
3) in calculations, mant times i see hfe as a term, and it would appear that 100 is the typical value used, however it is stated repeatedly that hfe dependent designs are poor designs.

I'm positive i will have 100 more questions before this is done but this should be a good start. For reference, i believe the treatment of emitter followers is towards the beginning of Chapter 2 in TAOE, i've been studying the first circuit example.

I would ultimately like to design this one bit at a time, but the total gist would be Emitter follower, to say a common Emitter stage. as far as i can tell at this point, what i need is:
1) 1Mohm input Z
2) maybe 40db gain
3) 1k output Z to drive 10k Z line input
...for now.

Any help, thoughts, ridicule.... all welcome. I so desperately want to learn this, i honetly just cant put my family on hold to spend 4 years in school.
Thanks in advance,
Ben.

 

[abbey road d enfer]

Okay so,
Initially i wanted to design a buffer for the front end of an INA 217 based micpre built around the circuit provided in the datasheet. I believe it was PRR who so eloquently asked why would i want to stick "a gitar in a mic hole?" Thats a great question that ultimately led me to dismiss that idea altogether. So the new idea would be this: a box i can plug said "gitar" into that can amplify my signal to drive a line input.

I own a copy of "the Art of Electronics" and i've been reading and re-reading this book for years. I still don't really understand, perhaps my friends here at prodigy-pro can help. With that said, please understand that i'm not looking to get this designed for me, i desperately want to do this on my own, i just REALLY need to be nudged in the right directions.

Here are my thoughts:
1. the treatment of emitter followers in TAOE states that input and output impedances are relatively easy to adjust. Based on the given example, biased from +20V there is a voltage divider setting the base bias. The apparent input impedance is dictated by the bias divider in parallel with re -the intrinsic emitter R. They seemingly arbitrarily choose 1mA quiesent current and that is the basis for the value of Re. Then there is the issue of Rc which is stated to set the gain.
So my questions start here.
1) how do you choose the quiesent current? What does this affect?

Mainly two things:
a) the output capability of the stage, i.e. the capability to drive the subsequent stage adequately.
b) the noise current of the transistor, which is a determining factor in the noise performance of the stage. The noise current must be chosen in relation with the impedance of the signal source. In the case of electric guitar pickups, it is a high impedance, which implies a very low quiescent current.

2) Rc was determined based on what?

Rc is the collector res? Irrelevant in an emitter follower. Are you switching from emitter-follower to common-emitter?

3) in calculations, mant times i see hfe as a term, and it would appear that 100 is the typical value used, however it is stated repeatedly that hfe dependent designs are poor designs.

They are not really poor designs. They have some issues like the fact that they probably need some tuning or adjustment, which is not compatible with accepted mass production paradigm. Generally, the higher the hfe, the more likely the product to work as expected. But excellent performance can be had with low hfe devices. As always, the performance of a circuit depends more on the designer's abilities than on the intrinsic parameters of the components.

I would ultimately like to design this one bit at a time, but the total gist would be Emitter follower, to say a common Emitter stage.

A wise decision, since it's almost impossible to have one single stage do what you expect.

as far as i can tell at this point, what i need is:
1) 1Mohm input Z

That would require an emitter-follower operating at about 0.2milliamp quiescent.

2) maybe 40db gain

This would probably require two stages with NFB to avoid distortion. But I think it is not right. An electric guitar p/u delivers about 100mV; with 40dB gain, you would end up with 10 volts output. You will need some gain control (not the guitar's pots because the preamp noise will be dominant).

3) 1k output Z to drive 10k Z line input

This is not very difficult. Use a 1k collector res.

I so desperately want to learn this, i honetly just cant put my family on hold to spend 4 years in school.

I understand that, but you'll have to put your project on hold, and learn about Ohm's law, Thevenin, basic circuit knowledge, and practice. When I started playing with electronics, I didn't know squat about hfe, noise factor...but I made some AM radios and headphones amps. I learned about the boring stuff later. You have one big advantage over me. A transistor costs about 10c; when I was a kid, I had to save for about two months to buy an OC71.

 

[lassoharp]

why would i want to stick "a gitar in a mic hole?"

Maybe different app here but this has been Ritchie Blackmore's secret weapon for a number of years.

 

[PRR]

> learn about Ohm's law, Thevenin, basic circuit knowledge, and practice.

+101

> it is stated repeatedly that hfe dependent designs are poor designs.

If you design so that hFE _MUST_ be 100 eXactly, you are doomed.

It is reasonably easy to design so that any part with hFE from 100 to infinity will work. On matchbook, you check the hFE=100 case and see it biases a bit low, hFE=1,000 and see it biases a bit high, neither one in trouble.

> a common Emitter stage. as far as i can tell at this point, what i need is:
> 1) 1Mohm input Z
> 2) maybe 40db gain
> 3) 1k output Z to drive 10k Z line input

What signal levels? 20mV in 2V out? 14V in 1,400V out?

40dB gain ALL the time? Strong arm, hot pickup, 500mV signal in, 50V signal out (to a 12V board)?

Spoiler: you can't get 1Meg in 40dB 1K out with any single standard transistor.

Why?

If it were unity-gain, then the impedance shift from 1Meg to 1K (or 10K) unity voltage gain suggests a total current gain of 1,000 (or 100). Biasing and loading mean you can't get jack-to-jack current gain as high as hFE.

You also ask 40dB or 1:100 voltage gain. Higher load voltage is higher load current. You need overall current gain of 100,000 plus bias effects.

Cheat: a TL071 will easily meet your specs. Why?

Recall we estimated you need a total current gain above 100,000. Using parts with naked hFE=100 we probably won't get signal current gain over 50 per part. To get 100,000 we need {Spoiler} [size=4pt]three stages with current gain of 46 per stage. Such chores are common. The TL071 has over a dozen transistors. Many are doubled-up, but it has four stages of current gain (and the first stage is FET with somewhat more usable audio current gain than a BJT). [/size]

 

[bachevelle52]

Thanks for the replies fellas,

Admittedly i was in a bit of a hurry when i posted this. I wasn't suggesting all of the above from a single device, my thought was along the lines of emitter follower coupled to common emitter stage, or multiple stage for gain. The 40db gain spec was more in haste as i hadn't yet figured how much i would actually need. The point is to drive a line input to line level. Please forgive me if i mis-speak here as i don't have my notes here with me. I think its in handbook for sound engineers in the section dealing with sound measurement i read about line levels. I cant remember if the subscript was dbV or dbU, however, it was referenced to a specific voltage output. Assuming an average pick-up output of 100mV (as stated above) it will be easy to estimate the actual gain I need. I have more to add, I have worked some of this out already. One question for now that is eating at me....the first response (by abbey rd.) suggested a 2mA quiescent current for my follower, how did you come to that conclusion? Just a formula would be great, i really do want to design this myself.
Thanks,
Ben.

As an aside, i have done an aweful lot of reading, i know a lot more than i realize until things start to fall into place. Some of these things for me are terribly counter intuitive. Also, i have 20 or so different books for beginners and it's quite amazing how different the info is from one book to another. I'm certainly not say "oh i know it all" because i don't, sometimes i just have to start applying theoretical knowledge for it to start making sense. I think this circuit, being relatively simple, is a good place to start. Again thank you for all the help and suggestions.

 

[PRR]

> eating at me....the first response (by abbey rd.) suggested a 2mA quiescent current for my follower, how did you come to that conclusion?

No, he wrote:

> Quote - 1Mohm input Z
> That would require an emitter-follower operating at about 0.2milliamp quiescent.

2mA is not the same as 0.2mA.

> Just a formula would be great

I don't think there is an all-situation formula.

Pulling round numbers out of my butt, I'd be thinking:

20V supply, 10V across emitter resistor.

Guru Abbey dropped the number 0.2mA.

10V at 0.2mA is 50K emitter resistor.

Assume next-stage loading is =>50K.

Then raw naked base impedance is hFE * 50K(or 25K) which for jellybean hFE=200 parts is about 10Meg-5Meg. We will need base-bias resistors. The naked base loading should be light on these bias resistors (so it is not hFE-fussy). 1Meg resistors will be suitable. 10Meg||1Meg is near-enough 1Meg. That won't be your Final Answer, and neglects a stray base-collector impedance, but shows that you can probably buffer 1Meg into 50K at unity gain. This can guide design of a second stage.

Note that a stage with 1Meg in and 50K loading gives a current gain of just 20, even though the part hFE was assumed to be 200. In general you can get current gain of 10 per stage. Maybe more if you are clever, and/or if you can afford to sort hFE=300-900 parts and discard all the hFE<500 parts. But 10 per stage is a good starting point.

Take a broader picture. 1Meg and current gain of 10 gives around 100K load. Next stage can work between 100K and 10K. To get to 1K you want a third stage.

You want voltage gain. Taking voltage gain generally conflicts with getting high input impedance, so buffer-first makes sense. OTOH any extra baggage before your first gain stage tends to increase noise. In this case the problem is mild, though the proof is beyond your present skills. Buffer, gain, and then a final buffer is the most-obvious plan.

Gain-control adds a new wrinkly and may suggest two low-voltage-gain stages rather than a hi-gain and a buffer. Work it out both ways.

> I cant remember if the subscript was dbV or dbU, however, it was referenced to a specific voltage output. Assuming an average pick-up output of 100mV (as stated above) it will be easy to estimate the actual gain I need.

0.775V, 1.00V.... what's the difference? Guitar does NOT play at any specific voltage. so little 30% differences vanish in the herbal haze.

Classic guitar amps only need 20mV for FULL output, OTOH can be turned down so that 500mV is not-quite full output.

Classic line inputs may have a mark for 0.775V, but in fact can accept 0.1V to several V and bring them all to nominal level. (Read and digest your mixers' specs.)

Strong guitar playing "could" just jack into a mixer's Line In, except modern mixers tend to 10K inputs and a guitar's impedance varies from under 10K in bass to near 100K in the critical overtone zone. Jacking into 10K load clobbers the guitar's brightness.

Guitar is historically used with strong EQ with mid-dip not readily found on all-purpose mixing consoles. There is a strong argument for stealing a Fender tube amp's first two stages and tone-stack, then figuring how to convert that down to modern line-input level and impedance. 'Course now you have to learn about Tubes enough to steal with respect.