Emitter bypass caps in a CE stage

[Novice62234]

I have a few questions, please bear with me if I demonstrate any lack of understanding…..


With regard to the emitter bypass cap in a common emitter stage –

Why is it that the rule of thumb is that Xc be 10% of the emitter resistor value? Why not anything lower? Like 1% or less to minimise phase issues in bass?

I tried searching this forum and also googled this - but I could not find an answer, but it has made me realize that this is back to basics for a lot of you here - your time and explanation will be much appreciated.

[AndrewT]

Hi,
the emitter impedance controls the gain of a CE stage.
That impedance will vary with frequency.
That means the gain will vary with frequency.
You have to decide what gain you need and what bandwidth you would like that gain.

[DF96]

I don't know where you found that rule of thumb, but it is misleading. The real issue is to compare Xc with the parallel combination of the emitter resistor and the emitter resistance. In many cases the emitter resistance dominates (it is the smaller of the two), so the emitter resistor is almost irrelevant. The LF response will then be -3dB (and phase 45 deg) when Xc is roughly equal to the parallel combination.

[Novice62234]

Thanks so much Andrew and DF96! So Xc should actually be equal to the emitter resistor Re and the intrinsic emitter resistance re in parallel and not be 10% of that value.

Two more questions

I gave it some more thought. If the emitter resistor is bypassed with an electrolytic and a smaller value film cap as is sometimes advised, the film cap would probably increase (unwanted) high frequency gain a lot (max hfe the transistor is capable of) and make it ring or oscillate. Same goes to low esr/impedance types. So am I right in saying that the best cap for this duty is a low voltage electrolytic with bad hf performance (not low esr/impedance caps)?

I have another issue: this is subjective as I don't have a 'scope. When the emitter bypass cap is added, there seems to be a small increase (peakiness) in the high midrange level as opposed to the smooth presentation although without bass impact when the emitter cap is taken out of circuit. Would it possibly be behaving like a resonant circuit?

[AndrewT]

I am not a designer so this is a layman's view.
Insert a 100r as the Re of the CE stage. bias the BJT so that it's internal RE is ~25ohms.
The total emitter resistance controlling the gain is Re+RE = 125r.
Now bypass this with a perfect enormous capacitor. This capacitor has an effective impedance of near zero down to low frequencies. This cap sets the gain of the CE stage for AC signals by using the RE as the total emitter resistance. i.e. AC gain is ~ 5times the DC gain.
Now add a series resistor to that bypass cap. set this extra resistor to 25r. The total impedance seen at the emitter is now 100r//25r+C = 20r. The total emitter resistance is 25+20 = 45r. The AC gain is ~ 125/40 ~ 3times the DC gain.
Now consider the bandwidth that you want this AC gain to operate over.
The LF limit is set by the capacitor value.
The HF limit is set by the transistor and the HF loadings that the transistors is connected to. This could easily go to MHz.

If you want to be able to design a CE stage then you must go and find out how the various components around the CE transistor affect it's amplifying performance, i.e. learn how and why it works. I usually state this as do your "homework".
Anecdotal internet advice is not the way to learn this. Go and find technical articles/papers on how a transistor operates (not at the atom level) and why different resistors define the amplifying characteristics.

[DF96]

It is possible for a film cap to resonate with the inductance of a bigger cap, so added bypassing can be a mixed blessing. This effect can be reduced by using a high ESR electrolytic. Alternatively, just use a decent electrolytic with an appropriate value and don't add an extra cap.

Bear in mind that electrolytics don't do much harm unless they have a signal voltage across them. This means that the electrolytic should not be used to establish the LF point - this job should be done by a non-electrolytic coupling cap earlier in the circuit. Keep the electrolytic decoupler about 5-10 times what is needed for the LF point. So, rule of thumb, Xc should be about 10-20% of the paralleled emitter resistances at the LF point. Don't make the decoupler too big, as it might start misbehaving at the upper end of the audio range.

[AndrewT]

Originally Posted by DF96

Bear in mind that electrolytics don't do much harm unless they have a signal voltage across them. This means that the electrolytic should not be used to establish the LF point - this job should be done by a non-electrolytic coupling cap earlier in the circuit.

You are one of the very few to preach this message. It is important and it does affect the resulting sound quality.
An electrolytic with no varying voltage across it cannot contribute to output distortion.

[Novice62234]

I think DF96 is right - the emitter bypass capacitor's function calls for a high esr high value low voltage cap that has bad hf performance but low distortion - like a Silmic or Cerafine.

The reason is that by bypassing AC the cap frees the transistor of AC feedback and lets the gain go high. If a low value film cap like .1 uF was used it will bypass a high (unwanted) frequency – for example like 100 KHz or more which could cause the transistor to oscillate or at least ring and also increase interaction with parasitics. Though films do good things at other locations, this is not one of them. Of course, unless someone who has measured this can attest to this I cannot confirm it right now till I buy the equipment needed.

When I brought up resonance I meant RC tank resonance.

I am going to get hold of some nice books soon. The Art of Electronics by Horowitz and Hill sounds good for now. Next purchase will be an oscilloscope and a signal generator instead of another speaker or preamp or audio equipment – this is so much more interesting!

[DF96]

I thnk you mean LC tank resonance.

When transistors oscillate it is usually at much higher frequencies than 100kHz - that is only just above audio.

[Novice62234]

Yes, I meant LC tank. Here though it's more like a RLC tank with the electrolytic's ESL playing the role.

Thank you Andrew and DF96. This discussion has definitely made me think more.

BTW, the Art of electronics book states that Xc of the bypass cap should be comparable to or less than only the intrinsic emitter resistance re at the - 3 db frequency which is derived in the book and shown to be 25/Ic. The value 25 is derived and constant for all transistors as it is temperature dependent.
So that way one can bias the transistor to more than 1 ma if needed and still obtain a re value.