 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
72.211.237.239
In Reply to: RE: Buffer - emitter resistor value? posted by andyr on December 23, 2024 at 21:03:36
The pull-down resistor sets the dc current in the follower and that affects the small signal output resistance and the large signal distortion performance. Assuming the base is biased mid-rail, the voltage across Re is either (12V-Vbe)/4.7k or (12V-Vbe)/1k ~ 2.4mA or 11.3mA. The output resistance of an emitter follower is (Vt/Ie + Rbase/Beta) // RE.
Vt/Ie (what we in the trade call 'little Re') is the incremental resistance of the base-emitter diode seen from the emitter side. Vt = kT/q ~ 26mV so with the two Ie options 'little Re' will be either ~ 10.8ohms or ~ 2.3ohm. Let's assume that the circuit driving the base has an output resistance of 1kohm (including the real part of the BJT base resistance, aka 'rbb') and the follower Beta = 100, then the contribution of the resistance on the base side as seen at the emitter is ~ 1k/100 = 10ohms, which is significant. With these assumptions you only get an improvement in output resistance from ~20.8ohms to 12.3ohms for a 5x increase in current (this is shunted by the pulldown resistance, aka 'Big RE', but >1k shunting 10s of ohms does not have much effect). The small signal output resistance tells you how well the follower will drive a low load resistance &/or a capacitive load.
Considering the distortion performance of a follower, that comes from the large signal performance and depends upon the current variation in the transistor. Assuming the signal at the base is un-distorted, the output follows that voltage offset by a Vbe. If the current in the transistor changes then Vbe changes in an exponential fashion given by Vt*ln(ratio of current change), e.g. if the current doubles the change in Vbe ~ 26mV*ln(2) = 18mV and this is distortion seen at the emitter that is not present at the base. With a resistive pulldown (RE) the current in the transistor varies with the signal voltage irrespective of any current supplied to the load e.g. if the base signal voltage is 2Vpk then the emitter current changes by either 2/4.7k = 0.15mApk or 2mApk. The effect on Vbe of the two cases is Vt*ln((2.4m+0.15m)/2.4m) = 1.6mV or Vt*ln((11.3m+2m)/11.3m) = 4.2m. Relative to a 2Vpk signal those are changes of 0.08% or 0.2%, which are small but not negligible if you want lowest distortion. This distortion can be all but eliminated by using a constant current source pulldown instead of RE so that the transistor current does not vary with signal.
So far I have ignored the effect of current supplied to the load as that is supplied by the follower and will also modulate Vbe. If you are driving 10k with a 2Vpk signal then the peak load current is 0.2mA, which is slightly more than doubles the current change for the 4.7k case but is only one tenth that of the current change in the 1k case. So, a 10k load resistance will not have a big effect on distortion performance. If the load was lower, say 1kohms, then the peak load current is 2mA. The impact on Vbe is now Vt*ln((2.4m+0.15m+2m)/2.4m) = 16.6mV (0.83% of the 2Vpk signal) or Vt*ln((11.3m+2m+2m)/11.3m) = 7.9mV (0.4% of 2Vpk).
In summary, the 1khom case allows you to drive a low load resistance with lower distortion but it will, counter intuitively, produce more distortion in a high load resistance case (though that can be eliminated by using a current source pulldown). One last thing is power dissipation, that will be either ~ 12.7V*2.4mA = 30mW or ~ 12.7V*11.3mA = 143mW so check the transistor rating.
Follow Ups:
Thank you very much for your detailed response, mate. :-))
Re. your last sentence - " In summary, the 1kohm case allows you to drive a low load resistance with lower distortion but it will, counter intuitively, produce more distortion in a high load resistance case ."
... as the Zin of the device following the phono stage is 10K - I take it that I will get lower distortion using a 4K7 ohm res than a 1K res?
Using my back-of-the-envelope analysis I get 0.18% (4k7) vs 0.23% (1k) so only a small difference either way. You could simulate to be sure (and validate the accuracy of my envelope) but, I think, if you were that worried about distortion you'd have a different circuit. The % numbers I calculated are the change in Vbe relative to a 2Vpeak signal - I think that should relate to THD as distortion caused by the Vbe change will be distributed across the harmonics, and an exponential characteristic generates all the harmonics from 2nd upwards. You could think about replacing the BJT with a FET - this will create distortion in the same way by modulation of the Vgs by the signal current and, because FETs have lower gain the change in Vgs will be much larger than the equivalent Vbe change for the same signal induced current change. But, the distortion will be predominantly 2nd order and that may give a different subjective effect (there will be higher orders too as there is no such thing as a perfect FET). Something to experiment with if you're curious.
Have a good Christmas on the beach:)
I've just changed the emitter resistors in my phono stage from 1k ... to 4k7; listening to the phono stage with the arm lifted off the LP ... this has considerably reduced the noise I hear. So that is a biiig win!
Thinking about it, I presume (with a +24v DC rail) there must've been 24ma flowing through the BC547C, when the emitter res was 1k. Only having 1/5th of that current flowing through it now (with 4k7) ... makes it quieter.
I would not expect the follower noise to be significant. In cascaded stages the source noise gets amplified down the chain and each stage adds its own noise but the noise floor is raised with each stage of amplification hence succeeding stage noise requirements are relaxed. The first stage is the most important and in communication systems the first stage is always a low noise amplifier (LNA).
For simplicity, let's ignore the amplifier noise floor and just think about the noise of the signal source. If it is a LOMC the source resistance is ~ 10ohms and the noise of that resisitance will get amplified. Assuming you have at least 40dB of voltage gain (100x), the noise power after amplification will be equivalent to a resistor of 10*100*100 = 100k. I cannot imagine your follower producing anywhere near that level of noise (datasheet says it has 2dB noise figure referenced to 2kohms that means it is bit noisier than a 2k resistor). So, a mystery that you could decipher using spice noise analysis or just enjoy another LP
you may well think it " a mystery that I could decipher using spice noise analysis or just enjoy another LP ".
All I know is that after I changed the 1k emitter resistor to 4k7 ... the noise that I was hearing from the spkrs with the stylus lifted off the LP was low enough to make listening to an LP, a pleasure. Whereas the noise that I heard from the spkrs when the 1k emitter res was in place ... made it impossible to enjoy listening!
And re. removing the 'M' from my description, as I obviously know very little EE theory ... Audioasylum require me to have that 'M' designation as, even though I am merely a retired bloke with a hobby, the fact that I make and sell a few components means I have to be called a 'manufacturer'!
Andy, sorry if I've offended. 'M' was a small elephant in my room in that someone can make stuff without really knowing how it works. But I get you're doing this as a hobby and it is great that you're selling something. Selling something makes you a more successful audio designer than me! I met the LKV guy a couple of times at shows in SoCal and he was a hobbyist during his lawyering career and then turned audio into a business when he retired. The second time I met him he told me he was at least breaking even, which is great if you can perform your hobby without going broke. I haven't seen anything of LKV in the audio press for a long time so I hope all is well.
That said, I am happy to provide insights as, IMHO, those are the most important things and they can take a long time to acquire (well, they have for me but I've worked with a lot of really smart people who pick them up quickly). I made the joke about 'M' because the answer I gave you about HOMC was mostly gleaned from a web search and wasn't anything you couldn't have found out yourself. FYI, there are probably better sites to pose technical questions such as DIYaudio and PFM. I read PFM and there are technical posters there. I don't go to DIYaudio as it is full of people nerdier than me and I'd have to fully engage or stay away.
And I wasn't having a dig about the spice noise analysis. As I said, I would not expect the output follower to set the noise floor so when you change the bias current you should not hear a difference. Understanding why things don't work as expected as where one can gain insight and I could be wrong (I make no claim to infallibility). Spice can rank the noise contributors in the circuit so you'd see where most of noise comes from. I don't think low noise design is that necessary in audio even for phono preamps because once the needle hits the record even the quietest vinyl will create more noise than from the electronics.
Talking of phono playback, I remember your skeletal LP12 from The Wam. Last year I upgraded to a Supatrac Blackbird tonearm on my LP12 and it has been the most wonderful improvement.
And I greatly appreciate your input.
The phono stage I build is certainly not a 'business'. I sell a few each year and my sell-price is based on parts cost + A$10 an hour for my labour (so it's by no means a 'profit maker')!
As such, it outclasses in SQ, phono stages up to 5 or 6x the price. I actually have a customer in SoCal, if you're interested in hearing it. :-))
I certainly agree with you that " once the needle hits the record even the quietest vinyl will create more noise than from the electronics ". However, achieving a lower noise floor is my current "windmill tilt" ... I want to achieve as low a noise-floor as my backup phono stage has (a Hagerman 'Euphonia').
Fascinating that you found a reference to my 'SkeletaLinn'. It gets gradually improved, as time advances - and someone has just built one in the UK; he reckons it already outclasses his LP12 ... and he's still got more tweaking to go. :-))
The Supatrac Blackbird seems to be a great tonearm; if I didn't already have my 12" 'Univector' (UP) - which was the reason I designed my 'SkeletaLinn' - I would probably get one, too. Unfortunately, the designer/builder of the Univector died about 5 years ago ... so they are no longer available.
Here's a pic of the current state of the SkeletaLinn' - on its recently installed air isolation platform:
The 12" Univector is at the front; a Magnepan 'Unitrac' (also a UP) is at the back right. Current chassis is made of a Panzerholtz-like ply ... which you may know Linn recently used for their new plinth.
I have the whole circuit (not just the buffer) in LTspice ... so I ran it with, first, 1k as the Emitter resistor - and then 4k7.Distortion results are as follows:
1k :
Total THD = 0.0654%
* 2nd: 6.541 e-4
* 3rd: 1.690 e-5
* 4th: 6.918 e-7
* 5th: 6.098 e-74k7 :
Total THD = 0.0580%
* 2nd: 5.802 e-4
* 3rd: 1.677 e-5
* 4th: 6.621 e-7
* 5th: 5.977 e-7So - as you say - a slight decrease in THD by increasing the Emitter res to 4k7 ... mainly the 2nd harmonic.
So I will go back to using 4k7 - and ponder the use of a JFET, instead of a BC547C.
Any suggestions as to a suitable JFET? I have heaps of:
* MPF102
* SJ103, and
* SK246.And it's a perfect day, here ... but no beach time planned. :-((
Edits: 12/24/24
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
