Addressing John Curl's concerns on low noise designs

Status
Not open for further replies.
Jump to Latest
#681
No, it boils down to base current, collector current, and their relationship. It does not boil to points of view how to call devices.

Again, let me repeat:

"it is obvious that an emitter current is the sum of collector current and base current. Base current of a transistor working in an emitter follower roughly equals to collector current divided by beta.That means, in case of even an ideal transistor base current depends on load resistance. It is basic and obvious. Since beta depends on collector-emitter voltage and collector current base current also depends on beta variations. As the result, an output current of emitter follower is reflected on it's input current, and reflected non-linearly. If you observe base charges this picture will be more complex, frequency dependent."

It is the reality, and it does not depend on how do we agree to call devices: voltage controlled, or current controlled. In order to help my students to grasp the whole picture, I taught them to think about power amplification. Such a way they would not miss neither currents, nor voltages, nor impedances, and their designs reflected the real picture.

Edit: when Scott Wurcer meant IC design and non-importance of Beta, he meant that beta of all transistorsa on the die will be the same if their geometry is the same, and using IC-specific topologies where beta of transistors are out of equations it is possible to design without beta involved in calculations, but it does not mean neither that beta don't exist, nor that it don't play the role.
 
#682
Wavebourn, in this case of triple EF with bootstrap there is nominaly a current gain of say 10^6 (exact details are a little more complicated) and the designed load is 600 Ohms. So at 10V out we are talking .4uV (.04ppm) at the input (~40mS gm) TOTAL error not even the nonlinear part. This can not be resolved from the other contributions and will be swamped by microscopic errors in PC board layout, etc. Yes it's obvious that base current changes with load, I don't feel making the point here is particularly productive.
 
#684
scott wurcer said:
Wavebourn, in this case of triple EF with bootstrap there is nominaly a current gain of say 10^6 (exact details are a little more complicated) and the designed load is 600 Ohms. So at 10V out we are talking .4uV (.04ppm) at the input (~40mS gm) TOTAL error not even the nonlinear part. This can not be resolved from the other contributions and will be swamped by microscopic errors in PC board layout, etc. Yes it's obvious that base current changes with load, I don't feel making the point here is particularly productive.
Click to expand...

Yes, cascading emitter followers you are decreasing input current, but are increasing it's variations by power of number of cascades. No free lunch in the real world. Russian saying: "Mice can find a free cheese, but in a mousetrap only".

Edit: I missed the word "Bootstrap". I like it. 🙂
 
#688
Copied, Xerox'd, and a memo sent to the parts logistics floor.
Thanks for sharing the entire shabang, Mr Ovidiu.

(to make sure that the first copycat production run floods in 3 months from now i'd like to add that the Patent Pending Copyright is lacking, easier for a non-Western search engine to find)
 
#689
Re: Site updated

syn08 said:
As promised, I have updated my site with all full schematics, Gerbers, descriptions, pictures and construction details for HPS 1.0, HPS 2.0 and HPS 3.1

www.synaesthesia.ca

Measurement results and listening tests/conclusions to follow soon.
Click to expand...


Hello Ovidiu

Just as an idea you can build reverse RIAA + attenuation and feed CD signal into it , and then compare it to the CD directly with the RIAA stuff bypassed. This is a good way of seeing how transparent your Phono stage is.

Regards
Arthur
 
#692
Re: Re: Site updated

PHEONIX said:
............Just as an idea you can build reverse RIAA + attenuation and feed CD signal into it , and then compare it to the CD directly with the RIAA stuff bypassed. This is a good way of seeing how transparent your Phono stage is..............
Click to expand...

Instead of building a reverse RIAA, you could also use LT Spice, draw a reverse RIAA and transform your WAV accordingly. Afterwards, feed it to the phono stage.

So you don't have to take in account non ideal caps, tolerances L/R etc.

Tino
 
#694
Re: Article on reverse RIAA filter

PHEONIX said:
Here is an artilce of what I am talking about by Lipshitz and Jung. I think that too often people dont realise that their RIAA stage can be easily sonically evaluated and how good it actually is can be easily determined by this bypassing back to the CD input.
Click to expand...

Thanks Arthur, I know the article and already played with this method. From a subjective listening test, I was unable to identify any kind of differences between HPS 1.0 2.0 and 3.0 (which probably only confirms I am stone deaf as most of the engineers 🙂. In fact, the only audible differences I was able to hear were due to the different headrooms (more about on my web site as soon as I'll have the time to put it together).

From a measurement perspective, I have a much better method. My DSA instruments have synthesis capabilities. That is, you can specify the poles and zeros of a whatever transfer function and the instrument will synthesize, store and display the frequency and time domain responses. Given a measured frequency response, I can calibrate the (say) 1KHz gain and substract the two responses. The residual is the measured difference to the idea RIAA frequency response. I can (and tried) to synthesize the reverse RIAA and get (after substraction) the residuals (there is no significant difference to the previous method).

Here's an example of such differences http://www.diyaudio.com/forums/attachment.php?s=&postid=1680457&stamp=1228866643 (this is without the 50KHz pole in the synthesized RIAA response). More about when I'll put together the Measurements page on my web site.
 
#700
Re: HPS measurements

Ovidiu - good work!

The noise data you show are at 1kHz if I understand the text correctly,
but it would be more interesting (and easier to compare with say Stereophiles measurements) I think to get a SNR value
with a 20-20k BW and both flat and A weighted values re 0.5mVrms.
Could you maybe use a soundcard to make such measurements?





Sigurd


syn08 said:
I have added the HPS measurements page to my web site. Conclusions and listening tests results to follow soon.

www.synaesthesia.ca
Click to expand...
 
Status
Not open for further replies.
Top Bottom