Modding (rebiasing) a 2-stage phono pre

I removed C03 and it did have a major impact on the input capacitance and impedance. Input capacitance dropped to 22pf and impedance to 44k. Both were welcome effects as I was wanting to reduce both for a better match with my cartridge. And no rfi issues to boot. I also ended up shunting C01 and based on listening tests I am going to keep it this way. Perhaps add a 50pf parallel at the input and see how it sounds.
but those ceramics need to go - as I was unsoldering one of them just crumbled and broke while I was pulling it out with a tweezer.

What do you guys think is the purpose of C05 between the transistors ?
 
A few small changes - and now I personally like the overall simulation result of the time-honored circuit.

There should be no instabilities and other spontaneous oscillations, C3 and C5 are not needed, but C9 must be reduced in size.
I would give it a try
🙂.


Openloop:

Vuo > 74dB
f1 < 25Hz
f2 > 36kHz

Closeloop (without C13, C11):

Vu = 56dB
f1 = 5Hz
f2 > 300kHz


and so on!
 

Attachments

  • percy_mm_eq.png
    percy_mm_eq.png
    7 KB · Views: 129
  • Like
Reactions: percy
I am guessing C03 and C05 are for preventing runaway oscillation or rfi interference so I left them alone.
Whether these capacitors are required or not - also depends on the transistor used.
I would remove them - absolutely. The reactance of the RIAA network in the negative feedback should ensure complete silence - no RF oscillations.

Try BC 550 B type as BJT.
 
  • Like
Reactions: percy
In short:

It's a really good question 🙂.
You might think that the AC short circuit should be realized with 47µF, 100µF or 470µF - but at this point you are mistaken. As soon as the loop is closed, a capacitor that is too large leads to a peak in the amplitude frequency response between 5Hz and 20Hz.

The passive low-pass filter at the output ensures that the falling edge continues to fall at -20dB per decade, because our negative feedback (and circuit) has a catch: above a certain frequency (let's talk about a fourth time constant) we can no longer go below 1. The amplification factor does not fall infinitely / finitely ..!

In a nutshell, the topic is more extensive than you might think.


Greethings,
HBt.
(a bit busy with CA 651P)
 
I have an update to share. Before jumping into making major changes I wanted to try my initial idea of reducing R05 to raise Vce of TR1. I dropped it to 100k from 220k (actually diff is 97k from 224k). Vce increased from 1.8v to 2.4v.
There was a very obvious improvement in sound quality especially in the midrange and some in treble. Obvious enough to make me repeat all measurements for FR, THD, IMD to find out where that difference was. I found no real difference in FR. Not much difference in THD either. No improvement in headroom either. But a huge improvement in IMD especially high frequency imd (tested with different combinations between 6khz and 9khz). Huge drop in harmonics and more improtantly huge drop in the spurious tones of imd which usually end up in the midrange (1khz-3khz). They all but disappeared!
Interesting side observation is that I was expecting the total gain of the preamp to drop (thinking gain of Tr1 will drop due to this change) but total gain did not change at all. Does this mean the global negative feedback was quite high to begin with ?
 
I don't even know how to answer because you should have some basic knowledge (back in mind). Blanket answers almost always ignore the core contexts and are therefore completely useless. The matter at hand is basically quite simple to answer: "Does this mean the global negative feedback was quite high to begin with?"

No, the entire negative feedback has not changed at all as a complex voltage divider.

A calculation example:

Model transistor
beta=297, rbe=4.85kOhm, rce=30.43kOhm

Vuo = (beta/rbe) * RC
k = RE / RC
Vuo' = Vuo / (k*Vuo +1)

The output stage amplifies in open loop by a factor of 612.4 -> 55.74dB. Nothing changes here for the time being.
The input stage amplifies in openloop mode by the factor
13472.2 -> 82.58dB. However, it is locally counter-coupled by RE = 330Ohm and everything changes. 13472.2 becomes 635.23 -> 56.06dB. You would think! But this stage is loaded by the second stage. A new and now also correct calculation must be made, we set the load resistance (dynamic input resistance) equal to rbe, as working resistance now follows RC1||rbe. The no-load gain drops to 290.6 (49.3dB), locally coupled -> 13.7 -> 22.74dB.

G_org = 22.7dB + 55.74dB = 78.44dB

Now we replace the RC1= 220kOhm by 100kOhm and calculate again; 283.33 -> 49.04dB coupled locally by RE and obtain -> 13.4 -> 22.51dB.

G_mod = 22.51dB + 55.74dB = 78.25dB

You have changed the operating point of both stages, but not the core gain and the overall negative feedback (which, by the way, is frequency-dependent and thus also the cool of the loop).

Your observations and measurements confirm the theoretical principles perfectly.


Best wishes,
HBt.
 
  • Like
Reactions: percy
Yes, the global feedback is high and so changes to the open-loop gain will not affect the closed-loop gain by much.
Two points:

1. the openloop gain of the two-stage system is virtually not changed at all by reducing the size of RC1, taking into account the load from the second stage (output stage, transistor 2). Because its dynamic resistance is much smaller than RC1 (220kOhm or 100kOhm).

2. what does "the overall negative feedback is high" mean?

I am missing the definition for high (in this case).


Greethings,
HBt.
 
If we look at the RIAA network in the bodediagram then we recognize three significant ratios for the kü of the loop considering the Vuo :
1 to 7.94 and
1 to 79.4 and
1 to 794 in percys case with the model BJT!

The resulting value of the loop, as a value or ration - is high' but only after 500Hz (better speak, think of 2122Hz).


Design

Basically, the aim is to achieve an openloop gain of around 80dB. This point is largely fulfilled -then you look for the optimum operating points for both transistors, taking into account the maximum and symmetrical signal level (deviation or span) -> this is actually the very first point to consider.

Analysis and synthesis always run in parallel,
HBt.
 
Last edited:
hbtaudio - My statement was general. When OLG >> CLG, then changes to OLG do not affect CLG by much.

As you showed, changing R05 from 220K to 100K does not affect the OLG by much. A more accurate calculation would consider Hfe at each operating point.

The amount of feedback is frequency dependent. I consider 40dB of feedback to be "high". The two-transistor amplifier starts to run out of loop gain at low frequencies.
Ed
 
  • Like
Reactions: hbtaudio