Phono preamplifier from Lineup & Fotios

[Salas]

Hi Salas

Did i get it all right then?
this will be a first

Yes, that's the ''head amp'' concept. The ''head'' rules the rest for noise spec.

 

[robmil]

From ARTISAND

dear fotios I am new to the site. I saw your phono amplifier today,and its very interesing.I'll suggest to you to replace all 1N4148 diodes by leds,replacing also emitter"s resistors with the approppriated values. After all these you will notice significant improvement in S/N and also exceptional clarity.

Going back to the Fotis/ Lineup opamp.... what is the advantage of replacing all 1N4148 diodes for LEDs? Does this technique work for all discrete opamp circuits?

rgs Rob

 

[fotios]

The difference between LED and diode is that: The voltage drop across the LED is stable at 1.8V so its If is very stable. The voltage drop across a diode is about 0.65V and presents small variations due to thermal dynamics, so its If is not so linear like a LED. For this reason, people claims that LEDs are less noisy from common diodes and this is by some way correct. Personally, if i had the intention to improve a little my discrette, i would like to replace all diodes with transistors that offer the BETA factor which gives the possibility of making CCS with exceptional linearity, unbeatable from diodes or LEDs.

 

[robmil]

Thank you Fotios. I think I understand. I take it then that the increase in voltage drops would mean recalculating the emitter resistors OR would it?

If replacing the diodes to transistors, is that simply connecting the transistors as diodes?

I think the reason I am asking is that I still see/read about new designs using 4148 diodes. If the advantages of 'diodes connected transistors' and LEDs replacing 4148s was measurable and audible why not just forget using 4148s in these positions altogether.

rgds Rob

 

[fotios]

Thank you Fotios. I think I understand. I take it then that the increase in voltage drops would mean recalculating the emitter resistors OR would it?

Because we need a fixed value constant current, then if voltage drop is increased emitter resistors should be recalculated. For example, two diodes connected in series present a total voltage drop of 1.3V. If replaced by one LED its voltage drop is 1.8V. If replaced by a transistor its Vb-e voltage drop is 0.65V.

If replacing the diodes to transistors, is that simply connecting the transistors as diodes?

No, in CCS the transistor that compensates the emitter current of the main transistor, is connected "as transistor". A transistor is connected "as diode" in current mirrors.

I think the reason I am asking is that I still see/read about new designs using 4148 diodes. If the advantages of 'diodes connected transistors' and LEDs replacing 4148s was measurable and audible why not just forget using 4148s in these positions altogether.

rgds Rob

For this issue there are two answeers:
1) A discrette implemented OPA, usually is connected as voltage follower, i.e. its Av=1. Only then it can presents excellent bandwidth up to 1MHz. From my experiments, a discrette implemented either with top BJTs like BC550C-BC560C or with JFETs presents same bandwidth, up to 1MHz, as far as is working with unity gain (Av=1). The main difference between them is that JFETs present very high input impedance compared to BJTs, which in practice is translated that we can connect in their input e.g. a volume control potentiometer of any value without degrading performance. Instead, BJTs can accept in their input potentiometers of 10KΩ as much.
2) Use of transistors as thermal compensation elements instead diodes, it increases the cost without any audible profit (IMHO). Though transistors have better noise figure than diodes, in a discrette implementation that works with Av=1 we have "by hand" a very - very low noise level. How much can be decreased the noise level more by substituting diodes with transistors? From -100dB to -102dB?