I still can not get my hands of the Echo. It simply sounds too good to be discounted as waiste. When i go trough the literature and look through the posts and contributions i have seen here this is the situation : Bias varies a lot per transistor in designs that claim low noise in a MC Head-Amp. This is what i found : Leach, 0.125mA, Self, 1mA (that what i used to do), Accuphase, 3mA, Syn08, 6mA. What i can contribute from experience is that transistors can have lower noise when the powersupply voltage is very low. I build a Leach type MC Head-Amp in 1983 (without knowing the Leach at that time) that ran on a single 1.5V battery, had 2 x BC550C and 2 x BC560C in parallel and had very low subjective noise. I think Quad did a circuit too that ran the transistors on very low voltage with quite a low Bias that is known for low subjective noise. Anyway, what i will try is to parallel 8 BC550C with a bias of 3mA each, a kind of compromise of all sorts. I will then measure the noise voltage at the speakers and listen how the noise sounds. Do not understand me wrong, the BC550C is not ideal for this job but most anybody here in Europe has them in the waistebin so this is a low cost attempt.
Correct SY, conventional RIAA stages (with cap and resistor elements) can be made not to contribute significantly to the noise and i have done that here. Still the inductive solution sounds different to my ears and some people apreciate that i try inductive because they find that interesting. One reason my inductive stages may sound different is the totally different topology i am using. The inductive stage has a 6dB falling response from 50Hz until it bottoms out at gain 1x at higher frequencies. The advantage could be that the open loop gain of an opamp also falls with 6dB octave so there is constant feedback availlable. The other advantage is that the second stage is presented with a constand velocity signal that is much less demanding in terms of slew rate. I found that stages can be easyly brought into problems with intense high frequency signals that can come from record blemishes and dirt. Anyway, i do it for the sound mostly.
Agreed, but... My Benz Micro Wood has 0.44nV/rtHz (12ohm). Joachim's Lyra Titan has 0.3nV/rtHz (5.5ohm). So, according to your criteria, HPS 4.1 at 0.28nV/rtHz is not good enough for either
I know YMV, but that's why I use 5mV Grado @ 400 Ohms even Salas style circuits are good enough. In fact even some LTP's can work.
EDIT - Just had an idea, maybe a figure of merit like noise-power/signal-power could be worked out. This is not a match terminated situation but there should be a common point of comparison.
Last edited:
And that's why you need a JFET input pre. 1nV/rtHz and zero current noise is better than any Grado owner will ever dream.
Edit: I have suggested as a MC pre metric (Gain @1KHz) X (Dynamic Range) X (S/N ref: 1mV, wideband)
HPS 4.1: 60.4 + 32 + 78.0 = 170.4dB
HPS 3.1: 60.8 + 32 + 77.0 = 169.8dB
HPS 2.0: 64.3 + 26 + 74.3 = 164.6dB
HPS 1.0: 64.3 + 26 + 67.1 = 157.4dB
Last edited:
From a 56dB gain 1 jfet input build non cascode, just Rload drain, I got this for noise floor ref 0dBV=1V out. I used a 1/200 10k series 50R parallel attenuator to obtain 0.5mV from 100mV Emu Tracker Pre usb box generator and I put the phono in its loop. Without Lpad, gen's output is rather noisy under 500mV and can't be controlled precisely under 50mV for asked output. That high frequency blip is an artifact, most of the times isn't there. Is this noise floor result possibly fake?
P.S. I found the output impedance of the card at 580 Ohm when I calibrated the levels for different loads.
P.S. I found the output impedance of the card at 580 Ohm when I calibrated the levels for different loads.
Attachments
Not sure how your attenuator is built, but for noise measurement purposes you'd better repeat the measurement with the input shorted to ground. The generator and the attenuator may add much more noise than the pre itself.
I just remembered that I did an S parameter complex noise figure spreadsheet 15 yr. ago that might apply here. The problem then was ultrasound transducers connected at the end of lossy coax. An Aglient box was used to extract S parameters at frequencies of interest. Here we could just use the lumped R/L/C and find the best front end termination/amp combination.
EDIT - Still HP at the time.
EDIT - Still HP at the time.
My commercial Goldstandard is balanced. The good thing is that this needs no ground wire so works even when the groundwire is connected to the tonearm and is not seperatetely put out like in some Rega turntables. Hum and noise in unbalanced Head-Amps can be very low nevertheless even with low impedance cartridges. The Goldstandard sounds very clean, but if that is the result of the ballanced connection i can not say. I have also an experimental ballanced stage with INA163 that works fine but the necessary gain resistor in the intrumentation amp reduces the noise performance to something like 1.4nVQHz.
I just remembered that I did an S parameter complex noise figure spreadsheet 15 yr. ago that might apply here. The problem then was ultrasound transducers connected at the end of lossy coax. An Aglient box was used to extract S parameters at frequencies of interest. Here we could just use the lumped R/L/C and find the best front end termination/amp combination.
EDIT - Still HP at the time.
Interesting idea, I'll look into this, it shouldn't be to difficult. I can do it directly, unfortunately only from 300KHz up to 6GHz, so not really useful here.
What was the reference impedance for your measurements?
That work was done at 50 Ohms. A series of equvalent two ports was very easy to set up, i.e. an X ft. piece of RG58 at a given f was a simple two port. For this problem I would start with simple lumped parameters. Several mirowave design texts have tables of two port equivalents for R-L-C networks even transformers. Voltage and current noise were simple sources.
The problem remains what do you do when the best noise performance has a cartridge time constant well into the audio band (and not one accounted for as with MM).
The problem remains what do you do when the best noise performance has a cartridge time constant well into the audio band (and not one accounted for as with MM).
Last edited: