0dbV = 1V , i hope we agree on this...
So an equivalent INPUT noise of -147dBV = 0.0000000446 V = 44.6 nV.
Is that the EIN of your preamp ?....
Those interested in an example of ultimate performance in low-noise power supplies can try the circuit given in Horowitz & Hill, section 6.21
This supply yields 6uV pk-pk, (0.1 to 10Hz) at the output. It does not use stacked references, but a single 10V buried-zener, amplified to 25V with low noise amplification and 2 stages of filtering, to eliminate noise gain.
This supply yields 6uV pk-pk, (0.1 to 10Hz) at the output. It does not use stacked references, but a single 10V buried-zener, amplified to 25V with low noise amplification and 2 stages of filtering, to eliminate noise gain.
"Well, it seems I KNOW BETTER. I am rather familiar with feedback theory et al, hence it is a complete mystery how anyone would declare a feedback system without build-out network to be stable into capacitive loads."
Methinks thow dost protest too loudly, sir.
Putting suitable cap (so one with some ESR) across the output of a regulator - and especially a three terminal one like the 317, is a well recognized technique for compensating them.
Back to the -147dBVA subject. Your claims are not what they appear to be . . . please give us the details of exactly how you calculated them and what the set -up was. Maybe we can learn something from you
Methinks thow dost protest too loudly, sir.
Putting suitable cap (so one with some ESR) across the output of a regulator - and especially a three terminal one like the 317, is a well recognized technique for compensating them.
Back to the -147dBVA subject. Your claims are not what they appear to be . . . please give us the details of exactly how you calculated them and what the set -up was. Maybe we can learn something from you
His "methodology" is quite simple it seems..
Say a 40dB gain preamp.
Inject 1V at its output wich yield 100V output.
Then calculate the ratio 100V/preamp Input noise in V.
If you want more S/N ratio just increase the gain...
In this case the real reference is 100V so the number should
be in -db100V , wich would be right but Thorsten erroneously
still talk of -dBV.
Last edited:
Hi,
Yes, A-Weighted. It is MC Phono.
BTW, my point was that Sy had declared that cascodes have no PSRR and I used this design which has independent objective measurements confirming my claims and which uses a (folded) cascode as example that a competently implemented cascode has in fact excellent PSRR.
I would not be able to get anywhere near these noise levels if it did not.
The Circuit is actually sensitive enough to show up the differences in the resistors used in the circuit as noisefloor changes.
Ciao T
Yes, A-Weighted. It is MC Phono.
BTW, my point was that Sy had declared that cascodes have no PSRR and I used this design which has independent objective measurements confirming my claims and which uses a (folded) cascode as example that a competently implemented cascode has in fact excellent PSRR.
I would not be able to get anywhere near these noise levels if it did not.
The Circuit is actually sensitive enough to show up the differences in the resistors used in the circuit as noisefloor changes.
Ciao T
Hi,
Well, well well. But a circuit that is stable does not to compensated additionally, otherwise it clearly cannot be considered stable.
It is a simple logic that "Not A" cannot be equal to "A".
I am not here to teach you. I am not here to debate independently verified performance claims. Especially not when the facts are well known to you.
I do not play this game. Go bother someone else.
Ciao T
Well, well well. But a circuit that is stable does not to compensated additionally, otherwise it clearly cannot be considered stable.
It is a simple logic that "Not A" cannot be equal to "A".
I am not here to teach you. I am not here to debate independently verified performance claims. Especially not when the facts are well known to you.
I do not play this game. Go bother someone else.
Ciao T
Hi,
Correct.
And hence when we have a variable gain device (in my case between 30dB and 72dB) to list the EIN allows SNR may be derived directly by comparing the input level to input noise level, without having to consider gain and any complications.
It is about the only way of specifying noise that is truely honest by it's very nature.
Ciao T
Correct.
And hence when we have a variable gain device (in my case between 30dB and 72dB) to list the EIN allows SNR may be derived directly by comparing the input level to input noise level, without having to consider gain and any complications.
It is about the only way of specifying noise that is truely honest by it's very nature.
Ciao T
Hi,
The methodology for deriving Equivalent Input Noise is indeed very simple and should be known to anyone who has practiced electronics at least at technician levels, never mind degreed EE's.
Look it up one day.
Ciao T
The methodology for deriving Equivalent Input Noise is indeed very simple and should be known to anyone who has practiced electronics at least at technician levels, never mind degreed EE's.
Look it up one day.
Ciao T
Whatever the source or the topology this number is quite good
and would amount to about 80 dB S/N ratio for a 0.5mV output MC
as well as 0.3nV/sqrt hz input noise density.
Although the RIAA curve allow to cut a lot of the noise
that s still better than many MM preamps that have the
same advantage.
Hi,
Indeed.
The circuit design adopted keeps this Ein for gains from 54dB to 72dB.
Yup. It was one of the design goals to avoid transformers and to be able to use the lowest output MC's (around 0.05mV @ 5cm/S) without excessively compromising noise. As the "full scale" on LP is 25cm/S we get an effective SNR of 75dB below "analogue full scale" or 60dB below 0dB.
Another way of looking at this is to say that with a 0.05mV @ 5cm/S output cartridge and 72dB gain the "analogue full scale" output will be 1V with 75dB SNR referred to "analogue full scale".
For a 0.5mV @ 5cm/S output cartridge with 60dB gain the "analogue full scale" output will be 2.5V with 95dB SNR referred to "analogue full scale".
Ciao T
Indeed.
The circuit design adopted keeps this Ein for gains from 54dB to 72dB.
Yup. It was one of the design goals to avoid transformers and to be able to use the lowest output MC's (around 0.05mV @ 5cm/S) without excessively compromising noise. As the "full scale" on LP is 25cm/S we get an effective SNR of 75dB below "analogue full scale" or 60dB below 0dB.
Another way of looking at this is to say that with a 0.05mV @ 5cm/S output cartridge and 72dB gain the "analogue full scale" output will be 1V with 75dB SNR referred to "analogue full scale".
For a 0.5mV @ 5cm/S output cartridge with 60dB gain the "analogue full scale" output will be 2.5V with 95dB SNR referred to "analogue full scale".
Ciao T
Hi,
Let us be clear.
Given is:
1) A Noiseless regulator circuit (for clarity I wish to disregard the circuit noise in this set of examples) that requires a reference voltage to operate.
2) A Circuit as above that has had the DC gain set according to the requires output voltage and the AC gain set to unity (one) within the bandwidth of interest.
3) A set of reference sources with a given DC Voltage and a given AC noise.
The three cases to evaluate are:
A) The DC Gain is also set to unity and multiple references are stacked to gain the required DC voltage. AC gain in the bandwidth of interest, as stated above, is unity.
B) The DC Gain is also set to unity and a single reference is used in conjunction with a transistor that has much lower noise and is connected as "amplified zenner" with unity gain in the bandwidth of interest to give the required DC Voltage. AC gain in the bandwidth of interest, as stated above, is unity.
C) The DC Gain is set to the value that is required to produce the required DC output voltage using a single reference, a single referenceis used. AC gain in the bandwidth of interest, as stated above, is unity.
Which of the above cases offers the lowest noise, which is second, which is the looser?
Ciao T
Let us be clear.
Given is:
1) A Noiseless regulator circuit (for clarity I wish to disregard the circuit noise in this set of examples) that requires a reference voltage to operate.
2) A Circuit as above that has had the DC gain set according to the requires output voltage and the AC gain set to unity (one) within the bandwidth of interest.
3) A set of reference sources with a given DC Voltage and a given AC noise.
The three cases to evaluate are:
A) The DC Gain is also set to unity and multiple references are stacked to gain the required DC voltage. AC gain in the bandwidth of interest, as stated above, is unity.
B) The DC Gain is also set to unity and a single reference is used in conjunction with a transistor that has much lower noise and is connected as "amplified zenner" with unity gain in the bandwidth of interest to give the required DC Voltage. AC gain in the bandwidth of interest, as stated above, is unity.
C) The DC Gain is set to the value that is required to produce the required DC output voltage using a single reference, a single referenceis used. AC gain in the bandwidth of interest, as stated above, is unity.
Which of the above cases offers the lowest noise, which is second, which is the looser?
Ciao T
A is noisy.
B is noisy.
C is noisy.
We just aren't talking the same language T. I just don't follow you.
How about D - a low noise reference, in a regulator with enough open loop gain to have very good error correction, when closed loop.
You can always filter the reference too, so it is quieter.
Scott,
As I am not immortal it is all AC to me! Of course not calling DC a "signal" is based on a short term of vision!
Stu,
That is an interesting way of looking at it. Often correct results but lacking the precision of the concept that the same impedance must also be involved when the voltage doubles resulting in the current doubling.
Jan,
Actually even in a stable environment adding a capacitor decreases the noise until the self resonant frequency, where the dielectric losses act as an interesting noise source.
There also is some vibration induced noise added by all charged capacitors, not just some ceramic types. Then there is thermo-electric issues if the capacitor has a temperature differential across it (Doe to one end being near the heatsink of the regulator.) Also there can be triboelectric issues in some cases.
A multiple stage approach where each stage not only decrease ripple but specializes in different parts of the noise spectrum may be required for the lowest noise. So when things get really quiet new issues pop their head up.
Have you ever seen a "Whack a Mole" game?
ES
Hi,
Well, we can quantify how, noisy right.
That would be case C.
Given the same filter, it does not matter. The same relative differences in levels of noise will be observed.
But I will readily agree that filtering a reference is a much better solution to low noise than stacking them.
Ciao T
Well, we can quantify how, noisy right.
That would be case C.
Given the same filter, it does not matter. The same relative differences in levels of noise will be observed.
But I will readily agree that filtering a reference is a much better solution to low noise than stacking them.
Ciao T
Hi,
I bought one like this a few weeks ago for my two years old daughter. It has worms in apple instead of moles, but the principle is the same...
Yes, this is a lot of a problem in electronics, hammer something on the head in one place and something else pops up in another...
Ciao T
I bought one like this a few weeks ago for my two years old daughter. It has worms in apple instead of moles, but the principle is the same...
Yes, this is a lot of a problem in electronics, hammer something on the head in one place and something else pops up in another...
Ciao T
- Status
- Not open for further replies.