usekgb
Well-known member
I love the old school, drawn with a pen, self etch! Very diy.
Still learning design preamp with LTspice(Some help please)
- Thread starter ForthMonkey
- Start date
Help Support GroupDIY Audio Forum:
ForthMonkey
Well-known member
What about transistors? I will use BC550C & BC560C.But if i can't find,can i use BC550B & BC560B? Only hfe different,i guess.B smaller than C.
BC550C-Forward current transfer ratio (hFE), min: 420
BC550B-Forward current transfer ratio (hFE), min: 200
BC560C-Forward current transfer ratio (hFE), min: 420
BC560B-Forward current transfer ratio (hFE), min: 240
What about it?
BC550C-Forward current transfer ratio (hFE), min: 420
BC550B-Forward current transfer ratio (hFE), min: 200
BC560C-Forward current transfer ratio (hFE), min: 420
BC560B-Forward current transfer ratio (hFE), min: 240
What about it?
maybe you could edit that parameter editing the component... if not look for another general purpose transistor with the specs you are looking, or try with the BC5#0B, for a first apron may work, it should be in the library in some other place, in another type of transistors or something...
JS
JS
"C" should be easy enough to find.ForthMonkey said:
Using "B" instead will not change much in performance, because the circuit is more important than the components in a heavily feed-backe'd application.
BUT, noise performance will be impaired.
Roughly, for the same operating conditions, the base current of the input transitors will double, which will produce a 3dB increase in Input Noise Current, which may be significant, or not, depending on the source impedance. In the particular case of a mic preamp, the optimum source impedance is about 5-10k. Increasing noise current may shift it down to 3-5k. You use a 1:10 input xfmr, which makes the source impedance 20k, which is probably already a little too high for the "C" version. Using "B" may make the Input Noise Current dominant over the Input Noise Voltage. (Optimum source Z is when Input Noise Voltage is equal to Input Noise Current multiplied by source Z). Anyway, in the end it's only a couple dB over the absolute minimum; in actual operation, it shouldn't make a big difference. In most cases, using the right mic and positioning it properly is more significant than the preamp's noise performance.
ForthMonkey
Well-known member
abbey road d enfer said:
Thanks!
Actually yes,"C" easy enough to find.But for now i can't find.I've looked every online markets in my country but i can find only "B".I will figure it out one way or another...And i will try "C" and "B" with this circuit.Then i can see what is difference and why should i use or why i shouldn't..."B" medium gain and "C" high gain,i guess.
BTW i'm trying to simulate "Noise".What about it?Can i trust result?Or how should i do simulate "Noise" to getting "real life" noise performance?
ForthMonkey
Well-known member
BTW i added noise simulation.Can someone explain this?
What was the gain for the noise sim?
It's showing about 80 nV/sqrt(Hz). If the gain is 4, that's 20 nV/sqrt(Hz) input noise at the opamp input.
Now the voltage source you are using in your sim has 0 output impedance. That means that the noise analysis only reflects the voltage noise component of the opamp noise. Use a realistic impedance like 200 ohm for a microphone (times 100 since you were using a 1:10 input transformer, right? So 20 k source impedance seen by the opamp.)
To optimize noise for a given source impedance, you should balance the current noise and voltage noise contributions. You do that by changing the current in the input stage: increasing the current gives you higher gain (transconductance or gm), which translates into lower voltage noise, but more current in the emitter also means in the base, and thus more current noise.
It's showing about 80 nV/sqrt(Hz). If the gain is 4, that's 20 nV/sqrt(Hz) input noise at the opamp input.
Now the voltage source you are using in your sim has 0 output impedance. That means that the noise analysis only reflects the voltage noise component of the opamp noise. Use a realistic impedance like 200 ohm for a microphone (times 100 since you were using a 1:10 input transformer, right? So 20 k source impedance seen by the opamp.)
To optimize noise for a given source impedance, you should balance the current noise and voltage noise contributions. You do that by changing the current in the input stage: increasing the current gives you higher gain (transconductance or gm), which translates into lower voltage noise, but more current in the emitter also means in the base, and thus more current noise.
ForthMonkey
Well-known member
VictorQ said:
Thanks for explanation.
How can i add impedance? Should i specify Rser=200? Or what?
keefaz
Well-known member
Just add a resistor in serie with V1
You should either edit V1 and specify internal resistance 20k or add a 20k resistor in series with V1.ForthMonkey said:
ForthMonkey
Well-known member
Thanks for advices! I edited first post and added noise.Now can someone explain?
Explain what? If your question is how to run a noise analysis in LTspice, I really recommend you post on the LTspice forum. In the meantime I suggest you look at some of the examples in the LTspice website and that you google wiki ltspice; you will find several tutorials.ForthMonkey said:
ForthMonkey
Well-known member
I've learned how to analysis noise in LTspice.But i need info about calculating signal to noise ratio.
I analysed highest gain.And it's result.
Noise about 1.882uV.
Now how can i calculate signal to noise ratio?
I used this link.
http://www.sengpielaudio.com/calculator-noise.htm
Noise voltage level (audio) Lu=-112.28912 dBu
Noise voltage level LV=-114.507608 dBV
So?
OK. I'll explain, but it's not the most useful figure, because it depends on the level of wanted signal. I'll tell you first about EIN (Equivalent Input Noise).ForthMonkey said:
No. Read carefully the graph and particularly the units on the vertical axis. It is in uV.Hz(exp1/2) or, in more practical language, in
microvolts per square root of Hertz. That is called noise density. In order to calculate the rms noise voltage, you need to integrate the noise density over the frequency range of interest. As you can see, the noise density varies significantly over the frequency rage, so integrating it takes a graphical integration method.
But if you concentrate on the audio frequencies, (conventionally 20Hz-20kHz)the noise density is rather constant, so the integration is quite simple and results in
vn=Noise density.sqroot of freq range
So the rms noise level is 1.8.10E-6.sqrt(19980)=253uV or -69.7dBu
This is noise measured at the output, correct?
So now you divide this by the gain of the preamp, 45dB(?)
And end up with EIN=-114.7dBu
Which is just impossible because the intrinsic noise of a 20k resistor is -109.7dBu
So there is something wrong in theses results.
Are you sure of the 20k source resistor? or the actual 45dB gain?
You should post the .asc file so we could easily check your circuit. I believe this site does not accept .asc files but you can rename it in .txt, or, alternatively, you could post in on the LTspice site in the Files-> Temp folder. Doing this would help us to help you.
Once you know the actual EIN, calculating S/N ratio is very simple.
You just divide the source signal by the EIN, or rather substract them in the log (dB) domain.
Let's say you end up, after redoing your simulation (or measurement) -100dBu EIN - without xfmr, then you add the 1:10 xfmr, which puts the EIN at the xfmr primary at -120. Let's say the actual mic signal is -58dBu, the S/N ratio then is 62dB.
So you can see by entering 200r that the intrinsic noise level of a 200r resistor is -129.7dBu. This is a figure that you should know by heart, like pi=3.14, and remember that it follows the sq root of the value in ohms and also follows the sq root of the frequency range.
ForthMonkey
Well-known member
Thanks for explanation abbey road d enfer
I'm still trying to understand that you wrote.Cause English is not my main language.
It's .asc...
https://dl.dropboxusercontent.com/u/91809016/urei.asc
I'm still trying to understand that you wrote.Cause English is not my main language.
It's .asc...
https://dl.dropboxusercontent.com/u/91809016/urei.asc
When you design a circuit with LTspice, it is saved as a file with the suffix ".asc". In fact it can be opened with a simple text editor and you will see the name of the circuit, the list of components with their value and the nodes to which they are connected and also the characteristics of the analysis (AC, transient, DC...). If you post this file in the Files-> Temp directory on the LTspice, the members of the list can open the file and run the simulation and help you solve your problem.ForthMonkey said:
ForthMonkey
Well-known member
abbey road d enfer said:
Uploaded. "urei.asc"
https://groups.yahoo.com/neo/groups/LTspiceFiles/files/TEMP/
OK. So now with gain at about 40dB, the output noise voltage computes at 0.26uV or -68.4 dBu, which in turn computes at 108.4dBu EIN or -128.4dBu reflected at the xfmr primary. This is not a bad figure at all, particularly if you can achieve it in practice. All sorts of gremlins will conspire to make this figure unachievable in real life.ForthMonkey said:
In the simulation, if you make tye source resistance very small (e.g. 1r) this will show the result of the Input Noise Voltage only (ignoring the contribution of the noise current into the source resistance). I see 0.4uV instead of 1.87. That shows that the contribution of the noise current totally dominates the overall noise performance.
If you increase the source resistance to the point where the noise voltage density increases by 3dB, that will be the optimum source resistance, where the contributions of Input noise Voltage and Input Noise Current are equal (because they are not correlated, they combine quadratically). I see it happens for Zsource=ca. 1500r.
Now it seems the simulation doesn't react well at changing the operating point of the input transistors.
