Low Noise Preamp Design Questiuon

Thats OK if you express the noise level in dB relative to 1 milliwatt, which gives the same figure regardless of the resistance. If you use a

50-ohm microphone and express the noise level in dBm, the calculation becomes much more complicated because dBm refers to 1 milliwatt into 600 ohms and is a voltage-based measurement.

I think you have to calculate the voltage across 50 ohms which corresponds to the noise power, then work out what power that would give if it were across a 600-ohm resistor. Then that power is compared to 1 milliwatt using a decibel scale.

[...]

I don't know if it is still true, but it used to be said that Nuvistors were better than any semiconductor device for low LF noise. They were particularly useful for the front ends of electroencephalograph machines and for low jitter V.F.O.s. at R.F.

Here's a little tool my son Aaron wrote for me when I was doing a WiFi repeater chip design and the RF guys were driving me nuts talking eveything in dBm...

You can enter dBm & impedance or Vp-p & impedance.

...Jim Thompson

Sorry, this is rubbish - I was thinking "dBv" and typing "dBm" [...]

Pity it doesn't work on a Mac. ...but I've written a calculator in ClarisWorks4 spreadsheet, for my own use, which does noise calculations for any combination of temperature, bandwidth and resistance.

If any Mac users are interested, I have put a copy at:

I have a macro in PSpice that does that, so I can display noise figure directly ;-)

...Jim Thompson

[...]

It's interesting to find out how many tools like that have been written by people for their own use. I have another one I use for time constants, which saves me no end of work (bundled in the above '.sit' file).

One problem I have never been able to crack is a method of automatically selecting series and parallel combinations of preferred-value resistors to give an accurate non-standard value. I have approached it with a spreadsheet showing a grid of the various possible combinations, but some sort of iterative program would be better.

Golly 'dern... it's binary :-(

Mine...

[MACROS] BETA(A)=I(A:c)/I(A:b) CAPINT=IMAG(-I(VAC))*IMAG(-I(VAC))-REAL(-I(VAC))*REAL(-I(VAC)) CAPRES=REAL(-I(VAC))/CAPINT CAPVAL=CAPINT/IMAG(-I(VAC))/(2*pi*FREQUENCY) CENTER(A)=A-MID(A) dBm(N,Z) = VdB(N)*sqrt(500/Z) GAIN(A,B)=(MAXX(A)-MINN(A))/(MAXX(B)-MINN(B)) GDB(A, B) = DB(PP(A))-DB(PP(B)) IMAG(A)=IMG(A) LoopGain=DB(RAWGAIN) LoopPhase=P(RAWGAIN*(-1)) MAGGAMMA=M(2*V(IN_NODE)-V(SRC_NODE)) MAGN(A)=M(A) MAXX(A)=MAX(A+100)-100 MID(A)=(MAXX(A)+MINN(A))/2 MINN(A)=MIN(A-100)+100 MODULATION(A)=(ENVMAX(A,1)-ENVMIN(A,1))/2 NF = 10*LOG10((V(INOISE)*V(INOISE))/VSRCSQ) NINETY(A)=A-0.9*MAXX(A)-0.1*MINN(A) OUTNOISE=SQRT(s(V(ONOISE)*V(ONOISE))) PF=AVG(-I(VPF)*V(VPF:+)/MAX(V(VPF:+)))*sqrt(2)/RMS(-I(VPF)) PH(A,B)=P(A)-P(B) pi=3.14159265 PL1 = V(LG)@1 PL2 = V(LG)@2 PP(A)=MAXX(A)-MINN(A) PTAT(A)=(273+TEMP)/300*A Peak2Peak(A)=ENVMAX(A,1)-ENVMIN(A,1) QL1 = I(VMAC)@1 QL2 = I(VMAC)@2 RAWGAIN=PL1*(QL1*PL2+QL2*(1-PL1))/(QL2*PL1*PL1-QL1*PL2*(1+QL1)) REAL(A)=R(A) RET_LOSS=20*LOG10(MAGGAMMA) ROOT4KTR(A)=129P*SQRT(A) S11=2*(V(IN_NODE)/V(SRC_NODE))-1 S21=2*V(OUT_NODE) SATMARGIN(M,THRES)=MAGN(VD(M)-VS(M))-MAGN(VG(M)-VS(M))+MAGN(THRES) SOAG(M,VDS,VGS)=(SGN(VD(M)-VS(M)-VDS)+1)*(SGN(VG(M)-VS(M)-VGS)+1)/4 TEN(A)=A-0.1*MAXX(A)-0.9*MINN(A) VSRCSQ=16.5757565E-21*V(RSRC:1,RSRC:2)/I(RSRC) VSWR=(1+MAGGAMMA)/(1-MAGGAMMA) VdBnorm(A,B)=VdB(A)*PWR(10,(B/20))

I don't have that problem... if I need a 137K resistor I just calculate the silicon length for the width I've chosen... also automated, I just build the foundries equations into my template generator ;-)

Like this...

*symbol XFAB-XB06-RPOLYH X-FAB XB06 PolyH Resistor, 1.25K ohms/sq | REFDES=R? TEMPLATE=X^@REFDES %1 %2 @MODEL PARAMS: R=@VALUE W=@W VALUE=10K USER1=POLYH MODEL=RPOLYHC LVS=X^@REFDES %1 %2 @MODEL PARAMS: R=@VALUE W=@W ; Design Length = @LDRAWN\\m W=2u RSH=1250 NCONT={1+((@W-2*0.4u)/1.2u-0.499)} RESRC={20/@NCONT} LDRAWN={(@VALUE-2*@RESRC)*@W/@RSH}

...Jim Thompson

It's a Stuffit file containing two Claris Works spreadsheets. Some PC un-Zipping programs will also handle Stuffit expansion and there is a PC version of Claris Works which will read the Mac versions if they are given a '.cwk' file extension. Better still just bung them into a Mac running Classic OS and CW4.

I tried saving them as Excel spreadsheets, but the differences are too great to be bridged by the translator.

...

Golly 'dern... it's ... it's ...well whatever it is, I can't run it.

All my jobs were one-offs on tagboard ...until recently, when the manufacturers stopped making my preferred type of tagboard. Now they are built on Veroboard (perforated stripboard), take five time as long to debug because of all the hidden short-circuits and are far less reliable.

It's simply Algebraic... substitute Q5 for A, then you get...

BETA(Q5)=I(Q5:c)/I(Q5:b) | | Collector current of Q5---^ ^-- Base current of Q5

Convenient for watching beta changes _during_ a transient analysis

I used to use this, in 1' x 1' pieces...

I'm considering making up a similar thing in around 4" x 6" pieces, but changing the general arrangement to handle leadless devices.

For G-jobs ;-)

...Jim Thompson

That was his parent's excuse, too. :(