So if you'd like this even lower noisefloor, you could sacrifice some PSRR and output impedance performance and go with denoiser, and you could compensate for PSRR with the capacitance multiplier. But I had the pcb on the desk, with no shielding. It measured at -182.13dBV which means 0.78nV/sqrtHz as opposed to the 0.84nV/sqrtHz of the dienoiser. I did measure with just the denoiser and the noise is a touch lower than with the dienoiser. Playing with the sim it seems that with either BC817 or ZTX851 you'd get the noisefloor at around 0.42 to 0.47nV/sqrtHz. So basically to switch from the dienoiser circuit to the denoiser one you just have to short the 560R resistor thus taking the second dienoiser transistor out of the circuit.
So apparently, opposed to LM317+denoiser, these discrete circuits have a really low noise with just the denoiser. Third photo is how I installed the BC8x7 on the BC3x7 footprints. Second photo is measurement of noisefloor with BC8x7. For some reason the negative versions don't have a lower noisefloor with the BC8x7 pair instead of BC3x7, but they do simulate a lower noisefloor with the ZTX pair.įirst photo is measurement of noisefloor with BC3x7. This seems to be valid for both positive versions of both circuits. The top one is the BC807 and the bottom one is the BC817. I made a photo of how I installed them on the BC3x7 footprints on the backside of the PCB.
So using the BC8x7 pair results in an even lower noise floor. I then measured with the BC8x7 pair and it came out at -181.5dBV which means 0.84nV/sqrtHz. I measured the board with the BC3x7 pair and it came out at around -178.3dBV which means 1.21nV/sqrtHz. LTSpice sim predicts 1.23nV/sqrtHz using BC327/BC337 pair for the denoiser, and it predicts 0.63nV/sqrtHz using BC817/BC807 pair. I measured the noise for the positive version of the first circuit.
0 kommentar(er)
