Hi, its because you only had 5 posts. Each of those automatically goes through the moderation queue and evaluated to avoid spam posting robots etc. The system releases at >5. Now at 6 you should be free for real time posting and allowed to pm.
I did some noise measurements of Salas Ultrabib regulator used by me for UGS modules (UGS Muse preamp).
Before Salas Ubib there is a main power supply. RCLC filter (1R0 - 22000uF - 100mH/1R5 - 22000uF) on each arm.
Salas Ultrabib (CCS 150mA) :
Vin=33Vdc Vout 24Vdc
Rf - drossel 220uH 0.38R 2.4A
C1 - 8200uF/56V Gold Audio
C2 - 470uF/50 Nichicon Muse
C3 - 33uF/50V Nichicon ES
M1 - IRF9610
J1, J2, J3 - 2SK117GR (matched about 6mA Idss), source resistors 220R for 1.4-1.6mA current. !!! Pinout different from PF5102 !!!
The measurements were made with my Ultra Low Noise Preamp and RMS Meter, powered by modified Jung Super Regulator (I use the sames for +/-15V Muses 72320, in the photos the regulators are next to the UBiB), HP3582A, Agilent 34410A.
Ultra Low Noise Preamp and RMS Meter
Bandwidth noise results:
-------------------------
1.73 µVrms (Bandwidth 0Hz-1MHz)
1.22 µVrms (Bandwidth 0Hz-100kHz)
0.102 µVrms (A-Weight)
noise floor measured with HP3582A:
----------------------------------
0.4µV
I do not claim serious and accurate measurements.
More like quick measurements made out of curiosity.
Before Salas Ubib there is a main power supply. RCLC filter (1R0 - 22000uF - 100mH/1R5 - 22000uF) on each arm.
Salas Ultrabib (CCS 150mA) :
Vin=33Vdc Vout 24Vdc
Rf - drossel 220uH 0.38R 2.4A
C1 - 8200uF/56V Gold Audio
C2 - 470uF/50 Nichicon Muse
C3 - 33uF/50V Nichicon ES
M1 - IRF9610
J1, J2, J3 - 2SK117GR (matched about 6mA Idss), source resistors 220R for 1.4-1.6mA current. !!! Pinout different from PF5102 !!!
The measurements were made with my Ultra Low Noise Preamp and RMS Meter, powered by modified Jung Super Regulator (I use the sames for +/-15V Muses 72320, in the photos the regulators are next to the UBiB), HP3582A, Agilent 34410A.
Ultra Low Noise Preamp and RMS Meter
Bandwidth noise results:
-------------------------
1.73 µVrms (Bandwidth 0Hz-1MHz)
1.22 µVrms (Bandwidth 0Hz-100kHz)
0.102 µVrms (A-Weight)
noise floor measured with HP3582A:
----------------------------------
0.4µV
I do not claim serious and accurate measurements.
More like quick measurements made out of curiosity.
Attachments
Last edited:
Hi, excellent documentation, photos, and results. Congratulations. I see 3.92Ω CCS resistor ~150mA limit. Can't see test load value & spare mA?
1.22μVrms over 100kHz bandwidth is equivalent to 920Ω resistor thermal noise at 20 degrees Celsius. Or 0.55μVrms over 20kHz audio bandwidth.
Translates to -125.3dBV level. 2dB close to your HP analyzer result in last picture. 83.6% of it is located below 1/F as the A-Weighted figure infers.
For spot noise (spectral noise density) 1.22μV across 100kHz BW gives 3.08 nV/(√Hz) if with usual 1-pole correction.
Not far away from Tombo56's 418nV 22kHz or 2.25 nV/(√Hz) spot noise on REW using his LNA + Focusrite interface.
Your HP Analyzer's 25kHz bandwidth -127.5dBV or 0.42μV result translates to 2.12 nV/(√Hz) spot noise nonetheless.
1MHz test mode makes lower spot figure than 100kHz mode. Maybe the Agilent bench multimeter cuts off before 1MHz?
So your tests are serious indeed despite you don't claim them as such.
1.22μVrms over 100kHz bandwidth is equivalent to 920Ω resistor thermal noise at 20 degrees Celsius. Or 0.55μVrms over 20kHz audio bandwidth.
Translates to -125.3dBV level. 2dB close to your HP analyzer result in last picture. 83.6% of it is located below 1/F as the A-Weighted figure infers.
For spot noise (spectral noise density) 1.22μV across 100kHz BW gives 3.08 nV/(√Hz) if with usual 1-pole correction.
Not far away from Tombo56's 418nV 22kHz or 2.25 nV/(√Hz) spot noise on REW using his LNA + Focusrite interface.
Your HP Analyzer's 25kHz bandwidth -127.5dBV or 0.42μV result translates to 2.12 nV/(√Hz) spot noise nonetheless.
1MHz test mode makes lower spot figure than 100kHz mode. Maybe the Agilent bench multimeter cuts off before 1MHz?
So your tests are serious indeed despite you don't claim them as such.
Thanks Salas!
Load resistor value 450R (current 53mA).
Agilent 34410A used only with Vdc measurement (not Vrms).
All measurement: Output LNA --> Input RMS meter (with LTC1968 RMS Meter) --> Output RMS Meter (Vdc) --> Agilent 34410A (Vdc).
Agilent True RMS specification 3Hz - 300kHz.
Today I will take measurements with LNA before spectrum analyser with gain 40dB (100 times). Perhaps slightly more accurate results will be obtained.
Load resistor value 450R (current 53mA).
LTC1968 has higher accuracy up to 500 kHz.
Agilent 34410A used only with Vdc measurement (not Vrms).
All measurement: Output LNA --> Input RMS meter (with LTC1968 RMS Meter) --> Output RMS Meter (Vdc) --> Agilent 34410A (Vdc).
Agilent True RMS specification 3Hz - 300kHz.
Today I will take measurements with LNA before spectrum analyser with gain 40dB (100 times). Perhaps slightly more accurate results will be obtained.
Last edited:
Here are today's results.
Strangely low noise figures with the HP3582A.
Something seems wrong to me. I repeated them many times and that's it.
With Anrutsu it is difficult to make correct magnitude measurement below 10kHz.
So I don't guarantee any accuracy.
Strangely low noise figures with the HP3582A.
Something seems wrong to me. I repeated them many times and that's it.
With Anrutsu it is difficult to make correct magnitude measurement below 10kHz.
So I don't guarantee any accuracy.
Attachments
Output noise measured with HP3582A is about right. My own measurement was 0.418 uV in the 22 Hz-22 kHz bandwidth.
However, LNA with 36 uV output noise, that can’t be right. My LNA has 66 nV total noise in the 22 Hz-22 kHz bandwidth.
Another error factor is with noise subtraction operations. Noise is added or subtracted by square root law.
However, LNA with 36 uV output noise, that can’t be right. My LNA has 66 nV total noise in the 22 Hz-22 kHz bandwidth.
Another error factor is with noise subtraction operations. Noise is added or subtracted by square root law.
36uV is correct. This is bandwidch noise (0Hz - 300kHz) at gain 40dB (100 times) measured to LNA output. Reffered to input this is 36uF/100 = 360 nV at this bandwidch.My LNA noise at bandwitch (20Hz-20kHz) is 129 nVrms. Look in my topic LNA
Вашият LNA има по-нисък шум.
@Salas
Thanks for simulation !
Attachments
Last edited:
Thank you tombo56.
What is the right way (formula) to subtract the RMS noise whitch is added by LNA?
Last edited:
For uncorrelated noise sources connected in series: Square each figure then add them. Lastly square root the result. For example 100uV+36uV noise makes 106.283uV.
Say you measured 106.283uV noise knowing your LNA factored in 36uV: Square each figure then subtract them. Lastly square root the result. Makes 100uV.
Say you measured 106.283uV noise knowing your LNA factored in 36uV: Square each figure then subtract them. Lastly square root the result. Makes 100uV.
@tombo56
Sorry, I wrote in Bulgarian without meaning to and now I saw this.
I wanted to say that your LNA has a lot less noise than mine.
Thank you, I lack the equipment to to test for that, is very helpful at to have a general idea.
I haven’t thought of using a Muse BP for a power cap until I started thinking it may be perfect for VREF on an AK4497.
Of course, it goes without saying that a dedicated high quality regulator would be the best…
I haven’t thought of using a Muse BP for a power cap until I started thinking it may be perfect for VREF on an AK4497.
Of course, it goes without saying that a dedicated high quality regulator would be the best…