Difference between revisions of "RF Noise Generator"

From Land Boards Wiki
Jump to navigation Jump to search
 
(93 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
[[file:NoiseGen-P783-720px.jpg]]
 
[[file:NoiseGen-P783-720px.jpg]]
  
== Design ==
+
== Usage ==
  
 
* Useful as a filter test signal source
 
* Useful as a filter test signal source
 
** Pass filters
 
** Pass filters
 
** Crystal filters
 
** Crystal filters
 +
* Would like to use as noise source for [[tinySA]]
 +
** [[tinySA]] has lower noise floor than [[NanoVNA]]
 +
 +
=== Others ===
 +
 +
* [https://www.youtube.com/watch?v=732ESoul088 7MHz calibration signal generator]
 +
** 0dBm and -60dBm outputs
 +
 +
== Design #1 ==
 +
 
* From Experimental Methods in RF Design
 
* From Experimental Methods in RF Design
 
** Fig 7.72
 
** Fig 7.72
Line 14: Line 24:
 
* Noise figure -50 dBm @ 10 MHz
 
* Noise figure -50 dBm @ 10 MHz
  
== Schematic ==
+
=== Schematic ===
  
 
[[file:ng-sch-prototype.PNG]]
 
[[file:ng-sch-prototype.PNG]]
  
== Prototype ==
+
=== First Prototype ===
 +
 
 +
==== Build / Enclosure (Original board design) ====
 +
 
 +
* Used very old, but high quality chassis mount BNC connector
 +
** Silver connector has a nice patina
 +
* Built onto Single Side PCB material
 +
** Acts as ground plane
 +
** Pads cut from PCB material and superglued down onto the Base PCB
 +
* Hot glued down into plastic enclosure
 +
* Ferrite toroids at top are not used
 +
 
 +
[[file:NoiseGen-P783-720px.jpg]]
 +
 
 +
[[file:NoiseGen-P790-720px.jpg]]
 +
 
 +
[[file:NoiseGen-P787-720px.jpg]]
  
 
* Unique nodes in red
 
* Unique nodes in red
Line 26: Line 52:
 
[[file:NG-layout.PNG]]
 
[[file:NG-layout.PNG]]
  
=== Meaurements ===
+
* OSHPark PCB
 +
 
 +
[[file:NoiseGen_P1855-720px.jpg]]
 +
 
 +
=== LTSpice Simulation ===
 +
 
 +
* Amplifier looks flat but the result wasn't
 +
 
 +
[[file:NoiseGen_Design1.PNG]]
 +
 
 +
=== Measurements ===
  
 
* 12V, 43mA current draw
 
* 12V, 43mA current draw
* Rigol DS1054Z
+
 
** 50 MHz scope with mods
+
==== tinySA Measurements ====
** UltraScope settings
+
 
*** SCAL 1.00 V
+
* Measure from 1-30 MHz
*** H 200nS
+
* Power Off
** FFT settings
+
* -90 dB noise floor
*** CF = 16 MHz
+
 
*** 2 MHz start, 50 MHz stop
+
[[file:NoiseGenOnTinySA_Off.PNG]]
 +
 
 +
* Power On
 +
* Measure from 1-30 MHz
 +
 
 +
[[file:NoiseGenOnTinySA_On.PNG]]
 +
 
 +
* 100 KHz-1 MHz
 +
* Power Off
 +
 
 +
[[file:NoiseGenOnTinySA_Off_100KHz-1MHz.PNG]]
 +
 
 +
* Power On
 +
* Lot more energy at ~400 KHz
 +
* -30 dB
 +
* Not too useable for filter testing
 +
* Might be OK for receiver noise source
 +
 
 +
[[file:NoiseGenOnTinySA_On_100KHz-1MHz.PNG]]
 +
 
 +
==== Rigol DS1054Z Measurements ====
 +
 
 +
* 50 MHz scope with mods
 +
* UltraScope settings
 +
** SCAL 1.00 V
 +
** H 200nS
 +
* FFT settings
 +
** CF = 16 MHz
 +
** 2 MHz start, 50 MHz stop
 
* RBW 416.6 KHz
 
* RBW 416.6 KHz
 
* Background noise
 
* Background noise
 
* Generator off
 
* Generator off
* Average ~ -64 dB
+
* Average ~ -64 dB noise floor too high
  
 
[[file:NG_OFF.PNG]]
 
[[file:NG_OFF.PNG]]
Line 51: Line 115:
 
* Measured performance per EMiRFD (predicted - 50 dB)
 
* Measured performance per EMiRFD (predicted - 50 dB)
  
== High Pass Filter ==
+
== Better Design #2 ==
 +
 
 +
* Replace Zener as noise source with 2N3904 B-E junction reverse biased
 +
* C1 is connected Emitter of Q1 to GND instead of +12V to reduce power supply noise
 +
* Higher noise than EMiRFD design
 +
** Usable from 1-10 MHz
 +
** -40 dBm at 1 MHz
 +
** -80 dBm at 30 MHz
 +
* Nearly zero power supply current
 +
 
 +
[[file:Better_Noise_Gen_schematic.png]]
 +
 
 +
[[file:NoiseGen_V2-P1852-720px.jpg]]
 +
 
 +
=== Better Design #2 - LTSPice Simulation ===
 +
 
 +
* Falls off in a straight line (in dBm) with frequency
 +
** 10 dB at 1 MHz
 +
** 0 dB at 10 MHz
 +
** -10dB at 100 MHz
 +
 
 +
[[file:LTSpiceSim_Better2.PNG]]
 +
 
 +
=== tinySA Measurements ===
 +
 
 +
* Off
 +
 
 +
[[file:V2-NoiseGenOnTinySA_OFF_1MHz-10MHz.PNG]]
 +
 
 +
* On, 1-30 MHz
 +
** Consistent with the LTSpice simulation
 +
 
 +
[[file:V2-NoiseGenOnTinySA_On_1MHz-30MHz.PNG]]
 +
 
 +
== Better Design #3 - Cascode Amp with output buffer ==
 +
 
 +
* Previous design is probably good into receiver but not as good for tinySA tracking generator
 +
** Want to be able to drive passive/crystal filters for testing
 +
** Can already use [[NanoVNA]] but I'd prefer the tinySA due to lower noise floor
 +
* Goal
 +
** Flat response across 1-30 MHz
 +
** Good parts of Design #2
 +
** Transistor seems to be a better noise source than a zener
 +
* Idea - replace amp stage with cascode output stage followed by emitter follower for better match to 50 ohm load
 +
 
 +
=== Example cascode design ===
 +
 
 +
[[file:CascodeExample-720px.jpg]]
 +
 
 +
* Replace R1 with 0.1uF in parallel with 470K (from the the previous)
 +
* Remove R2
 +
 
 +
=== Better Design #3 - LTSpice Simulation ===
 +
 
 +
*  Output stage is high gain, but limits bandwidth
  
* Output series coupling 10 nF cap and 50 Ohm Load resistor make a High Pass Filter at 312 KHz
+
[[file:Cascode2.PNG]]
* No real point in making a better filter, but why not?
 
  
=== Design Inputs ===
+
== Better Design #4 ==
  
* High Pass filter
+
* #3 output stage was limiting the bandwidth
* 5th Order Chebyshev filter
+
** Remove output stage
** Fc = 1 MHz
 
** 3 Inductors, 2 Capacitors
 
* [https://www.66pacific.com/calculators/toroid-coil-winding-calculator.aspx Toroid Coil Winding Calculator]
 
  
[[file:Filter DesignInputs.PNG]]
+
[[file:Cascode2_Sch.PNG]]
  
=== Schematic - 5% Standard parts ===
+
* PCB design
  
[[file:Filter DesignSchematic.PNG]]
+
[[FILE:NoiseGen_V3_3D.png]]
  
=== Predicted Response ===
+
[[FILE:RF_Noise_Cascode2_PCB.PNG]]
  
[[file:Filter DesignGain.PNG]]
+
{| class="wikitable"
 +
! Qty
 +
! Value
 +
! RefDes
 +
!
 +
! Qty
 +
! Value
 +
! RefDes
 +
|-
 +
| 2
 +
| 100K
 +
| R2,R5
 +
|
 +
| 3
 +
| 0.1uF
 +
| C1-C3
 +
|-
 +
| 1
 +
| 150K
 +
| R3
 +
|
 +
| 3
 +
| 2N3904
 +
| Q1-Q3
 +
|-
 +
| 1
 +
| 470K
 +
| R1
 +
|
 +
| 4
 +
| TestPoint
 +
| TP1-TP4
 +
|-
 +
| 1
 +
| 560
 +
| R4
 +
|
 +
|
 +
|
 +
|
 +
|-
 +
|}
  
=== Toroid winding ===
+
* < 3 dB down at 100 MHz
  
* T37 cores, approx .5" per turn, plus tails
+
[[file:RF_Noise_Cascode2.PNG]]
  
* [https://www.66pacific.com/calculators/toroid-coil-winding-calculator.aspx Toroid Winding Calculator]
+
=== Cascode Design Videos ===
  
[[file:Toroid-6.8uH.PNG]]
+
<video type="youtube">18F_LL6KiUw</video>
  
[[file:Toroid-3.9uH.PNG]]
+
<video type="youtube">r5Z5W_7B8Go</video>
  
=== High Pass Performance ===
+
<video type="youtube">jnK0lSekFg4</video>
  
[[file:InputToHighPassFilter.PNG]]
+
<video type="youtube">SZW0qLmUPeY</video>
  
[[file:OutoutFromHighPassFilter.PNG]]
+
<video type="youtube">zAoLwpkbXmE</video>
  
 
== Assembly Sheet ==
 
== Assembly Sheet ==
  
 
[[RF Noise Generator Rev 1]]
 
[[RF Noise Generator Rev 1]]

Latest revision as of 19:22, 10 October 2021

NoiseGen-P783-720px.jpg

Usage

  • Useful as a filter test signal source
    • Pass filters
    • Crystal filters
  • Would like to use as noise source for tinySA

Others

Design #1

  • From Experimental Methods in RF Design
    • Fig 7.72
    • Described as "not flat"
  • "Junk box" parts
    • (2) 2N3904 transistors
    • (2) 5V (nominal) zener diodes
  • Noise figure -50 dBm @ 10 MHz

Schematic

Ng-sch-prototype.PNG

First Prototype

Build / Enclosure (Original board design)

  • Used very old, but high quality chassis mount BNC connector
    • Silver connector has a nice patina
  • Built onto Single Side PCB material
    • Acts as ground plane
    • Pads cut from PCB material and superglued down onto the Base PCB
  • Hot glued down into plastic enclosure
  • Ferrite toroids at top are not used

NoiseGen-P783-720px.jpg

NoiseGen-P790-720px.jpg

NoiseGen-P787-720px.jpg

  • Unique nodes in red
  • Prototype on single sided copper clad PCB
  • KiCAD zener packages have backwards silkscreen (will need to fix on OshPark PCBs)

NG-layout.PNG

  • OSHPark PCB

NoiseGen P1855-720px.jpg

LTSpice Simulation

  • Amplifier looks flat but the result wasn't

NoiseGen Design1.PNG

Measurements

  • 12V, 43mA current draw

tinySA Measurements

  • Measure from 1-30 MHz
  • Power Off
  • -90 dB noise floor

NoiseGenOnTinySA Off.PNG

  • Power On
  • Measure from 1-30 MHz

NoiseGenOnTinySA On.PNG

  • 100 KHz-1 MHz
  • Power Off

NoiseGenOnTinySA Off 100KHz-1MHz.PNG

  • Power On
  • Lot more energy at ~400 KHz
  • -30 dB
  • Not too useable for filter testing
  • Might be OK for receiver noise source

NoiseGenOnTinySA On 100KHz-1MHz.PNG

Rigol DS1054Z Measurements

  • 50 MHz scope with mods
  • UltraScope settings
    • SCAL 1.00 V
    • H 200nS
  • FFT settings
    • CF = 16 MHz
    • 2 MHz start, 50 MHz stop
  • RBW 416.6 KHz
  • Background noise
  • Generator off
  • Average ~ -64 dB noise floor too high

NG OFF.PNG

  • Noise Generator On
  • Falls off with frequency, but better than -50 dB

NG ON.PNG

  • Measured performance per EMiRFD (predicted - 50 dB)

Better Design #2

  • Replace Zener as noise source with 2N3904 B-E junction reverse biased
  • C1 is connected Emitter of Q1 to GND instead of +12V to reduce power supply noise
  • Higher noise than EMiRFD design
    • Usable from 1-10 MHz
    • -40 dBm at 1 MHz
    • -80 dBm at 30 MHz
  • Nearly zero power supply current

Better Noise Gen schematic.png

NoiseGen V2-P1852-720px.jpg

Better Design #2 - LTSPice Simulation

  • Falls off in a straight line (in dBm) with frequency
    • 10 dB at 1 MHz
    • 0 dB at 10 MHz
    • -10dB at 100 MHz

LTSpiceSim Better2.PNG

tinySA Measurements

  • Off

V2-NoiseGenOnTinySA OFF 1MHz-10MHz.PNG

  • On, 1-30 MHz
    • Consistent with the LTSpice simulation

V2-NoiseGenOnTinySA On 1MHz-30MHz.PNG

Better Design #3 - Cascode Amp with output buffer

  • Previous design is probably good into receiver but not as good for tinySA tracking generator
    • Want to be able to drive passive/crystal filters for testing
    • Can already use NanoVNA but I'd prefer the tinySA due to lower noise floor
  • Goal
    • Flat response across 1-30 MHz
    • Good parts of Design #2
    • Transistor seems to be a better noise source than a zener
  • Idea - replace amp stage with cascode output stage followed by emitter follower for better match to 50 ohm load

Example cascode design

CascodeExample-720px.jpg

  • Replace R1 with 0.1uF in parallel with 470K (from the the previous)
  • Remove R2

Better Design #3 - LTSpice Simulation

  • Output stage is high gain, but limits bandwidth

Cascode2.PNG

Better Design #4

  • #3 output stage was limiting the bandwidth
    • Remove output stage

Cascode2 Sch.PNG

  • PCB design

NoiseGen V3 3D.png

RF Noise Cascode2 PCB.PNG

Qty Value RefDes Qty Value RefDes
2 100K R2,R5 3 0.1uF C1-C3
1 150K R3 3 2N3904 Q1-Q3
1 470K R1 4 TestPoint TP1-TP4
1 560 R4
  • < 3 dB down at 100 MHz

RF Noise Cascode2.PNG

Cascode Design Videos

Assembly Sheet

RF Noise Generator Rev 1