Download An RF Analog Design from Concept to First Prototype

An RF Analog Design from
Concept to first Prototype
I was asked to give a talk about something ‘hardware based’ for the
QCWA quarterly dinner this November.
So I’ll take you on a journey through my current project, a linear RF
power amplifier for the 24-23cm band (1240-1300MHz). I’ll talk
about the process I’ve followed to create this device starting with
a concept then on through to the first hardware prototype. The
steps on this quest will include a look at what other people have
done, and how that influenced the set of requirements. I’ll talk
about the process I go through to conceptually realize a circuit
that achieves the objectives. I’ll also outline the design process,
including component selection, schematic capture, and printed
circuit board design. I’ll then go on to the challenges packaging
it all up. At the time of this writing the project isn’t finished but I
do hope to end the talk with some performance test results.
Wayne VE3CZO
An RF Analog Design
from Concept to first
Prototype
Wayne Getchell
VE3CZO
Mar12
Recipe for PCB Prototyping
2
Topics Covered
Design start
• Conjure up and specify requirements
Design Phase
•
•
•
•
Circuit Design
Component selection
Refine requirement specification
Schematic capture and PCB layout
Prototype Construction
• The physical layout tasks
Amplifier Testing
Considerations for next iteration
Nov 12
Recipe for PCB Prototyping
3
Requirements Specification
Start by envisioning the amp’s functions & features
• RF power amplifier covering the 24 – 23 cm amateur band
– Capable of covering 1240 to 1300 MHz for use with either ATV at
the low band end or SSB / CW operation at the high end
• Power from a12V battery. Needs to operate over 10.5 to 15V.
• Moderate size for portable operation
• Input power between 6 & 10 dBm for full output power to
interface with most transverters and ATV modulators
• Linear output power about 20 watts – so about 36dB gain
• Input attenuator for gain adjustment up to about 10dB loss.
• Low power use when not in transmit – less than 5 mA with LEDs
• LED indicators for power and transmit
• Provisions for antenna relay switching
– Single port to an ant, separate ports to an exciter O/P and Rx
• Output power modulation envelope detector
Nov 12
An RF HW Design
4
See what’s out there
Then search the internet to find out what
others have done.
• Is there something available that comes close to meeting
all requirements?
• If not gather ideas for a new design
Nov 12
An RF HW Design
5
See what’s out there
G6ALU 23cm 18W Power Amplifier
• Mitsubishi RA18H1213G
• Bias Adjust
• Transmit Switch
– Active high or low
Nov 12
An RF HW Design
6
So what’s out there?
DEMI 2330 30W 1240-1300 MHz Linear Amplifier
•
•
•
•
Nov 12
Mitsubishi RA18H1213G
Res divider bias adjust
Transmit Switch
Output power &
modulation detector
An RF HW Design
7
So what’s out there?
Over a dozen solutions using the RA18H1213G
• For ATV & SSB / CW
PE1RKI
F1GE
DB6NT
GB3TM Digital ATV
Nov 12
An RF HW Design
8
Data Sheet Info
Key Info
• Covers 1.24 to 1.3GHz
• 18 Watts output
• VDD 12 to 17 Volts
• Adjustable bias for gain linearity
• Gain ~ 25dB peaks at 33dB @1270
• I/P 20 - 23dBm for full O/P pwr
• Efficiency 25 - 30%
Nov 12
An RF HW Design
9
RA18H1213Module
Nov 12
An RF HW Design
10
RA18H1213G Module
Additional info from datasheet & web articles
• Robust output at 18 W tolerates 20:1 VSWR
• Most data sheet testing is done @ Pin = 23dBm and Vgg of 5V.
• Maximum input power 300mW or 24.8 dBm is quite close to the
23 dBm needed for the specified output power.
• Gain changes over frequency in ‘linear’ RF output range can be
a challenge
– 33dB at 1270 versus 25dB at 1240 and 1300.
• Vgg must not exceed 6 V or the module will be damaged
• Stability has been a problem for a number of users
– The heatsink flange is the only ground so you can’t put insulating
heatsink compound under the screw portion of the mounting flange
• This also makes thermal management at the high RF output
powers that amateurs like to run a challenge. Power dissipation
can exceed 80Watts! For portable use a ‘smallish’ heatsink with
a fan will be a must…so fan control is added to the requirements
Nov 12
An RF HW Design
11
st
1
Stage Amplifier
The RA18H1213G module doesn’t have enough gain for full output
with less than 8 dBm input so an additional amp will be nice
Target characteristics
• P1dB > 23dBm (200mW) – the input power used for most of the
datasheet specs
• Shouldn’t be able to blow up the module input so output should saturate
at less than 25dBm
• Maximum gain 17dBm for 23dBm output with 6dBm input 13 to 15
would be just about right
• From factor SOT89 – moderate power capable thermal package
Started with SGA 7489
• Spec…gain 20 dB & P1dB 19dBm – gain quite high, P1 a bit low…
• ….but I have some on hand so will start with this one
Try SXA 389BZ …these needed to be ordered… eBay again!
• P1dB 25dBm gain about 15dB
• Vdd=6V max 115mA typical bias current
• Opt for passive resistor biasing so I can limit output power while
maintaining linearity hopefully up to around 23dBm.
Nov 12
An RF HW Design
12
Circuit Design Considerations
Voltage Range
• 10.5 to 15V - covers most battery chemistries. 13.8V nominal
Input Switch
• Should have an active low PTT function that uses little current
• Switches off bias to the module & 1st stage amp for low Istandby
• Switched supply output available for aux uses i.e. coax relay
Internal regulated voltage
• 9.0V LDO to insure good regulation with 10.5V battery. 9V for 1st
stage RF gain block passive biasing
• SOT223 heat tab package for good thermal management
• Provides bias for RA18H module, IPA, and fan thermostat IC
• Note that 9V can destroy the RA18H Vgg input so this voltage
must be limited before it gets to the module!
Thermostat
• Found an LM56 to provide a temperature controlled switch with
hysteresis for the fan.
Nov 12
An RF HW Design
13
Schematic Capture
Thermal Switch
LDO Regulator
PTT & I/P Sw
PA module
1st Stage RF Amp
Envelope Detector
Nov 12
An RF HW Design
14
PTT Switch & Regulator
• Adjustable LM1117 LDO in SOT 223 power pack set for 9.0V
• PTT switch is a 30V 2.3A Pch with 150 mOhm in SOT23 pack
Nov 12
An RF HW Design
15
LM56 Thermal Switch
• Dual temp sense fan then over temp
• Calculator input trip points & R1 - R3
Nov 12
An RF HW Design
16
PA Module Section
• Thorough
bypassing
required
• Bias Set
via R6
• D5
protects
Bias input
Nov 12
An RF HW Design
17
Envelope Detector
• Same circuit as used for Sequencer1
• Coupling probe is installed above the PCB
Nov 12
An RF HW Design
18
1st Stage RF Gain schematic
• TL segments are back annotated into the schematic after layout.
They show the section length in mm & deg at 1270 MHz
Nov 12
An RF HW Design
19
Interstage Matching
LL Smith Chart
matching from
RF Dude
• Start with SGA
7489 S22 and
work toward
RA18H S11
• Use Zplots to get
S11 or S22 from
Touchstone
SP1/SP2 format
compatible files
Nov 12
An RF HW Design
20
Interstage Matching
LL Smith Close-up
Nov 12
An RF HW Design
21
PCB Layout
Careful attention to grounding
• Many vias around grounding points
• Careful placement of bypass caps
All components on the top side no tracks on bottom
• Bottom is ground plane only, can be seated against heatsink
ground for lowest ground plane impedance to the module.
Connectorized for ease of assembly
• JST PH 2mm connector an inexpensive solution
Extract 50 ohm TL segments from layout
• Back annotate into schematic with length in mm & deg
• TL segment info used for Smith chart matching
• Interactive as additional components may be needed for
best match
Nov 12
An RF HW Design
22
PCB Layout
Nov 12
An RF HW Design
23
PCB Layout
Coplanar waveguide transmission line traces
• Wcalc used to determine dimensions for 50Ω on FR4 PCB
• Corner calculator ● Surface mount SMA
• Special library parts for TL resistors
– 1mm pads
– no place
Nov 12
An RF HW Design
24
Wcalc CPW on FR4
Nov 12
An RF HW Design
25
TL
Corner
Calc
Nov 12
An RF HW Design
26
Surface Mount SMA
No off the shelf surface mount right angle parts
• Jigs created to cut ground lugs and center conductor
• Grounds penetrate holes but don’t protrude through PCB
• Center conductor filed to be a bit proud of seated ground
lugs, soldered then with rework station heated and pushed
against PCB to seat connector making solid contact
Nov 12
An RF HW Design
27
PCB build & test
First fill ground plane vias with solder
• Cover the bottom side ground plane with Kapton tape
• Push solder paste
into vias & melt to
create a solid ground
plane area
Then populate the power supply and test
• Tx switch
• 9.0V LDO regulator output voltage correct & stable
Next populate then test the 1st stage amplifier
• Surface mount the SMA connectors
• Power and test the amplifier without the module
• Check gain P1dB, S11, S22 over frequency.
Nov 12
An RF HW Design
28
st
1
Stage Amp Tests
• AV 19.5dBm
• Quite wide band
Nov 12
An RF HW Design
29
st
1
Stage Amp Tests
• Input return loss 15db
• P1 20 – 21 dBm
SGA7489 Gain vs Pout
20.4
20.2
20.0
Gain (dB)
19.8
19.6
19.4
19.2
19.0
18.8
18.6
0.0
5.0
10.0
15.0
20.0
25.0
Pout Pwr (dBm)
Nov 12
An RF HW Design
30
PCB build & test
Then populate and test the Fan thermostat
• Verify turn on temperature and tweak.
• Was initially 40 but reset to 37 then 35 deg C
• Dual thermostat capability so will add over
temperature shut down feature for the next iteration
Populate all remaining components
• Modulation envelope detector
Then mount the RA18H1213G module on the
heatsink & wire it up to the PCB
Nov 12
An RF HW Design
31
RA18H Thermal Challenges
Info from
datasheet
• Efficiency is < 30%
• For Po = 34W heat
dissipated at
Vcc=13.8V is 87W
• For reliability case
temperature should
not exceed 60 deg C
and 90 deg C in
extreme conditions
Nov 12
An RF HW Design
32
RA18H Thermal Challenges
What to do with a module that has flanges?
• Center of module is approx 0.5mm above heatsink
• Standard thermal compound won’t work across gaps
Nov 12
An RF HW Design
33
RA18H Thermal Challenges
What others have done with the flanges…
• DEMI took a belt sander to them
• Others filed the flanges flat then sanded the bottom smooth
Nov 12
An RF HW Design
34
RA18H Thermal Challenges
Buy Why is the mounting surface flanged?
• For reliability according to Mitsubishi G2K-R-051201
Nov 12
An RF HW Design
35
RA18H Thermal Challenges
Reliable operation means slow temp changes
• Ok if the thermal impedance flange to heatsink is very low
Nov 12
An RF HW Design
36
RA18H Thermal Challenges
Mitsubishi Recommended
thermal compound
• AN-GEN-001-B App Note –
Use ShinEtsu G746
Nov 12
An RF HW Design
37
RA18H Thermal Challenges
Mitsubishi Recommended thermal compound
• Application note AN-GEN-042-D
• Minimum acceptable flange area coverage… 80%
Nov 12
An RF HW Design
38
RA18H Thermal Challenges
ShinEtsu
G746
• Thermal
Conductivity
1.6*10E-3
cal/cm-secdeg.C is
about 0.7
W/m.k
Calculation assumes 100% flange area coverage
There is no compound in the flange fastener area
Nov 12
An RF HW Design
39
RA18H Thermal Challenges
Laird Tgrease 880
• Will not dry out settle or harden
• Fills microscopic irregularities
• Supplied in 0.5 1 or 3 kg containers
Luckily it’s
also
Available
from DigiKey in 30cc
containers
Nov 12
An RF HW Design
40
RA18H Thermal Challenges
Thermal
Calculator
• Laird
Tgrease
880
• 3.1 W/m.K
Calculation assumes 100% flange area coverage
There is no compound in the flange fastener area
Nov 12
An RF HW Design
41
RA18H Mounting
Heatsink mounting flatness recommendation
• From datasheet & failure analysis G2K-R-051201-1
– Heat sink flatness must be less than 50 μm (a heat sink that
is not flat or particles between module and heat sink may
cause the ceramic substrate in the module to crack by
bending forces, either immediately when driving screws or
later when thermal expansion forces are added).
Mounting Surface Ok
Causes Cracks
Nov 12
An RF HW Design
42
Heatsink Preparation
Surface Preparation
• Smooth surface Initially with orbital sander & 120 emery
Nov 12
An RF HW Design
43
Heatsink Preparation
• Final smoothing & planarization by taping emery cloth to a
flat surface. First with 400 (23.6u grit) then 600 (16u grit)
Nov 12
An RF HW Design
44
Module Preparation
• Apply a generous coating of thermal compound, enough that
a small amount squeezes out when the module is mounted
Measured 19.2
deg C rise with
34W out &
130W DC in
Nov 12
An RF HW Design
45
Transmission Line testing
Traces on test PCB
• 50 ohm lines are
connectorized &
terminated
• No solder mask
ρ = -26mU or 47.5Ω
(RL=32dB)
• With solder mask
ρ = -45mU or 45.7Ω
(RL=27dB)
Nov 12
An RF HW Design
46
Module Initial Functional test
Module & PCB mounted on
heatsink
• Initial power up – done without a
fan or antenna relay
• Set RA18H Vgg adjusting pot for
minimum voltage & limit supply
current to about 0.5A
• Result – supply goes into current
limit, voltage a few volts…well at
least it wasn’t 0V.
• Cause – bias supply voltage higher
than anticipated as Igg low
– Recalculate voltage divider
and try again
Initial test results
• IPA and PA work, but…
Nov 12
An RF HW Design
47
Module Initial Functional test
Initial test results
• PA isn’t stable @ Ibias >2A
Nov 12
An RF HW Design
• Gains over 55dB
(350,000) a bit
tricky to handle!
• PCB was mounted
on standoffs to best
mate with module
pins to reduce lead
inductance.
• That didn’t work!!
• But the module was
stable with PCB flat
against the heatsink
as shown in picture
48
Module Gain too high
Initial test results
• PA is stable but still there’s too much gain especially at
the modules peak near 1270MHz
Amplifier Av vs Pout @ Ibias -6.2A
56.0
55.0
54.0
53.0
Gain (dB)
52.0
51.0
50.0
49.0
48.0
47.0
46.0
45.0
25.0
30.0
35.0
40.0
45.0
50.0
Pout (dBm)
Nov 12
An RF HW Design
49
Replaced
st
1
Stage amp
SGA7489 replaced with the just arrived SXA389
• Lower gain 14 vs. 19.5 dB and higher P1dB, 24 dBm vs. 20 dBm
XSA389BZ Gain vs Output Power
14.50
The P1dB was lowered a bit by
reducing the bias current to help
keep the module input power
below the module’s 24.8 dBm
maximum.
Gainn (dB)
14.00
13.50
13.00
12.50
12.00
11.50
5.00
10.00
15.00
20.00
25.00
30.00
Output Power (dBm)
Nov 12
An RF HW Design
50
Interstage Matching
SXA389 output to RA18H1213G input
• Nothing to do!
RA18H1213G S11
The P1dB was lowered a bit
by reducing the bias current to
help keep the input power
during transient below the
module’s max
SXA398 S22
LLSmith
showing
both Z’s
Nov 12
An RF HW Design
51
Now Measure the Module
RA18H1213G Module Gain vs Bias Current
RA18H1213G module
40.0
• Av vs. Ibias
• Av vs. Frequency
35.0
30.0
Gain (dB)
25.0
20.0
RA18H1213G Gain vs Frequency normalized to 1300MHz
15.0
7.0
10.0
6.0
5.0
5.0
0.0
1000
2000
3000
Bias Current (mA)
1240MHz
1270MHz
4000
Gain (dB)
0
4.0
5000
6000
3.0
1300MHz
2.0
1.0
0.0
1240
1250
1260
1270
1280
1290
1300
Frequency (MHz)
Pin= 4dBm Ibias=6.2A
Nov 12
An RF HW Design
Pin= 4dBm Ibias=1A
Pin=23dBm Ibias=6.2A
52
Then Measured Module
Amplifier Linearity at two bias settings
1270MHz Input v.s. Output Power
15.0
10.0
Input Power (dBm)
5.0
0.0
-5.0
-10.0
-15.0
-20.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
50.0
Output Power (dBm)
Igg=1A
Nov 12
Igg=6.2A
An RF HW Design
53
Amp Gain Linearity vs. Ibias
Amplifier at 1300MHz
1300MHz Amplifier Gain vs Output Power
45.0
44.0
43.0
•
42.0
41.0
•
40.0
Av (db)
39.0
Saturated Pout
43.5dbm 23W
Max Linear Pout
42.5dBm 18W
38.0
37.0
36.0
35.0
34.0
33.0
32.0
31.0
30.0
15.0
20.0
25.0
30.0
35.0
40.0
Ibias=1.8A
Ibias=2.5A
45.0
Pout (dBm)
Ibias=1A
Nov 12
Ibias=1.2A
Ibias=1.5A
An RF HW Design
Ibias=6.2A
54
Amp Gain Linearity vs. Ibias
Amplifier at 1240 MHz
1240MHz Amplifier Gain vs Output Power
45.0
44.0
•
43.0
42.0
41.0
•
40.0
Av (db)
39.0
Saturated Pout
46.5dbm 45W
Max Linear Pout
44.5dBm 28W
38.0
37.0
36.0
35.0
34.0
33.0
32.0
31.0
30.0
15.0
20.0
25.0
35.0
40.0
Ibias=1.8A
Ibias=2.5A
30.0
45.0
Pout (dBm)
Ibias.1A
Nov 12
Ibias=1.3A
Ibias=1.5A
An RF HW Design
Igg=6.2A
55
Amp Gain Linearity vs. Ibias
Amplifier at 1270 MHz, the module’s gain peak
Av (db)
1270MHz Amplifier Gain vs Output Power
50.0
49.0
48.0
47.0
46.0
45.0
44.0
43.0
42.0
41.0
40.0
39.0
38.0
37.0
36.0
35.0
34.0
33.0
32.0
31.0
30.0
•
•
15
20
25
30
35
40
Saturated Pout
46dbm 40W
Max Lin Pout
44dBm 25W
45
Pout (dBm)
Ibias=1A
Nov 12
Ibias=1.5A
An RF HW Design
Ibias=6.2A
56
I/F the amp into a system
With a modulator or transverter that has Pout > 0 dBm
• If Pin to the amplifier is over 23dBm
– Use an external attenuator to get the power down to 15 dBm.
• If Pin is greater than 15 dBm but less than 23 dBm
– Don’t use the 1st stage amp. Jumper it out with a zero ohm resistor.
– Use the input attenuator to reduce the signal to 15dBm (0 to 8 dB)
• If Pin is less than 15 dBm but greater than 8 dBm
– Set the module bias to 1.5A
– Use the input attenuator to reduce the input signal to 8 dBm (0 to
7dB)
– This is the lowest power option, easiest on batteries for portable ops
• If Pin is less than 8dBm but greater than 0dBm
– Set the module bias to 6.3A
– Use the input attenuator to reduce the input signal to 0dBm
Nov 12
An RF HW Design
57
Physical Assy I/O Connections
Determining I/O
• SMA connectors in N out
● PowerPole for Power
• Mini-Din for control
● LED’s for Pwr, Tx, & Over temperature
• Antenna relay to be mounted into the enclosure
Nov 12
An RF HW Design
58
Ant Relay Coax Switch
SMA Coax Relay
• 12 - 28V boost converter – a previous project
• Minor activation delay noted 2-3mS
• Next amp PCB iteration will add a Tx delay timing option.
Nov 12
An RF HW Design
59
Physical Assembly
PA enclosure is a Hammond 1590Y
• PCB was sized to just fit inside
Nov 12
An RF HW Design
60
Final Assembly
Enclosure
installed on
heatsink
Nov 12
An RF HW Design
61
Demod coupling Adjusted
• Envelope demodulator’s coupling factor is adjusted by raising or
lowering the wire loop above the PCB output transmission line
track
Nov 12
An RF HW Design
62
Final Assembly
• The feet above the fan will be
removed after final testing. Until
then they allow air into the heatsink
as the unit is inverted for test.
Nov 12
An RF HW Design
63
What’s Next?
Iterate the PCB
• About a dozen design improvements
– Add over temperature indicator and PA output power cut-off
– Add transmit relay delays to prevent hot switching when the
amplifier isn’t used with a sequencer
– Tweak the transmission line impedance
– Add two additional ground screw points
Get an RA18H1213G module from another
manufacturing batch to determine
performance consistency
Mar12
Recipe for PCB Prototyping
64
Bits & Pieces mentioned
• Wcalc transmission line calculator
http://wcalc.sourceforge.net/index.html
• LLSmith for impedance matching
http://tools.rfdude.com/RFdude_Smith_Chart_Program/RFdude_smith_chart_program.html
• Zplots for displaying SP2 files and a lot more…
http://ac6la.com/zplots.html
• Cadsoft Eagle for Schematic capture and PCB layout
http://www.cadsoftusa.com/
Mar12
Recipe for PCB Prototyping
65