INTERSIL ISLA216P20

Application Note 1837
Author: Michael Steffes
Ultra High Performance Broadband 12 to 16-Bit Data
Acquisition Platform
ISLA214P50-55210EV1Z High
Speed ADC/AMP Evaluation
Board
1. ISLA214P50 High Speed, High Performance ADC (14-bit,
500MSPS)
2. ISL55210 High Performance, Low Power, Fully Differential
Amplifier (FDA)
3. Compatible with existing Intersil high speed ADC
evaluation platforms
4. Optional response measurement port from ADC inputs to
board edge
5. Pin compatible family of 12-to-16 bit ADC’s can be used on
this board
Performance Range
1. Clock rate range: 80MSPS to 500MSPS
2. 283mVP-P input (-7.0dBm) for -2dBFS at ADC inputs
(1.6VP-P at ADC)
3. ±0.8dB flat response from 100kHz to 100MHz
Evaluation Platform Overview
This ISLA214P50-55210EV1Z is an evaluation platform
featuring Intersil’s ultra-high dynamic range fully differential
amplifier (FDA), the ISL55210, and the High Speed, High
Performance, 14-bit, 500MSPS ADC, the ISLA214P50. This
PCB daughterboard mates to Intersil’s existing high speed ADC
evaluation platform allowing for easy performance
measurement and analysis (see Intersil Application Notes
AN1433, and AN1434 for more information). The ADC
evaluation platform consists of custom designed hardware
and software supporting a wide range of ADC daughterboards
on the KMB-001 motherboard. The function of the hardware is
to provide power to the ADC daughterboard, manage the
communication to the ADC internal settings, accept clock and
signal inputs, and buffer the digital outputs for communication
to a host PC. The Konverter software is required to configure
the ADC for initial operation, to modify the device functionality
or parameters, and to process and display the captured digital
data. Konverter software version 1.22c (or later) supports the
ISLA214P50 family and the ISLA214P50-55210EV1Z PCB.
CONTACT THE FACTORY FOR ASSISTANCE IN USING THE
KONVERTER SOFTWARE TO MODIFY THIS BOARD TO A
DIFFERENT ADC IN THE PIN COMPATIBLE FAMILY.
4. Typical SNRFS (30MHz input, 500MSPS): 71.8dBFS
(vs 72.6dBFS for ADC only)
5. Typical SFDR (30MHz input, 500MSPS): 89dBc (vs 84dBc
for ADC only)
TABLE 1. PIN COMPATIBLE HIGH PERFORMANCE ADC FAMILY
MAXIMUM
SAMPLE RATE
(Msps)
POWER
CONSUMPTION
(mW)
250
785
PART NUMBER
RESOLUTION
(Bits)
1. ISLA214P50-55210EV1Z Evaluation Board
ISLA216P25
16
2. KMB-001LEVALZ Intersil Motherboard (+5V supply
provided with motherboard)
ISLA216P20
16
200
720
ISLA216P13
16
130
615
System Requirements
3. Intersil Konverter software
http://www.intersil.com/content/intersil/en/products/dat
a-converters/high-speed-a-d-converters/hs-adc-evaluationplatform.html
ISLA214P50
14
500
835/900 (Note)
ISLA214P25
14
250
450
ISLA214P20
14
200
410
4. Low jitter clock source
ISLA214P13
14
130
360
5. Bandpass filters
ISLA214P12
14
125
310
6. PC running Windows XP operating system with Konverter
software installed
ISLA212P50
12
500
823/892 (Note)
ISLA212P25
12
250
440
ISLA212P20
12
200
405
ISLA212P13
12
130
355
Board Numbering Note
The original board marking was ISLA214P50/55210EV1Z. For
ordering purposes, that has been changed wherever possible
to ISLA214P50-ISL55210EV1Z. Any reference to
ISL214P50/55210EV1Z would be the same end item as the
final ordering number version with the dash instead slash.
May 3, 2013
AN1837.0
1
NOTE: I2E disabled/enabled.
CAUTION: These devices are sensitive to electrostatic discharge; follow proper IC Handling Procedures.
1-888-INTERSIL or 1-888-468-3774 | Copyright Intersil Americas LLC 2013. All Rights Reserved
Intersil (and design) is a trademark owned by Intersil Corporation or one of its subsidiaries.
All other trademarks mentioned are the property of their respective owners.
Application Note 1837
USB
To Host PC
Mezzanine
Connector
REF
IN
Analog Input
Test
Bandpass
Filter
Low-Jitter
RF Generator
Daughter
Card
SRAM
Optional
Attenuator
AMP
10MHz
Reference
Motherboard
KMB001
+5V
USB
ADC
FPGA
Analog Input
Low-Jitter
RF Generator
REF
OUT
Clock
(200-500MHz)
SRAM
Optional
Measurement
Port
Clock Inputs
FIGURE 1. TYPICAL CHARACTERIZATION SETUP USING THE INTERSIL KMB-001 MOTHERBOARD AND KONVERTER SOFTWARE
OPERATING PRECAUTIONS:!!
IT IS STRONGLY RECOMMENDED TO INSERT THE +5V PLUG AT
THE MOTHERBOARD PRIOR TO PLUGGING IN THE AC ADAPTER
TO REDUCE THE POSSIBILITY OF POWER SURGES WHICH CAN
DAMAGE THE PCB. PROBING ON THE PCB SHOULD BE DONE
WITH CARE USING PROPER ESD TECHNIQUES WHILE HANDLING.
Hardware Description
There are two components in the hardware portion of the
evaluation platform. The daughtercard and the motherboard
(Figure 1). The FDA and ADC are on the daughtercard, which
accepts power from the motherboard and contains the analog
input circuitry, clock interface, and supply decoupling. The
daughtercard interfaces to the motherboard through a
mezzanine connector. The motherboard contains a USB
interface, an FPGA and SRAM. The motherboard serves as the
interface between the host PC and the ADC daughtercard. Most
of the ADC functionality is controlled through the motherboard by
the Konverter software. The FPGA accepts output data from the
ADC and buffers it to the SRAMs before passing it to the PC at
the lower data rate required for post-processing. The maximum
buffer depth is approximately one million words.
The designer must supply a low jitter RF generator for the clock
input to achieve the SNR reported here. Some possible options
are shown in “Appendix A: Low Phase Noise RF Generators” on
page 15. An alternate to a signal generator for fixed 500MSPS
clock rates would be the 3.3V supply, SMA barrel, RFPRO33-500
from Crystek. A slight degradation in SNR might be expected
using this device vs the best low phase noise RF signal
generators and a bandpass filter on the clock. Using a bandpass
filter on the clock will reduce clock jitter and improve SNR while
using a bandpass on the signal source is normally required to
eliminate harmonics while testing the board performance. Most
RF generators that might be used as a test analog input source
have very poor harmonic distortion and require a bandpass
postfilter to see the full performance of the ADC’s FFT.
Software Description
The software component is the Konverter Analyzer, a graphical
user interface (GUI) created with MATLAB™. A MATLAB
Component Runtime engine is supplied, which executes a
2
compiled version of the m-files. Therefore, a separate version of
MATLAB is not required to run the Konverter Analyzer.
The GUI controls the ADC configuration through its SPI port, reads
data from the motherboard and performs the post-processing and
display of digitized data. Data can be viewed in either the time or
frequency domain, and can be saved for later processing. Critical
performance parameters such as SNR, SFDR, ENOB, etc. are
calculated and displayed on the screen when FFT output is
selected and a dominant single frequency is being applied.
Initial Start-Up
Referring to Figure 3, connect the daughtercard to the
motherboard by aligning the two matching mezzanine connectors.
Four screws on the motherboard align with the mounting holes on
the daughtercard. Next, connect the clock source (≈12dBm level
into 50Ω) which will be required for communication to the
Konverter software. Then connect a test source or signal of interest
coming from your signal channel at a maximum input level
<350mVP-P (or <-5.1dBm for single tone). With the RF signal
generators delivering a clock and input signal to the daughtercard,
and the +5V supply jack plugged into the motherboard, plug the
AC power plug into a wall socket. The daughtercard is powered
from linear regulators on the motherboard through the mezzanine
connector. The USB cable should now be connected from the
motherboard to the PC. Be sure to use the same USB port that was
originally used when the Konverter software was installed on the
PC to insure it is recognized. Now launch the Konverter software on
the PC where it should recognize the motherboard and proceed to
taking an FFT.
Motherboard
The only connections to the motherboard are the +5V supply power
and the USB cable to the PC with the Konverter software loaded.
No additional configuration of the motherboard is required. IT IS
STRONGLY RECOMMENDED TO INSERT THE +5V PLUG AT THE
MOTHERBOARD WITH THE DAUGHTERBOARD ATTACHED PRIOR
TO PLUGGING IN THE AC ADAPTER TO REDUCE THE POSSIBILITY
OF POWER SURGES WHICH CAN DAMAGE THE PCB. PROBING ON
THE PCB’S SHOULD BE DONE WITH CARE AND PROPER ESD
PROCEDURES USED WHEN HANDLING THE BOARDS.
AN1837.0
May 3, 2013
Application Note 1837
Software Start-Up
The FPGA clock is derived from the ADC output clock, and the
FPGA clock must be active for the software to operate correctly.
Therefore, it is critical to have a convert clock present and the
board powered up before the Konverter software is launched. It is
not necessary to have an analog input signal present.
The compressed MATLAB files are unpacked the first time the
GUI is invoked after installation. This will slow the start-up the
first time the evaluation system is used but it will run more
quickly in subsequent startups. Complete information can be
found in the KMB-001 Installer manual at:
http://www.intersil.com/converters/adc_eval_platform/
The main Konverter Analyzer window is shown in Figure 16. The
application opens in FFT mode by default, but other modes can
be selected using the radio buttons in the lower left corner. In
each mode, relevant parameters are displayed in the data box
the left side of the window.
The following data is displayed in all operating modes –
1. Fsamp: Sample clock frequency, automatically detected.
2. Ffund: Input frequency, automatically detected if input is a
single tone or dominant tone waveform
3. Fund: Amplitude of the dominant tone in dBFS
4. Samples: Record length, defaults but can be updated
Data Acquisition
Normally, the first step to an FFT data display is to open the
“Setup” key in the upper left and pick a windowing function to
limit spectral leakage. All the plots and data here used the
Blackman Harris 4 term windowing. It is also possible to change
the record length for the FFT in this screen. Normally, the
continuous calibration and background keys are selected and, if
desired, an averaging of the FFT outputs may be selected in the
lower left of the Konverter screen set up for FFT display (default
at start-up). Then, press “Run” in the lower left and the FFT will be
updating. The number of averaged sweeps defaults to 10, but
can also be changed in the “Setup” conditions dialog window.
Menu items and the toolbar buttons may not function properly if
data is being captured continuously. Press “Stop” in the lower left
before selecting a menu item or using a toolbar button.
AN1433 should be consulted for more information on the
options provided in the Konverter windows. AN1434 offers help
for first time installation if needed.
Hardware and Analog Signal Path
Description
Figure 2 shows a close-up of the ISLA214P50-55210EV1Z
daughtercard.
The SMA at the top left is the analog input connection, the next
one down is an optional sense port for an attenuated version of
the differential signal being presented to the ADC input pins,
while the lower SMA is the clock input pin. Several optional
inputs are not populated, like the amplifier disable control line at
the lower left of the card and some alternate clock input paths on
the bottom. The V TEST path SMA is not populated on the board as
delivered but may be easily added.
FIGURE 2. ISLA214P50-55210EV1Z DAUGHTERBOARD
3
AN1837.0
May 3, 2013
Application Note 1837
FIGURE 3. ISLA214P50-55210EV1Z DAUGHTERBOARD ON KMB-001 MOTHERBOARD
The analog input signal comes in through 2-MiniCircuits
transformers then into the ISL55210 then into an interstage
passive RLC filter to the ADC. The blue potentiometer in the
middle of the board is a common mode voltage adjustment for
the average DC voltage applied to the ADC inputs for this AC
coupled signal path. This board plugs into the KMB-001
motherboard to appear as in Figure 3.
Here, the input signal is connected and the 500MHz clock is
being applied through a TTE 500MHz Bandpass filter. The power
is being applied to the motherboard as shown by the active green
light just under the +5V power connector in the upper right.
Again, the +5V board connector is plugged in first, then the AC
plug inserted to the power allowing the power brick to filter the
+5V power up transient. Similarly on power down, unplug the +5V
power at the AC plug side.
While the full signal path schematic is shown in Figure 46, it is
best to break it into pieces for discussion of design, performance,
and options. In general, the board offers numerous optional
connections that are indicated in green on the schematic and/or
by DNP for the standard board build. The basic signal path is
intended to:
1. Terminate a single ended input with an AC-coupled,
broadband, 50Ω impedance. This is of course expecting the
source to also be a broadband 50Ω source and that
impedance does get reflected through the input transformers
to be part of the signal chain flatness characteristic.
2. Convert the single ended input to the required differential
signal at the ADC inputs centered within the converter’s
desired common mode voltage range.
3. Provide amplification from a nominal -7dBm (283mVP-P)
single tone input to a -2dBFS (2dB below the 2VP-P full scale
4
of the ADC) or 1.58VP-P at the ADC input pins with extremely
low noise and distortion. This 5.6V/V gain (15dB) gain is a
combination of the input transformer step up, ISL55210 gain,
and various insertion losses in the transformers and
interstage filter from the ISL55210 to the ADC.
4. As part of the interstage filter design, this board includes a
VCM control loop that senses the average DC voltage at the
two ADC inputs and servo’s a control voltage into the filter
design to match a reference voltage applied to the servo loop
op amp. The board is delivered set to 0.96V as that has shown
improved SFDR for the filter source impedance and the
ISLA214P50 ADC. Changing ADC selections and/or filter
designs in this board might benefit from a different nominal
VCM control target for best SFDR and the VCM adjust pot
allows and easy means to test that.
AMPLIFIER POWER SUPPLY DECOUPLING ISSUES
The portion of the signal path schematic of Figure 4 shows one
example of good power supply decoupling for this single 3.3V
supply amplifier. Where possible, a large valued capacitor
isolated towards this 4GHz amplifier with a high frequency ferrite
and then another 1µF element provides a PI filter on the board.
Right at the 2 device supply pins, X2Y capacitors are used to get
the best high frequency decoupling. These 0.1µF elements
actually are two capacitors connected in parallel to give 0.2µF
decoupling with much lower ESR and higher self resonant
frequency than typical ceramic SMD capacitors. Standard SMD
0.1µF ceramic capacitors can be used instead but may increase
the even order harmonics at higher frequencies as they go self
resonant. Those are provisioned on the board as optional
elements but not populated as shown in Figure 4 by the green
capacitors C1003 and C1008.
AN1837.0
May 3, 2013
Application Note 1837
FIGURE 4. AMPLIFIER POWER SUPPLY DECOUPLING AND INPUT SIGNAL UP TO THE AMPLIFIER INPUTS
INPUT SIDE CIRCUIT DESIGN AND OPTIONS
As delivered, the interface uses an input transformer that is very
flat to low frequencies followed by a common mode choke
transformer that is a DC short for biasing with <0.3dB insertion
loss to >1GHz for the differential signal. Either may be removed
and replaced by shorts or resistor elements using the bypass
resistors.
The input signal is coupled through a 1μF blocking cap (to protect
against accidental DC shorts) to an input step up transformer
and then directly into a common mode choke transformer. The
1:2 turns (1:4Ω) ratio step up transformer offers many
advantages combined with an FDA as shown here. The second
transformer provides a differential signal short but a common
mode open circuit. Differentially, the two 100Ω gain resistors
appear directly at the output of the input step up transformer.
Briefly, the 2-100Ω series resistors feed into a differential virtual
ground at the FDA summing junctions (R1009&R1010 in
Figure 4) and form a 200Ω differential termination impedance
for the ADT4-6T. That is input referred as a broadband 50Ω
termination impedance out of the 1μF blocking cap. Scaling
these resistors up, while still maintaining a 200Ω secondary
termination, can be done using the Rterm1 element. This is
sometimes useful at lower gain targets to allow the feedback
resistors (R1008 & R1011) to be scaled up when they might be
adding significant loading to the output stage of the ISL55210.
Detailed specifications for 4GHz gain bandwidth product,
0.85nV/√Hz input noise ISL55210 may be found at:
http://www.intersil.com/products/ISL55210
5
The ISL55210 provides duplicates of the differential outputs on
the input side for tighter signal path layout. The feedback
resistors are the 2 -500Ω elements where the connection back
into the inverting summing junctions are the 0Ω elements.
Neglecting transformer insertion losses, the gain from the input
port to the amplifier outputs should be 2 (in the first transformer)
X 5 (in the FDA) = 10 (or 20dB). The ADT4-6T was selected
primarily for its excellent flatness and distortion down to 100kHz.
Its measured response showed a -1dB flatness span from 40kHz
to 178MHz when driven from 50Ω to 200Ω load with a -0.18dB
midband insertion loss. This is a suitable frequency span for the
intended 100kHz to 100MHz digitizer bandwidth in this board. It
does show a bit of rolloff at 100MHz which is partially equalized
by the 10pF capacitors to ground at the summing junctions in
Figure 4. Adding those capacitors does start to peak the output
noise of the ISL55210, but this stage will be followed by a
passive filter rolling that noise off. A detailed discussion of the
input referred noise figure for this transformer coupled FDA
topology may be found at:
http://www.edn.com/design/analog/4400484/Accuratelypredict-measured-noise-figures-for-transformer-coupleddifferential-amplifiers--Part-1-of-2-For the default configuration on this ISLA214P50-55210EV1Z
board using the components shown in Figure 4, the estimated
noise figure will be 7.2dB from a 50Ω source. Converting that to
an input referred spot noise voltage including a 50Ω source noise
gives a very low 1.02nV/√Hz. This is only for the amplifier stage,
and not including any noise in the original source signal.
Delivering this to the ADC inputs through the full signal path gain
AN1837.0
May 3, 2013
Application Note 1837
of 5.6V/V yields a 5.7nV/√Hz differential spot noise. Combining
this with the various noise elements within the ISLA214P50 will
give a slight degradation in the resulting SNR in the FFT. Those
calculations are described in this article: “Deliver the lowest
distortion and noise in a low power, wideband, ADC interface –
Part 2 of 4”
http://www.planetanalog.com/document.asp?doc_id=528177
The second ADTL1-12 common mode choke transformer
provides a very broadband, low insertion loss, element that
forces balance in this differential signal path. Testing with and
without this element showed a significant improvement in the
FDA output 2nd harmonic distortion at higher frequencies. This is
an optional element in the design and can be bypassed with the
optional shorts, but the best SFDR will be achieved with this
element included as it is in the standard board build.
ELEMENTS CONTRIBUTING TO THE PASSBAND
FLATNESS AND HIGHER FREQUENCY CUTOFF
Each of the elements in the signal path have fine scale rolloffs
that need to be considered to achieve the final ±0.8dB flatness
through the 100kHz to 100MHz intended digitizer range for this
example design board.
The ADT4-6T input transformer was selected mainly for its low
frequency performance. While specified as -1dB flat from
150kHz to 200MHz, typical devices measure to have a -1dB
flatness span when driven from a 50Ω source to a 200Ω load of
40kHz to 180MHz. This far exceeds the Mini-Circuits specified
flatness region on the low frequency side which is very typical for
these wideband baluns.
Figure 5 shows a comparison to measured and modeled
transformer response with a 50Ω source to 200Ω load. Since
there is limited data at low frequencies in the vendor data sheet,
no comparison is made to that.
6.0
5.5
MEASURED
5.0
MODELED
4.5
4.0
3.5
3.0
100k
1M
10M
100M
1G
FREQUENCY (Hz)
FIGURE 5. ADT4-6T RESPONSE CURVES
The measured curve is showing about -0.5dB at 100kHz and
-0.3dB at 100MHz. The Spice model (used in subsequent
simulations) is only attempting to match the high and low F-3dB
frequencies and the midband gain including the measured 0.2dB
insertion loss. That modeling approach is described in this
article: “Measuring and modeling wideband baluns for
application to ADC input stages”
http://www.planetanalog.com/author.asp?section_id=434&doc
_id=558824&
For a higher frequency range design, the MA/COM
MABA-0096-CF48A0 measures in the same configuration to have a
-0.5dB flatness span from 300kHz to 220MHz typically which would
make it a good choice for 1MHz to 200MHz analog input span.
The ADT1-12 common mode choke following this actually has
0dB insertion loss in this configuration at low frequencies. This
increases to -0.2dB midband with a -1dB point at >1GHz with
these higher 200Ω source and load impedances used at this
point in the signal chain.
The amplifier will have its own frequency response from these
source impedances and gain settings. Having good simulation
models for each of the elements in the design allow easy
comparisons of options. Setting up an iSim PE circuit for the
input stage of Figure 4 gives a simulation circuit of Figure 6.
FIGURE 6. SIMULATION CIRCUIT FOR THE INPUT STAGE PART OF THE ISLA214P50-55210EV1Z BOARD
6
AN1837.0
May 3, 2013
Application Note 1837
This is set up with the source V1 at a “2” amplitude to generate
the response from the input to C1 to VOUT as shown in Figure 7.
The amplifier circuit is expecting a 50Ω source impedance for
best flatness. The extrinsic one shown in Figure 6 will produce a
6dB loss to the board input of Figure 4 but simulating with a
source set to “2” will remove that matching loss from the VOUT
plot. The simulated response shows exceptional flatness down to
100kHz and about -1dB rolloff at 100MHz. That rolloff will be
equalized a bit with a slight peaking in the interstage filter
design.
20
18
16
(dB)
14
INTERSTAGE FILTER FROM THE FDA OUTPUTS TO
THE ADC
The signal path from the ISL55210 to the ADC is AC coupled,
allowing the amplifier and ADC to operate at the common-mode
voltage that optimize each device’s performance. As delivered,
the differential output of the ISL55210 operate with a common
mode voltage that is left to default to the internal 1.2V value.
This can be adjusted to different set point via an optional path.
The interstage passive circuit provides an AC coupled, 3rd order
low pass filter. Built into this filter are an ADC common mode
voltage servo loop which controls the common-mode DC voltage
delivered to the ADC input pins and a wideband passive sense
path going differential to single ended to directly measure the
response shape to the ADC inputs. The circuit as delivered is
shown in Figure 8.
12
10
8
6
4
10k
100k
1M
10M
100M
1G
FREQUENCY (Hz)
FIGURE 7. PREDICTED RESPONSE SHAPE TO VOUT FOR A 50Ω
SOURCE TO THE INPUT OF C1
FIGURE 8. OUTPUT INTERFACE FROM ISL55210 TO THE ISLA214P50 DIFFERENTIAL INPUTS
7
AN1837.0
May 3, 2013
Application Note 1837
FIGURE 9. SIMULATION CIRCUIT FOR THE INTERSTAGE FILTER AND VCM SERVO LOOP
Numerous options exist for providing this ADC input VCM voltage
for different designs. The two offered on this board, along with
several others, are detailed in this article: “Advantages to Precise
Input Common Mode Voltage Control to High Performance High
Speed ADC’s”
http://www.edn.com/design/analog/4389814/Advantages-toprecise-input-common-mode-voltage-control-to-highperformance-high-speed-ADCs?page=0
Critical to understanding the response shape are the estimated
internal ADC elements as shown in the simulation circuit for this
interface in Figure 9 (this element numbering here does not
follow the build schematic of Figure 8)
At the far right the ADC is modeled as 2-clock rate dependent
current sources (1.3mA here for the 500MSPS case) with an
internal lumped element 16pF in parallel with 200Ω. The probes
show the DC operating voltages where the 2.22V at the
ISL28113 outputs gets back to the targeted 0.96V at the ADC
inputs as those Icm currents pull down through the DC
impedances from the output of the ISL28113. The internal ADC
elements combine with the external RC elements to give the
simulated frequency response shape from the ISL55210 outputs
to the ADC inputs shown in Figure 10.
-2.5
dBV (|2-pos - |1-pos)/dB
Again, the green elements are optional and not populated. The
non-populated elements connecting into C1005 would be the
FDA VCM adjustment if desired. As delivered, C1005 simply
decouples the ISL55210 VCM control pin which defaults
internally to 1.2V on 3.3V supply. Not shown is a jumper on J2
from pin 2 to 3 to connect in the servo loop ADC VCM control
path. The ISLA214P50 ADC uses an unbuffered sample and hold
and will therefore sink a sample rate dependent common-mode
current which will give a sample rate dependent voltage drop
from the midpoint of resistors R1015 -- > R1018 and R2011,
R2012. The servo loop is used primarily to counteract the
sample-rate dependent voltage drop to deliver a fixed commonmode voltage to the ADC input pins across all sample clock rates.
An alternate connection uses pin 2 to 1 on jumper J2 and
populates R1019 to provide a fixed Thevenin source for the ADC
VCM control. This provides a simpler solution when the design is
known to be a fixed clock rate design.
-3.0
-3.5
-4.0
-4.5
-5.0
-5/5
The differential signal at the outputs of the ISL55210 proceeds
from left to right in Figure 8 through the 40.2Ω resistors to a
differential 5.6pF capacitor and then into the 1μF DC blocking
capacitors. Those level shift the DC operating voltage from the
FDA outputs to the required common mode voltage at the ADC
inputs. The rest of the passive filter from there is pair of series
82nH inductors then into a parallel RC network comprised of the
4 resistor network feeding the differential to single ended sense
path at the output of T4, an external 1pF differential capacitor
and then the internal RC elements of the ADC. A final circuit
element senses the average common mode voltage at the ADC
inputs using the 2-20kΩ resistors and feeds that into a low
frequency servo loop amplifier using the ISL28113 which then
feeds a DC control voltage to the center of the 4-resistor string
that acts to control the ADC common mode operating voltage to
the reference voltage applied at the ISL28113 V+ input.
8
-6.0
20M
40M
60M
80M 100M
150M
FREQUENCY (Hz)
FIGURE 10. TARGETED RESPONSE SHAPE IN THE INTERSTAGE
FILTER TO THE ADC
This slight peaking is intended to equalize some of the rolloff up
to the FDA outputs but then bandlimit quickly above 100MHz.
The V TEST of Figure 8 provides an easy means to verify the
frequency response shape from the board input to the ADC. The
4 resistor network feeding T4 in Figure 8 shows a about a 25Ω
source to each leg of the 1:1 transformer while its total
impedance across the signal path is part of the filter design. This
path will have considerable insertion loss (≈-31.8dB) but an
accurate replica of the response shape as shown for 2 boards
measured in Figure 11.
AN1837.0
May 3, 2013
Application Note 1837
-10
-10
-11
BOARD #1
-20
-12
-25
dBFS
MEASURED S21 (dBc)
-15
-30
-35
BOARD #2
-13
-14
BOARD #2
-40
-15
BOARD #1
-45
-50
1M
1.0M
100M
FREQUENCY (Hz)
1G
FIGURE 11. OVERALL FREQUENCY RESPONSE SHAPE FROM THE
BOARD INPUT TO THE ADC
The ADC also can be used to measure the signal path frequency
response by holding a constant input power while stepping the
frequency and recording the change in the dBFS out of the FFT.
This is shown on a linear frequency scale in Figure 12 targeting a
single tone at -12dBFS out of the FFT at 30MHz, then holding
constant input power and measuring the drop in dBFS as the
frequency is stepped up.
Clearly the overall response shape is doing a good job of
providing approximately 15dB gain from board edge with <-1dB
rolloff to 100MHz. It is important for distortion reasons to stay
away from the rolloff regions of the input step up transformer.
The intended minimum frequency in this application is 100kHz,
well above the 40kHz -1dB measured on t he ADT4-6T while the
maximum intended frequency is 100MHz which is also well
below the measured 180MHz -1dB frequency on the ADT4-6T.
The interstage filter bandlimits the broadband noise out of the
ISL55210 to reduce SNR degradation through the ADC while also
providing a bit of HD2 and HD3 attenuation from the FDA outputs
to the ADC inputs. For instance, a single tone 80MHz at the FDA
output pins will have an HD2 at 160MHz and an HD3 at 240MHz.
The response shape of Figure 11 suggests that HD2 term will get
about 4dB attenuation while the HD3 term will get 13dB
attenuation to the ADC inputs.
CLOCK AND CONTROL OPTIONS
The board offers several optional features that are in some cases
not fully populated. The clock options and two other control
inputs are shown in Figure 13.
The populated path for the ADC clock is the lower right input
through the TC4-19G2 transformer. It is this path that must have
a valid clock input (usually a filtered sine wave) when the
Konverter software is started. Lab tests here were at 10dBm to
14dBm input levels at J4.
An alternate clock path is through an ADI – ADCLK905
differential output ECL clock driver. That path is populated by
adding the SMA at the buffered clock input point at J5, adding
the ECL clock buffer chip, populating the coupling caps (C46 and
C47) and removing the coupling caps from the transformer input
path (C28 and C29). This alternate clock path allows much lower
power sine wave inputs into what is essentially a very low jitter
differential output comparator. Its inputs are also differential, but
9
-16
30
40
50
60
70
80
90
100
110
120
130
FREQUENCY (MHz)
FIGURE 12. ZOOMED IN RESPONSE SHAPE USING THE ADC TO
MEASURE FLATNESS
one side is biased to the midpoint threshold to run single ended
input. A very low phase noise sine wave inputs as low as -2dBm
will generate the necessary output clock transition times to drive
the ADC clock inputs. The clock chip includes internal 50Ω
termination for the sine wave source.
A similar signal path is shown in Figure 13 just below the
alternate clock path that provides an ADC sync operation if
populated. To use this, add an SMA connector at J7 and the clock
chip at U9. Refer to the ADC data sheet for the operation of this
control path.
The amplifier may also be disabled through a high speed
interface by populating the Pd SMA input through U5, a CMOS
inverter. As delivered, the DC coupled 50Ω termination resistor
holds the input at ground providing a 3.3V output to the disable
control line on the ISL55210. This holds the amplifier in the
enabled mode while connecting the SMA and driving that signal
to a logic high, will disable the amplifier. It is important to
recognize that even in the disable mode a signal path to the ADC
will be present through the feedback resistors. It will be
significantly attenuated from the active mode, but it will not be
an open circuit. The ISL55210 includes two desirable features
when disabled.
1. There are internal back to back diodes across the input
summing junctions to limit the amplitude of high overdrive
signals when disabled (or when active as well). In disable, this
limits the maximum differential voltage available across the
inputs to a diode voltage which is then all that can feed
forward to the ADC – at an attenuated level.
2. A low power monitor circuit holds the output VCM voltage at
the same set point during disable as for the active mode. This
prevents a long turn on time (or AC coupled common mode
voltage spikes) through blocking caps in the output interface
circuit as the amplifier cycles through enable/disable modes
– further protecting the ADC from out of range inputs.
Contact the factory for assistance in exercising these clock and
control options.
AN1837.0
May 3, 2013
10
Application Note 1837
FIGURE 13. CLOCK AND CONTROL PORTION OF THE SIGNAL PATH SCHEMATIC
AN1837.0
May 3, 2013
Application Note 1837
COMBINED FFT PERFORMANCE FOR THE INPUT
INTERFACE CIRCUIT AND THE ADC
The starting point for the dynamic range of the
ISLA214P50-55210EV1Z board would be the reported
performance of the ADC only in its typical 2-transformer input
evaluation board circuit. The typical ADC only EVM is described
here:
http://www.intersil.com/content/dam/Intersil/documents/isla/
isla214ir72ev1z_schem_layers.pdf
While the ISLA214P50 data sheet is available here:
http://www.intersil.com/content/dam/Intersil/documents/fn75
/fn7571.pdf
The SNR shown in Figure 2 from the ADC data sheet (and
repeated here) is actually SNRFS.
SNR (dBFS) AND SFDR (dBc)
95
90
85
4. HD3 ≈< -80dBc
The SNRFS will certainly drop from this 72.6dB by some amount
due to the added integrated noise presented to its inputs over a
simple transformer input test. The HD2 and HD3 will show a
complex result when combined with the ADC where that testing
will be done here at -2dBFS. Recognizing that the midrange input
frequencies will get little interstage filter help on the HD2 and
HD3 terms, the 40MHz FFT of Figure 16 is showing that the
intrinsic dynamic range up to the FDA outputs is so good as to
produce almost no degradation in ADC only operation.
3. HD2 has improved to -89dBc from ADC only of approximately
-85dBc
65
4. HD3 has improved to -93dBc from a typical -88dBc looking at
Figure 15 at 40MHz
60
0
100
200
300
400
INPUT FREQUENCY (MHz)
500
600
FIGURE 14. SNRFS AND SFDR vs F IN USING 500MSPS CLOCK
AND TARGETING -1dBFS
And then the swept frequency on just those HD2 and HD3 terms
from Figure 3 in the ISLA214P50 data sheet.
-55
HD2 AND HD3 MAGNITUDE (dBc)
3. HD2 ≈< -84dBc
2. SFDR is bit better than typical at 89dBc vs the ADC only
82dBc to 84dBc
SNR
70
55
2. SFDR ≈ 82dBc to 84dBc
1. SNRFS has dropped from 72.6dBc to 71.9dBc
SFDR
75
1. SNRFS ≈ 72.6dB
For this 283mVP-P single tone input at 40.07MHz that is
bandpass filtered and delivered to the SMA input, the:
SFDR (EXCLUDING H2, H3)
80
Combining the information in these plots along with the
specification table suggests the following typical numbers for the
ADC only up through 100MHz inputs:
The swept frequency HD2 performance actually improves slightly
over the ADC only plots. This might be attributed to the -2dBFS
target but certainly shows no degradation using the very high
dynamic range input interface circuit implemented on this board.
The swept frequency HD3 also seems slightly improved over the
typical ADC only plots.
-60
-65
-70
HD2
-75
-80
HD3
-85
-90
-95
Testing two boards for swept input frequency SNRFS gives the
plot of Figure 17 where the result has dropped about 1dB from
the ADC only data.
0
100
200
300
400
INPUT FREQUENCY (MHz)
500
600
FIGURE 15. HD2 AND HD3 vs F IN USING 500MSPS CLOCK AND
TARGETING -1dBFS
11
These plots are showing some degradation in SNRFS, and a
slight improvement on the HD2 and HD3 performance. Here,
however, the -2dBFS input level is only 283mVP-P at board edge
vs a much higher level for the typical ADC characterization
circuit. The harmonic distortion performance is the combined
result of many elements in the design. Hence, it is very difficult to
make too strong a claim on the worst case distortion. The limited
testing here seems to indicate a <-80dBc performance is
certainly being delivered where even <-85dBc seems possible.
While the signal path circuit here may seem a bit involved, it is
converting from single ended input and delivering a bandlimited
response to the ADC with precise common mode control and
exceptionally low noise and distortion using <120mW total.
Since the FFT’s are showing nearly as good performance as the
ADC itself, this solution is equivalent to a 14-bit, 500MSPS ADC
requiring only 300mVP-P single ended input for -1dBFS.
AN1837.0
May 3, 2013
Application Note 1837
FIGURE 16. 40MHz INPUT, 500MSPS FFT FROM THE KONVERTER SOFTWARE
730
-85
BOARD #1
-90
BOARD #2
BOARD #2
72.0
-95
-100
BOARD #1
71.5
71.0
HD2 (dBc)
SNRFS (dB)
72.5
-105
0
20
40
60
80
-110
100
0
20
INPUT FREQUENCY (MHz)
40
60
80
100
INPUT FREQUENCY (MHz)
FIGURE 17. 500MSPS SNRFS RESULTS FOR 2
ISLA214P50-55210EV1Z BOARDS
FIGURE 18. 500MSPS HD2 RESULTS FOR 2
ISLA214P50-55210EV1Z BOARDS
-80
BOARD #1
HD3 (dBc)
-85
-90
BOARD #2
-95
-100
-105
0
20
40
60
80
100
INPUT FREQUENCY (MHz)
FIGURE 19. 500MSPS HD3 RESULTS FOR 2 ISLA214P50-55210EV1Z BOARDS
12
AN1837.0
May 3, 2013
Application Note 1837
Tested Performance Over ADC Input VCM
Setting
-78
-80
THD (dBc)
Using the available ADC common mode voltage servo loop
feature, it is an easy matter to move the DC operating voltage at
the ADC inputs around and verify the range of good performance.
Using the same basic targets of -2dBFs with a fixed Fin at 30MHz,
one test board was swept from 0.9VCM to 1.1VCM. The figure of
merit here was the THD as the various spurious are moving
around a lot with VCM but the overall THD is relatively constant.
Figure 20 shows this test at two clock frequencies.
-82
THD
-84
-86
-83
-88
450
THD (-dBc)
-84
490
510
530
550
FCLK (MHz)
FIGURE 21. THD vs FCLK AROUND THE 500MSPS SPECIFIED
MAXIMUM CLOCK RATE
-85
2-Tone, 3rd Order IM3 Testing
-86
450MSPS
Since the board passes frequencies to 100MHz, duplicating the
70MHz IM3 performance reported in the ISLA214P50 data sheet
will show the combined performance for the ADC and the interface
circuit. The plot from the ISLA214P50 data sheet (Figure 16 there)
is shown in Figure 22.
-87
-88
-89
0.90
470
500MSPS
0.95
1.00
1.05
1.10
1.15
1.20
VCM SETTING AT ADC INPUTS (V)
0
IMD2
IMD3
2ND HARMONICS
3RD HARMONICS
FIGURE 20. THD vs ADC INPUT COMMON MODE VOLTAGE
The 500MSPS data is relatively insensitive to VCM input over this
range showing very robust performance to varying VCM input
voltages. Overall improved spurious performance has been
observed with this ADC at lower clock rates and the 450MSPS
data shows a bit more sensitivity to the ADC input VCM setting. In
this test, the DC operating points through the FDA are not
changing, none of the response shapes are changing up to the
ADC, the only variable is the DC average input voltage for the
signal being delivered to the ADC input pins. This is exercising
fine scale input impedance nonlinearities in the ADC against the
source impedance of the filter. While very robust over a relatively
wide input VCM range, the plot above suggested a 0.96VCM set
point for this board and that is the delivered condition. Changing
the filter design and/or ADC might suggest a reset on that target
ADC VCM voltage. This is easily accomplished using the VCM
servo loop feature.
Tested Performance with Fixed FIN and
Narrow Clock Range Around 500MSPS
Since it seemed the FFT improved somewhat in dropping just
below 500MSPS, a ±50MSPS range around 500MSPS was
evaluated with a fixed 50MHz input generating a -2dBFS in the
FFT. Looking again at the THD since the various spurious are
moving around a lot with each test, gives the example
performance of Figure 21.
This is indeed showing a pretty rapid improvement in THD
dropping below 500MSPS and a good guardband above 500MSPS
for acceptable performance. While it is not suggested that the ADC
be operated above 500MSPS, this plot does show a good margin
above that before catastrophic falloff in the THD. This is intended
to add over temperature margin in the ADC performance.
13
AMPLITUDE (dBFS)
-20
-40
-60
-80
IMD3 = -88dBFS
-100
-120
0
50
100
150
200
250
FREQUENCY (MHz)
FIGURE 22. ISLA214P50 IM3 PLOT AT 70MHz AND 71MHz
INPUTS FREQUENCIES
This is reporting a -88dBFS 3rd order intermodulation spurious
for the 2 close in spurs at ±3ΔFIN around the midpoint – that
would be at 69MHz and 71MHz here. Converting this dBFS to dBc
gives -80dBc for the IM3. For this broadband test, the IM2 is also
apparent at 141MHz and 1MHz. Duplicating this set up with
slightly lower carriers (-8dBFS vs -7dBFS on the ADC data sheet)
at 69.5MHz and 70.5Mz gives the wideband FFT of Figure 23.
In this case, with 2 test tone inputs, the reported SNR does not
compute correctly. It is easy to see here that the IM2 at 140MHz
has been suppressed quite a lot by the combined excellent even
order suppression in the interface circuit and the interstage filter.
The other IM2 at 1MHz is also lower. Zooming in on a 65MHz to
75MHz range in Figure 24 shows exceptionally low 3rd order
terms in this solution.
AN1837.0
May 3, 2013
Application Note 1837
FIGURE 23. FULL NYQUIST SPAN FFT FOR A 70MHz IM3 TEST
FIGURE 24. ZOOMED IN FFT AROUND THE CARRIER FREQUENCIES
14
AN1837.0
May 3, 2013
Application Note 1837
The data markers are showing -11dBFS on the carriers and
-105dBFS on the 3rd order intermodulation spurious terms. This
is the raw data where the actual carriers were each -8dBFS and
are reduced 3dB in the data by the 4-term Blackman Harris
windowing being used. In any case this is showing ≈-94dBc IM3
performance at 70MHz. The bandpass filter shape can also be
seen easily in the noise floor. This -94dBc far exceeds the ADC
performance shown in Figure 22 which might be attributed to a
lower spurious input test signal using the 15dB gain in front of
the ADC here or perhaps poorer IM3 in the interface elements on
the ADC only EVM board. The ISL55210 is very nearly unmeasurable for OIP3 at 70MHz and that is clearly reflected in the
significantly improved performance of Figure 24. Dropping the
test power levels showed an intercept performance in the FFT
where dropping only 3dBm on the two test powers dropped the
spurious 9dB into the noise floor.
Appendix A: Low Phase Noise RF
Generators
Some examples of low phase noise generators suitable for high
resolution ADC clock and source signal generation in test
include:
1. Rohde & Schwarz: SMA100A
2. Agilent 8664A
3. Gigatronics 6080A
TESTED PERFORMANCE OVER A WIDE RANGE OF
CLOCK FREQUENCY AND FIN
As the clock rate is reduced several slight changes in the
response can be expected.
1. The Icm current into the 2 ADC inputs will decrease. Using the
servo loop amplifier will act to hold the ADC input VCM voltage
constant as the clock rate is changed.
2. The ADC input resistance will increase slightly. The 200Ω
internal value shown in the simulation circuit of Figure 9 is a
combination of an extrinsic 300Ω element and the effective
resistance of a sampling cap. That impedance is
approximately 1/(Fs*3.3pF). As the clock rate decreases, this
impedance will increase moving the apparent input
resistance up. By 200MSPS the total ADC internal resistance
is ≈250Ω. This shift will slightly change the response shape of
the interstage filter.
3. Reducing the clock rate gives every operation internal to the
ADC a bit more time to settle and improved dynamic range
over the analog input frequency is observed.
The following figures summarize the swept input frequency
dynamic range vs Fsample repeating the 500MSPS data for
comparison to 450MSPS. These are all targeting a -2dBFS using
very low phase noise sources and bandpass filtering on both the
clock and Fin. In general, at clock rates ≤450MSPS the HD2 and
HD3 terms hold below -85dBc.
For fixed 500MSPS clock operation, the Crystek
RFPRO33-500.00 offers a simple solution. This device operates
within an SMA body and requires a 3.3V supply to produce the
required clock to operate this EVM. Bandpass filtering on the
clock always helps the SNR performance for any of these
sources. Most of the data here was taken with the 8664A which
seemed to give the best SNRFS results.
15
AN1837.0
May 3, 2013
Application Note 1837
450MSPS
730
73.0
72.5
72.5
72.0
BOARD #2
71.5
BOARD #1
71.0
0
SNRFS (dB)
SNRFS (dB)
500MSPS
72.0
BOARD #2
71.5
20
40
60
80
71.0
100
BOARD #1
0
20
INPUT FREQUENCY (MHz)
40
60
80
100
80
100
80
100
INPUT FREQUENCY (MHz)
FIGURE 25. 500MSPS SNRFS
FIGURE 26. 450MSPS SNRFS
-85
-85
BOARD #1
-90
-90
BOARD #1
HD2 (dBc)
HD2 (dBc)
BOARD #2
-95
-100
-95
-100
-105
-105
-110
-110
BOARD #2
0
20
40
60
80
100
0
20
INPUT FREQUENCY (MHz)
40
60
INPUT FREQUENCY (MHz)
FIGURE 27. 500MSPS HD2
FIGURE 28. 450MSPS HD2
-80
-80
BOARD #1
-85
-85
-90
HD3 (dBc)
HD3 (dBc)
BOARD #1
BOARD #2
-95
BOARD #2
-95
-100
-100
-105
-90
0
20
40
60
INPUT FREQUENCY (MHz)
FIGURE 29. 500MSPS HD3
16
80
100
-105
0
20
40
60
INPUT FREQUENCY (MHz)
FIGURE 30. 450MSPS HD3
AN1837.0
May 3, 2013
Application Note 1837
350MSPS
73.0
73.0
72.8
72.8
72.6
72.6
72.4
72.4
SNRFS (dB)
SNRFS (dB)
400MSPS
72.2
BOARD #1
72.0
71.8
71.6
72.0
BOARD #2
71.8
71.6
BOARD #2
71.4
71.4
71.2
71.2
71.0
0
20
40
BOARD #1
72.2
60
80
71.0
100
0
20
INPUT FREQUENCY (MHz)
40
60
80
100
80
100
80
100
INPUT FREQUENCY (MHz)
FIGURE 31. 400MSPS SNRFS
FIGURE 32. 350MSPS SNRFS
-85
-85
BOARD #1
BOARD #1
-90
HD2 (dBc)
HD2 (dBc)
-90
-95
BOARD #2
-100
-95
-100
-105
-105
-110
BOARD #2
-110
0
20
40
60
80
100
0
20
FIGURE 33. 400MSPS HD2
60
FIGURE 34. 350MSPS HD2
-80
-80
-85
BOARD #2
-90
BOARD #1
-85
HD3 (dBc)
HD3 (dBc)
40
INPUT FREQUENCY (MHz)
INPUT FREQUENCY (MHz)
-95
-90
-95
BOARD #1
-100
-100
BOARD #2
-105
-105
0
20
40
60
INPUT FREQUENCY (MHz)
FIGURE 35. 400MSPS HD3
17
80
100
0
20
40
60
INPUT FREQUENCY (MHz)
FIGURE 36. 350MSPS HD3
AN1837.0
May 3, 2013
Application Note 1837
300MSPS
250MSPS
73.0
73.0
72.8
72.6
BOARD #1
72.2
SNRFS (dB)
SNRFS (dB)
BOARD #1
72.5
72.4
72.0
71.8
BOARD #2
72.0
71.6
71.5
71.4
BOARD #2
71.2
71.0
0
20
40
60
80
71.0
100
0
20
INPUT FREQUENCY (MHz)
40
60
80
100
80
100
80
100
INPUT FREQUENCY (MHz)
FIGURE 37. 300MSPS SNRFS
FIGURE 38. 250MSPS SNRFS
-85
-85
BOARD #2
90
BOARD #2
-90
HD2 (dBc)
HD2 (dBc)
BOARD #1
-95
-100
BOARD #1
-100
-105
-105
-110
-95
0
20
40
60
80
-110
100
0
20
60
FIGURE 40. 250MSPS HD2
-80
-80
-85
-85
BOARD #2
HD3 (dBc)
HD3 (dBc)
FIGURE 39. 300MSPS HD2
-90
40
INPUT FREQUENCY (MHz)
INPUT FREQUENCY (MHz)
-95
-90
BOARD #1
-95
BOARD #1
-100
-105
-100
0
20
40
60
INPUT FREQUENCY (MHz)
FIGURE 41. 300MSPS HD3
18
80
100
-105
BOARD #2
0
20
40
60
INPUT FREQUENCY (MHz)
FIGURE 42. 250MSPS HD3
AN1837.0
May 3, 2013
Application Note 1837
200MSPS
73.0
-85
BOARD #1
72.5
BOARD #2
HD2 (dBc)
SNRFS (dB)
BOARD #1
-90
72.0
-95
BOARD #2
-100
71.5
-105
71.0
0
20
40
60
80
-110
100
0
20
INPUT FREQUENCY (MHz)
40
60
80
100
INPUT FREQUENCY (MHz)
FIGURE 43. 200MSPS HD2
FIGURE 44. 200MSPS HD2
HD3 (dBc)
-85
-90
-95
BOARD #1
-100
BOARD #2
-105
0
20
40
60
80
100
INPUT FREQUENCY (MHz)
FIGURE 45. 200MSPS HD3
Board Options
TABLE 2. PIN COMPATIBLE HIGH PERFORMANCE ADC FAMILY
While this board shows a complete example of a low power, high
dynamic range, single to differential amplifier stage, numerous
options can be implemented on this board. Principally, the gain
in the amplifier can be easily changed modifying the feedback
resistors up or down (R1008, R1011). The output filter can be
re-designed for a different passband. L1004 is included
(Figure 8) to implement bandpass filters on this board as well.
The ISLA214P50 is part of a large pin compatible ADC family.
Those can be dropped into this board replacing the ISL214P50
but will require some redesign in the filter for different ADC input
impedances and a reprogramming through the Konverter
software for the specific ADC. Contact the factory for assistance
with this. Table 2 summarizes the pin compatible, high
performance, ADC family supported by this single daughtercard.
These span a large range in bits and maximum clock rate where
lower clock rates run lower power in each family of devices.
PART NUMBER
RESOLUTION
(Bits)
MAXIMUM
SAMPLE RATE
(Msps)
POWER
CONSUMPTION
(mW)
ISLA216P25
16
250
785
ISLA216P20
16
200
720
ISLA216P13
16
130
615
ISLA214P50
14
500
835/900 (Note)
ISLA214P25
14
250
450
ISLA214P20
14
200
410
ISLA214P13
14
130
360
ISLA214P12
14
125
310
ISLA212P50
12
500
823/892 (Note)
ISLA212P25
12
250
440
ISLA212P20
12
200
405
ISLA212P13
12
130
355
NOTE: I2E disabled/enabled.
19
AN1837.0
May 3, 2013
Full Signal Path Schematic
The full schematic from input to ADC pins is shown in Figure 46.
20
Application Note 1837
AN1837.0
May 3, 2013
FIGURE 46. FULL SIGNAL PATH SCHEMATIC
Application Note 1837
The following BOM is for the entire ISLA214P50-55210EV1Z board. It includes some elements not described here that are associated
with the ADC operation common to its EVM board – the ISLA214P50IR72EV1Z. The main focus of this board is to add a very high
linearity input interface circuit that has been thoroughly described here. Elements in the following BOM that do not have a description in
the comments column are not part of the signal path but common to the ADC only EVM board.
ISLA214P50-55210EV1Z Bill of Materials
PART NUMBER
REFERENCE
QTY UNITS DESIGNATOR
ISLA214P5055210EZRVBPCB
1
ea
160X14W473MV4T-T
2
ea
GRM188R71E105KA12D-T 4
H1044-00100-50VR25-T
COMMENT
DESCRIPTION
MFR.
MFR. PART
PWB-PCB, ISLA214P5055210EZ, REVB, ROHS
IMAGINEERING ISLA214P50INC
55210EZRVBPCB
C1004, C1009 ISL55210 supply
decoupling
CAP-X2Y, SMD, 0603,
0.047µF, 16V, 20%, X7R,
ROHS
JOHANSON
DIELECTRICS
INC
160X14W473MV4T
ea
C1007, C1010, Signal path blocking
C1012, C1019 caps
CAP, SMD, 0603, 1µF, 25V,
10%, X7R, ROHS
MURATA
GRM188R71E105KA12D
2
ea
Cterm1,
Cterm2
Response shaping at
inverting summing
junctions
CAP, SMD, 0402, 10pF, 50V, AVX
0.25pF, NP0, ROHS
04025U100CAT2A
H1044-00101-50V5-T
1
ea
C1022
ISL28113 supply
decoupling cap
CAP, SMD, 0402, 100pF, 50V, MURATA
5%, C0G, ROHS
GRM1555C1H101JZ01D
H1044-00102-16V10-T
4
ea
CAP, SMD, 0402, 1000pF,
C28, C29, C48, ADCLK905 supply
16V, 10%, X7R, ROHS
C58
decoupling & ADC
transformer clock path
H1044-00103-16V10-T
3
ea
C30, C1020,
C1021
H1044-00104-16V10-T
17
ea
a) C4, C5, C20, Various supply
C22, C24, C27, decoupling caps,
C33, C34, C42 largely on the bottom
of the board.
CAP, SMD, 0402, 0.1µF, 16V, VENKEL
10%, X7R, ROHS
C0402X7R160-104KNE
H1044-00104-16V10-T
0
ea
b) C43, C44,
C45, C50, C56,
C113, C114,
C115
Various supply
decoupling caps,
largely on the bottom
of the board.
CAP, SMD, 0402, 0.1µF, 16V, VENKEL
10%, X7R, ROHS
C0402X7R160-104KNE
H1044-00105-10V10-T
1
ea
C1014
Vcm feedback point
decoupling cap.
CAP, SMD, 0402, 1.0µF, 10V, MURATA
10%, X5R, ROHS
GRM155R61A105KE15D
H1044-005R6-50VR5-T
1
ea
C16
3rd order interstage
differential filter cap
CAP, SMD, 0402, 5.6pF, 50V, MURATA
0.5pF, NP0, ROHS
GRM1555C1H5R6DZ01D
H1044-DNP
0
ea
a) C8, C9, C10,
C11, C12, C13,
C14, C15
CAP, SMD, 0402, DNP-PLACE
HOLDER, ROHS
H1044-DNP
0
ea
b) C25, C35Optional clock and
C38, C46, C47 control signal path
caps not populated
CAP, SMD, 0402, DNP-PLACE
HOLDER, ROHS
H1045-00010-50VR1-T
1
ea
Cdiff
3rd order interstage
differential filter cap
CAP, SMD, 0603, 1.0pF, 50V, TDK
0.1pF, NP0, ROHS
C1608C0G1H010B
H1045-00102-50V10-T
1
ea
C32
Clk input to
transformer
CAP, SMD, 0603, 1000pF,
50V, 10%, X7R, ROHS
AVX
06035C102KAT2A
H1045-00103-16V10-T
2
ea
C52, C54
ADCLK905 input signal CAP, SMD, 0603, 0.01µF,
coupling
16V, 10%, X7R, ROHS
VENKEL
C0603X7R160-103KNE
21
Blank Board
Shaded rows unpopulated
Vcm servo loop caps
around ISL28113 &
clock transformer
centertap AC gnd
CAP, SMD, 0402, 0.01µF,
16V, 10%, X7R, ROHS
TDK
C1005X7R1C102K
TDK
C1005X7R1C103K
AN1837.0
May 3, 2013
Application Note 1837
ISLA214P50-55210EV1Z Bill of Materials
PART NUMBER
REFERENCE
QTY UNITS DESIGNATOR
COMMENT
Shaded rows unpopulated (Continued)
DESCRIPTION
MFR.
Various supply and bias CAP, SMD, 0603, .1µF, 25V, MURATA
line decoupling caps. 10%, X7R, ROHS
MFR. PART
GRM39X7R104K025AD
H1045-00104-25V10-T
6
ea
C2, C3, C53,
C55, C1005,
C1023
H1045-00105-50V10-T
4
ea
C49, C51, C57, Supply decoupling on
C59
the ADCLK905's
CAP, SMD, 0603, 1µF, 50V,
10%, X5R, ROHS
H1045-DNP
0
ea
C1003, C1008, Alternate ISL55210
C1017, C1018 supply decoupling &
Vtest path coupling
CAP, SMD, 0603, DNP-PLACE
HOLDER, ROHS
H1065-00105-16V10-T
1
ea
C1002
CAP, SMD, 1206, 1µF, 16V,
10%, X7R, ROHS
H1112-00336-16V10-C-T
4
ea
C1, C6, C7, C31 Supply decoupling
H1121-00475-10V10-B-T
1
ea
C1001
0603CS-82NXGLU
2
ea
L1002, L1003 Signal path 3rd order
filter inductors
COIL-RF INDUCTOR, SMD,
0603, 82nH, 2%, 400mA,
ROHS
142-0701-851
1
ea
INPUT
Analog input signal
end launch SMA
CONN-RF, SMA JACK, 50Ω,
JOHNSON
142-0701-851
BMT, TAB-END LAUNCH, ROHS COMPONENTS
22-28-4360-1X3
1
ea
J2
Jumper to select ADC CONN-HEADER, 1X3,
BRKAWY-1X36, 2.54mm,
Vcm set path,
0.240X0.125, ROHS
defaulted to use the
servo loop with pins 2
& 3 shorted.
5002
1
ea
TP1
Sense point for
average CM at ADC
input pins
5004
1
ea
Amplifier +3.3V Sense point for the
3.3V supply to the
signal path amplifier
5011
1
ea
55091-1875
1
70280-0458-2X2
87832-1420
Main 3.3V supply line
decoupling
TDK
C1608X5R1H105K
PANASONIC
ECJ-3FB1C105K
CAP-TANT, SMD, C, 33µF, 16V, AVX
10%, ROHS
3.3V supply decoupling CAP-TANT, LOW ESR, SMD, B, KEMET
to ISL55210
4.7µF, 10V, 10%, 3.5Ω, ROHS
COILCRAFT
TAJC336K016RNJ
T491B475K010AT
0603CS-82NXGLU
MOLEX
22-28-4360
CONN-MINI TEST POINT,
VERTICAL, WHITE, ROHS
KEYSTONE
5002
CONN-MINI TEST POINT,
VERTICAL, YEL, ROHS
KEYSTONE
5004
GND
GND connector tap into CONN-MULTI-PURPOSE TEST KEYSTONE
for setting Vcm voltage PT, BLK, ROHS
with DVM
5011
ea
J6
Mezzanine connector
on board bottom to
mate to motherboard
2
ea
JP1, JP2
Headers to select ADC CONN-HEADER, 2X2,
BRKAWY-2X50, 2.54mm,
options. Only JP1 is
ROHS
used with jumper
connected set of pins
closer to Input
1
ea
J1
CPLD connector
22
MOLEX
55091-1875
MOLEX
70280-0458
CONN-HEADER, SHROUDED, MOLEX
SMD, 14P, 2mmPITCH,
CENTER SLOT, ROHS
87832-1420
CONN-HEADER, SMD, DUAL
ROW, 180P, 0.635PITCH,
6mmSTACK, ROHS
AN1837.0
May 3, 2013
Application Note 1837
ISLA214P50-55210EV1Z Bill of Materials
PART NUMBER
REFERENCE
QTY UNITS DESIGNATOR
COMMENT
Shaded rows unpopulated (Continued)
DESCRIPTION
MFR. PART
901-144-8RFX
1
ea
J4
default clock input
SMA to XFMR path
SPC02SYAN
2
ea
JP1-Pins 3-4,
J2-Pins 2-3.
Connecting jumpers on CONN-JUMPER, SHORTING,
the JP1 and J2 posts 2PIN, BLACK, GOLD, ROHS
HZ1206E601R-10-T
2
ea
L1001, L1005 +3.3V supply line
ferrites to ISL55210
and ISL28113
FERRITE-EMI CHIP, SMD,
LAIRD
HZ1206E601R-10
1206, 600Ω, 500mA, ROHS TECHNOLOGIES
MMZ2012R102A-T
14
ea
L3, L4, L8,
L10-L20
FERRITE BEAD, SMD, 0805,
1k, 0.5A, 100MHz, ROHS
24FC128-I/SN
2
ea
U2 U3
Back side components IC-I2C SERIAL EEPROM, 8P,
SOIC, 128 KBIT, ROHS
ISL28113FHZ
1
ea
U10
ADC Vcm control servo IC-RRIO OP AMP, 5P, SOT-23- INTERSIL
loop op amp
5, ROHS
ISL28113FHZ
ISL55210IRTZ
1
ea
U7
Signal path fully
differential amplifier
IC-DIFFERENTIAL AMP, 16P,
TQFN, 3X3, ROHS
INTERSIL
ISL55210IRTZ
ISLA214P50IRZ
1
ea
U1
ADC
IC-14-BIT, 500MSPS A/D
CONVERTER, 72P, QFN,
10X10, ROHS
INTERSIL
ISLA214P50IRZ
SN74AHC1G04DBVR-T
1
ea
U5
ISL55210 disable
control line inverter
IC-SINGLE INVERTER GATE, TEXAS
SN74AHC1G04DBVR
2V-5.5V, 5P, SOT-23-5, ROHS INSTRUMENTS
XC2C64A-7VQG44C
1
ea
U4
3299W-1-502LF
1
ea
R1043
ADC Vcm adjust pot
POT-TRIM, TH, 5k, 0.5W, 10%, BOURNS
3P, 3/8, ROHS
3299W-1-502LF
H2510-00200-1/16W1-T
1
ea
R1031
Isolating resistor into
ISL28113 summing
junction
RES, SMD, 0402, 20Ω,
1/16W, 1%, TF, ROHS
PANASONIC
ERJ2RKF20R0
H2510-003R9-1/16W1-T
2
ea
R2011, R2012 Kickback isolation at
ADC input pins
RES, SMD, 0402, 3.9Ω,
1/16W, 1%, TF, ROHS
VISHAY/DALE
CRCW04023R90FKED
H2510-00R00-1/16W-T
7
ea
Various shorting paths RES, SMD, 0402, 0Ω, 1/16W, VENKEL
R53, R54,
5%, TF, ROHS
R1009, R1010, in signal path circuit
R1118, R1020,
R1035
CR0402-16W-00T
H2510-01000-1/16W1-T
4
ea
RES, SMD, 0402, 100Ω,
R55, R56,
Input termination &
1/16W, 1%, TF, ROHS
R1006, R1007 gain elements, and
ADCLK905 output load
CR0402-16W-1000FT
H2510-01001-1/16W1-T
21
ea
a) R1, R4-R13,
R30, R36, R38,
R42, R52, R57,
RES, SMD, 0402, 1k, 1/16W, VENKEL
1%, TF, ROHS
CR0402-16W-102JT
H2510-01001-1/16W1-T
0
ea
b) R58, R59,
R60, R1041.
RES, SMD, 0402, 1k, 1/16W, VENKEL
1%, TF, ROHS
CR0402-16W-102JT
23
CONN-RF, SMA JACK, 50Ω,
PCB MNT, STRAIGHT, ROHS
MFR.
IC-CR-II CPLD, SMD, 44P,
VQFP, 64MACROCELLS,
10X10, ROHS
AMPHENOL
901-144-8RFX
SULLINS
SPC02SYAN
TDK
MMZ2012R102A
MICROCHIP
TECHNOLOGY
24FC128-I/SN
XILINX
VENKEL
XC2C64A-7VQG44C
AN1837.0
May 3, 2013
Application Note 1837
ISLA214P50-55210EV1Z Bill of Materials
PART NUMBER
REFERENCE
QTY UNITS DESIGNATOR
H2510-01002-1/16W1-T
2
ea
R46, R1034
H2510-02000-1/16W1-T
2
ea
R27, R1022
H2510-02002-1/16W1-T
3
ea
H2510-02320-1/16W1-T
1
H2510-040R2-1/16W1-T
COMMENT
Shaded rows unpopulated (Continued)
DESCRIPTION
MFR.
MFR. PART
RES, SMD, 0402, 10k,
1/16W, 1%, TF, ROHS
PANASONIC
ERJ-2RKF1002X
RES, SMD, 0402, 200Ω,
1/16W, 1%, TF, ROHS
PANASONIC
ERJ-2RKF2000X
R1032, R1033, R1032, R1033 part of RES, SMD, 0402, 20k,
R1042
Vcm servo loop
1/16W, 1%, TF, ROHS
PANASONIC
ERJ2RKF2001
ea
R1021
RES, SMD, 0402, 232Ω,
1/16W, 1%, TF, ROHS
VENKEL
CR0402-16W-2320FT
2
ea
R1013, R1014 Differential 3rd order
filter first resistors
RES, SMD, 0402, 40.2Ω,
1/16W, 1%, TF, ROHS
PANASONIC
ERJ-2RKF40R2X
H2510-054R9-1/16W1-T
2
ea
R48, R49
RES, SMD, 0402, 54.9Ω,
1/16W, 1%, TF, ROHS
VISHAY/DALE
CRCW040254R9FKED
H2510-07151-1/16W1-T
1
ea
R1044
RES, SMD, 0402, 7.15k,
1/16W, 1%, TF, ROHS
PANASONIC
ERJ-2RKF7151X
H2510-095R3-1/10W1-T
2
ea
R50, R51
RES, SMD, 0402, 95.3Ω,
1/16W, 1%, TF, ROHS
PANASONIC
ERJ-2RKF95R3X
H2510-DNP
0
ea
a) R3, R35,
R16, R14 R15,
R40, R41,
R1019
H2510-DNP
0
ea
b) Rbypass1Rbypass4,
Rterm1,
H2511-00R00-1/10W-T
3
ea
R32, R33,
R1000
RES, SMD, 0603, 0Ω, 1/10W, VENKEL
TF, ROHS
CR0603-10W-000T
H2511-01001-1/10W1-T
2
ea
R62, R63
RES, SMD, 0603, 1k, 1/10W, PANASONIC
1%, TF, ROHS
ERJ-3EKF1001V
H2511-01002-1/10W1-T
1
ea
R34
RES, SMD, 0603, 10k,
1/10W, 1%, TF, ROHS
KOA
RK73H1JT1002F
H2511-026R7-1/10W1-T
2
ea
R1016, R1017 Sense path divider
output to 1:1
transformer
RES, SMD, 0603, 26.7Ω,
1/10W, 1%, TF, ROHS
PANASONIC
ERJ-3EKF26R7V
H2511-04701-1/10W1-T
6
ea
R2, R29, R31,
R37, R39, R61
RES, SMD, 0603, 4.7k,
1/10W, 1%, TF, ROHS
YAGEO
9C06031A4701FKHFT
H2511-049R9-1/10W1-T
1
ea
Rterm3
RES, SMD, 0603, 49.9Ω,
Disable input
1/10W, 1%, TF, ROHS
(ISL55210)
termination to ground
at logic gate
VENKEL
CR0603-10W-49R9FT
H2511-05490-1/10W1-T
2
ea
R1015, R1018 Sense path divider
output to 1:1
transformer
RES, SMD, 0603, 549Ω,
1/10W, 1%, TF, ROHS
VENKEL
CR0603-10W-5490FT
H2511-DNP
0
ea
R1001, R1002, Alternate Vcm setup
RES, SMD, 0603, DNP-PLACE
R1004, R1005, for fixed clock and
HOLDER, ROHS
Rterm2
divider to disable logic
input
RR0510P-4990-D-T
2
ea
R1008, R1011
SUSUMU CO.,
LTD
RR0510P-4990-D
24
R1022 part of Vcm
servo loop.
ADC Vcm output
pulldown
Sync output series
elements
Sync output shunt
elements
RES, SMD, 0402, DNP, DNP,
DNP, TF, ROHS
Optional bypass
RES, SMD, 0402, DNP, DNP,
elements around input DNP, TF, ROHS
transformers
RES, SMD, 0402, 499Ω,
1/16W, 0.5%, THINFILM,
ROHS
AN1837.0
May 3, 2013
Application Note 1837
ISLA214P50-55210EV1Z Bill of Materials
PART NUMBER
REFERENCE
QTY UNITS DESIGNATOR
COMMENT
Shaded rows unpopulated (Continued)
DESCRIPTION
MFR.
MFR. PART
B3FS-1000P-T
1
ea
SW1
ADC Reset button
SWITCH-PUSH, TH, 6MM, OFF- OMRON
MOM, SPST-NO, 100GF, ROHS
ADT1-1WT+
1
ea
T4
Vtest path output
transformer
TRANSFORMER-RF, SMD, 6P, MINI-CIRCUITS ADT1-1WT+
CASE CD542, 0.5W, 30mA,
ROHS
ADT4-6T+
1
ea
T1 *(PIN 1 AT
UPPER LEFT
CORNER)
Input step up
transformer
MINI-CIRCUITS ADT4-6T+
TRANSFORMER, SMD, 6P,
7.8X5.5, 50Ω, 0.06-300MHz,
ROHS
ADTL1-12+
1
ea
T2 *(PIN 1 AT
UPPER LEFT
CORNER)
Input common mode
choke transformer
TRANSFORM-RF, SMD, 6P,
CASECD542, 20-1200MHz,
2W, ROHS
TC4-19G2+-T
1
ea
T3
Transformer for clock
input
TRANSFORMER-RF, SMD, 6P, MINI-CIRCUITS TC4-19G2+
AT224-3, 50Ω, 1/4W, 30mA,
ROHS
5X8-STATIC-BAG
1
ea
Place assy in
bag.
DNP
0
ea
J5, J7
SMA
DO NOT POPULATE OR
PURCHASE
DNP
0
ea
L1004
Optional filter for BP
design to ADC
DO NOT POPULATE OR
PURCHASE
DNP
0
ea
Pd Vtest
SMA
DO NOT POPULATE OR
PURCHASE
DNP
0
ea
R1003
2kΩ 10 turn pot for the DO NOT POPULATE OR
ISL55210 Vcm control PURCHASE
path.
DNP
0
ea
R22-R26, R28
DNP
0
ea
U6, U8, U9
LABEL-SERIAL NUMBER
1
ea
AFFIX LABEL TO
BOTTOM OF
PCB
BAG, STATIC, 5X8, ZIPLOC,
ROHS
B3FS-1000P
MINI-CIRCUITS ADTL1-12+
INTERSIL
212403-013
DO NOT POPULATE OR
PURCHASE
ADCLK905 for U8, U9 DO NOT POPULATE OR
PURCHASE
LABEL-FOR SERIAL NUMBER INTERSIL
AND BOM REV #
LABEL-SERIAL NUMBER
Intersil Corporation reserves the right to make changes in circuit design, software and/or specifications at any time without notice. Accordingly, the reader is
cautioned to verify that the Application Note or Technical Brief is current before proceeding.
For information regarding Intersil Corporation and its products, see www.intersil.com
25
AN1837.0
May 3, 2013