Delivering OCXO-grade PTP Performance with MEMS Precision TCXO Nazariy Tshchynskyy WSTS 2017
Delivering OCXO-grade PTP
Performance with MEMS Precision TCXO
Nazariy Tshchynskyy
WSTS 2017
2
Introduction
• Ethernet is becoming increasingly popular method for transferring data in a
mobile backhaul, but it’s asynchronous in nature
• LTE BS (eNodeB) requires 1.5 us time/phase synchronization
• PTP can be used to achieve sub-microsecond time synchronization over a
packet network, like Ethernet
• High quality local oscillator is required for best PTP performance
• OCXO’s are considered higher performance than TCXO’s because they are
less sensitive to airflow and offer tighter frequency stability
• Do you need a frequency stability of OCXO if you can get the same
level of ΔF/ΔT performance with precision MEMS TCXO?
3
Factors affecting PTP Accuracy
• Local oscillator quality
• Sensitivity to external conditions (for example, slope of frequency over
temperature, sensitivity to VDD change)
• Wander
• Control loop design
• Control loop bandwidth/transfer function
• Packet delay variation
• Network load
• Network architecture
• Use of network devices with PTP support (Transparent Clocks, Boundary Clocks)
• Time stamping accuracy
• Typically hardware based time stamping used to avoid software delays
• Hardware timestamping resolution is usually in nanosecond range
4
Servo Loop BW is a Tradeoff between
PDV filtering and Oscillator Noise
PDV Servo Loop is LPF for PDV
Filtered time
+tau
ADEV
Dominated by Temperature
sensitivity
Servo Loop is HPF for OSC
Time_error (s)
time (s)
• Time Error is a combination of Network Performance and Oscillator Noise
• Tradeoff between PDV filtering and Oscillator Noise is defined by Servo
Loop Bandwidth
5
Oscillator performance determines the
limits of PDV filtering capability
• Lower servo loop bandwidth → better PDV filtering
• Requires oscillator with good ADEV at long tau (dominated by temperature effects)
• ΔF/ΔT of an oscillator (slope) is a Temperature to ADEV conversion factor
• For best PTP performance ΔF/ΔT of an oscillator should be minimized
Oscillator characteristics
determine the limits of
PDV filtering capability
Note: servo loop algorithm may use higher bandwidth in unlocked state to ensure fast lock time and reduce the bandwidth
once locked to improve filtering performance
Packet selection and time stamping
Low pass filter (eliminates rapid
transients)
Servo Loop(Low BW to filter
PDV jitter)
TCXO
Tuning
CLK
GMII
6
Properties of the Oscillator that affect PTP
Performance
• Sensitivity to temperature changes (defined as Frequency Slope)
• Dominating contributor to Time Error
• Ambient temperature variations translate to oscillator output frequency change
• Short term aging (1-day aging)
• Has little impact on PTP performance if 1 ppb/day or better
• Native oscillator wander
• In good quality TCXO’s is small enough and doesn’t impact µs-level Time Error
performance
• Important for achieving <100 ns Timer Error performance level
7
Simulation methodology of Local
Oscillator impact on PTP performance
Time_error (s)
time (s)
Time error (t)
1.5us
-1.5us
PN (dBc/Hz)
f offset (Hz)
Phase noise and ADEV
Phase Noise → Time Jitter
+
f (ppb)
time (s)
TCXO frequency in time
∫0
tΔf(t) Δφ(t) /(2πfc)
PLL responsemodel
f
H(f)
f (ppb)
T (°C)
f (ppb)
time (s)
1 ppb/day
f (ppb)
time (s)
Frequency vs Temperature 1 day aging
Temperature ProfileT (°C)
time (s)
T1 T2
T1
T2
Frequency change in time (due to temperature)
Dwell time = 15 minRamp rate = 0.5 °C/min
+
PLL responsemodel
f
H(f)
8
PTP Performance with 1 ppb/°C and
10 ppb/°C TCXO (Time Constant 10 min)
9
PTP Performance with 1 ppb/°C and
10 ppb/°C TCXO (Time Constant 1 min)
10
Frequency Slope over Temperature
• Frequency Slope over Temperature is a measure of frequency change due
to temperature change by 1°C and is typically expressed in ppb/°C
MEMS-Based TCXO Frequency Stability – Measured
Results
-100
-50
0
50
100
-50 -10 30 70 110
Fre
qu
en
cy S
tab
ility
(p
pb
)
Temperature (C)
20 ppb
100 ppb
ΔT
ΔF
Slope = ΔF/ΔT [ppb/°C]
11
Which part is better? 50 ppb or 100 ppb?
-100
-50
0
50
100
-50 -10 30 70 110
Fre
qu
en
cy S
tab
ility
(p
pb
)
Temperature (C)
± 50 ppb
± 100 ppb
12
MEMS Precision TCXO delivers OCXO-
level Frequency Slope Performance
1 ppb
- 1 ppb Better than 1ppb/°C
OCXO-level performance
13
Time Error Measurement Setup
Master
Oregano Systems syn1588® VIP
Ethernet Slave
Oregano Systems syn1588® VIP
Slave
Oregano Systems syn1588® VIP
Ethernet
Ethernet
Unmanaged Switch
Frequency CounterAgilent 53230A
Frequency CounterAgilent 53230A
1pps output
CH 1 CH 1
CH 2 CH 2
- -CH 1 – CH2 CH 1 – CH2
1pps output
1pps output
OCXO
MEMS TCXO
Quartz TCXO or OCXO
14
Time Error Measurement Data
(Temperature Transient)
Hot air 50°C applied to
oscillators
15
Short Term Holdover is a Reflection of the
Oscillator Performance
• Short term holdover may range from few seconds to few hours
• During holdover servo loop freezes the TCXO tuning at the last
known good value
• Holdover performance is a reflection of the oscillator characteristics
• Possible causes may include
• Master change – few seconds to few minutes
• Equipment failure or reconfiguration – up to few hours
• During the holdover clock should maintain Time Error within the
specified limits while running of the local oscillator
16
Short term Holdover Simulation
f (ppb)
T (°C)
f (ppb)
time (s)
1 ppb/day
f (ppb)
time (s)
Frequency vs Temperature 1 day aging
Temperature ProfileT (°C)
time (s)
T1 T2
T1
T2
Frequency change in time (due to temperature)
Dwell time = 15 minRamp rate = 0.5 °C/min
f (ppb)
time (s)
+
TCXO frequency in time
∫0
tΔf(t) Δφ(t) /(2πfc)
PN (dBc/Hz)
f offset (Hz)
Phase noise
Phase Noise → Phase Jitter
+
Time_error (s)
time (s)
Time error (t)
1.5us
tholdover
17
Short term Holdover Simulation
18
Short term Holdover Measurement
Slave lost network
connection
19
Conclusions
• PTP devices require high quality oscillators to achieve good accuracy
• Better stability oscillators allow tuning servo loops for better PDV filtering
• Frequency Slope impacts PTP performance not Frequency Stability
over full operating temperature range
• TCXO’s with the same Frequency stability spec may have significantly
different Slope over Temperature
• MEMS-based precision TCXO’s have been designed to minimize
Frequency Slope over Temperature (5x to 20x improvement comparing to
Quartz TCXO’s) and can be used to replace OCXO’s in PTP applications
• SFP modules is an example of an application where MEMS-based precision
TCXO’s provide OCXO-level performance while saving critical space and
power
20
Thank You!
Questions?