Folie 1 - RWTH Aachen University
Download
Report
Transcript Folie 1 - RWTH Aachen University
Compilation of Dis-/Advantages
of DC-DC Conversion Schemes
Katja Klein
1. Physikalisches Institut B
RWTH Aachen University
Power Task Force Meeting
December 16th, 2008
Advantages: Grounding
• Standard grounding scheme
Module ground potentials are all the same
Common ground reference for bias, analogue and digital voltage for
whole substructure (rod, petal)
Bias voltage ground reference is the same for all modules
Easier for slow controls (difficult in SP to sense voltages)
Katja Klein
Discussion of DC-DC Conversion
2
Advantages: Communication
• Readout and control scheme is very standard
AC-coupling of communication not needed
Control chips can be supplied independently of modules
Katja Klein
Discussion of DC-DC Conversion
3
Advantages: Start-Up & Selective Powering
• Easy start-up
Control chips can be powered on first
If one converter per module, single modules can be powered on/off
In scenario w/ charge pump per chip, single chips can be powered on/off
Katja Klein
Discussion of DC-DC Conversion
4
Advantages: Different Voltages
• Different voltages can be provided
Buck-type converters: the same converter chip can be configured for
different output voltages
Via a resistive bridge
Two conversion steps can be combined
No efficiency loss (in contrast to linear regulation in SP)
Can cope with Vopto > Vchip
Can cope with Vana ≠ Vdig
Charge pumps: only integer conversion ratios, defined by configuration
Katja Klein
Discussion of DC-DC Conversion
5
Advantages: Flexibility
• Great flexibility with respect to
combination of modules with different load
Different numbers of readout chips
Trigger modules vs. standard modules
power groups with different number of modules
TEC vs. barrel
• In contrast, with SP current is fixed to highest current needed by any
chain member chains must be uniform to avoid burning power in regulators
Katja Klein
Discussion of DC-DC Conversion
6
Advantages: Changing Loads
• Compatibility with changing loads, relevant for
pixel detector
load is driven by occupancy
trigger modules
• SP: the highest current potentially needed must always be provided
inefficiency
Katja Klein
Discussion of DC-DC Conversion
7
Disadvantages: Chip Technology
• Need for a “high voltage“ tolerant process (10-12V)
Good candidate identified, radiation hardness still to be proven
Strong dependency on foundry: support of process over years?
Any changes in process must be followed closely and irradiation
tests be repeated
Katja Klein
Discussion of DC-DC Conversion
8
Disadvantages: Converter Efficiency
• Converter efficiency will be around 80%
(ESR of passive components, Ron of transistors, switching losses)
Local generation of heat cooling of DC-DC converters needed
Local efficiency decreases with lower conversion factor (Uout/Uin)
Local efficiency decreases with higher switching frequency
In two-step schemes efficiencies multiply
Katja Klein
Discussion of DC-DC Conversion
9
Disadvantages: Currents in Cables
• Cannot compete with Serial Powering
Currents in power group with DC-DC conversion = I0nr
I0 = current of a single module
n = number of parallely powered modules in the power group
r = conversion ratio = Uout/Uin
Current in Serial Powering chain = I0, independent of n
E.g. for 20 modules in power group need r = 20 to compensate
Higher efficiency in SP (at least up to FE) less cooling needed
Cables inside tracker volume can be thinner with SP
Katja Klein
Discussion of DC-DC Conversion
10
Disadvantages: Risks
• We have to stick with parallel powering
Multiplicity (modules per cable) as today or higher
Open connections (e.g. at PP1, PP0) lead to loss of power group
Short on module leads to loss of power group
Protection needed? Use DC-DC converter to switch off module?
Converter can break: can imagine isolated failures (loss of
regulation...) and failures that lead to loss of power group (short)
More risky if one converter powers several modules
Do we need redundancy?
This adds mass
Katja Klein
Discussion of DC-DC Conversion
11
Disadvantages: Material & Space
• Material budget and space considerations
Amount of copper in cables scales with current = I0nr
I0 = current of a single module
n = number of modules in the power group
r = conversion ratio = Uout/Uin
Air-core inductor (even if integrated into PCB, it needs a lot of copper)
Filter capacitors, maybe other filter components
With regulation (buck etc.), PCB traces can be narrow
Without regulation (charge pumps), input voltage must be exact
Linear regulator or rather solid input traces
Is shielding needed? How to design good low mass shielding?
Katja Klein
Discussion of DC-DC Conversion
12
Disadvantages: Material Budget
Simulated components:
Kapton substrate with 4 copper layers
Copper wire toroid
Resistors & capacitors
Chip
TEC
1 converter / module
Motherboards
Analog
OptoHybrids
Kapton
circuits
FE-hybrids
Katja Klein
Discussion of DC-DC Conversion
13
Disadvantages: Material Budget
Total gain for strip tracker with 1 converter per module and a conversion ratio of 8;
with power cables and motherboards modified accordingly:
Strip tracker
Katja Klein
Discussion of DC-DC Conversion
14
Disadvantages: Noise
• DC-DC converters are undoubtedly noise sources (by design)
Conductive noise through cables
Ripple on output voltage: switching frequency (1-5MHz) + higher harmonics
are in the bandpath of the amplifier
Switching leads to high frequency noise (tens of MHz, not so critical)
Both CM and DM contributions
Radiated noise
From inductor near field via inductive (and capacitive?) coupling
From cables
Has to be taken into account for all aspects of electronics system design:
readout chip, FE-hybrid, grounding & shielding, motherboard, layout ...
Not clear what to prepare for: noise depends on implementation
For same chip, noise emission can be rather different depending on PCB etc.
Scalability from lab system to complete detector not obvious
PS noise requirements to be understood
Katja Klein
Discussion of DC-DC Conversion
15