And now for something completely different
description
Transcript of And now for something completely different
And now for something completely different
First results from the DC-DC StaveletPeter & Ashley
with Giulio & John
2
DC-DC Stavelet First Module
Unstuck?
Power ONPower ON
Power OFF
SE4445 is used to attach modules tostavelets. At room temperature, this takes a long time to cure. When the module was first powered, it bowed as shown. It’s notquite as bad now, but it does still bow.• Do we need to remove and re-attach this module or is it OK to leave it?• If we use SE4445 for further modules, do we need to allow a longer cure before application of power?
H0H1
3
4
Chip & Strip Numbers
32 33 34 35 36 37 38 39 40 41
64 65 66 67 68 69 70 71 72 73
BCC
Edge
DC-D
C Ed
ge
Channel 0 Channel 1279
Strictly, the hardware channel numbering is contiguous only for the upper column. Chip 64 for example has channel 127 at the sensor
edge – but in all plots which follow this has been corrected in software to match the above labels.
5
ShieldConnections
(IN USE)
ShieldConnection
(NOT IN USE)
BackplaneConnection
(H1 only)
Shield Referencing
Option toLink LVDS& SensorShields
6
H0 Original ShieldH1 Original Shield
Both hybrids at Once
H0 Original ShieldH1 Unpowered
H0 UnpoweredH1 Original Shield
10/06/2011, 10V 3.0A limit, H0 DC referenced to shield, H1 AC referenced to shield
H0 Outer Column
H0 Inner Column
H1 Inner Column
H1 Outer Column
ENC MeasurementsCh 0 -> by BCC
Ch 1279 -> by DC-DC
710/06/2011, 10V 3.5A limit, H0 DC referenced to shield, H1 AC referenced to shield
H0 Wrapped ShieldH1 Wrapped Shield
Both hybrids at Once
H0 Wrapped ShieldH1 Unpowered
H0 UnpoweredH1 Wrapped Shield
ENC MeasurementsCh 0 -> by BCC
Ch 1279 -> by DC-DC
8
Cursory investigation of B fieldalong axis through converter centres
0 5 10cm
Converter PCB now closer to the hybrid edge than for the
single module DC-DC tests (~2mm from
~7mm) and it’s also now in the same
plane. Therefore the scope for pickup is
increased.
9
HMS3000 Typical Screenshot
Shields wrapped in Cu tape, both converters running, 3cm point
10
Shields Removedboth converters running
0 2 4 6 8 10 12 14 16
-90
-80
-70
-60
-50
-40
-30
1.96 MHz3.91 MHz5.87 MHz7.85 MHz9.80 MHz
NoiseFloor
cm
dBmNot Calibrated: relative measurement only
11
Shields Removed H0 converter running, H1 powered down
0 2 4 6 8 10 12 14 16
-90
-80
-70
-60
-50
-40
-30
1.96 MHz3.91 MHz5.87 MHz7.85 MHz9.80 MHz
NoiseFloor
cm
dBmNot Calibrated: relative measurement only
12
Shields Removed H0 powered down, H1 converter running
0 2 4 6 8 10 12 14 16
-90
-80
-70
-60
-50
-40
-30
1.96 MHz3.91 MHz5.87 MHz7.85 MHz9.80 MHz
NoiseFloor
cm
dBmNot Calibrated: relative measurement only
13
Shields Wrapped in Copper Tapeboth converters running
0 2 4 6 8 10 12 14 16
-90
-80
-70
-60
-50
-40
-30
1.96 MHz3.91 MHz5.87 MHz7.85 MHz9.80 MHz
NoiseFloor
cm
dBmNot Calibrated: relative measurement only
14
Shields Wrapped in Copper TapeH0 converter running, H1 powered down
0 2 4 6 8 10 12 14 16
-90
-80
-70
-60
-50
-40
-30
1.96 MHz3.91 MHz5.87 MHz7.85 MHz9.80 MHz
NoiseFloor
cm
dBmNot Calibrated: relative measurement only
15
Shields Wrapped in Copper TapeH0 powered down, H1 converter running
0 2 4 6 8 10 12 14 16
-90
-80
-70
-60
-50
-40
-30
1.96 MHz3.91 MHz5.87 MHz7.85 MHz9.80 MHz
NoiseFloor
cm
dBmNot Calibrated: relative measurement only
16
Field leaks through gap between shield and converter PCB
Will a “skirt” help?fitted over existing
layer of copper tapeH0
H1
Also revised hybrid referencing to make both DC referenced
17
Effect of “Skirt” on B field
These measurements were made using the smallest magnetic field probeplaced immediately adjacent to the converter shield of hybrid H0 of the
single DC-DC module M5 (not stavelet module M13).
WITHOUT SKIRT WITH SKIRT
18
H0 Wrapped ShieldH1 Wrapped Shield
Both hybrids at Once
H0 Wrapped ShieldH1 Wrapped Shield
One hybrid at a time
H0 Wrapped + SkirtH1 Wrapped Shield
Both hybrids at Once
H0 Wrapped + SkirtH1 Wrapped Shield
One hybrid at a time
21/06/2011, 10V 5.0A limit, Both hybrids DC referenced
19
H0 Wrapped ShieldH1 Wrapped Shield
Both hybrids at Once
H0 Wrapped ShieldH1 Wrapped Shield
One hybrid at a time
H0 Original ShieldH1 Wrapped Shield
Both hybrids at Once
H0 Original H1 Wrapped Shield
One hybrid at a time
21/06/2011, 10V 5.0A limit, Both hybrids DC referenced
20
TEST 1: 5.0AWrapped Shields
Both hybrids at Once
TEST 2: 3.5AWrapped Shields
Both hybrids at Once
TEST 3: 3.0AWrapped Shields
Both hybrids at Once
TEST 4: 5.0AWrapped Shields
Both hybrids at Once
22/06/2011, 10V variable current limit, Both hybrids DC referenced
TEST 1: 21/06/11Sensor Shield DC
@ BCC edgeBoth hybrids at Once
TEST 2: 21/06/2011Sensor Shield DC
@ BCC edgeEach hybrid separately
TEST 3: 21/06/2011Sensor Shield DC
@ both edgesBoth hybrids at Once
TEST 4: 21/06/2011Sensor Shield DC
@ both edges Each hybrid separately
Wrapped shields used throughout, TTi supply at 10V 5A limit
22
Double Trigger Noise @ 0.75fC
TTi TSX3510P @ 10V, wrapped shields, both hybrids at once (yes it really is this clean)
23
Double Trigger Noise @ 0.50fC
TTi TSX 3510P @ 10V, wrapped shields, both hybrids at once
24
Summary• Increasing the TTi TSX3510P PSU current limit above 3.0A has no
effect upon the ENC noise– Avoid values below this
• supply will not trip but may current limit
• Referencing changes have had a greater effect than increasing the mass of the shields– Extra 2 bonds per hybrid at DC-DC edge helped reduce the noise
• Beyond that, all that’s present in the design is the option to connect the sensor shield pad to the LVDS shield.
• Next : further reduce inductance of the shield connection– How can we improvise Cu or Al foil connections between Al shield and
adjacent DC-DC power return line?• See over for first attempt....
25
Lowering the Inductance of the Shield Connection
1. Remove kapton tape from power bus2. Partially remove kapton cover from sensor shield3. Add a dozen bonds to connect them together=> Sensor Shield is connected directly to power GND
TEST 3: 21/06/2011Sensor Shield DC
@ both edgesBoth hybrids at Once
TEST 4: 21/06/2011Sensor Shield DC
@ both edges Each hybrid separately
NEW TEST 5: 29/06/2011Sensor Shield DC & to power return
Both hybrids at Once
TEST 5: 30/06/2011Sensor Shield DC
& to power Return Each hybrid separately
Wrapped shields used throughout, TTi supply at 10V 5A limit
27
Future Plans
• Continue to lower the inductance of the shield connection
• What if this does not work?
????
Backup
29
Naked Converters!Converter 06: BC0000000476043A Converter 14: 7C0000000455D23A
Q) Does the precise physical orientation of each coil have any great significance?
30
M13 Single Module Result
(SP @ Liverpool)
M13 Stavelet Bestrunning both at once
Wrapped Shields
M13 Stavelet Bestrunning one at a time
Wrapped Shields