|
|
# Heatsinks for the FMC Fine Delay
|
|
|
# Temperature issues
|
|
|
|
|
|
This page summarizes the heatsink testing results. All tests made with
|
|
|
version EDA-02267-V6-0.
|
|
|
Despite rather low thermal dissipation (approx. 7.5 W), the cards are
|
|
|
getting quite hot in installations. This page summarizes the Fine Delay
|
|
|
heating problems and proposes solutions to mitigate them:
|
|
|
|
|
|
## Fine Delay on SVEC (lab test)
|
|
|
|
|
|
*Test evironment:**
|
|
|
|
|
|
\- crate: 20-slot Wiener VME64x
|
|
|
|
|
|
\- crate contents: single SVEC in VME slot 5 with the card under test in
|
|
|
FMC slot 2 and FMC TDC in FMC slot 1
|
|
|
|
|
|
\- fan speed: 1200 RPM (minimum)
|
|
|
|
|
|
\- ambient temperature: 20 deg Celsius
|
|
|
\- warm-up time: 8 minutes
|
|
|
|
|
|
*Results:**
|
|
|
| **Serial No** | **Configuration** | **Temperature (degrees Celsius)**
|
|
|
|
|
|
|
| CR37 | no heatsink | 59.7 |
|
|
|
| CR09 | small heatsinks on delay and PLL chips | 51.7 |
|
|
|
| CR01 | big heatsink in the rear of the card | 58.6 |
|
|
|
| CR03 | both heatsinks | 52.3 |
|
|
|
|
|
|
*Conclusions:**
|
|
|
|
|
|
\- big heatsink gives very little temperature drop (1 degC can be
|
|
|
considered measurement error). See the section with thermal camera
|
|
|
measurements for details.
|
|
|
\- small heatsinks glued on the chips give improvement of 8 degrees C.
|
|
|
- fit a heatsink
|
|
|
- increase air flow (fan speed) to improve heat transfer
|
|
|
- replace LVPECL delay lines with low-power LVDS equivalents
|
|
|
|
|
|
## Detailed temperature measurements on different platforms
|
|
|
|
|
|
*Note:** All temperatures in degrees Celsius.
|
|
|
*Note:** All temperatures in degrees Celsius. All thermal tests made
|
|
|
with PCB version EDA-02267-V5-0.
|
|
|
|
|
|
### Test environment
|
|
|
|
... | ... | @@ -52,7 +29,7 @@ before taking measurements. The first stable sample (with less than 1 |
|
|
degree temperature change over 1 minute) was used for the statistics.
|
|
|
\- Unused slots were covered to ensure maximum airflow through the FMCs.
|
|
|
|
|
|
### Data :
|
|
|
### Data (cards without heatsinks):
|
|
|
|
|
|
*Wiener 21-slot VME64x crate, slots 4-10:**
|
|
|
|
... | ... | @@ -291,36 +268,20 @@ Slots 10-15: |
|
|
*Elma 2-slot VME64x crate**
|
|
|
| **Fans** | **Slot/FMC** | 2/1 | 2/2 |
|
|
|
| Auto | **Temp** | 61.06 | 64.62 |
|
|
|
| Max (3600 rpm) | | 56.38 | 60.62 |
|
|
|
*Kontron KISS 6U industrial PC**
|
|
|
| Max (3600 rpm) | | 56.38 | 60.62 |
|
|
|
|
|
|
<table>
|
|
|
<tbody>
|
|
|
<tr class="odd">
|
|
|
<td><strong>Slot</strong></td>
|
|
|
<td>1</td>
|
|
|
<td>2</td>
|
|
|
<td>3</td>
|
|
|
<td>7</td>
|
|
|
<td>8</td>
|
|
|
</tr>
|
|
|
<tr class="even">
|
|
|
<td><strong>Temp</strong></td>
|
|
|
<td>88.19</td>
|
|
|
<td>81.94</td>
|
|
|
<td><span class="red">89.12</span></td>
|
|
|
<td>83.0</td>
|
|
|
<td><span class="blue">71.75</span></td>
|
|
|
</tr>
|
|
|
</tbody>
|
|
|
</table>
|
|
|
*Kontron KISS 6U industrial PC**
|
|
|
| **Slot** | 1 | 2 | 3 | 7 | 8 |
|
|
|
| **Temp** | 88.19 | 81.94 | 89.12 | 83.0 | 71.75 |
|
|
|
|
|
|
### Summary - without heatsink
|
|
|
### Summary
|
|
|
|
|
|
<table>
|
|
|
<tbody>
|
|
|
<tr class="odd">
|
|
|
<td><strong>Platform</strong></td>
|
|
|
<td><strong>Fans</strong></td>
|
|
|
<td><strong>Heatsink</strong></td>
|
|
|
<td><strong>Min temp</strong></td>
|
|
|
<td><strong>Max temp</strong></td>
|
|
|
<td><strong>Mean temp</strong></td>
|
... | ... | @@ -328,88 +289,12 @@ Slots 10-15: |
|
|
<td><strong>Ambient</strong></td>
|
|
|
<td><strong>Mean-ambient</strong></td>
|
|
|
</tr>
|
|
|
<tr class="even">
|
|
|
<td>Wiener (1200 rpm)</td>
|
|
|
<td>48.7</td>
|
|
|
<td>74.3</td>
|
|
|
<td>61.5</td>
|
|
|
<td>25.6</td>
|
|
|
<td>28</td>
|
|
|
<td>33.5</td>
|
|
|
</tr>
|
|
|
<tr class="odd">
|
|
|
<td>Wiener (3600 rpm)</td>
|
|
|
<td>39.3</td>
|
|
|
<td><span class="blue">58.8</span></td>
|
|
|
<td>48.5</td>
|
|
|
<td>19.5</td>
|
|
|
<td>28</td>
|
|
|
<td>20.5</td>
|
|
|
</tr>
|
|
|
<tr class="even">
|
|
|
<td>Elma 18-slot</td>
|
|
|
<td>44.5</td>
|
|
|
<td>73.7</td>
|
|
|
<td>58.7</td>
|
|
|
<td>29.2</td>
|
|
|
<td>28</td>
|
|
|
<td>30.7</td>
|
|
|
</tr>
|
|
|
<tr class="odd">
|
|
|
<td>Elma 6-slot</td>
|
|
|
<td>60.1</td>
|
|
|
<td>90.0</td>
|
|
|
<td>75.3</td>
|
|
|
<td>29.9</td>
|
|
|
<td>28</td>
|
|
|
<td>47.3</td>
|
|
|
</tr>
|
|
|
<tr class="even">
|
|
|
<td>Elma 6-slot (excl slot 6)</td>
|
|
|
<td>60.1</td>
|
|
|
<td>82.6</td>
|
|
|
<td>72.6</td>
|
|
|
<td>22.5</td>
|
|
|
<td>28</td>
|
|
|
<td>44.6</td>
|
|
|
</tr>
|
|
|
<tr class="odd">
|
|
|
<td>Kontron</td>
|
|
|
<td>79.5</td>
|
|
|
<td><span class="red">98.8</span></td>
|
|
|
<td>89.2</td>
|
|
|
<td>19.3</td>
|
|
|
<td><strong>20</strong></td>
|
|
|
<td>69.7</td>
|
|
|
</tr>
|
|
|
</tbody>
|
|
|
</table>
|
|
|
|
|
|
Note: the ambient temperature on the Kontron measurement is 20 degrees.
|
|
|
It is 28 degrees on the other measurements.
|
|
|
|
|
|
### Conclusions
|
|
|
|
|
|
- Cards in the middle slots (8-10) are the worst (25 degC difference
|
|
|
over slots 7 and 12). Uneven air flow in the crate?
|
|
|
- FMC1 is approximately 8-10 degrees cooler than FMC2. FMC1 is closer
|
|
|
to the fans and obstructs air flow to FMC2.
|
|
|
- Large temperature spans in VME crates suggest uneven air flow
|
|
|
between different slots.
|
|
|
- Keeping the fans speed at a maximum (Wiener) results with reasonable
|
|
|
cooling (worst temperature: 59).
|
|
|
- On Elma 6U and Kontron PC platforms, the temperatures are
|
|
|
unacceptable and exceed the components rating (-40 ... +85).
|
|
|
- Not using slot 6 on Elma 6U, only on Kontron PC platforms, the
|
|
|
temperatures are unacceptable and exceed the components rating (-40
|
|
|
... +85).
|
|
|
|
|
|
## Thermal imaging
|
|
|
|
|
|
tbc
|
|
|
|
|
|
## Thermal camera measurements
|
|
|
|
|
|
## Replacing the LVPECL delay lines with LVDS counterparts
|
|
|
|
|
|
We tried replacing the SY89295 chips (and their LVPECL terminations)
|
... | ... | @@ -426,7 +311,37 @@ use the new chip, for production purposes a new revision of the PCB |
|
|
should be made. Also, the new LVDS delay line is much harder to obtain
|
|
|
than the PECL version.
|
|
|
|
|
|
-----
|
|
|
## Conclusions
|
|
|
|
|
|
Short version:
|
|
|
|
|
|
- **Forced airflow is absolutely necessary**. Set the fans to the
|
|
|
maximum level in all VME installations.
|
|
|
- **Kontron PCs must be equipped with additional fan units**.
|
|
|
- **Heatsink is not effective if the air flow is too slow**. In such
|
|
|
case, the heat is exchanged through radiation with the cards in
|
|
|
immediate neighbourhood, causing collective heating of all
|
|
|
mezzanines/carriers (the ones in the middle are the worst).
|
|
|
- **Using an LVDS delay line chip gives little thermal advantage.**
|
|
|
- **Slot 6 in 3U Elma crates should be avoided.**
|
|
|
|
|
|
More notes:
|
|
|
|
|
|
- Cards in the middle slots (8-10) are the worst (25 degC difference
|
|
|
over slots 7 and 12). This suggests uneven air flow in the crate.
|
|
|
- FMC1 is approximately 8-10 degrees cooler than FMC2 in 3U and 1U
|
|
|
crates (opposite on Wiener/big Elma). FMC1 is closer to the fans and
|
|
|
obstructs air flow to FMC2.
|
|
|
- Large temperature spans in VME crates suggest uneven air flow
|
|
|
between different slots.
|
|
|
- Keeping the fans speed at a maximum (Wiener) results with reasonable
|
|
|
cooling (worst temperature: 59).
|
|
|
- On Elma 6U and Kontron PC platforms, the temperatures are
|
|
|
unacceptable and exceed the components rating (-40 ... +85).
|
|
|
- Not using slot 6 on Elma 6U, only on Kontron PC platforms, the
|
|
|
temperatures are unacceptable and exceed the components rating (-40
|
|
|
... +85).
|
|
|
\---
|
|
|
|
|
|
15 August 2014
|
|
|
|
... | ... | |