I have carried out some more temperature measurements at two different current levels on the amplifier board. The original current amplifier had insufficient heatsink originally, and I improved the cooling by thermally bridging the heatsink to the aluminium case. This has improved the situation in the original amplifier for up to a 1A load current. Beyond that, a bit too much heat gets transferred into the case.
The test board does not have a case, so heatsink modifications needed to be made. As I also wanted to get more current out to test with the coils, a heatsink was also added to the rectifier.
More reading of the OPA548 data sheet, revealed a calculation for the amount of heat dissipation, that has allowed me to build up a table for the potential heat output.
Maximum continuous output of the amplifier is 3A, equivalent to 27W heat dissipation. If the output is shorted, this doubles to 54W, however, in my design, I have a 1 Ohm resistor in series with the output, so it should never receive a direct short circuit.
Knowing a bit more about the heat dissipation allows me to make a better decision regarding the heatsink requirements. However, there will be mechanical constraints to the sizing of the heatsink that will limit my choices. The initial issue will be the width of the heatsink. On the PCB design, I placed a ceramic capacitor alongside the amplifier and this will block making the heatsink wider. To overcome this aspect, I cut a small block of aluminium to act as a spacer and push the heatsink further away from the amplifier to allow the width to be increased. Now one of the large smoothing capacitor prevents widening the heatsink. I also don't want the heatsink to get too close to the capacitor and cause unnecessary heat transfer over to it and shorten the capacitors life.
I know that adding the aluminium spacer will also add more resistance to transferring the heat from the amplifier onto the heatsink itself. A photograph from above shows the end result of the layout.
Pushing the heatsink back, takes it away from its original mounting holes. No real problem as it is just empty space on the board, so two new mounting holes can soon be drilled. However, these will go through the ground plane copper, therefore I need to add in a mica insulating washer between the amplifier and the aluminium block. The tab on the amplifier is connected to the negative supply terminal, so I don't want to risk a short circuit from the tab to the ground plane via the heatsink.
There are plenty of heatsink calculators available online to aid with choosing the characteristics.
This particular one is from Heatsink Calculator website and is their free version, so is a little limited. The calculation specifies a heatsink with a thermal resistance of 1.42 C/W, which is physically quite a large heatsink, that I just wouldn't be able to fit within the current design.
As the current amplifier test board, is primarily for investigating the use of the polymer capacitors agains the electrolytic and then to investigate parallel and serial amplifier operation and test out some more coils, I opted for a smaller heatsink with a higher thermal resistance that would physically fit much easier. I ended up selecting a 5 C/W heatsink from Fischer Elektronik, specifically for TO220 packages. Unfortunately, due to the use of the aluminium spacer, I couldn't utilise the bespoke mounting clip and instead drilled and tapped the heatsink for an M3 screw. The particular heatsink I used is 37.5mm wide, which was just about the maximum space available when keeping the heatsink on the PCB.
As a little compromise, I did install a fan on the heatsink to provide some more cooling, but I will basically overcome the undersized heatsink by only running the amplifier on high output currents for short durations.
I measured the temperature of the heatsink after 2 minutes supplying a 2.5A load current.
The thermography pictures shows how the heat is dissipated across the device, spacer and heatsink. The amplifier gets to a temperature of 105 Deg C, but the aluminium spacer only reaches a temperature of 73 Deg C. This shows quite a high resistance to conducting the heat between the amplifier and the spacer. From the spacer to the heatsink, there is only a drop of 6 Deg C, showing that the transfer of heat is much more successful. A further 7 Deg C is then dropped across the length of the heatsink, which will be due to the heatsink's own built in thermal resistivity.
The thermal image doesn't show any heat rise through the fan, so I am not sure how effective it actually is. The air above it, is quite cool and doesn't really reflect the temperature of the heatsink.
Due to the proposed increase in load current, there is also a need to install a heatsink onto the rectifier. The data sheet for this gives a dimension for an aluminium plate of 50mm x 50mm x 1.6mm copper plate for the full 6A rating. I will only be taking this to 50% and copper plate is a bit exotic for me, so I purchased another Fischer Elektronik heatsink, that was a bit larger to accommodate the larger format of the rectifier.
Again an aluminium spacer was used between the rectifier and the heatsink to allow the mounting holes in the PCB for the heatsink to be located within the ground plane. The rectifier is an insulated package, so there was no need for an insulating washer, but again, the heatsink was drilled and tapped with an M3 thread to allow the rectifier to be bolted to it.
Due to the location of the rectifier, it was a bit harder to capture the thermal image. In the side on view, I tried to capture the rectifier, spacer and heatsink, but it is difficult to tell them apart.
A picture captured at a different angle capture the rectifier and the heatsink, but the aluminium spacer is hidden under the heatsink fins.
The rectifier reached a temperature of 32.5 Deg C. A 7 Deg C, drop is then seen from the rectifier, through the aluminium spacer, to the heatsink. This seems to be more efficient than the amplifier, with the extra insulating washer. The top of the heatsink reaches 24.4 Deg C. This set-up seems to be more than adequate for the rectifier and the load currents I plan to draw.
SP4 at 31.7 Deg C, is one of the poly fuses. I am not sure if the temperature displayed is genuine, or if it is due to a reflection from the different surface it has in comparison to the heatsink and rectifier body. The poly fuse can be seen in the picture below , just in front of the aluminium spacer. Unfortunately, the FlirOne thermography came, does not appear to be as versatile as a Canon DSLR, and I struggled to replicate this picture as a thermal image.
As there was a minimum order quantity on the rectifier heatsink, I used a further two of them for the 1 Ohm load resistors. The resistors are rated at 25W with a heatsink, and 9W without one. Drawing a 3A load, would therefore put me on the maximum rating of the resistor, so a heatsink seems prudent.
The resistor reaches a maximum temperature of 39 Deg C. There is then a 5 Deg C drop onto the heatsink itself and the top of the heatsink reaches 32 Deg C. This is well below the maximum 110 Deg C body temperature for this resistor.
As the heatsink and resistor are mounted on a pice of steel right-angle, some heat will also be dissipated away by this.
Overall, the temperature control methods seem to be reasonable for the tests that I want to carry out on the current amplifier. More work would need to be done on the heatsink requirements for the amplifier on a second build, but this would also involve a new PCB and would therefore open up a lot more options for lower thermal resistivity heatsinks.
Top Comments