Embedded Files
I decided to re-purpose the large aluminum box I had previously used for my low-distortion oscillator. Since moving the oscillator to a smaller box, it freed up the larger box and I hate wasting things.
The box I had used previously. It measures 68 x 145 x 250mm. The front panels will be replaced with custom-designed aluminum PCB panels. I got the box from Amazon but can't really recommend it now since the price has doubled to over $45. There are probably equivalents on eBay or other China sources.
In order to fit a high wattage load into the box, a heatsink and fans will be required. This type of heatsink looks like a good choice. There are no specs on the dissipation/heat rise given, but I am guessing two 40mm fans sucking air across it will be suitable. I chose one which is 180 x 69 x 36 mm (picture not to scale), which is about the largest that can fit into the box and allow clearance for the fans and connectors.
These 40mm fans (two of them) placed in the back of the box should be sufficient. These are 5 volt USB powered. Power to the fans will be switched by a N.O. thermostat mounted to the heatsink so they won't be on all the time. These are claimed to have over 7 CFM airflow each, so 14 CFM total. From Amazon.
I have a bunch of these so I will use two for the input from the amp.
This Neutrik combo jack (XLR + RTS 1/4" phono) will serve as a balanced output to measurement devices - like my stepped attenuator. Since the downstream devices will be differential, common-mode noise should be reduced.
As usual, I designed the panels in KiCad and ordered them in aluminum from JLCPCB. The front has a bunch of ventilation holes to allow airflow into the front. I might glue some screening on the backside to cut down on dust incursion.
The back has mounting holes for the two 40mm fans that will exhaust hot air out the back and a cutout for a female USB panel-mount connector which will supply 5 Volts to the fans through a thermostatic switch.
These Caddock non-inductive film resistors look ideal. They have exposed ceramic pads on the back which can be mounted against a flat heatsink surface. I will mount these on the underside of the heatsink and use spacers to mount the heatsink to provide clearance above the base of the enclosure.
They are available up to 100 Watts dissipation and are 1% tolerance. I'll get two 4 Ohm units wired in series with a switch to select 4 or 8 Ohms loads. With 2x4 Ohm 100 Watt resistors amp testing up to 200 Watts will be possible (at 8 Ohms).
These thermostatic switches look like a good choice for turning on the fans. They have a similar package to the load resistors, so are low-profile and one can be attached to the bottom of the heatsink to switch the fans on at 55 degrees C (about 130 F) and this seems about right. I anticipate that this will rarely be necessary, except under extended test runs.
Front panel mocked up with connectors and switches. The 4/8 Ohm selector switch has center off position - this will allow amp output to be measured no load easily.
Mounting the 100 Watt Caddock resistors to the bottom of the heatsink. I used some thermal paste, fender and lock washers.
Wiring some 14 gauge flying leads to the Caddocks which will run to the front panel. Also mounted the thermostat at the rear of the heatsink.
Heatsink mounted on standoffs. Rear panel with fans partially attached to the chassis.
Wiring completed
Checking heatsink temperature with bench supply connected - enclosure open. The fans come on at 56 C, within 1 degree of the thermostat rating. Good!
Testing with 30.76 Volts into 8 Ohms - about 120 Watts for 1/2 hour - temp hits 68 C and fans hold it steady - enclosure closed up. Fans run until temp drops to 33 C.
Fini!
The back. I am thinking that cooling efficiency could be improved by adding baffles to the heatsink to direct the airflow better, but the design seems adequate for now.
Note: at least one DIYer has commented that thick film resistors such as I have used here may introduce their own distortion signatures at high power levels and there may be better (and more costly) alternatives.
Page updated
Google Sites
Report abuse