1. Please take 30 seconds to register your free account to remove most ads, post topics, make friends, earn reward points at our store, and more!  

Diy d-class amp, cooling question

Discussion in 'Amps and Cabs [BG]' started by Matthijs, Aug 13, 2019.

  1. MobileHolmes

    MobileHolmes I used to be BassoP

    Nov 4, 2006
    My band uses a couple EV powered speakers and they have a metal heat sink on the back, but it it looks like the newer version has a multi speed fan
  2. Matthijs


    Jul 3, 2006
    Well I did do so research on how these are used. On most of the more high end designs I’ve seen, like in the Hevos amplifiers, there’s no fan. The amp module is cooled with a heatsink or a substantial piece of aluminium as part of the housing. The Smps has free airflow in those designs, but not nescesarily in great abundance or directed at the smps itself. The main purpose seems te be to prevent heat from the smps reaching the amp module. That was why I was originally contemplating to seperate the two modules in two compartiments: one for the amp, heatsinked on the outside. And the other possibly closed for the smps, but you can consider the closed part scrapped.

    My inital plan was to build a very simple amp in terms of functionality: input, output, powerswitch and signal light. Maybe also just the one input sensitivity switch if needed. And I had a form in mind like some of those old Quad power amps: a brick with just cooling fins and a signal light on the front. But I’m prepared to let form follow function.
  3. john m

    john m Supporting Member

    Jan 15, 2006
    Probably a case of increased wattage creating more heat.
    MobileHolmes likes this.
  4. agedhorse

    agedhorse Supporting Member Commercial User

    Feb 12, 2006
    Davis, CA (USA)
    Development Engineer-Mesa, Product Support-Genz Benz
    The separation may also have to do with EMC management...
    Matthijs likes this.
  5. agedhorse

    agedhorse Supporting Member Commercial User

    Feb 12, 2006
    Davis, CA (USA)
    Development Engineer-Mesa, Product Support-Genz Benz
    I believe some of the EV amps were linear, which generate more heat.
  6. beans-on-toast

    beans-on-toast Supporting Member

    Aug 7, 2008
    Since your components were designed to operate in free air without a fan, at the very least, run the system unpackaged and see how much heat is given off in free air. If you have an infrared thermometer you can use it to estimate the temperature of the heatsinks. This is your baseline, go from there.

    There are Class D self powered speakers where the amp is inside the cabinet. There is no venting, it depends on the design.
  7. agedhorse

    agedhorse Supporting Member Commercial User

    Feb 12, 2006
    Davis, CA (USA)
    Development Engineer-Mesa, Product Support-Genz Benz
    There are temperature sensitive components that do not use (conventional) heatsinks as well, some use the thermal mass of other components, some use PCB copper and some use nothing at all. Some of these parts have thermal monitoring that uses the temperature of the component to describe say average current, so artificially cooling these parts can greatly skew the information that the supervisory circuits receive.
  8. beans-on-toast

    beans-on-toast Supporting Member

    Aug 7, 2008
    In this case, the manufacturer should be able to provide the necessary guidance. Cooling or overheating parts can throw off thermal time constants.

    As an example, even a single component such as certain types of transistors can be temperature sensitive and perform differently. When taking readings amplification (hfe) can change by simply touching the case.
  9. agedhorse

    agedhorse Supporting Member Commercial User

    Feb 12, 2006
    Davis, CA (USA)
    Development Engineer-Mesa, Product Support-Genz Benz
    Most manufacturers are not able or willing to provide that level of support, it's normally reserved for OEM customers buying thousands of parts.

    In designs these days, many measurements of operating parameters are made indirectly. For example, monitoring current through a power FET is often done by measuring the gate to source voltage and extrapolating the current data from the voltage. If a new (different) temperature variable is introduced, the coefficients may be different and the association may no longer be accurate.
  10. BogeyBass


    Sep 14, 2010
    Separation or 2 compartments for amp/power supply is good idea. For noise interferance.

    But heat kills everything. Biggest issue is diodes and actual switching devices in the power supply. Those need to be well cooled.

    A commercial smps unit or off the shelf unit. Will use the case as the heatsink. The case itself is also a RF shield and is also usually vented with a small fan.

    It's not difficult to design heatsinks. Literally foolproof if you make them as large as possible. There is no such thing as a heatsink that is too big.

    Failed designs include heatsink that is too small.

    Plenty of plug n play smps that could work. But would be absolutely essential. To have inlet and outlet vents that allow the included fan to vent very very very well.

    Did I mention that power supply cooling is very very important... lol.

    Like I mentioned. Off the shelf smps mounts switching mosfet to the case. Not very good heatsink. And absolutely relys on the fan to be successful.

    After testing and doing data entry on over 4 thousand failed smps supply's. Number one cause of death was fan failure. Number 2 was capacitor failure. All related to the initial thermal issues.

    To solve the problem. Most were incouraged to develop " fanless" designs.
    Meaning if the fan failed the device could maintain thermal stability without a fan.

    Easy design. Like I mentioned earlier. The switching devices are mounted on a huge oversized heatsink. Likewise the diodes are not crammed right up against any capacitors. And the switching heatsink and diodes are mounted so there is a tunnel" so air flows over the mosfet and diodes.

    So yes do you get the " fanless" pun lol.
    The absolutely still use a fan. But if failed it will survive for longer duty cycle.

    Hence for diy. I'd assume removing the rf sheild/ heatsink. Mount the switching devices to a much much much larger heatsink. Then incorporate your own vented RF sheild. And add a larger fan or either way have very good inlet/outlet for very very good airflow for the fan

    Obviously the simple problem is off the shelf units draw air from above or below blowing downward. Your amp vents need to be on top and bottom of the case. Not good way for airflow sitting on a cabinet. Airflow is better from side to side. Or front to back. Like many various musical instrument amplifier wind tunnels have used.

    Plenty of fanless smps designs that work. For most part biggest problem or the only solution is very large heatsink, and still requires very good venting and vented RF sheilding. Not using garbage 85c caps crammed right next to diodes or heatsinks
    Last edited: Aug 15, 2019
  11. agedhorse

    agedhorse Supporting Member Commercial User

    Feb 12, 2006
    Davis, CA (USA)
    Development Engineer-Mesa, Product Support-Genz Benz
    Actually, high reliability SMPS use deeply integrated supervisory circuits to maintain reliability, not all the garbage that you suggested. They also design for minimum radiated RF generation through a variety of techniques rather than try to "fix the problem" by shielding.

    If you logged 4000 entries for failed SMPS, I would suggest that perhaps the design was not as good as it should have been, or the number of shipped parts was astronomical.
    StatesideRambler likes this.

Share This Page

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.