[star882]

Has anyone tried to use active cooling devices (such as Peltiers and compressors) with audio hardware? As temperature is reduced, Johnson-Nyquist noise decreases significantly.

For a small device like a DAC or headphone amp, the cheapest option is a Peltier. You can buy one for just $20 at Fry's. Then add a heatsink and power supply. For relatively little money, you can really cool down the circuits.

For a larger device like a speaker amp, you can use a compressor cooler. It's a far bigger investment than Peltiers but not much as far as high end audio is concerned. Compressors are far more suited for relatively constant loads, so while it's tricky to use one with a class D amp (excess liquid refrigerant may enter the compressor during low load and cause thermal stress), it'll be perfect for a class A amp (relatively constant load).

With some spending hundreds or even thousands on just cables, has anyone considered something cheaper that would (in theory) make a greater change?

 

[Garbz]

The Johnson–Nyquist noise formula is (4kTRB)^1/2, where k is boltzmann's constant, T is temp in kelvin, R is resistance, and B is bandwidth. Unless you can super cool the resistor it won't make a difference, and the fan will most likely be louder. Furthermore a 10k resistor at room temp is 1.3uV over 10khz so is it really an issue?

Also I have a DAC where the source resistor on the output FET is a 1W resistor that has 0.997Watt across it. The idle operating temperature of the resistor is 104 degrees C. Yes the DAC is dead silent, the noise is below that of my amp when turned all the way up.

If you have noise problems I doubt this is their source.

 

[blaken]

Star with enough giberish and overcharging the "audiophiles" will buy anything. So write up a semi-accurate paper, build your super cooler and make a fortune.

.

I had thought of liquid cooling my projector but the wife had a fit.

 

[star882]

Quote:

Originally Posted by Garbz The Johnson–Nyquist noise formula is (4kTRB)^1/2, where k is boltzmann's constant, T is temp in kelvin, R is resistance, and B is bandwidth. Unless you can super cool the resistor it won't make a difference, and the fan will most likely be louder. Furthermore a 10k resistor at room temp is 1.3uV over 10khz so is it really an issue? Also I have a DAC where the source resistor on the output FET is a 1W resistor that has 0.997Watt across it. The idle operating temperature of the resistor is 104 degrees C. Yes the DAC is dead silent, the noise is below that of my amp when turned all the way up. If you have noise problems I doubt this is their source.

What about the semiconductors?

 

[Garbz]

The formula is still pretty much the same on a physics level. Aren't noise stats normally displayed in semiconductor datasheets anyway?

Lets just say the IRF610B (which admittedly is one of the quieter MOSFETs) is the source follower I mentioned above. That resistor puts about 25mA into it, and it has no heatsink. It's operating temperature is roughly 85 degrees C give or take, so on the borderline of my comfort zone for an active device but not out of spec. The output is still quiet as the night. Also the whole I/V stage converts a +/-2.45mA signal into 1.2Vrms so the gain is pretty high too.

Do you have a specific noise problem with equipemnt? Check grounding and inductive pickup first.

 

[star882]

Even MOSFETs benefit from active cooling. Did you know that you can get over a 50% performance boost on a Pentium D 805 with compressor phase change? And Pentium Ds are CMOS!

 

[randytsuch]

Then give it a try, and see if it makes a difference. You should be able to use a walwart to power it, so it would be simple to implement.

I am thinking about building a class A amp to power my K1000's, and was thinking about adding a fan if things get too hot. I know most 'pliles don't like fans, but you can get really quite fans, and undervolt them so they are basically silent. The silent PC guys taught me this trick.

Randy

 

[psilosome]

Quote:

Originally Posted by star882 Even MOSFETs benefit from active cooling. Did you know that you can get over a 50% performance boost on a Pentium D 805 with compressor phase change? And Pentium Ds are CMOS!

The Pentium perfomance increase is through cooing to allow a higher clock rate. Performance of a CPU is measured very differently from performance of an audio circuit.

 

[Garbz]

If there's a change to be had it will be in the datasheet graphs. If you get one that shows improved linearity or significantly reduced noise vs temperature then by all means.

 

[star882]

Note that lots of very sensitive analog circuitry (like low noise amps and CCDs in professional video cameras) are active-cooled (usually with Peltiers) to reduce noise.

 

[motherone]

If you're convinced it will make a difference, why don't you try it out, document the results, and post them here? I personally doubt that it's going to make much of a difference, as CCDs and overclocked Pentium 805D's are vastly different beasts from Digital to Analog Converters.

 

[Garbz]

Oh agreed some things need it. Charged Coupled Devices, Infra red spectrometres, ultra high precision crystal ocillators etc. These are however more for precision circuits where the goal is often to eliminate drift and create equipment with a reference capable of measuring devices.

I am sure that industrial network analysers would use parts that exhibit drift or noise measured in the sub 10ppm/C, but do we need equipment with a +130dB signal to noise ratio? The 96dB typical of 16bit cds and much better is easily achieved without temperature control or compensation.

But by all means go ahead. Bolt a Dynahi too it too that may actually help there

 

[tangent]

Quote:

Originally Posted by blaken So write up a semi-accurate paper, build your super cooler and make a fortune. .

Can I get it with a solid cherry veneer?

 

[Torm]

I run a phase change cooler on my FX57.
Trust me - you dont want to cool anything audio with such a beast unless you have a really long cord and is doing the listening in a different room!

 

[star882]

Quote:

Originally Posted by Torm I run a phase change cooler on my FX57. Trust me - you dont want to cool anything audio with such a beast unless you have a really long cord and is doing the listening in a different room!

So use a Peltier with a fanless heatsink or watercooling. It'll be completely silent. Not a practical option for a large amp, though.
Has anyone watercooled their amp before?