Do you measure component temperatures?

[Greg Erskine]

I have an InfraRed Thermometer and for the last year, whenever I have the top off one of my amps I take a look at the temperature of the various components. I am trying to get a feel for what is normal.

I realise the differences between classes and how hard the amp is being driven will make a difference.

On my AKSA 55 amp, the temperature of the components are what I'd expect. All active components are elevated a few degrees above ambient, maybe up to 10 or 12 degrees.

But...on some of my amps, some components (input pairs, some resistors) are over 50 degrees C and that's just sitting on my test bench.

Looking at the datasheets it indicates that these devices behave differently (sometimes better) at these higher temperatures. Do designers decide on an operating temperature of the various components? I suppose temperature is probably just a function of voltage and current (power) so there may be no need to focus on temperature.

regards

 

[Wavebourn]

I would love to have a camera based temperature measurement device, but it is too expensive yet for me to afford. Looking at the picture in dynamics is very revealing. For example, you can see oscillations of termocompensations that can't be caught easily; you can see how to improve layout, and so on.

I mean, something like this one:

http://www.mikroninfrared.com/Catalog.aspx?id=1074&ekmensel=c580fa7b_8_24_1092_1

 

[Greg Erskine]

Thanks Wavebourn,

That Mikron M7800 Infrared Imaging Camera looks really cool.

I had never really thought of thermal imaging in relation to audio before. It would be interesting to see a movie of a complete amp goig through its heating and cooling cycle.

The only problem I can see would be, a normal temperature for an output devices might be dangerous for a capacitor. So the 'red' on the image would not always be a problem, but I guess anyone with experience could be able to work out the problem areas will little effort.

regards

 

[wrenchone]

I'm lucky - we have an $8k thermal imaging camera used to record temperature profiles for switching power supply prototypes. It's a little hand-held thing that stores data on a standard memory card. What makes it especially valuable is that you can use the software included with the camera to look over the image data at your leisure to scope out device temperatures after the fact. The thing has paid for itself already in uncovering problems with components where one would normally not think of pasting a thermocouple.

I haven't used the camera for any of my amp prototypes yet. I usually just let my fingers do the walking - 50C is ~10 sec uncomfortable, 55C is 5 sec uncomfortable, and 60C is "get your finger off right away". Your fingers are sensitive nonlinear thermometers developed over long years of evolution....

 

[Wavebourn]

Is it possible to borrow it somehow? I would bring my amps to take series of pictures.

 

[Mr Evil]

I use my thermal sensing digit. I have nine spares.

Designers should be paying attention to how much power everything dissipates, and also to where hot components are in relation to everything else, e.g. no heat sources near input stages.

 

[pacificblue]

On my AKSA 55 amp, the temperature of the components are what I'd expect. All active components are elevated a few degrees above ambient, maybe up to 10 or 12 degrees.

But...on some of my amps, some components (input pairs, some resistors) are over 50 degrees C and that's just sitting on my test bench.

When the amp is closed, those temperatures will be different. They will be higher in general and more similar to each other from component to component.

One of my profs gave the rule of thumb that at 25 °C ambient the inside of an enclosure with electronic components would heat up to between 60 and 70 °C. And that we should therefore always calculate the components for an ambient of at least 70 °C, better 80 °C for a safety margin. Higher, if the apparatus was to be exported to hotter countries. That was at a time, where surface mount did not exist yet, and the possible packing density turned out to lead to that temperature range. Heat dissipation for resistors is usually given at 70 °C. Coincidence?

We will probably have to raise that figure soon to make up for global warning.

Looking at the datasheets it indicates that these devices behave differently (sometimes better) at these higher temperatures. Do designers decide on an operating temperature of the various components?

[insert text block]Some do, others don't.[/insert text block] That could be one of the differences between good and bad designs, and the reason, why many audio components sound better, when they have warmed up.

 

[AndrewT]

On my AKSA 55 amp, the temperature of the components are what I'd expect. All active components are elevated a few degrees above ambient, maybe up to 10 or 12 degrees.

I think that is AKSA's design expertise showing through in a practical example.

I would expect, if second guessing Hugh is allowed, that he has taken account of many and various factors that affect sound quality and looked for methods that maximise SQ.

I would not be surprised if he came out and said that temperature modulation of semiconductor junctions and maybe to a lesser extent resistors too, do affect SQ and that minimising the temperature rise of each of the devices, particularly in the front end of the voltage amp stage also reduces temperature modulation.

My own thoughts are to run everything as cool as practical and fit heatsinks if significant heat must be dissipated, even in non power devices.

 

[Steve Dunlap]

For about the last 10 years I used thermocouples epoxied to the device to be monitored. Sometimes there could be as many as 20 thermocouples in one piece of gear. You then monitor the temperature with the component running normally, again in a 35C oven and again in a 50C oven. One hour at each temp is usually enough.

 

[Greg Erskine]

Re: Re: Do you measure component temperatures?

When the amp is closed, those temperatures will be different. They will be higher in general and more similar to each other from component to component.

One of my profs gave the rule of thumb that at 25 °C ambient the inside of an enclosure with electronic components would heat up to between 60 and 70 °C. And that we should therefore always calculate the components for an ambient of at least 70 °C, better 80 °C for a safety margin....%<

Thanks pacificblue.

Yes I understand about when the amp is closed, but the degree of change will be depend on physical design of the amp. In my case, the cases are 3U with lots of space between PCBs, the main heatsink is oversized and external. A lot of commerical gear seems to be more tightly packed and use internal heatsinks, so the temperature difference between closed and open will be greater I imagine. Also, equipment designed with fans may actually run hotter with the lid off, I know this applies to some PCs.

Thanks for your profs rule-of-thumb values, it gives me a little bit more confidence. It looks like everything I have measured so far is well within your profs rule-of-thumb.

regards

 

[Greg Erskine]

Re: Re: Do you measure component temperatures?

My own thoughts are to run everything as cool as practical and fit heatsinks if significant heat must be dissipated, even in non power devices.

Hi Andrew,

That is a very practical point of view and will result in a very stable and reliable amp. I guess I am trying to quantify terms "cool as practical" and "significant heat".

We could put heatsinks on all active components to be safe, but that makes PCBs bigger and is there "really a gain".

regards

 

[Greg Erskine]

For about the last 10 years I used thermocouples epoxied to the device to be monitored. Sometimes there could be as many as 20 thermocouples in one piece of gear. You then monitor the temperature with the component running normally, again in a 35C oven and again in a 50C oven. One hour at each temp is usually enough.

Hi Steve,

I guessed the professionals would do this kind of testing. I imagine the main objective of these tests was to determine the stability of the amp of various ambient temperatures.

Did you log the temperature of the components over time or just the final temperature after the hour soak?

I've been looking at a PIXAXE and DS18B20 digital temperature sensor to make a cheap monitor. Adding datalogging increases the complexity and costs.

Were you ever able to listen to your amps at the different temperatures? I'd be interested if there was a difference in sound because there is a reasonable difference in the operating parameters of devices between room temperature and 50 degrees.

regards

 

[Wavebourn]

I miss the testing laboratory in Tomsk I had access to. We could test vibrations observing resonances, we could test the gear in Peltier camera with very stable and wide temperature diapason, we could test for accelerations, humidity, and many other things... But we did not have such wonderful video camera to observe things in dynamics!

 

[Steve Dunlap]

Hi Steve,

I guessed the professionals would do this kind of testing. I imagine the main objective of these tests was to determine the stability of the amp of various ambient temperatures.

Did you log the temperature of the components over time or just the final temperature after the hour soak?

I've been looking at a PIXAXE and DS18B20 digital temperature sensor to make a cheap monitor. Adding datalogging increases the complexity and costs.

Were you ever able to listen to your amps at the different temperatures? I'd be interested if there was a difference in sound because there is a reasonable difference in the operating parameters of devices between room temperature and 50 degrees.

regards

We used a chamber that would go from -40C to 150C where we datalogged to observe rate of change. for the one hour soaks only the final temperature was measured.

I did listen to my amps at the different temperatures. The difference between the temps was very minor if it even existed. What was clearly audible was the transition time between temperatures. It didn't matter if the temp was going up or down, it sounded much worse until the temperature stabilized.

 

[Conrad Hoffman]

I was able to borrow a $20k thermal imaging system for a brief period. It was the neatest thing in the world. You could see temperature changes propagating thru parts, and see the rate of heating and cooling as power levels changed. Unfortunately, as interesting as it was, the information just wasn't valuable enough to justify buying the thing. Sooner or later somebody has to figure out how to do it at a far lower cost. For home I have a thermocouple meter and a thermister meter. If something seems unusually warm with the finger test, I might measure it. More likely I'll just put a bigger heat sink on it.

 

[Wavebourn]

The thermal imaging camera is #1 priority in my measurement equipment list. Guess, what is the number last?

 

[wrenchone]

Thermal cameras are way too expensive for the average amateur - you could outfit an entire lab for the cost of one. They can also be fooled by low emissivity metal parts, which need a piece of tape or some other treatment to make them read correctly. I also haven't found the cheap non-contact thermometers to be very useful, though their prices have come way down since they were introduced.

A far more sensible choice for the average lab is a multichannel thermocouple-based meter. The old green 10-channel Fluke meters were excellent, though it's hard to find one these days with the thermocouple insert in place. Fluke replaced these relatively inexpensive and easy to use thermometers with the Hydra, a multichannel instrument that is much more expensive and suffers hugely from "feature paralysis". The Hydra can do a lot, if you can ever remember how to program the thing - they have an opaque user interface that rivals HP instruments. Doric Trendicators are another possible choice for the cash-strapped amateur.

 

[nigelwright7557]

Components temperatures are important and I always go around my designs touching each component to make sure it is not getting hot.

Semiconductors will work upto 150 degrees C so they can work at very hig htemperatures reliably. However they wont dissipate as much power at higher temperature levels so care must be taken.

I tend to use a rule of if its too hot to touch then it needs a heatsink or a bigger heatsink or a fan.

My amps tend to be overdesigned with large heatsinks and dual fans to ensure they dont cook even at very high power levels.

 

[Nelson Pass]

There's a nice little hand-held battery operated IR temperature meter
at Radio Shack for a few bucks. F or C, works great, bought several.

 

[tiefbassuebertr]

There's a nice little hand-held battery operated IR temperature meter
at Radio Shack for a few bucks. F or C, works great, bought several.

I am not able to find that type from Radio Shack you mean. Please let me know the model number - thank you.
By searching on the web I have found this:
http://www.sperscientific.com/images/broc_pdf/800100-4.pdf
http://www.sperscientific.com/800101-2-i.pdf

I have an InfraRed Thermometer and for the last year, whenever I have the top off one of my amps I take a look at the temperature of the various components. I am trying to get a feel for what is normal.
I realise the differences between classes and how hard the amp is being driven will make a difference.
On my AKSA 55 amp, the temperature of the components are what I'd expect. All active components are elevated a few degrees above ambient, maybe up to 10 or 12 degrees.
But...on some of my amps, some components (input pairs, some resistors) are over 50 degrees C and that's just sitting on my test bench.
Looking at the datasheets it indicates that these devices behave differently (sometimes better) at these higher temperatures. Do designers decide on an operating temperature of the various components? I suppose temperature is probably just a function of voltage and current (power) so there may be no need to focus on temperature.
regards

This is a very good idea.
Actually this task must been perform from all professional manufacturers for home audio. But even brands like ML or Krell Industries haven't investigate the temperature of their components while use.The proof of this fact this is very simple: Have a look to the main PCBs of the models No 23 (No23 ML-23, VAS aera) and Krell KBL (both outdoor power supply and main pcb of Krell KBL line preamp). You will find dark burned aeras (hot spots). It is shameful to me that such things happened in this class. By first generation of Linn (pre)amps similar things are to observe, but by the later generations of Linn products this problems are completly solved. The envelope (cabinet, chassis, enclosure) serves as heatsink - easy but effective.

An "OFF-OPIC" question to Greg Erskine: Where I can find all audio manufacturers from Australia and New Zealand?
about http://audiotools.com/
I have found only the follow:
http://www.pliniusaudio.nzld.com/contactus/index.asp
http://www.slinkylinks.net/ (accessories)
http://www.perreaux.com/home/
http://www.image-loudspeakers.co.nz/index.asp only Speaker
http://www.keaaudio.com/index.html only Speaker
But I cannot believe, that there are no more brands from New Zealand.
Perhaps there are URLs like "audiotools.com" but sorted by countries - thank you for your advice.