Its interesting to note that as you say the cooling air through the chassis will have less cooling effect dependant on ambient temp..
NB Very noticeable with OTL..
Temp rise in equipment can easily exceed cap working temperatures. NB no different to a car in summer.
So the point is to design to hottest ambient not average room temp. Also taking into account sun light falling on equipment due to location.
Just as a point of interest I had a session playing a guitar with my tube Marshall with its back to an open fire and after about two hours the supply caps blew on over heat. Just something I hadn't considered or thought about..(I didn't think it was very close the tolex was just hot to the touch..)
Regards
M. Gregg
Last edited:
No, the air will have exactly the same cooling effect.M Gregg said:
If a circuit dissipates 50W and air going in at 20C exits at 40C, then on a hotter day when air goes in at 25C it will come out somewhere around 45C. Same cooling effect: 50W means 20C rise in temp. What you need is a greater cooling effect when the air is hotter; for that you may need a temp-controlled fan.
I was going to say that it sounded like you were becoming contentious. However, in reading the thread, it is not so. DF96, I have been following your comments for years… what I like the most about them is that they're unerringly rooted in pragma, physics and balance. Thanks, and keep up the good work!
GoatGuy
PS: all those little degree-signs next to the temperature I did not insert manually. They result in passing the comment thru a ''formatting upgrade filter'' I wrote some years back.
Last edited:
But not in percentage terms. If the warmed-up interior is usually 50 degrees, then one day the ambient is 5 degrees warmer, the interioir will increase to 55 degrees. A 10% change.
But if you naively supposed that the interior was always similar to ambient then a change from 20 romm temp to 25 degrees would be a 25% increase.
The fact that the interior is heated by internal components helps to stabilise its performance against ambient changes.
Proper engineering takes that into effect, and specifies a warm up period before stabilization and measurement.
Once it is warmed up, it should track ambient. This is consistent with every UL/FM Witness test I've seen over the last 30 years (Quite a few).
Agreed.
A given temperature change relative to zero degrees Celcius, is a smaller percentage change if measured in degrees Fahrenheit relative to zero, and a much smaller percentage change if using Kelvin relative to zero, and therefore is not a useful measurement.
Last edited:
Actually not totally irrelevant, LEDs do have a tempco and large increases in internal ambient temperature could result in a small but significant and unexpected shift in operating point over temperature - in most cases probably not a big deal.
Pro Audio Design Forum • View topic - LED Temperature Coefficients When Used as Bias Sources
Pro Audio Design Forum • View topic - LED Temperature Coefficients When Used as Bias Sources
A person would have to be very naive to base their thermal calculations on a zero point arising from the freezing point of a common solute at a particular pressure. No physicist could do this; no engineer should do this. The only zero which matters is the real zero.Merlinb said:
No! That is the very point I was making. Only if the internal dissipation varies significantly with temperature (in the right way) will stability happen. Most electronics will vary, if at all, in the wrong direction - which is why we have to think about thermal runaway.
They show 1.5mV/K for red. That means a 20K shift would correspond to 30mV bias change. How significant is that?
If the temp change is much larger than that at the LED position, thermal management is badly needed to ensure reliability, especially if there are electrolytic capacitors about.Brad's anecdote in the linked thread is amusing and disturbingly familiar!
True enough. A simple resistor (positive tempco) is more stable, in percentage terms, when running in a heated interior, whereas as the forward voltage of a semiconductor junction (negative tempco) is the opposite. I should have thunk before I spake.
Look at the chart that started this fork in the discussion and you'll see the V. delta for any operating current over a 40F (20C) range is lower than the change you'll get in biasing from everyday primary voltage drifting. The curves are symetric. The change is not meaningful from a danger to circuit perspective and so most folks would set the biasing based on a warmup, first. This is really just OCD hand wringing. If an amp is biased so close to unwanted limits then it's time to rebias or accept life in the danger zone. The diodes are stable enough that it isn't an issue with everyday usage. The change would be so gradual it wouldn't even be noticed.
Last edited:
One minor issue with the report is that the test performed is a pulse test.
I'm certain that this was done to avoid self heating and avoid having to temp probe every LED to match die temp, however we are not using the LEDs in a pulse mode.
Self heating will shift the curves. This can become critical in poorly ventilated enclosures with LEDs biased near max.
Bottom line is, do your Worst Case Analysis on the circuit, including die temperature calculations.
I had a temp chamber run away over night once and when I cam in in the morning the chamber was up to 80C. The LEDs I was testing had shifted color (Die temp exceeded 100C).
The color shift was permanent, along with the change in forward voltage drop.
I'm certain that this was done to avoid self heating and avoid having to temp probe every LED to match die temp, however we are not using the LEDs in a pulse mode.
Self heating will shift the curves. This can become critical in poorly ventilated enclosures with LEDs biased near max.
Bottom line is, do your Worst Case Analysis on the circuit, including die temperature calculations.
I had a temp chamber run away over night once and when I cam in in the morning the chamber was up to 80C. The LEDs I was testing had shifted color (Die temp exceeded 100C).
The color shift was permanent, along with the change in forward voltage drop.
At the risks of being chided, ignored, lambasted, etc., I'll point out that whatever the perceived shortcomings of a fixed negative bias supply, LED bias has the same nearly-fixed voltage shortcomings in the same circuit. … along with the additional shortcoming of reducing some of the cathode-to-anode quiescent voltage. Perhaps not by much, but still…
With real-world valves, with real-world aging, real-world tube-to-tube variability, the fixed grid bias (whether by way of a C- supply, a string of LEDs or a shunt regulator) will follow the tube worse than the old fashioned, oft derided, but eminently adaptable cap-bypassed-resistor supply. When mu (proxy: gm) falls as tube age, the reduced current flow thru the cathode resistor partially compensates, dropping cathode voltage, which raises the tube current flow in reverse of its ageing direction. This can not be said about either fixed negative bias supplies OR LED bias, given how low their effective DC impedance is.
[Goat hides behind a rock, waiting for the aromatic pelting of old vegetables…]
GoatGuy
With real-world valves, with real-world aging, real-world tube-to-tube variability, the fixed grid bias (whether by way of a C- supply, a string of LEDs or a shunt regulator) will follow the tube worse than the old fashioned, oft derided, but eminently adaptable cap-bypassed-resistor supply. When mu (proxy: gm) falls as tube age, the reduced current flow thru the cathode resistor partially compensates, dropping cathode voltage, which raises the tube current flow in reverse of its ageing direction. This can not be said about either fixed negative bias supplies OR LED bias, given how low their effective DC impedance is.
[Goat hides behind a rock, waiting for the aromatic pelting of old vegetables…]
GoatGuy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.