Oh no, I'm not intending them to run at 100°C under normal use. I would just like them to stay at 100°C or less with a 50°C ambient temperature. You know, if some idiot leaves the unit on, leaves the house for the day with no cooling and the ambient temperature goes up to 50°C. If their house gets any hotter than that, they can pay for the repairs.
se
Just use a dab of cyanoacrylate to glue the thermocouple tip to the plastic case. When you are done, use a heat gun to melt the CA and remove the thermocouple. Be careful not to inhale the fumes though. Nasty stuff.
That's actually what I used when I was doing thermal testing on this amp a couple years ago (I've since designed a different case and heatsink for it). I'd drilled a small hole in the side of the transistor case and glued the probe tip in with cyanoacrylate. I just didn't (and don't) know how accurate a measure I was getting of the case temperature. This time around, since it will go into production, I want more certainly.
se
Ok, the amp's been idling for a little over an hour. It's a single ended design so it's dissipating maximum power at idle. My thermocouple thermometer is giving me an ambient room temperature of about 20°C. I took the IR thermometer and pointed it at the inner transistor a little over an inch away. Moving it around a bit, the hottest temperature I'm measuring is about 65°C. If I touch the thermocouple probe to the transistor case, the highest temperature I'm able to get is just a bit under 60°C. For what it's worth.
se
se
I would use NiCrNi-junction, fix it with hi-temperature tape to the case, put some styro foam above for thermal insulation and then wait until temperature is stable.
This should give little thermal gradient, i.e. give a useful result close to chip temperature.
I suppose measuring Vbe of DUT including calibrating same DUT in thermal bath comes closest to junction temperature. This is the same idea measuring transformer temperature using tempco of copper winding.
This should give little thermal gradient, i.e. give a useful result close to chip temperature.
I suppose measuring Vbe of DUT including calibrating same DUT in thermal bath comes closest to junction temperature. This is the same idea measuring transformer temperature using tempco of copper winding.
Since it all is just a (thermal) resistance string, really, I would think that if one side of the chain (ambient) changes, the 'voltages' (temp) along the chain change as a function of the impedance ratios. So if ambient changes 30 deg, I would the expect the case and the junction to change less. Don't know if the difference is significant in this app though.
I am having a similar issue as yours, trying to develop a means to get the junction temp of a device for SOA limiting purposes.
Some form of analog computing circuit is what I am looking at. Measure the heatsink temp (easy), and feed that to a circuit that feeds the device dissipation into a thermal model of junction-to-case thermal resistance (case here being the metal part fixed to the HS). That would give me delta-T from junction to heatsink.
It's untrodden terrain for me but fascinating.
Jan
Jan
As the old saying goes, 'a fool with a tool is still a fool'. My view is that if a cable requires group delay measurements then it's a faulty cable at audio frequencies. Likewise TDR unless you are Ed.
It is all matter of money. You are not ready to pay high price for connectors, and you force you to think that they not get any sound improvement. At the same time you are probably ready to pay the same amount for a meal in a better restaurant.Without trying something all else is empty story.
DIY is not for people who turn twice every coin before spending it.I used to buy 20 dollars Vishay bulk foil resistors which gave zero difference in sq. But it is a part of my building experience.
Last edited:
Thermal resistance from junction to case of a BD139 is 10C/W
Derating factor is the reciproke value, 1/Rth(j-c) = 0.1W/C
Max dissipation is when the case temperature is at 25C, and the die at 150C.
150-25=125C
Rth(j-c) times 12.5W (Pd) = 125C
Zero dissipation is when both the case temperature and die are at 150C
150 -150 = 0
Rth(j-c) times 0W = 0C
Aka, the temperature in the derating curve is the one of the case
(semiconductor-thermo 101)
[a relic would use a $1 temperature sensor in a TO92 package, LMx35 series]
That can happen too. Money doesn't have to be involved, though. However, in the case of money, as you describe, people tend to put money into different mental accounts. As one type of example, saving $50 on a new car is seen very differently from saving $50 on a $75 audio cable. But, rationally speaking, saving $50 either way still saves you the same amount.
I would pay high price to get something in return for the money. A BNC or XLR gets you an engineered connector. £50 on a WBT is lipstick on a pig!
There is a theorem in computer science which says that every program has to either halt or go into a loop. We have been in the loop for some time now.marce said:
I have said that before. I will probably say it again.
Anyone using an FFT is using calculus. Some might not realise this is happening. If EEs knew more maths (and were more comfortable with it) then they might understand their physics better too (this is aimed primarily at UK EEs, where standards seem particularly low).RNMarsh said:
Yes. The field equation solution separates into a geometric factor and a time-dependent factor. The geometry depends on the frequency of the time-dependent part - assuming sinusoidal time variation. For a complex waveform use Fourier and superposition, provided that the metal is linear (which it normally will be).keantoken said:
Oh dear - the dreaded 'engineering model'! In other words, "I can't, or won't, understand the physics (what is actually happening) so I want to invent a simple picture which enables me to quickly draw false conclusions and then blame 'theory' when reality turns out to be different".RNMarsh said:
I thought the aim of marketing was to be misleading in a hugely substantial way, but without saying anything which a lawyer would recognise as a lie? The usual method is to fail to explain any technical concepts but to use lots of words which may look 'scientific' to someone who knows no science. In other words, the very opposite of what you say.
You can't explain skin effect to a consumer. All you can do is state that it occurs and describe its consequences (normally, none for audio).
It is all a matter of physics. You are not ready to put in the effort to understand circuit theory and how an audio interconnect is just a trivial potential divider, so you are forced to spend unnecessary money on connectors and then post-justify that by attempting to get others to join you in your folly and ignorance.kamis said:
A good meal may bring pleasure. An overly expensive connector brings unnecessary pain. The whole point of civilisation, education etc. is to enable wise people to learn from the mistakes of foolish people, so they don't have to try things which don't work or are harmful.
- Status
- Not open for further replies.