Balanced F5 question

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
> http://www.semicon.toshiba.co.jp/eng...28g_chap03.pdf

Take a look at Figure 2.2 first.

Even with a heatsink temperature of 67°C, I calculated that

for 32W per MOSFET + Keratherm 86/82 TO247 insulator, junction temperature will be just below 100°C.
No reliability issue for me.
And there are still ways to reduce the peak sink temperature.

Average sink temperature will stay 57°C.

:cheers:


Patrick
 
Last edited:
thanh1973, You stated the Pd was approx. 135W. with what transistor? What Tj-case? What interface? I'm no expert but it came to mind that you could have a really sheety interface and Tj-c and there is no where near 135W being disapated in those sinks? It could all be in a 175C junction instead?
I'm not trying to criticise your work or efforts but I think we need that data too, in order to get a real working condition estimate? :Pawprint:
However, the last big GB on these sinks kinda came up with the conclusion that 100W was a cool number :cool:
 
> I'm no expert but it came to mind that you could have a really sheety interface and Tj-c and there is no where near 135W being dissipated in those sinks?

First I have a Ph.D. in mechanical engineering.

At steady state, ALL heat is dissipated through the heat sink, however "sheety" the interface is.
If you would believe me, and I know what I am talking about in mechanical things, then we have no transistor reliability problems even with 67°C sink temperature. If you want to believe what people give out here as holy rules and not look at the science, then as I said, use a Hifi2000 case or something of similar size.

When I am eventually finished with the case design, the sink temperature will drop further.

There is nothing to criticise in thanh's experiment IMHO. The results agree with manufacturer's data 100%.
Please kindly take your time to read post #301, 302 and the reliability handbook from Toshiba.


Best regards,
Patrick
 
Last edited:
It took me a while to reconcile today's calculations of 100C junction temperatures with what remembered calculating a while ago. I was looking at an AX with 4 outputs and derated the sink for lower delta-T perhaps too aggressively. I concur with Patrick that the junction temperatures should be around 100C and cause no problems.

Thinking further, since the extremities of the sink are relatively cool it stands to reason that the tips of the fins would be touchable, and this would be a warm but usable solution.

Funny that sometimes the gut reaction doesn't stand up to engineering scrutiny. Nothing says you can't overbuild if you wish...
 
My famous AXJ100 Tower has been running at > 60°C sink temperature for the last 4 years without the slightest problem.
There is still a thermal path from sink to junction which one can do something about.
There are good reasons why I do not use mica & goop but Keratherm for Class A.

And I don't sit on my amp when I listen to music, so what is the problem with a 67°C sink, even if it were.

I shall not disclose any further technical details until the design is finished and the prototype is tested, if it continues like this.


Patrick
 
Last edited:
Patrick, I hope I haven't offended you. My sincerest apologies if I have. I should have said my gut feelings sometimes don't stand engineering scrutiny - it was supposed to be a joke at my own expense.

I don't sit on my amp, either. The cat wants to now that the weather is turning cooler. ;) Sorry if I am a bit of a burn safety nut. A bicycle accident has left me very slow to sense heat and susceptible to burns before I feel warm.
 
Last edited:
"Why don't you knock it off with them negative waves? Why don't you dig how beautiful it is out here? Why don't you say something righteous and hopeful for a change?" - Oddball, Kellys Heroes

Edit: If you have seen the movie, then you will know this is meant to be a light hearted joke. Relax everyone. Have a little faith in Euvl.
 
Last edited:
First I have a Ph.D. in mechanical engineering.

At steady state, ALL heat is dissipated through the heat sink, however "sheety" the interface is.

None is disapatted to the surrounding atmosphere through the uninterfaced surface of the package? Or the leads?Hmmm?

If you would believe me, and I know what I am talking about in mechanical things, then we have no transistor reliability problems even with 67°C sink temperature.

Well, You started this post with a attempt at quoting me, therby addressing my question to thanh but, I said nothing about a transistor reliability problem. I did have some problem with my post not actually being posted when I hit the button? Therfore I did not see your posts #s 301 and 302.

If you want to believe what people give out here as holy rules and not look at the science,

Wait a minute. Is not the reason thanh is doing this because the abitlty of controling all the variables is difficult and going by manufacturer's data is not always proven by actual application. I'm tempted to go back a few pages to find you Patrick saying something to that effect. Now however you come out with calculations that you seem to want to make a "holy rule" after it has been proven by bench verification?

...then as I said, use a Hifi2000 case or something of similar size.

This type of statement is getting old. Pardon my audacity but it is starting to sound like the Blowtourch thread.

When I am eventually finished with the case design, the sink temperature will drop further.

I have respected your desire to not be bothered by us pee-ons. Sorry, your post did quote me.

There is nothing to criticise in thanh's experiment IMHO.

First
First I have a Ph.D. in mechanical engineering.

IMHO? I could start over but...

>I'm not trying to criticise your work or efforts but I think we need that data too, in order to get a real working condition estimate? (Me)
I think it is strange you elude to the interface conditions but I cannot ask thanh the details of his work?

Please kindly take your time to read post #301, 302 and the reliability handbook from Toshiba.

Thank you, for a while I was having alot of trouble finding anything good in your post. As I said I've got some problem now and then where my posts are showing up hours after I type them??? I have no problem with your calculations, estimations or proven examples. My problem is they way you rudely addressed me answering my post directed at someone else. But you did not answer my question? Certainly not "kindly".
I guess I'll shut-up now. I'm probably getting myself in trouble saying what I said. I need to go out in the yard now :Pawprint:
 
Hi flg
For the thermal pads I use Bergquist SP2000-0.015-00-122.
The mosfets used were IRFP240/9240.
Intially I determined there was not an issue with Tj-case so I didn't bother to record it.
I still need to do more tests here to try and spread the heat more evenly across the heatsink and bring down the temp around the mosfets, so I can record the temperatures of the mosfets in these followup tests.

Heat transfered from the mosfet casing directly to air is limited by the surface area of the mosfet, ambient temp and convection. Since the surface area of the mosfet is small relative to the heatsink surface area (as a guess 1:1000) the heat dissipated to the air directly from the mosfet is very small (you could probably say insignificant) compared to the heat dissipated via the heat sink to the air
 
No, Bob, I was not offended at all.
I am actually glad that you posted, and had looked at my arguments and concurred that 100°C junction is fine.
Not everyone can free himself from preconceptions.

Nor am I offended by anyone else in particular.
I am however offended by posts (in general and no single one in particular) that create fear without solid facts and scientific foundations.
The problem with flg was unfortunate timing with his and my posts. I apologise openly for my over-reaction.

It was my own fault in the first place, because I encouraged thanh to post his results without warning him upfront to present a balanced overview (e.g. the average temp is 57°C, the result agrees with Conrad data) to calm fear.

I learn my lessons and we shall in the future only present results when they are mature, and not on the fly.

But I also stand by my statement -- if you wish to have margins, you need a larger case.
I shall stay on my own course, and we shall build a prototype case soon (at our own cost) to verify performance and to be able to publish photos.
We shall not publish drawings or orthogonal views.

So I shall not have much to add in the next 4 weeks or so while both PCB and case design / prototyping is in progress.

The forum is all yours. ;)


Patrick



PS

thanh,

There is no issue with the MOSFET-sink interface in your experiment because we were not interested in junction temperature, but only heat sink performance. The only thing that was "not perfect" was the unfavorable heat source positions. But we have already discussed how that can be improved. So take your time with further experiments.

.
 
Last edited:
Thank you thanh. I was just getting a little sick of the behaviour and comments from some people here :confused: Respect for science and respect for all the fellow forum members along with respect for the hobby is not the same. I'm struggling to maintain myself right now...
I would like to apologise to those Tee'ed off by me going off on Patrick... I thought maybe he needed it?
He is no doubt a sharp cookie. He demands a little more than most of us can deal with and I couldn't, earlier today. Some of that is whatever is going on when I post, while my heat soaking VCore multiphase pwr supply demo boards are waiting for an IR image of the power stage to be takin. I'm no doctor but I obviously know, if you aren't doing what you just did with the Conrad Sinks, you must be in marketing or a fool who thinks he can sell something with authoritative sounding words. I am supprised Patrick has commited to doing what he is doing. He is very capable, I'm sure. And, a benifit to all of us but, like N.P. short on descriptive words? Unlike N.P., and much like me, short on patience. Not necessarily a bad thing but difficult.
I don't like speaking my mind and being jumped all over. I have trouble believing he's actually becoming so involved but, I really don't want to **** him off for everyone's sake. I guess i didn't do to bad beacuse I'm not experiencing the sin-bin???
I did not mean to exagerate so much regarding possible errors in your data in my post to you and that may have also have been an issue. But, I believe, you were right. Especially when you said:
"Why don't you knock it off with them negative waves? Why don't you dig how beautiful it is out here? Why don't you say something righteous and hopeful for a change?"
BTW, GiGi the dog<<< went to bed... This was me...
 
Patrick, now that I have read your post :eek: Fine, to bad we can't :cheers: been ther tonight to much already. Obviously you understand my position. And, thank you. I was very hesitant to blast you too. I knew I would regret it and I do. Sorry bud. I'm sure you just read my post to Thanh and everyone and I openly apologise to you also. I really hope you keep up the good work here and do not become frustrated with "us pee-ons". No need to work in the background with this project but, hey. We should take it a little easier eey. No management wanting results by the end of the bissness day here...
Thanks guy's. This wasn't originally a EUVL thread but I don't think there is any objections to continuing with the timely progress and discussion of this project ...
 
why do you use 3 pairs for the test
aren't the pairs 2 for channel

although It seems that pcb design Is for 2 rows of 2 Fets
can we place all 4 Fets at one row to get better cooling

Because that is what I had ready to go, and I wanted to spit out some results as fast as possible.
Next I will rotate the orientation 90 degrees. This may take some time because I need to drill and tap new holes to do this.
The F5X will be better than what I have because the mosfets are spaced further apart which will take advantage of the temperature gradient across the heatsink.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.