Possible heatsink for F5 ???
I've been looking at heatsinks and I was wondering about using this one... well two of them in each monoblock. Can anyone tell me based on the information provided if these look plausible to use? I don't know anything about thermal dynamics... but many here seem to. They are used for high output led lighting, which I know gets very hot too. They are almost 3 inches thick and about 6 inches around.
eddie
I've been looking at heatsinks and I was wondering about using this one... well two of them in each monoblock. Can anyone tell me based on the information provided if these look plausible to use? I don't know anything about thermal dynamics... but many here seem to. They are used for high output led lighting, which I know gets very hot too. They are almost 3 inches thick and about 6 inches around.
eddie
Attachments
If you have many of them, and fore free, mayb, if you find a proper way to mount devices with good heat transfer
But dont think I would actually buy them
Inner surface to transfer heat to periferical finning is smaller, but if its good or not I dont know
And height of 3" is only half of whats usually considered minimum fore this
In a world of no miracles I dont think its effective enough
But it might be nice fore classAB
Or maybe with fan cooling
Btw, the most effective heatsinks I have seen are with "welded" finning
But dont think I would actually buy them
Inner surface to transfer heat to periferical finning is smaller, but if its good or not I dont know
And height of 3" is only half of whats usually considered minimum fore this
In a world of no miracles I dont think its effective enough
But it might be nice fore classAB
Or maybe with fan cooling
Btw, the most effective heatsinks I have seen are with "welded" finning
You have to put the "P" infront of the numericals.
The 2SJ74 is "obsolete" but one manufacturer is working on a high gm P-channel JFET.
The Fairchild FQA12P20 and FQA19N20C are also, seemingly, "obsolete".
Mouser stocks the 2sj201 and Digi-key has the 2sk1530. Both are in stock as of this morning.
Yes the fairchild numbers are showing as obsolete , I have the Jfet input stuff already , just need power outputs.
Thanks ,
had forgotten about them .... Fairchilds or Vishay any preference here?
Thanks
Hello Hags,
Are those substitute prt numbers? I will take a look ..
I will have a number of Toshiba 2SK1530 and 2SJ201 in hand in about two weeks.
I bought a batch because I want to match some for my own use, and some friends.
After that I am offering them up here - will post to the Swap or GB areas.
Those who are using the usual single device F5 do not really need matching at all - you don't get matched when you buy Fairchilds or IRFs either. Those who will use 2 per rail do need/want matching.
(matching is nice even for single devices per rail, since in theory the + & - transfer curves will match more closely increasing the inherent linearity of the output stage...)
I am unsure how many matches will appear out of my relatively small batch (a batch of 100pcs min is often required to get a number of good matches), and I am unsure how P & N will compare either. So, I'll be making up whatever matches I can out of the lot first, and then the remainder of perfectly good devices (still tested) will be set out as best matches... I use a Tek or Fairchild curve tracer - not static Vgs testing.
The matched sets, IF there are enough to let go will cost more. I'll sell the others (best matches possible) priced more or less in batches ranging from "very good" down to "whatever" with the lowest group going off at more or less cost. I only want to recover the money I put out up front.
As far as testing/curve tracing, it remains to be seen how they will match in terms of curves from device to device and from P to N ch. The static match tells you something about the bias point, but not about the actual linearity - important if you are using pairs per rail, and much much less important if you are using single devices per rail.
Either way, I'll be able to say if the gains/curves look like they're rather complementary or not (they are supposed to be).
I'm not in competition with the luvdunhil effort, since he static matches, is out of stock and may buy a substantially greater number if and when he gets back into it - I think that there are a great number of folks who will build or are building the F5 and want the Toshiba devices, so this should work out fine. If not, these are super devices as far as I can tell, and I'll use 'em for other projects...
That's my story on this.
Contact me as you wish, nothing formal until I have the devices in my paws. They're ordered, you know how that can go...
_-_-bear
I bought a batch because I want to match some for my own use, and some friends.
After that I am offering them up here - will post to the Swap or GB areas.
Those who are using the usual single device F5 do not really need matching at all - you don't get matched when you buy Fairchilds or IRFs either. Those who will use 2 per rail do need/want matching.
(matching is nice even for single devices per rail, since in theory the + & - transfer curves will match more closely increasing the inherent linearity of the output stage...)
I am unsure how many matches will appear out of my relatively small batch (a batch of 100pcs min is often required to get a number of good matches), and I am unsure how P & N will compare either. So, I'll be making up whatever matches I can out of the lot first, and then the remainder of perfectly good devices (still tested) will be set out as best matches... I use a Tek or Fairchild curve tracer - not static Vgs testing.
The matched sets, IF there are enough to let go will cost more. I'll sell the others (best matches possible) priced more or less in batches ranging from "very good" down to "whatever" with the lowest group going off at more or less cost. I only want to recover the money I put out up front.
As far as testing/curve tracing, it remains to be seen how they will match in terms of curves from device to device and from P to N ch. The static match tells you something about the bias point, but not about the actual linearity - important if you are using pairs per rail, and much much less important if you are using single devices per rail.
Either way, I'll be able to say if the gains/curves look like they're rather complementary or not (they are supposed to be).
I'm not in competition with the luvdunhil effort, since he static matches, is out of stock and may buy a substantially greater number if and when he gets back into it - I think that there are a great number of folks who will build or are building the F5 and want the Toshiba devices, so this should work out fine. If not, these are super devices as far as I can tell, and I'll use 'em for other projects...
That's my story on this.
Contact me as you wish, nothing formal until I have the devices in my paws. They're ordered, you know how that can go...
_-_-bear
Last edited:
Yes, those are the same parts, but if you are using a pair per rail, the Vgs needs to be matched by some means. Imo, adjusting Vgs per device does not get me where I want to be.
The fellow with the round PC processor heatsinks - those are designed to be blown with a fan (noise) at a very high CFM, assuming that is what they are... otherwise they are not likely to be able to get rid of enough heat.
EDIT - oh, wait they are for LED lighting - find out the relatively efficiency of the LED, watts of light OUT vs HEAT, that will tell you if these will do the job. The graph shows watts of LED... although there is no spec on the heatsink, IF these graphs apply to this heatsink this idea will work - but I think the graphs merely show the spec for the heatsink that is required for a given LED wattage.
_-_-bear
PS. no "list" until I post to Swap or GB, and actually have the parts.
The fellow with the round PC processor heatsinks - those are designed to be blown with a fan (noise) at a very high CFM, assuming that is what they are... otherwise they are not likely to be able to get rid of enough heat.
EDIT - oh, wait they are for LED lighting - find out the relatively efficiency of the LED, watts of light OUT vs HEAT, that will tell you if these will do the job. The graph shows watts of LED... although there is no spec on the heatsink, IF these graphs apply to this heatsink this idea will work - but I think the graphs merely show the spec for the heatsink that is required for a given LED wattage.
_-_-bear
PS. no "list" until I post to Swap or GB, and actually have the parts.
Last edited: