Hi
According to the DIYaudio store, the 4U/300 comes with heatsinks each rated at 0.31 C/W
If you push 100W on these in a well ventilated enclosure you will theoretically have a temp rise of 0.31 x 100 ; 31C.
If your ambient is 25C then your heatsink will be at 56C.
Eric
According to the DIYaudio store, the 4U/300 comes with heatsinks each rated at 0.31 C/W
If you push 100W on these in a well ventilated enclosure you will theoretically have a temp rise of 0.31 x 100 ; 31C.
If your ambient is 25C then your heatsink will be at 56C.
Eric
I'm remembering just one thing , and everything else I'm concluding from that:
4U/400 Modushop Dissipante - perfect for 80W of heat per side (sum 160W), and my OCD dictating use of Babysitter during summer months (I tend to imagine sizzling caps inside)
anyone is free to have different opinion, that's according to air quality in my neck of wood
4U/400 Modushop Dissipante - perfect for 80W of heat per side (sum 160W), and my OCD dictating use of Babysitter during summer months (I tend to imagine sizzling caps inside)
anyone is free to have different opinion, that's according to air quality in my neck of wood
Agreed absolutely. That’s just a bare minimum. I’m still curious if any builders have measured the temps at the heatsinks although it’s hard to be accurate. I bet it’s 60 degrees plus…
5U/300 it is! 0.18 degC/watt, only 18 degrees rise, and the front panel should help while having a cover on…doesn’t. I almost always add some heatsink grease between the heatsink and front panel in hopes that it participates thermally…
Best,
Anand.
Don't forget ratings are for heat transfer when the heatsink temperature us 85 degrees C ABOVE ambient.
You want about 25 degrees above ambient. Multiply heatsink rating by about 1.4 to get correct heat transfer characteristic.
I linked a paper on this in my Aleph 2 rebuild thread today and calculated 4U 400mm capacity.
You want about 25 degrees above ambient. Multiply heatsink rating by about 1.4 to get correct heat transfer characteristic.
I linked a paper on this in my Aleph 2 rebuild thread today and calculated 4U 400mm capacity.
Thank you all for your input. I have some options to consider. Since my F3 won’t only be an F3, it will have a small gain stage ahead of it, either FE2022 or Wayne BA2018. I want to stick with a discrete front end, no opamp based or transformer based FE this time around, as such I will need to accommodate for that in the enclosure. It will have both balanced and unbalanced inputs. At this time, I do have one pair of Wayne’s BA2018 stereo boards (the blue ones) so I would like to put them to good use. I will obviously drop the gain on the BA2018 to the lowest it can go. The F3 on it’s own is about ~ 13dB or so.
Oh…if it isn’t obvious, I don’t have a preamp although the Iron Pre minikit just recently arrived at my doorstep
.
Thanks!
Anand.
Oh…if it isn’t obvious, I don’t have a preamp although the Iron Pre minikit just recently arrived at my doorstep
.Thanks!
Anand.
Last edited:
Hi Everyone,
I've returned to some projects that I never finished properly. This includes an F3, which I got working, but wanted to rebuild since the heatsinks were insufficient. (And because I ruined a Lovoltech fiddling with it.) I used Nelson's schematic (attached) without any changes, except the rail voltage is a little higher, at about 49-50V. (I haven't checked at full load since rebuilding). The PSU was discussed in an old thread here,
https://www.diyaudio.com/community/threads/clc-power-supply-for-f3-in-apple-computer-case.365009/
although that is probably not very relevant except for the higher output voltage. The Lovoltechs are matched and on xrk's adaptor boards (the jfets are from Nelson, through the related thread - Many thanks!) and I'm fiddling with the value of R5 and the setting of P1. I understand from this thread and others on the site that the 1.1V and 3.5V on the jfet are critical. (This makes sense to me, since you'd want the jfet to operate at the right bias point, as in the Zv9 article.) I also understand that the 21V on the +ve side of the output caps is important.
My question is why the particular value 21V? Is it because it's about half the 41V shown after the cap multiplier? If so, then with a slightly higher rail voltage I should probably be looking for something nearer 23V, right?
Thanks for any input!
Best
Nigel
I've returned to some projects that I never finished properly. This includes an F3, which I got working, but wanted to rebuild since the heatsinks were insufficient. (And because I ruined a Lovoltech fiddling with it.) I used Nelson's schematic (attached) without any changes, except the rail voltage is a little higher, at about 49-50V. (I haven't checked at full load since rebuilding). The PSU was discussed in an old thread here,
https://www.diyaudio.com/community/threads/clc-power-supply-for-f3-in-apple-computer-case.365009/
although that is probably not very relevant except for the higher output voltage. The Lovoltechs are matched and on xrk's adaptor boards (the jfets are from Nelson, through the related thread - Many thanks!) and I'm fiddling with the value of R5 and the setting of P1. I understand from this thread and others on the site that the 1.1V and 3.5V on the jfet are critical. (This makes sense to me, since you'd want the jfet to operate at the right bias point, as in the Zv9 article.) I also understand that the 21V on the +ve side of the output caps is important.
My question is why the particular value 21V? Is it because it's about half the 41V shown after the cap multiplier? If so, then with a slightly higher rail voltage I should probably be looking for something nearer 23V, right?
Thanks for any input!
Best
Nigel
Attachments
A few hours fiddling around and I'm pretty sure I can get the voltages where they should be, although I still need to finalize things.
However I've spotted a problem. The chassis I've bolted together should be plenty big enough to dissipate the heat, but I'm using boards from Peter Daniels that are a tight fit, and all the mosfets and the Lovoltech are very near either the top or the bottom. Not ideal, but it isn't a problem according to my hand and infrared thermometer, except for Q2, which is in an unfortunate position and is too hot for my comfort. (I haven't tried anything terribly accurate to measure temp.)
I can solve this (I think) by moving the mosfet off the board and using flying leads. These would be 2 to 3 inches long if I put Q2 where I have in mind. I imagine moving R11 (the gate stopper) is a good idea - I can just solder it directly to the mosfet leg. Are there any other issues I haven't thought of?
O mighty ZM: What say you? Good idea? Bad idea?
Best
Nigel
However I've spotted a problem. The chassis I've bolted together should be plenty big enough to dissipate the heat, but I'm using boards from Peter Daniels that are a tight fit, and all the mosfets and the Lovoltech are very near either the top or the bottom. Not ideal, but it isn't a problem according to my hand and infrared thermometer, except for Q2, which is in an unfortunate position and is too hot for my comfort. (I haven't tried anything terribly accurate to measure temp.)
I can solve this (I think) by moving the mosfet off the board and using flying leads. These would be 2 to 3 inches long if I put Q2 where I have in mind. I imagine moving R11 (the gate stopper) is a good idea - I can just solder it directly to the mosfet leg. Are there any other issues I haven't thought of?
O mighty ZM: What say you? Good idea? Bad idea?
Best
Nigel
How about some pictures even though not done yet? Here's some pics of work in progress.
The amp has an external PSU, feeding separate L and R units. (I hesitate to call them monoblocks when the PSU is external. Seems wrong, somehow...) PSU is CLC, plenty of filtration, and has been discussed elsewhere. I had previously tried these boards with different heatsinks which turned out to be inadequate, hence the redesign. The heatsinks were cheap from ebay; three of them that turned out to have the fins running the "wrong way". (I didn't look carefully enough. Not the vendor's fault at all.) So I cut them in half and milled the ends to give me three shorter ones per channel. Hence the design.
The little white board for the Lovoltech came from member xrk971 s group buy a while ago. Very nice. The boards were expecting the legs from the jfet with their normal spacing, though, so little wires were needed. If I hadn't already soldered the larger pins on the board I might have done things a little differently. There's a little gate stopper on the board.
The first and second pictures shows the rear of one channel, without back panel while I'm fiddling around. You can see Q2 has been moved from the top left of the board to the left "wall" as discussed above. It's really much, much happier there. So far temps seem to be very modest, to my relief.
I've most of the way to having the right voltages. I'll post again when there's something prettier to look at, or something goes wrong again...
Best
Nigel
The amp has an external PSU, feeding separate L and R units. (I hesitate to call them monoblocks when the PSU is external. Seems wrong, somehow...) PSU is CLC, plenty of filtration, and has been discussed elsewhere. I had previously tried these boards with different heatsinks which turned out to be inadequate, hence the redesign. The heatsinks were cheap from ebay; three of them that turned out to have the fins running the "wrong way". (I didn't look carefully enough. Not the vendor's fault at all.) So I cut them in half and milled the ends to give me three shorter ones per channel. Hence the design.
The little white board for the Lovoltech came from member xrk971 s group buy a while ago. Very nice. The boards were expecting the legs from the jfet with their normal spacing, though, so little wires were needed. If I hadn't already soldered the larger pins on the board I might have done things a little differently. There's a little gate stopper on the board.
The first and second pictures shows the rear of one channel, without back panel while I'm fiddling around. You can see Q2 has been moved from the top left of the board to the left "wall" as discussed above. It's really much, much happier there. So far temps seem to be very modest, to my relief.
I've most of the way to having the right voltages. I'll post again when there's something prettier to look at, or something goes wrong again...
Best
Nigel
Attachments
I think I'm done fiddling around with tweaks, at least for the present, . Rail voltage under full load is a little lower than I expected, but after the cap multiplier I have 43 - 43.5V on both channels. On one channel I measure 21.43V just before the output caps, 1.137 on the source of the jfet and 3.587 on the drain, giving Vds of 2.45V. On the other channel I measure 21.41 on the output caps, 1.129 on the source and 3.554 on the drain, so a Vds of 2.425V. I have little to no confidence in the third decimal place of these measurements, BTW; my DMM is pretty good, but I could really use a second one for a double check, and in any event the last digit bounces around a bit even while measuring.
I've also found that it takes a long time for these voltages to stabilise - around an hour to an hour and a half - has anyone else seen that? Is something, maybe the Lovoltech, sensitive to temp? (That's about how long it takes for the heatsinks to reach stable temperature. Which downstairs in my little basement workshop is perfectly OK. Hottest thing I can find is the screw and washer holding Q2, and I can easily keep my finger on that indefinitely. It may be warmer up in the real world - we'll see.
Unless anyone has a good reason to continue tweaking these voltages I'm going to see what it sounds like, and live with it a bit. I can always adjust R5 again if I think it needs it, and if it looks like a keeper I'll pretty up the chassis a bit. And put in the connector for the DC voltage, which the sharper-eyed among you may have noticed is missing...
Best
Nigel
I've also found that it takes a long time for these voltages to stabilise - around an hour to an hour and a half - has anyone else seen that? Is something, maybe the Lovoltech, sensitive to temp? (That's about how long it takes for the heatsinks to reach stable temperature. Which downstairs in my little basement workshop is perfectly OK. Hottest thing I can find is the screw and washer holding Q2, and I can easily keep my finger on that indefinitely. It may be warmer up in the real world - we'll see.
Unless anyone has a good reason to continue tweaking these voltages I'm going to see what it sounds like, and live with it a bit. I can always adjust R5 again if I think it needs it, and if it looks like a keeper I'll pretty up the chassis a bit. And put in the connector for the DC voltage, which the sharper-eyed among you may have noticed is missing...
Best
Nigel
Attachments
Yes, agreed that 3rd digit is nonsense. I understand about temperature equilibrium being the point, too, but does 60-90 minutes sound like a reasonable amount of time for stability? I remember my F5 taking about that long for the DC offset on the output to stabilize, but in this case the voltages on the drains move over 1/2 a volt over 15 -20 minutes, from above 4V in both channels and take a long time to get to their stable values. Not really a problem, I guess, unless I'm not seeing something.
What do you think - do the voltages otherwise look close enough?
Listening to it now and so far it sounds good.
What do you think - do the voltages otherwise look close enough?
Listening to it now and so far it sounds good.