Balanced F5 question

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I did some simulation and extrapolation from previous builds and come to a heat sink temperature rise of 45~55°C above ambient for dissipation of 145W, which is indeed a bit high. So the concern was well justified.

The MOSFET itself has a thermal resistance of 0.83°C/W, or 27°C rise at 32W. Give another 5°C for Kerafoil insulation. For ambient temperature of 25°C, the junction temperature would be 112°C. So too high for my own taste.

Need experimental verification. Not easy to solve without increasing cost (larger heatsinks and case) or reducing wife acceptance factor (fans on the outside).

Back to the drawing board.

:(


Patrick

Worst case scenario is, we have to buy 2 cases and build monoblocks.
It might cost a bit more but in the end, the result will be worth it.
It also allows margin for increased power in the future.
 
OK, possibly a simple solution.

Add 2 more smaller heatsinks on the inside surfaces of the front and rear plate, something like 250 x 150 x 25mm.
E.g. Fischer SK49-150. Costs 40 Euros extra, but gets the temperature down to 35°C above ambient.
Small price to pay than to go to 400mm sinks (and hence larger housing in all dimensions)

Test on Conrad still useful.


Patrick

PS
Any of the previous mentioned commercial housings from Schuro & Hifi200 has essentially the same heatsink as Conrad MF35-151.
So none of them will do in any case.

That sounds like a good idea regarding the extra heatsink internally.

I will try and test the heatsinks some time next week.
 
> Worst case scenario is, we have to buy 2 cases and build monoblocks.

You have a high bandwidth power amp.
You don't want long wires between mosfets mounted on opposite heatsinks in a monoblock running over your transformer.

So only solution is ..... Mini Tower. (The original PCB layout was also designed for this.)
Expensive solution.

And my task was (is) to design a (sort of universal) case that will suit other DIYA Class A projects and not just the F5X.

I'll stick to inner sinks for now, and wait for your measurements.


Patrick
 
Last edited:
As far as i could experiment with my aleph j biased at 2 x 1.3A dissipating 125w per channel, two heatsinks measuring 400mm x 140mm x 80mm do the job perfectly, even at 35°C ambient..

A stereo case costs anyway less than monobloks. And moderately high heatsinks are more efficient than tall ones.

The only drawback is the weigh.

I will cannibalise my alephJ case to give it a try.
 
Last edited:
hifi2000 can make special cases on order for a reasonable quantity (lets say 50 units)
A custom "all alu" Pesante Dissipante 4U or 5U/300mm with a nice 10 mm engraved front panel and some bigger heatsinks such as Fischer SK568 or similar (300 * 60 mm height 165-210 mm) can do the job.
Its only matter to resize rear panel and covers, not a difficult task indeed.
 
I checked out the Fischer online catalog.

I am afraid that SK568 is too small and will not do. It is listed with Rth 0.3°C/W at 150mm, but this corresponds to a temperature rise of 80°C. If you apply the correction factor for 35°C rise, you end up something like 0.4~0.45. Realistically you have about 150W dissipation per F5X, which means dT close to 60°C.

So you need to search for a catalog Rth value of about 0.2°C/W per heatsink, which was why I chose the Conrad in the first place.

Realistically if you do not want to use the inner sinks (i.e.sinks on all 4 sides), you should be looking at a sink of about 400x200x50mm or 350x150x60mm. The first you can buy (2x SK90 or 4x SK88 side by side), cost a bit more, but need larger plates all round, and higher shipment cost. The second will fit the design without significant changes, but I cannot find one as standard catalog item. Probably 3x SK88 with 15mm between sinks is the closest we can get. No idea about costs. Fischer is not cheap, and no special machining unless you order > 500 pieces. Which means shipping to somewhere else to machine and anodise. I doubt we will sell 50 cases.

No easy option I am afraid.

If Hifi2000 can provide a solution for less than 500 Euros including world wide shipment and a good donation to DIYA, you are welcome to take over from me.


Patrick
 
Last edited:
If Hifi2000 can provide a solution for less than 500 Euros including world wide shipment and a good donation to DIYA, you are welcome to take over from me.
Patrick
I would love a couple of your towers in my lounge, but hifi2000 is hard to beat for value. Pics at bottom of the page are quite nice for 250 euro. I dont think I could make and anodise a case for this.

modushop.biz
 
You are right.

I cannot beat them for value, because I don't have local source of cheap heatsink extrusion.
Feel free to bulk order from them. The 4pcs 250x210x40mm heatsinks looks big enough for 2x F5X.
The handles in the picture are not included in the price.
And I wonder what they charge for shipment (I would guess 80 Euros).

Do watch vibrations on the bottom plate when you put heavy Trafos on.
Make a good drum especially if you are unlucky that the resonance frequency of Tx + Btm Plate combo get close to low order harmonics of 50/60Hz.
But you can always special order a thicker bottom plate of say 6mm.

As I said all along, aluminium has its price. You can always save cost by using thin plates.
And he has basically no machining costs.
Front plate can be sawed; the rest is sheet metal (plunge or laser cut).


Patrick
 
Last edited:
Increasing the length of the Conrad from 150 to say 200mm brings very little, much less than the 30~40% reduction in temperature rise we need.

I will offer a design with 6pcs Fischer SK88-150 per case, which will take 2x F5X at about 40°C temperature rise. I have hands-on experience with the SK88, so I know what to expect. And the cost is not much different to the Conrad's. I shall also provide the option of at least a rear cooler using Fischer SK49-150, again at minimum extra cost.

But for those who think Hifi2000 is adequate, they have my blessing to go for them. I think they can fulfill the basic functions. We have decided right at the beginning not to compete with commercial players who design their product to a price. We only want to offer something which is design to fulfill certain functions to the level we consider acceptable, then uncompromising cosmetics (always a matter of taste), and only then try to minimise cost to make them affordable. We shall never be able to make a commercially viable product for 250 Euros for the quantities that we are talking about.

So for the few of you who want to stay with my stubborn approach, you will get a solution from me, and we'll still make sure DIYA gets sufficient profit or donation in the exercise.

The rest I cannot reveal right now, because Variac has the right to know first, and we also will not publish the design before we are ready with everything.
You will get the design copied within weeks by people waiting already on the line, before we can even start detailing.

Hope you would all understand.


Patrick
 
Last edited:
Thank you for the support.

I can tell you meanwhile the overall dimensions :

Width 350mm
Depth 400mm including lifting handle
Height 168mm including feet

Thermal resistance for complete case based on 6x SK88-150 and 1x SK49-150 at 35°C rise is calculated as 1.22°C/W.
i.e. it can dissipate 290W for the same temperature rise.

Values for Conrad 35-151 to be determined experimentally next week.


Patrick
 
Member
Joined 2004
Paid Member
Tower build

Patrick,
What about these HS for builders in the US. Seller will cut them to any height.
They are 10" wide and could be used for a tower. Claimed as °C/W/3" Approximately: 0.80, so 0.2 for 12" height? One per channel?

Barrredboss

http://cgi.ebay.com/Heatsink-Aluminum-Extrusion-10-080-Wide-x-12-Long-/170555438990?pt=LH_DefaultDomain_0&hash=item27b5e5778e
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.