- Status
HBrown
Active Member
I soldered all the components on the board except Q1-Q5, then calculated and bent the transistor legs as required, then I screwed Q1-Q5 down to the heatsink plate to hold them steady then slipped the board down onto the legs and then soldered Q1-Q5 from the top. Then it can be removed from the heatsink adapter plate to solder the legs backsides and check joints. Then in final assembly the silpads are installed under the devices. That insures everything is aligned and there is minimal residual stress on the board joints when it is bolted together in the final assembly. Then, the whole module can be fastened down to the 780's heatsink with thermal compound and the original STK screws.
Last edited:
That is what we came up with, sure, everyone who does EE/ME will come up with their own version. If you can find an easier/better/cheaper way, by all means, go for it.
Really impressive work guys.
I can see where getting Q1 ~ Q5 lined up (and getting the damn sil-pads to stay in place*) might be a PITA, but the end result looks awesome.
Only concerns:
1. With a double-sided through-hole board, pretty much impossible to desolder a component (destroy-to-remove is now in effect). But the compactness of the board probably made that necessary. So the warning is, get the parts installed right the first time.
2. Aluminum heat-spreader. 3/16" thick. Over-tighten any of the screws for those mounted transistors and you have a real problem. Any reason steel was not considered? Threads would be much more durable.
*Fully-encapsulated parts, anyone?
I can see where getting Q1 ~ Q5 lined up (and getting the damn sil-pads to stay in place*) might be a PITA, but the end result looks awesome.
Only concerns:
1. With a double-sided through-hole board, pretty much impossible to desolder a component (destroy-to-remove is now in effect). But the compactness of the board probably made that necessary. So the warning is, get the parts installed right the first time.
2. Aluminum heat-spreader. 3/16" thick. Over-tighten any of the screws for those mounted transistors and you have a real problem. Any reason steel was not considered? Threads would be much more durable.
*Fully-encapsulated parts, anyone?
HBrown
Active Member
Thanks EW. One could get a bit too aggressive and probably strip the threads. I initially worried a bit about the threads, but I chose 6061-T6 aluminum (much stronger than alloys most heatsinks are made of) and used a tap drill size for maximum thread egagement condition. Careful assembly and torquing is warranted, but the fastening seems entirely adequate.
Aluminum just seemed the correct choice for the plate- adequate strength, like thermal conductivity to the heatsink, easier to fabricate than steel. Heat transfer from devices through the spreader to the heat sink seems to be efficient.
Aluminum just seemed the correct choice for the plate- adequate strength, like thermal conductivity to the heatsink, easier to fabricate than steel. Heat transfer from devices through the spreader to the heat sink seems to be efficient.
HBrown
Active Member
I talked with my guy about pim inserts on the plates I had outsourced. Adds considerable cost to the part and you're taking conduction surface area away from right under the device when you counterdrill for the insert.
Honestly, If I managed to strip one I'd just take a drill and shoot thru both spreader and heatsink itself, get a longer screw and put a nut on backside of heatsink. I've only assembled 6 modules so far- but thats 30 tapped holes assembled with no problems. It really doesn't take that much torque to get the necessary tensioning on the devices. A conical spring washer would be a good idea as well, but so far I have just used flat washers. I could take the engineering to aerospace levels, but so far no problems have manifested with the KISS aproach.
The cleanest way to do this really is to layout and drill the heatsink itself and forgo the spreader. The spreader is basically there to allow a completed module to mount using the original STK mounting holes and screws for those who were unable/unwilling to modify the original heatsink.
Honestly, If I managed to strip one I'd just take a drill and shoot thru both spreader and heatsink itself, get a longer screw and put a nut on backside of heatsink. I've only assembled 6 modules so far- but thats 30 tapped holes assembled with no problems. It really doesn't take that much torque to get the necessary tensioning on the devices. A conical spring washer would be a good idea as well, but so far I have just used flat washers. I could take the engineering to aerospace levels, but so far no problems have manifested with the KISS aproach.
The cleanest way to do this really is to layout and drill the heatsink itself and forgo the spreader. The spreader is basically there to allow a completed module to mount using the original STK mounting holes and screws for those who were unable/unwilling to modify the original heatsink.
If you are concerned about stripping the threads on the aluminium plate, you can use nylon screws so they will strip before the aluminium will. Also acts as a screw torque control, so if you over torque the screw, it will slip and strip its threads.
Aluminum is the logical choice for the plate. steel is not a very good conductor of heat as it is.
i also thought that you could counter bore the back side of the Al plate so you can use a nut placed in the counter bored recessed out area instead of tapping threads. best to use a drill bit without a flue, to bottom out in the bore but they are not common. Just thinking out some options, but HBrown's implementation is just fine as it is. We could discuss this to death and still be no further ahead.
Aluminum is the logical choice for the plate. steel is not a very good conductor of heat as it is.
i also thought that you could counter bore the back side of the Al plate so you can use a nut placed in the counter bored recessed out area instead of tapping threads. best to use a drill bit without a flue, to bottom out in the bore but they are not common. Just thinking out some options, but HBrown's implementation is just fine as it is. We could discuss this to death and still be no further ahead.
Another thought and something I have done before for removal of comps on PTH pcb's without damage, is to use a piece of bare Cu wire to bridge across all the device leads. This way you can have the soldering iron heat up all the leads at once, flow molten solder onto each lead and easily remove the device, After you can solder suck/wick the holes clean of solder for the new device to be fitted.
Actually seen/read this being used in a Elektor mag. I have used this technique on both TH and SM components. Just have to form the Cu wire appropiately for each application. In many cases it works better than the hot air tool except for leadless pkgs that is.
Last edited:
ethangsmith
Well-Known Member
Will these be eventually sold in kit or assembled form? I have a 780 that I would love to do this to before the original packs fail.
petehall347
the brandy coffee man
on an engineering way of thinking studs screwed into the heat-sinks then fastened with nuts would ensure the threads in the aluminium would survive no problem ..
qprhooligan
Well-Known Member
I ordered some from pcb way.com. I couldn't get the for $13 per 10 though. It cost me about $2 each board when all was said and done. Hope to try them soon.
Arthur Smith
Super Member
There are some of these units, like mine, with perfectly functioning power packs, that are still original. They are not all like ticking time bombs. I have a couple of sets of spares, but still, the panic set off by these things online is blown way out of proportion.
No one has stepped up to bat to offer a kit as the labor costs out way what most people are willing to pay.
The idea is to order your own pcb's, parts and build your own using some simple tools. It is a high school level project.
The deal at pcbway is 100x100mm for $13.00 using the default process, of course not including shipping.
So the unit cost me for below using HK Post. You did better than I!!
Today I see the price down to 10$, just keeps getting better.
Description Unit price Qty Amount
PCB Order Group NO :20388
$28.00 USD 1 $28.00 USD
Subtotal $28.00 USD
Total $28.00 USD
Payment $28.00 USD
Charge will appear on your credit card statement as "PAYPAL *PAYMENT"
From amount $38.39 CAD
To amount $28.00 USD
Exchange rate: 1 Canadian Dollar = 0.729360 U.S. Dollars
People are stupid, and do stupid things. The list of stupid things that the STK's in the 780 will not survive is a long one, hence the demand for the 0050 devices outstripped the supply of authentic original parts long ago.
- Status