sorry for seeming disparaging, my fault. but i truly wish to help you to a good resolution with this.
and not to be interpreted as condescending but fuses are slow compared to transistors as fuseable elements!!!
Yes, but in this case nothing happens before the fuse blows. As mentioned at the beginning. Get a little pop and then the circuit is isolated at the mains input.
The case serves as a radiator, and the case must have a thermal stabilization transistor on the back side.
I haven't seen any details of the chassis here and there are no instructions available for building it anyway. At at the risk of repeating what's been said already, Naim's NAP models have always used the chassis as a heatsink but with heat spreaders and not just any chassis. It's a minimum 3mm (4 mm plate runs cooler) thick bottom plate and sides with a heatspreader bar of about 25x 4 mm aluminium, sandwiched between the power transistors + insulators and the chassis. If you're usin mica insulators, both faces of the bar and insulator are coated with a thin layer of heatsink compound before being bolted down and the assembly clamped up tight with two bridge clamps, each spanning both power transistors of either channel. As you'd expect, this is a better method than a 3mm bolt at either end of the transistors and much better than a single, offset mounting bolt used with TOP3, TO247 etc. power transistors. This is what the big slot in the PCB is for and on the original NAP200 and similar models, it's a neat fit. If you're using silicon rubber pads though, dont use any grease at all - none - nada!
I've seen several shoddy DIY builds of the Chinese kits without any spreader, even using just a thin aluminium or steel chassis which is all but useless for cooling when the power is up above 10W or so but since there are no building instructions, you can't expect much better from inexperienced constructors using blurry pictures as a specification and probably cutting corners to make an easy job for themselves rather than learn what cooling is required for a reliable 100W/4R stereo amplifier.
The mention of CPU cooler paste should set off alarm bells. Some compounds (the highest rated) still use conductive silver powder in the paste and this soon makes an electrical path between the power transistor collector and ground.....bang!...another smoking hot, dead transistor. Simply don't buy compound intended for CPUs which are actually in insulated cases anyway. Use the proper white stuff from electronic parts distributors and you won't have any trouble with live transistor cases or backplate shorts again.
If anyone is in doubt about the odd arrangement, bias control is regulated by 2 transistors mounted on the bottom side of the PCB. This brings them closer to the case (heatsink) which reduces the thermal lag somewhat and hence improves performance. If you blindly follow the overlay and mount them topside, the amplifiers won't work properly.
I've seen several shoddy DIY builds of the Chinese kits without any spreader, even using just a thin aluminium or steel chassis which is all but useless for cooling when the power is up above 10W or so but since there are no building instructions, you can't expect much better from inexperienced constructors using blurry pictures as a specification and probably cutting corners to make an easy job for themselves rather than learn what cooling is required for a reliable 100W/4R stereo amplifier.
The mention of CPU cooler paste should set off alarm bells. Some compounds (the highest rated) still use conductive silver powder in the paste and this soon makes an electrical path between the power transistor collector and ground.....bang!...another smoking hot, dead transistor. Simply don't buy compound intended for CPUs which are actually in insulated cases anyway. Use the proper white stuff from electronic parts distributors and you won't have any trouble with live transistor cases or backplate shorts again.
If anyone is in doubt about the odd arrangement, bias control is regulated by 2 transistors mounted on the bottom side of the PCB. This brings them closer to the case (heatsink) which reduces the thermal lag somewhat and hence improves performance. If you blindly follow the overlay and mount them topside, the amplifiers won't work properly.
yes, it has the transistors on the bottom.
The plate is 6mm
The thermal paste is non-conductive but I'm guessing it got through the paste.
The plate is 6mm
The thermal paste is non-conductive but I'm guessing it got through the paste.
The transistor will save the fuse. Transistors are fuse protection devices.
A fuse costs around a dollar. A big Sanken power transistor costs around $12. Makes sense, right?
A fuse costs around a dollar. A big Sanken power transistor costs around $12. Makes sense, right?
The paste isn't relied on to insulate, it is just to fill air gaps and hence improve thermal conductivity.
The term "thermal compound" got messed up when CPUs required silver particals in the paste. Don't use anything like that for mica washers or ceramic pads on transistors. Use Dow Corning 340 which doesn't have metal particals in it. I don't know how you tell the difference without reading the whole datasheet. But the brand "arctic" is a good tip off that the compound has silver in it.
If you live in Canada, do NOT use farnell as a supplies. The canadian website apparently ships from the US, they always use UPS, and UPS charges a ~$20 customs origination fee when they cross the border. In Canada use digikey that runs a truck from Minnesota across the border on their own, and there are no customs at all.
100 W using a case as a heat sink is very marginal. My amp had a 120 W rating, which was for an hour. when I tried to use it at 20W for 3.5 hours in a choir rehearsal, the base resistors caught fire and the output transistors shorted. I've since added fins, and put 2 PCAT fans blowing on them through the perforated cover, like mickey mouse ears. If you run more wattage than 10 for more than an hour you had better re-engineer the heat flow. An aluminum plate will act as a spreader, but won't actually get the heat out of the case into the air. Fins or air flow do that, or both. Hint, fins, the slot has to point vertically, for air to flow up along them.
Happy diyaudio.
If you live in Canada, do NOT use farnell as a supplies. The canadian website apparently ships from the US, they always use UPS, and UPS charges a ~$20 customs origination fee when they cross the border. In Canada use digikey that runs a truck from Minnesota across the border on their own, and there are no customs at all.
100 W using a case as a heat sink is very marginal. My amp had a 120 W rating, which was for an hour. when I tried to use it at 20W for 3.5 hours in a choir rehearsal, the base resistors caught fire and the output transistors shorted. I've since added fins, and put 2 PCAT fans blowing on them through the perforated cover, like mickey mouse ears. If you run more wattage than 10 for more than an hour you had better re-engineer the heat flow. An aluminum plate will act as a spreader, but won't actually get the heat out of the case into the air. Fins or air flow do that, or both. Hint, fins, the slot has to point vertically, for air to flow up along them.
Happy diyaudio.
Both suggestions were quite to the point.
1) in your situation a dim bulb tester (google it) is WAY more useful than a plain fuse.
Unless you can take useful voltage measurements within the milliseconds it takes for a fuse to blow.
2) the fire extinguisher suggestion does not seem too out of place considering you mentioned a visibly toasted wooden floor.
Which was not prevented by fuses.
3)
NOW it blows instantly, because of the sheer short, but it did not while amp smoked and toasted the floor.
4) not sure about your power amplifier output rating, but given the huge power transistors, I guess high power so high dissipation.
Sorry but completely inadequate as heat sinks.
Click this image until fully expanded, notice the brushed chassis finish, typical of Aluminum (and not of steel).
The impressive Naim amplifier has its chassis made out of thick aluminum (I guess 2mm to 2.5mm thick), a large single piece bent in an "U" or "C" shape.
To boot, transistors are not bolted straight to it but to a heat spreader bar, also made out of black aluminum.
Since chassis is "outside", in contact with free air, it dissipates heat well, no need for fins, it already has large enough surface.
Doubt it gets more than warm to the touch.
Now your own, which I guess is a Naim clone, is mounted on a recycled cheap PA thin steel case, completely different.
The flat aluminum plate you provided is not enough: it´s way smaller than the full Naim chassis but much worse, it´s sandwichd between PCB and steel floor, inside an**enclosed** space, no air circulation.
My best suggestion, which will solve the main problem definitively?: contact a sheet metal shop , nothing sophisticated, in my beginnings I used local sheet metal fabricators, those making custom rain gutters, HVAC ducts, farm feeding pails for pigs and chickens, etc. , which had metal shears and folders, and have a "U" shaped chassis made out of 2mm or 2.5mm Aluminum.
Buy "middle hardness", suitable for punching and folding, metal seller will understand, because "hard temper" is brittle and cracks when folded 90 degrees, and "soft temper" is meant to make pans and kettles, way too soft.
You drill, sand, paint it on your own.
It will be easier to work with home tools than the steel chassis you repurposed.
Or you *might* find a ready made aluminum chssis similar to what I suggest.
Plan B: add a small fan to the case (you already have the cutout)
To minimize noise, get a 12V fan but feed it 9V or so, or a 24V one and feed it some 16V (rectified 12VAC)
It will not really *blow* air but gently (and noiselessly) *move* it, which is enough, you do not want heat buildup inside that chassis.
Plan C: have some metal shop cut a large "window" in your current chassis bottom and cover the huge hole with a thick aluminum plate pop riveted to it, maybe the current one is enough, the idea is that it´s exposed to outside air.
Well, that settles any heat dissipation issues 😉 No need for grease either, as long as you've got decent, thin silicone washers from Bergquist, for example. The thick, cheap import pads may be OK at low power levels but who knows how bad they can be. Some I've received aren't just a bit inferior - in my experience, they've been little better at heat conduction than plain silicone rubber.
It was the 'arctic' paste that I used. It says non-conductive on the side of the tube. But I think we can pretty safely say the transistor bottoms were more than able to pass through the non-conductive paste.
Incidentally, I once knew a doctor (medical) who liked to tinker. He built himself a new PC from new shiny parts but couldn't get it to boot. Called me round in desperation (my main line of work is IT) and he had assembled the CPU with no paste between the CPU and heatsink. He went off shopping, got the paste, I put it in and everybody lived happily ever after.
We'll have to see about airflow and volumes of aluminium involved, ect, ect... I'm going to stick with what I have as I think it'll be ok.
Double-up as in use 2 pads sandwiched together? No, you want the least possible thickness material if you want good heat transfer. If your mounting surface and transistor backplates presumbly are dead flat, you shouldn't need thicker insulators anyway, as genuine Silpads and even their imitators are going to be quite good electrical insulators by design.
I thought I posted this article here but it was in another thread. It seems to me you should read it 🙂 ESP - Heatsink design and transistor mounting