VVR3 (fried the MOSFET ... not thinking)

[Reeltarded]

If there were a cutout in the headbox just outside the chassis edge under the area where you mount the heatsink, the thermal draw of fresh air might make the place where you have it the best place.

I am speculating only. All based on experiences with other hot electronics. The small fan would absolutely solve your problem, if not annoy you a little bit while it runs. I have disabled fans on Mesas that weren't what I would call at a bedroom level.

7 7 7 7 7 7 7 7 are my prefered setings on Mk II and III amps. heh

[doveman]

If there were a cutout in the headbox just outside the chassis edge under the area where you mount the heatsink, the thermal draw of fresh air might make the place where you have it the best place.

I am speculating only. All based on experiences with other hot electronics. The small fan would absolutely solve your problem, if not annoy you a little bit while it runs. I have disabled fans on Mesas that weren't what I would call at a bedroom level.

7 7 7 7 7 7 7 7 are my prefered setings on Mk II and III amps. heh

That first idea might the simple winner. Also, my amp for 30 years (1980-2010) was the Original Boogie ... 7's were plentiful on mine too. My son is still using that amp.

[Reeltarded]

Ooh! Mk 1b! GREAT AMPS!

[doveman]

I fixed the VVR3 today. It was in fact the MOSFET/Diode. While I did try to move the MOSFET to another place on the chassis, it added some noise. So for now ... I'm going to leave it in the original location and avoid full volume and a VVR setting of 5 (stresses the mosfet the most at 5). I have proved that does not work. I also have a backup mosfet and diode. I would like to make it bullet proof ... so will be thinking about heat dissipating options over the next few weeks.

Sure does sound good though ...

[markr14850]

You might ask Dana if he has any recommendations on paralleling two MOSFETs to share the load.

[Mark]

Heat would have been the issue that killed the mosfet. Do search on the part number and look for spec sheets. The manufacturer should give you the spec for a suitable heatsink for this device

[Structo]

Or perhaps a higher current rated MOSFET?

[doveman]

I think Dana purposely chose this mosfet because it could take a lot ... work on a lot of amps. My problem is simply limited choices for heat sinking to the aluminum chassis. I think it will boil down to:

• A heat sink (that doesn't collect heat from the rectifier)
• Maybe a small fan
• Interesting idea to add a vent on the underside of the head cab directly underneath the mosfet. Let convection draw in a bit of cool air naturally.
• A combination of the above

So with two mosfets ... I'd still have to put them in similar locations. BTW ... I did try to mount it in a good spot up near preamp tubes, but it created some weird hums and noises in the signal. I think I'm stuck in the area of the amp where it's located ... just a few options.[/list][/list]

[markr14850]

Yeah, I think the MOSFET Dana chose is awesome. It will do continuous 14A / 275W - if properly heat-sinked.

Tell us about the mounting of your MOSFET. What insulator did you use? Did you use thermal paste? Any problem in the mounting could easily be the cause of the initial failure.

When you tried relocating the MOSFET and found noise, did you move the VVR board too, or just run jumpers to the MOSFET? (Just jumpers would worry me a bit, because you should have a 200+ ohm gate stopper resistor mounted right at the MOSFET pins.)

For two MOSFETs, you would just parallel them (each with it's own gate stopper), and try to have them a few inches apart. However, you should really check with Dana on this, as parallel MOSFETs might not play nicely with the current limiting he has set up.

[doveman]

I used the silicon wafer that comes with it. Talked to Dana about nuts and bolts ... did both amps the same way. It did not look like it shorted ... removed all burrs where the hole was drilled before hand ... silicon wafer was still in the right place. The new one is working in the same place ... old one worked until I dimed everything. I still might be OK if I don't dime everything and have the VVR set a 5.

I think I read a post here where Dana advised against using two mosfets. I moved the mosfet only ... expected noise ... got it ... but it was worth a try.

[Tillydog]

I used the silicon wafer that comes with it.

Are you sure it was silicone, because in the picture in your first post, it looks a bit like a mica one. (Silicone = floppy rubber that can be used without compound; Mica - hard, brittle & needs heatsink compound). AFAIK, the NTE2973 is in an electrically insulated package, anyway, so doesn't need the mica, just the heatsink compound/pad.

I don't know KT66s, so I'm not sure what you biased your amp at - guessing at no more than 50mA per tube say?? roughly 450V B+ ?? that's 45W from the PSU - 50W, say with the preamp tubes.

From the NTE2973 datasheet, it has a maximum junction temperature of 150°C, and a thermal resistance from junction to case of 0.45°C/W

The Mosfet should (I think) only ever see about half of the 50W, but taking worst case, if it was dissipating the full 50W, it would live, provided the case stayed at 150 - (0.45 * 50) = 127°C. Taking off another 0.5°C/W for the silicone washer assuming that's what it is) means the chassis will be at 127 - (0.5 * 50) ~ 100°C - Spit would sizzle on your amp chassis before the MOSFET blew if it was in good thermal contact with the case - is that how it ran?

To specify a heatsink, I would aim for a 125°C max junction temperature (a bit conservative). For 25W dissipation (the likely 'real' case) this means a max case temperature of 125-(0.45 * 25) = 108°C. The ambient temperature might be ~38°C (for round numbers), this means any heatsink needs to limit temperature rise to 70°C above ambient for 25W dissipation, or 2.8°C/W. Allow 0.5°C/W for the thermal resistance of the mounting arrangements, and any heatsink needs to be better than ~ 2.3°C/W.

Your chassis probably wouldn't be that good, especially if it is thin-ish metal. If it was ~3mm , my gut feel would be that it *might* be OK. I don't think moving the MOSFET would make a tremendous difference, personally. (Adding a second one in parallel, mounted on a completely different part of the chassis might work, though).

The sort of thing you would be looking at for a stand-alone heatsink would be very roughly a 4" x 4" finned heatsink , vertically mounted, with the fins also vertical and in free air. (Many other permutations are possible.)

Have a look at what's available and pick something suitably rated to fit into whatever space you have available - manufacturers quote the °C/W figure for their heat sinks, but pay attention to the orientation that they're rated for, and the temperature rise above ambient that they assume (if it's lower than the 70°C above ambient assumed above it's OK). If you mount the MOSFET inside the chassis, with the heatsink outside, you will gain some heatsinking from the chassis provided you use thermal compound / silicone pads either side of the chassis.

If you're prepared to go to fan cooling, you'll be laughing.

HTH

Andy

[Super_Reverb]

One thing you might also try is to buy a 1/4" thick or so strip of Al and bolt it to the inside of your chassis where the MOSFET is bolted and then bolt your MOSFET to this strip. This strip of metal will help spread the heat to the remainder of the chassis.

When you install the device, use the white thermally conductive grease on both sides of the mica insulator. The thermal resistance from the MOSFET die to chassis is a series combination of the Rjunction/case, and the Rcase/chassis, so it's important to get a low thermal resistance when you mount the transistor.

There won't be a problem paralleling two MOSFETS if the threshold voltages are fairly close. Placing MOSFETS in parallel is like using a larger MOSFET. The current doesn't have to split perfectly between the two devices, for there to be a significant reduction in power/device. This circuit uses this device in a source follower. The "power" control is a pot that varies the gate voltage (tapped from drain voltage) and the source follows it. It functions as an adjustable voltage drop. My point here is it's a very simple circuit, similar to a cathode follower, but using a field effect transistor. It is inherently stable. Adding some capacitance (100s - 1000s pF) from gate to ground should quiet the circuit.



cheers,


rob

[doveman]

I stand corrected ... it's the mica wafer that comes in the package. I think I've seen Dana say the grease hasn't been necessary for him on Train Wrecks ... just uses the mica. I first worried the mica may have slipped out and shorted but it had not. Maybe the grease would be helpful.

The aluminum chassis is super thick ... Dr Z apparently uses thick aluminum. My problem is that the best place to put a heat sink would be in outside open air on the opposite side of the chassis. Not possible where it is now. If i move it to where i suggested others have correctly pointed out the heat sink might soak up more heat from the rectifier than it drains away from the mosfet.

446v I think ... biased at about 43mA - Golden Lions are a 30w tube.

The chassis was not extremely hot when it failed. I had it out of the cabinet and only been playing it a for less than a minute. I had played it for hours before in the cabinet. Just not at dimed settings. I certainly think I was close to being ok ... I may just be slightly out now and just need a little help with heat dissipation.

All the suggestions are appreciated and food for thought.

[Tillydog]

I stand corrected ... it's the mica wafer that comes in the package....


....The chassis was not extremely hot when it failed.

The mica washer on its own is a disaster, thermally (with all respect to Dana ). One **needs** to use heatsink compound with mica to get good thermal contact. A silicone pad is easier and less messy.

If you didn't use any compound, then I suspect your MOSFET died because it wasn't in thermal contact with the chassis. (Your chassis looked quite thin, to me, in the first picture, sorry).

FWIW, I would just replace the parts where they are now. Take care that the MOSFET sits completely flat against the chassis and that the mounting faces are flat and clean. Just a little heatsink compound (both sides of the mica washer) - it should ooze out all round when you bolt the MOSFET down (or use a silicone pad). Monitor the temperature of the chassis when it's in use. If you can bear your hand on it at all, then it's not too hot for the MOSFET. If it hisses when you spit on it, then you're pushing the envelope

Andy

[doveman]

The chassis is 1/8" thick aluminum I think. Seemed pretty thick to me as aluminum chassis go. I guess talking in measurements would make more sense.

I did see that Dana worries that some grease is conductive ... causing the mosfet to short out ... or maybe get on something else by accident. So I guess if you were sure it was non-conductive that would help. My understanding is that it's to remove air bubbles ... so laying it on thick is not as important as getting it even. But it sounds like it might help.

Just taking in what everyone is saying ... all I know is it sounds great ... just need to make it a bit more towards bullet proof.

ANOTHER THOUGHT:

I could run short leads through a gromet and put the MOSFET on the top of the chassis next to the power transformer. Keep it away from the rectifier ... how hot would that area get I wonder?