RG Amplified Zener - heat

[lowell]

I've implemented this wonderful little device in series with the OT center tap and in series with the screens and the screen supply node. IE both the plate and screen supplies drop about 200v with this in the circuit. Up till now I've mounted these Mosfets on the chassis for heatsinking. I've recently designed a circuit board with some relays and these amped zeners in order to minimize wiring. I'm concerned about the heatsinking. Is there anyway I can implement the circuit board and use floating heatsinks or is this simply necessary that the chassis be used as heat is just too much of an issue here. FYI the power amp is push-pull with 2 6V6 and 2 6L6.

[loudthud]

What package are the MOSFETs in? Do you intend to insulate the MOSFETs from the heatsinks or let the heatsinks have live voltages on them (bad idea)? Do you know how many watts will be dissipated?

[Steve Conner]

If the chassis is made of steel, it won't be a very good heatsink no matter how big it is.

If it's aluminium then it's a free heatsink.

Live heatsinks are bad from a safety point of view, but they boost performance. Those insulating washers hurt the heat transfer, and if you leave them out you can get by with a smaller heatsink or a smaller transistor. So the heatsinks are often live in TVs, SMPS and so on. The maker should put a Danger Live Heatsinks label on there, and so should you if you use them.

If you mount a heatsink inside an amp chassis, you have to make ventilation holes so that air can flow over it. My old Ninja Toaster amp took this to extremes, it was a mess of heatsink fins inside with a blower that drew air through the chassis and blew it out over the power tubes.

[Wilder Amplification]

My old Ninja Toaster amp took this to extremes, it was a mess of heatsink fins inside with a blower that drew air through the chassis and blew it out over the power tubes.

Hence the name right?

[Steve Conner]

No, Ninja Hairdryer would have been more appropriate there

It once got expelled from a recording session because the fan noise was bothering the engineer. I had to put the amp head outside the door and use a long guitar cord and speaker cable. Of course, you're supposed to play it loud enough that you can't hear the fan, but that day I was using it with a dummy load to record some bass parts.

[J M Fahey]

You should have told that engineer that when you mike a dummy load, said mic often picks some other noises too, it's normal.

[Steve Conner]

Oh, so that's what we were doing wrong! Maybe he should have plugged the dummy load into a DI box instead of trying to mic it.

On the original topic:
2x6L6 + 2x6V6 = about 450mA of current cranked up full. But they won't be cranked up full if you drop the supply by 200V. So let's say the current draw drops to 250mA when the amp is cranked in low power mode.

200V * 0.25A = 50 watts of heat.

I like to operate transistors at a heatsink temperature no hotter than 75'c. It gives some headroom for the thermal resistance of the insulating washer, and the transistor's own junction-to-case thermal resistance. So if your ambient is 25'c that means you need a 1 C/W heatsink. (50 watts, 50 degree rise.)

If you can't fit that on the PCB, and can't make ventilation holes to give it a good airflow, then you'll have to use the chassis.

Now pentodes and beam tubes are constant current devices. That means dropping the plate voltage won't make much of a difference to the power output: it's the screen voltage that does it. You could leave the plate voltage and only drop the screen voltage, the power reduction would be the same and the heat problem would be much reduced.

[lowell]

Thanks for the info everyone.

When the amp is in low power mode I also have a relay adjusting the bias to make all 4 power tubes at 90-100% dissipation. My goal with this setting on the amp is to have earlier power tube distortion.

The mosfets are IRF644 and are in a TO-220 package. IRF644PBF Vishay Semiconductors MOSFET Power

The chassis IS aluminum so that's good and I've used silpads and insulating washers. Up to this point they are in 2 amps that I've been playing 4 nights a week with for 2 years. The chassis mount has proven to be stable, however I'd like to rid of those 6 extra wires and 12 solder points.

I'm not quite clear on how to calculate that heatsink size. What is 1 C/W? and 50w at 50' rise? I think that 1 C/W is how to calculate the heatsink size and/or mass right?

[Steve Conner]

Degrees C per watt is how heatsinks are sold in electronics catalogs. It's a measure of the efficiency of the heatsink: how much its temperature rises above ambient when 1 watt of heat is dumped into it. The better the heatsink, the smaller the figure.

A heatsink that gives 1 degC/watt in free air with no fan, fins vertical, is about the size of a couple of packets of cigarettes. If it's inside a chassis with the fins horizontal and no holes for air flow, it would need to be bigger. How much bigger I can't say.

If you have an aluminium chassis, I would use it. How long does it take to make those 6 wires and 12 joints? How much does a 1'C/W heatsink cost? Time is money, right? And you know the chassis mount solution already works, it has a 2 year track record, but changing to separate heatsinks is introducing an unknown.

There is an equation somewhere to calculate the thermal resistance of an aluminium chassis when it's used as a heatsink, but I forget where.

[lowell]

Ok thank Steve I think I get it. I'll look into how big the hs would need to be. My 6V6 are cathode biased and the 6l6 are fixed bias. I think that may change the figures you calculated above. Also the 90% dissipation factor even when in low power mode. How would you calculate the heat dissipation? 200v * screen current for 4 tubes? And 200v * plate current for 4 tubes?

[J M Fahey]

Exactly like that=
(200V * screen current) + (200V * plate current)= "X" watts

[R.G.]

There is an equation somewhere to calculate the thermal resistance of an aluminium chassis when it's used as a heatsink, but I forget where.

I've been through that calculation, and it was so wildly off that I actively forgot it. It's by far simpler to put a power resistor or the transistor you're going to use in place on sink, heat it with DC volts and amps to some convenient power close to the actual, and measure the temperature with a thermistor, LM35, or other temp measuring device on the transistor/resistor. Faster, more accurate. The only disadvantage is that the answer doesn't help you much on the next problem, but it gets the job done.

For instance... the U-channel heat sink inside the Thomas Vox Beatle amp is about 7" long, 3" wide and 1" deep, and has a measured thermal resistance of 2.7C/W in free air. It's a miserable excuse for a heat sink, even if its surfaces are oriented vertically; it's chief virtue is that it was cheap. A flat horizontal surface is much worse for convection heating unless there's a fan or some source of chimney/convection to drive air over it. Otherwise, air is heated in a film over the surface and rises in a column in the approximate center of the horizontal area, and the hottest part of the sink gets the least moving air.

[lowell]

OK so my 6v6 screens probably draw around 25ma each=50ma. The 6l6 screens probably draw around 50ma each=100ma. I'm guessing at the 6v6 figures based on their 14w rated maximum dissipation, and using a figure from another thread where in a fender amp the 6l6 screens were measured at 50ma.

so for the screen supply zener that makes 150ma nominal. 200*.150=30w
If ambient temperature is 25'c and maximum temperature is 76'c that gives:
76-25=51
51/30=1.7
so is this then my ideal heatsink? - 1.7/W

[lowell]

OK so my 6v6 screens probably draw around 25ma each=50ma. The 6l6 screens probably draw around 50ma each=100ma. I'm guessing at the 6v6 figures based on their 14w rated maximum dissipation, and using a figure from another thread where in a fender amp the 6l6 screens were measured at 50ma.

so for the screen supply zener that makes 150ma nominal. 200*.150=30w
If ambient temperature is 25'c and maximum temperature is 76'c that gives:
76-25=51
51/30=1.7
so is this then my ideal heatsink? - 1.7/W

[J M Fahey]

Yes, your math is right.
The problem is you'll *never* have 25ºC inside a tube chassis .
I'd stick a thermal probe to my multimeter and measure actual chassis tempereture to begin with, then proceed.
As RG said, it pays to actually measure what you have and use that "real" data.
I always take that path, because I can't get useful data for many things.
Anyway, a mix of theory plus measuring rules.
If you wish, I can tell you how to make an accurate temperature sensor, about 2$ cost, but you'll have to calibrate it yourself.(no big deal)