"In God we trust; all others bring data" - Mike Bloomberg
I am speaking of course about tube guitar amps that hang their tubes upside down, and would seem to cook their chassis. Except they don't. We know they don't because this layout has survived for 70+ years while others have fallen away. But people have opinions and feelings. They don't seem to want to test them though. There is a whole new generation of enthusiasts though that have not caught on to the idea that "thermionic" refers to heat. These things need to be hot to work properly. But they are not halogen bulbs. They are not dangerously hot, and they are certainly not very fragile in the sense of the glass needing to be clean. Think of all the grimy hands that have shoved tubes into an amp on stage or in a van over the past 70 years after all. Vic who taught me basic repairs in his shop used to put on an oven mitt and change the pre-amp tubes while they were hot. He was on the clock after all. The borosilicate glass they use is good up to a max of about 514 deg F (Wikipedia). These amps don't need fans or holes cut in their cabinets by well intentioned folks who also don't have data. Ok, maybe the Vox AC100.. (which of course uses right side up tubes...)
Years ago I met a guy named Ted Weber on a Prodigy BBS about guitar amps. He was a powertrain engineer for GM, an electrical engineer by training. Kinda brilliant if I do say so. He posited the idea that this was poor design; I took the other side since I felt like I had history on my side. Reams of data. We have a saying in tech that in theory, theory and practice are the same but in practice they are different. This is one of those times. And he and I debated a bit and he conceded that despite seeming intuitively bad, it was in fact alright - the evidence is overwhelming that it is not only safe, but works so well that these devices soldier on decade after decade. I would also mention now that in my work doing repairs, there have been a number of occasions where I have deliberately left tweed amps on all night long (or more) to bake their fiber boards dry after being poorly stored led to issues. I have never seen damage from this of any sort. No failures, no fires, no nothing.
But rather than blather on about *my* feelings, here are some real world numbers. I have an embarrassing amp collection here, so I grabbed a couple of samples to test with. A '65 AC30TB combo, a '65 Super Reverb, a replica of a 1959 5F4-A Super Amp and a real 1952 Pro Amp - which has been hanging its tubes upside down for 70 years now.
I took their temp with an IR thermometer which was probably not super accurate; it had trouble with the chrome plating on the tweeds for one thing. I looked for the hottest spots on the chassis - typically where the output tubes were. This could be 40 or 50 Deg F hotter than the other end of the chassis. Hopefully these are close enough, and I did try to find the hottest readings I could. I did not sandbag this. I took a reading each hour for 3 hours to see how hot they got during a typical "rehearsal" or "gig".
All amps started out at 64 Deg F, ambient. The tweeds were measured top and bottom of chassis. Top was the hottest spot (near power switches) and bottom was in between the output tubes.
The AC30 was measured on the top panel in the hottest spot, near the middle.
The Super Reverb was measured on the rear panel above the output tubes
| 11AM | 12PM | 1PM |
AC30 | 102 F | 159 F | 160 F |
Super Rvb | 159 F | 159 F | 159 F |
Super Amp | 85 / 159 | 92 / 154 F | 92 / 174 |
5F6-A Bassman | 88 / 160 | 89 / 159 | 90 / 160 |
52 Pro Amp | 89 / 151 | 97 / 164 | 98 / 165 |
All temps in Deg F. 159 F is 70.5 Deg C
I did not have a way to measure the ambient temp inside the chassis, but I would like to do that next time with a thermocouple or something. it would also be helpful if others tried to duplicate this kind of test. Would the numbers be similar?
What I think I see in the data are that the Fenders all have large enough chassis heat-sinking the radiation from the tubes that they stabilize quickly and don't really get much hotter over time. The radiation is going out through the glass, 360 around the tube - where there is air. The chassis is warmed by the rising hot air and then any heat transmitted via the base and pins.
The Vox is an EZ Bake oven, but still does not get very hot in my case, although I am running sane B+ and dissipation. So YMMV.
Modern capacitors are rated at 85 Deg C, which is 185 deg F. (written right on them usually, or check the Mouser catalog) The very hottest spots on the outside did not reach this temp after hours, and I am very confident that the internal temp was significantly cooler. The caps are not touching the chassis, they are soldered to the boards which don't conduct well. Plus the pre-amp end of the chassis stayed around 80-90 deg F. Caps do de-rate at high temps so it would be good to understand more.
Modern resistors are also rated to 85 Deg C. Same as above I suppose.
Modern transformers are insulated to withstand a range of temps depending on the insulation used: (from Belling-Lee, p#43. linked below)
Class A: 105 Deg C (221 Deg F)
Class B: 130 Deg C (266 Deg F)
Class H: 200 Deg C (392 Deg F)
This is rated by the "hottest spot" temp on the inside. Insulation material life degrades with temp, so operating cooler is good for longer life. Insulation life is further reduced by vibration, humidity and corona (electrical, but the beer might not help either). Because all insulation is organic, there will be pores in it that should be blocked. This generally necessitates multiple layers of insulation. Electrons are lazy, they will take the easiest path out of wherever they are. All of our devices are simply frameworks to force them down the path we want in such a way that they do some work.
In general higher temps are bad but it depends on the insulation class. Industrial units can handle max internal temps of over 400 Deg F. If you have ever been near a 440 3-phase transformer they run hot and vibrate loudly. They are designed for this. But what can the one in your 1966 Bandmaster handle? It is one reason OEI picked Nomex (to resist high temps.) My working assumption here to be conservative is assume class A is used in retail/prosumer gear.
If I was a good engineer (and I am not), I would figure out the mass of the chassis vs the heat energy of the tubes being given off and figure out if there was something to understand there in the way or ratios or size or something. I am pretty sure there is a way to calc this nicely. I am also pretty sure (I have feelings too I guess!) that with modern CAD design we should be able to build in a nice "flue" effect in amps like the AC30 to draw enough air in the bottom to keep them cool via convection. We do it for laptops after all.
5F6-A calculation
First, how much heat do we think we are emitting? We need to look at how much power we draw at idle and convert that to BTUs.
We could add up all of the tube filaments and plate dissipations, OR we can measure the current draw at idle in aggregate and knowing with no input signal that every single watt is going into the atmosphere as heat; well that does seem easier.
Electricity | 1 kilowatt hour = 3,412 Btu |
1 BTU | heat to raise 1lb of liquid water 1 degree F |
5F6-A (Homebrew) B+ secondary is 315-0-315 for 120VAC input.
Mains Current draw: 2 Amps
Watts: (converted) V X I = W so 120 x 2 = 240 Watts
Watts to BTUs of heat: 1 Watt = 3.412 BTU. 240 Watts X 3.412BTU/Watt = 818.88 BTU
Now it is a little tricky because most of that heat is coming off of the rectifier tube and the output tubes and it doesn't come out in 1 direction or 1 spot. It is being radiated in 360 deg around the tubes, and above and below them. But then the radiated heat is rising. So we are losing a bunch to the air and the cabinet, but some of that hot air is rising. The older I get the more i am convinced that "It depends" is the only correct answer.
"The specific heat of a material is a measure of the amount of heat energy required to raise the temperature of a unit mass of the material by one degree Celsius (or one Kelvin). For steel, this property can vary slightly depending on the specific alloy and its composition. However, for most types of steel, the specific heat is typically in the range of 0.10 to 0.13 British Thermal Units per pound-degree Fahrenheit (BTU/lb°F) or approximately 420 to 500 Joules per kilogram-degree Celsius (J/kg°C)." (Link below)
So for a 3.6 lb steel chassis (https://www.mojotone.com/Tweed-Bassman-Style-5F6-Chassis see Specs tab) to get raised from 65 Deg F (room temp) to 159 Deg F what might we be able to infer?
Total energy being radiated: 818.88 BTU
Total mass to heat: 3.6 Lbs
Temperature change over 60 minutes*: +94 Deg F
I should do a new study and see how long it takes to create this rise in temp.
To be conservative we'll take the higher number, .13 BTU per Lb Deg F.
.13 BTU per pound X 3.6 Lb = .468 BTU per chassis Deg F
94 Deg F temp rise = 43.992 BTU
Now we have ~819 BTU available so this is only 5.3% of the available heat energy going into the chassis. That feels kind of low. There is another ~8-10lbs of transformers bolted to the chassis so that will factor in to some degree, although how well heat transfers into the chassis, aorund the to the back and up the 4 mounting bolts and then into the iron core is a question mark for sure.
If we pretend the transformers act as part of the chassis then:
Total mass to heat becomes ~13lbs and we need 1.69 BTU per chassis degree F for a total of 158.86 BTU to move it up 94 Deg F - which is up to 19.4% of the available heat energy. That feels closer to something intuitive. We'll need to measure this and see if it holds up. I am really not clear on the best way to estimate how much of the total heat we are actually sending to the chassis. Backing into that may be the cleanest method given all of the variables. I feel like these calcs are missing time and the amount of surface area the chassis has to get rid of its heat (it is being heat sinked by the air in the room it is in)
I am working with a new Flir thermal camera and starting to get some numbers but it will take a little time. The output tubes are somewhere around 150 Deg C or 302 Deg F around the plate. I'll get some newer better numbers up soon.
Sources:
Belling-Lee Book Download (also on Transformer Basics page with some other great books)
Specific Heat of Steel: https://www.proprep.com/questions/what-is-the-specific-heat-of-steel
Comments