I just realized that the typical traces on a 1oz copper PCB board dont even come up to having as much copper as a 29 AWG wire! If building an amp with point-to-point wiring we would never even consider using 29 gauge wire, we'd use 20 or 22 gauge typically. Here are the numbers I see for AWG vs PCB trace cross sectional area: First take a look at this chart of wire gauges and their cross sectional area (last column is the area first column is the gauge)
1 oz copper PCB board has a copper thickness of .0347 mm (1.37 mil). If we have a typical trace on the board that is lets say 2mm in width. 2 x .0347 gives a copper cross sectional area of only .0694 mm. I see many PCB's here that use 1oz copper and have traces of 2mm or even less for tube projects. OK now here's the kicker... Note that a cross sectional area of .09694 in the example above translates to 29 AWG wire! Huh? Is my math wrong? How can this be? A tube amp wired up with 29 gauge wire seems ridiculous, I cant even strip 29 gauge wire without damaging it. And nobody would consider doing a point-to-point amp with wire that thin. Lets see how much PCB copper trace we need to equal a 22 or 20 gauge wire... 2oz copper has a thickness of 2.8 mil (.07112 mm). So a 2mm wide trace would have a cross sectional area of 2 x .07112 = .14224 but that only puts us up to 25-26 gauge wire! OK, so I have to increase my track width to a little over 4.5 mm using 2oz copper to get me to the equivalent copper of a 22 gauge wire? Now at JLCPCB the price triples when you go from 1oz to 2oz copper, and increases again going to 3oz.
What is going on here? Is PCB copper somehow different than wires? Most of the PCB's offered here by folks are 1oz copper, shouldn't we all be using 3oz copper boards for tube work?
1 oz copper PCB board has a copper thickness of .0347 mm (1.37 mil). If we have a typical trace on the board that is lets say 2mm in width. 2 x .0347 gives a copper cross sectional area of only .0694 mm. I see many PCB's here that use 1oz copper and have traces of 2mm or even less for tube projects. OK now here's the kicker... Note that a cross sectional area of .09694 in the example above translates to 29 AWG wire! Huh? Is my math wrong? How can this be? A tube amp wired up with 29 gauge wire seems ridiculous, I cant even strip 29 gauge wire without damaging it. And nobody would consider doing a point-to-point amp with wire that thin. Lets see how much PCB copper trace we need to equal a 22 or 20 gauge wire... 2oz copper has a thickness of 2.8 mil (.07112 mm). So a 2mm wide trace would have a cross sectional area of 2 x .07112 = .14224 but that only puts us up to 25-26 gauge wire! OK, so I have to increase my track width to a little over 4.5 mm using 2oz copper to get me to the equivalent copper of a 22 gauge wire? Now at JLCPCB the price triples when you go from 1oz to 2oz copper, and increases again going to 3oz.
What is going on here? Is PCB copper somehow different than wires? Most of the PCB's offered here by folks are 1oz copper, shouldn't we all be using 3oz copper boards for tube work?
I fail to see the point of your post. Design requirements are always there, p2p or pcb traced. You are having problem with the pricing or?
You have to keep the current handling and voltage drop in mind when laying out the heater traces. For all other traces, clearances and insulation are usually a bigger issue than track width. I used the KiCad calculator a lot to check required trace widths and required clearances when I laid out a PCB for a valve circuit.
Creepage and clearance are the normal issues in tube circuits, not the pcb copper thickness, due to small plate currents. Gauges for tube circuit component leads and interconnecting wires are chosen more for ruggedness and convenience
than for current handling. I would never use smaller than 22 gauge wire, except in tone arms.
than for current handling. I would never use smaller than 22 gauge wire, except in tone arms.
Last edited:
One important issue is mounting of power tubes. Circuit board mounting has several issues, heat will degrade the
board, inserting and removing tubes will make large stress on the soldering and board while off board
mounted tube sockets will take stresses better and will allow socket replacements easier. In addition
off board sockets will allow other mouthing where heat issues is more manageable.
board, inserting and removing tubes will make large stress on the soldering and board while off board
mounted tube sockets will take stresses better and will allow socket replacements easier. In addition
off board sockets will allow other mouthing where heat issues is more manageable.
OK respectfully I'm not asking about any of these things. All I'm doing is computing the cross sectional area of the typical PCB trace of 2mm wide by 1oz copper. And compared that number to the cross sectional area of the nearest wire gauge from the chart. And I came up with that a 2mm wide by 1oz copper PCB trace is equivalent to a 29AWG cross sectional area! Then I said to myself whoa! You mean a 2mm wide trace on a PCB board would be the same as if I used a crappy piece of 29AWG wire to connect those same parts with point-to-point. We would have to go out to 3oz copper boards and wider traces just to get to the equivalent of 22AWG doing point-to-point.
As long as the current level is within the current handling capacity of the wire, the size is fine.
Many have used wire wrap wires for tube prototyping, which is normally 28 - 30 gauge.
So thin that you need a special wire stripper for it. The voltage would be the primary concern then,
so use sleeving, or keep the wires away from conductors or metal with differing voltages present.
There's nothing crappy about WW wires, they've gone to the moon.
Many have used wire wrap wires for tube prototyping, which is normally 28 - 30 gauge.
So thin that you need a special wire stripper for it. The voltage would be the primary concern then,
so use sleeving, or keep the wires away from conductors or metal with differing voltages present.
There's nothing crappy about WW wires, they've gone to the moon.
Last edited:
The topic is quickly going off the rails subject of cross sectional area of wire vs traces... But for tube sockets I find it better to route a large hole in the PCB and use a chassis mount socket on standoffs, then wire from the pins to pads on the board. This way no heat from the pins transmits to the board and the big hole under the socket gives lots of airflow, and you can bolt the chassis socket saddle directly to the chassis so from the outside it looks like a point-to-point build but the socket is really on standoffs over a PCB.
Why 2mm? Is your board crowded?
You often see unnecessarily narrow traces on a board with vast spaces between. The copper is already on the board. You might as well leave on whatever lowers resistance - unless you want it of course.
The bigger problem with tube sockets on the boards is that with a hard connection to the board, the pin connectors have no play and it becomes a lot harder to insert glass based tubes without cracking the glass. If you do want to solder the socket directly, it's a good idea to first insert a tube with straight pins into the socket and then mount and solder the socket with tube installed. Then at least the socket receptacles are in the right positions.
You often see unnecessarily narrow traces on a board with vast spaces between. The copper is already on the board. You might as well leave on whatever lowers resistance - unless you want it of course.
The bigger problem with tube sockets on the boards is that with a hard connection to the board, the pin connectors have no play and it becomes a lot harder to insert glass based tubes without cracking the glass. If you do want to solder the socket directly, it's a good idea to first insert a tube with straight pins into the socket and then mount and solder the socket with tube installed. Then at least the socket receptacles are in the right positions.
I just pulled 2mm out of my hat yes, because many boards I see run traces that do leave vast amounts of spaces between. Even if I picked 3mm or 4mm traces I'm still unable to hit the equivalent of 22AWG wire which is kind of the standard for point-to-point. Again just thinking of cross sectional area comparison of traces to wire. I see from charts that 22AWG can carry 7 amps in a chassis and 29AWG is still 1 amp or so. Both are well within the range for a typical circuit. I guess I just know all too well the frailty of 29 gauge wire.
People are right its all about the current handling of the trace. I was just taken aback because I never calculated the equivalent wire gauge of a known PCB trace before now. I was really surprised by how thin the equivalent wire is on PCB's, eye opening.
Cross-section is only a potential issue for heaters: tube circuits tend to work at high voltage/low current conditions (there are exceptions, in transmitters, welders, etc. but they are outside of the DIY realm).
Creepage and heat management are more of a problem, but they are perfectly manageable: TV sets from the sixties or seventies did it, not perfectly admittedly, but they were consumer products.
Professional products, from HP for example, had lots of tubes and semi's on a single two-layer PCB, and it didn't pose any particular problem.
Creepage and heat management are more of a problem, but they are perfectly manageable: TV sets from the sixties or seventies did it, not perfectly admittedly, but they were consumer products.
Professional products, from HP for example, had lots of tubes and semi's on a single two-layer PCB, and it didn't pose any particular problem.
The high voltage and very close traces can also be an issue. We’re talking very high voltages! Under 500 won’t be much of an issue. Make sure heater traces are big enough for the higher currents needed. You might get tiny capacitances from parallel traces too. If you are worried about traces being too thin you can always solder coat the traces to beef them up.
It is unlikely to be about the copper. We pick copper first for the insulation we want to use, and second for the increased loss at hgh temperature. High performance speaker coil wire can double in Ohms when working rock concerts; we don't like that because power compression and generally double the rack-weight.
Even so, the trace could melt? Being stuck to a board, even poor paper-board, can carry off a lot of heat. Also most round wire insulation is PVC and most PCB-stuff is allowed higher temps. (And depending what AWG chart you have, the round-wire may be derated for multiple conductors in one conduit.
Also we pick road-worthy wire gauge for mechanical robustness. In electronics we rarely get down to minimum electrical gauge. On PCB, mechanical strength is taken care of in the board.
The voltage drop could be significant, mostly if the length is long. But PCB stuff is so expensive that wire traces are typically short, less than a foot. Notable exception for large TTL logic boards from the early 1970s, but these did have wide buses and even metal jumper-bus.
Even so, the trace could melt? Being stuck to a board, even poor paper-board, can carry off a lot of heat. Also most round wire insulation is PVC and most PCB-stuff is allowed higher temps. (And depending what AWG chart you have, the round-wire may be derated for multiple conductors in one conduit.
Also we pick road-worthy wire gauge for mechanical robustness. In electronics we rarely get down to minimum electrical gauge. On PCB, mechanical strength is taken care of in the board.
The voltage drop could be significant, mostly if the length is long. But PCB stuff is so expensive that wire traces are typically short, less than a foot. Notable exception for large TTL logic boards from the early 1970s, but these did have wide buses and even metal jumper-bus.
I would have to say that I am in the same camp as you. 2oz traces and thick board is what will get me to buy a product if it has to be on a circuit board. I may feel differently about boards using only SMD devices though.
Take a look at the ampacity table at this site:
https://learnmetrics.com/wire-gauge-chart-amp-wire-sizes
30 gauge wire is rated for 860mA.
https://learnmetrics.com/wire-gauge-chart-amp-wire-sizes
30 gauge wire is rated for 860mA.
I did this on the ground bus of a power supply board, when I designed the board I removed the mask off that one ground trace. Then used hot air and regular solder to coat the whole trace. It's a good thing the board company tin plated the whole exposed trace otherwise it would be hard to coat bare copper directly sticky-wise, you'd need a lot of flux and heat to layer a solder coating onto an untinned trace.
Are you finding yourself in situations where you find the traces to be inadequate for the signal currents you are running? If not it's pretty immaterial if they do the job and don't degrade the circuit or safety margins.
For those interested in mental estimation of track resistances.
https://www.edn.com/counting-square...e of 0.5 mΩ per square, and,x 0.5 mΩ = 3.0 mΩ.
So if a track is on 1 oz copper and the trace is 1 unit wide and 10 units long it will have a resistance of 0.5 milliohms x 10 so 5 milliohms.
If the track is passing 4 amps it will dissipate (power= current squared x resistance) power = 0.08 watts, the voltage drop will be V=IR so Vdrop = 20mV
Ken K
https://www.edn.com/counting-square...e of 0.5 mΩ per square, and,x 0.5 mΩ = 3.0 mΩ.
So if a track is on 1 oz copper and the trace is 1 unit wide and 10 units long it will have a resistance of 0.5 milliohms x 10 so 5 milliohms.
If the track is passing 4 amps it will dissipate (power= current squared x resistance) power = 0.08 watts, the voltage drop will be V=IR so Vdrop = 20mV
Ken K