Bridged T Attenuator Resistor power ratings

[Williamblake]

Dear tubeheads

i always wanted but now repairing a 150W no master volume beast i have to have a power attenuator. I would like to have the output adjustable and i wouldn't like to have to build another attenuator again. So this will become a 100W+ adjustable attenuator. I would go with the nicely drawn schematic taken from https://www.electronics-tutorials.ws/attenuators/bridged-t-attenuator.html but just like the other schematics i found it doesn't  mention any power ratings. Of course i couldn't come up with a formula for current or power dissipation so i was just guessing where to put in the big resistors. Begging you to let me stand corrected. Please look at my assumptions in the attached picture.

Regards
Jens

[Williamblake]

Maybe you already guessed it but if i am not totally mistaken one variable resistance goes down while the other goes up.

[jjasilli]

Personally, for bench work, I would build a purely resistive dummyload, possibly with a line out. 


The line out would go to a listening amp or a powered speaker like a computer speaker.  When probing an amp under test it makes noises, some good some bad.  It's helpful to hear this, but at low volume.

[sluckey]

Personally, for bench work, I would build a purely resistive dummyload, possibly with a line out. 


The line out would go to a listening amp or a powered speaker like a computer speaker.  When probing an amp under test it makes noises, some good some bad.  It's helpful to hear this, but at low volume.

+1... KISS applies here.

[Williamblake]

The right R1 will be 10W not 25W. I don't see why this is too complex since it wouldn't need external amplification. The 10-Way switch seems overkill but convenient, too and i already have it.

[jjasilli]

OK.  N.B.:  An amp can "put out" 2X its rated wattage with fully distorted output, in this case 150 X 2 = 300W.  The attenuator should be rated to handle at least 2X that = 600W.  Even more for cool running (or maybe use a fan).  Consider that a speaker might accidentally not be connected, causing the attenuator to handle the full load of the amp's output.  Also I'm suspicious of the attenuator's -0- position for that reason.  (Murphy's Law.)

The good news is that each resistor needs to handle only its share of that total wattage.  The bad news is that there's a lot of resistors, in various series-parallel configurations.  You need to compute the voltage drop / wattage for ea resistor in each of the SW positions.  A PIA, but doable.  Then you'll know the minimum wattage rating needed for each R.

Note that with the attenuator accidentally switched to a wrong (loud) position, a test signal through this amp into a speaker could permanently damage your hearing, and that of any bystander or pet.  There are a number of reasons to use a dummyload.


(Also, I haven't verified the design of this attenuator and am merely assuming it works as advertised.)

[shooter]

+1
as an experimenter myself, I've found "things like this", once it's working, a month from now, you'll probably only use 2, possibly 3 settings.  So I changed my experiments philosophy to "what I need", then modify for wants

[Williamblake]

I second the thought there are too many positions to actually use, but which are they? On the positive side this would be one (resistive) attenuator to rule them all and all the small 5W resistors are cheap and share power dissipation.

A load that is always present when the speaker output is not connected would be on the safe side and would also enable line out only operation but to be on the failprove side i would have to add a resistor with the full power rating. Maybe i will omit the 0dB setting and add a "burn it all switch" using the existing 100W resistor. Would be nice if the speaker out jack could switch that but then again you would have all the power on a switched speakerjack. Also everybody and their dog knows you have to have a load connected but could still trip on the speaker wires …

I want this attenuator to handle 125W from the resistor ratings. But all the resistors will be mounted to a heat sink or metal chassis so they will be able to handle more than that and also i can safely check on the actual power dissipation. I wouldn't play a 100W amp and burn it's power with this, the amps i would attenuate will be 5W to 15W but i want to plug this in a a 150W speaker output to have a load while testing for scratchy pots and stuff, i wouldn't do this continually.

I went through the schematic assuming individual resistor power ratings by simplifying and exaggerating load and below 5W i didn't investigate further.

I updated the schematic, right R1 is changed.

[PRR]

> The bad news is that there's a lot of resistors, in various series-parallel configurations.

A LOT of resistors in 8 different configurations.

What you need is "grad students". Low/no pay workers. This is sheer drudgery.

Most simulators don't tell you dissipations without excess shenanigans. The simplest way I see is to pretend the source is DC, do a DC analysis, in a sim which will show DC op-points on the schematic, still need to subtract two voltages, square, divide by R.

Poking at a few likely worst-cases, I don't see over 25W in any resistor. I likely missed a case. A KISS approach would be to build spread-out with all 25W parts (mostly gross overkill), plus a 8r 100W load, and fed it with a 24V 4A AC transformer or DC power supply. Run a while then put your nose over it, the hottest parts will warm your nose-hairs. (An IR thermometer is more 21th-century.) If any are HOT, consider a bigger part there.

[Williamblake]

I went with the numbers and build a resistor sub-chassis.

[Williamblake]

With the switch wired in i better check what i did before it gets too complicated.

[Williamblake]

So this is what it`s looking like. It´s a 4x5 rotary switch so i added a DPDT switch. Maybe someday i will give this better labelling and matching screws and there is some space left for maybe a speaker simulator and line out. But it already is very convenient and my neighbours and me are glad i built it. I will be getting a decent DMM with temperature sensor and give this some testing with 150W but as of yet it has only seen 5W.