1971 Marshall Plexi Cooks Grilled Cheese Sandwich | AI Design
I'm turning a 1971 100-watt Marshall Plexi into a grilled cheese press by running its speaker output into a cooking surface instead of a speaker cab.
The problem: a stove burner puts out about 1,000 W and the Plexi makes about 100 W on a good day. If the load ever opens, the amp's 50-year-old output transformer is toast.
The first design used three 300 W aluminum-clad resistors on an insulated aluminum plate, for about $270 and 10 to 15 hours of work. Then Claude found a better load: a thrift-store 1500 W griddle. Its nichrome element measures about 9.5 ohms, close enough to the 8-ohm tap, and it's already built to run hot and bonded to a cooking surface. The plan is to pull the thermostat and pilot light so the load never opens, add a speaker jack with a 100-ohm backup resistor across it, and clip on a thermocouple to watch plate temperature. A resistor divider and op-amp line out feed an impulse response so you still hear the guitar. With 100+ W going into an insulated plate, that's about 10 minutes per sandwich. Total is around $50 and 3 to 5 hours, tested on a cheap tube amp first.
Should I build it? Comment BUILD IT if you want to see the real thing. I'm not building it without comments.
The real builds: https://lardrocket.com
The AI in this video is Claude by Anthropic. Her voice (ElevenLabs) and the avatar (HeyGen) are AI-generated. The engineering conversation is real and unscripted.
Chapters
- 0:00 Marshall Plexi Sandwich Press
- 0:31 100 Watts vs. a Stove Burner
- 1:55 Plate Mass and Convection Losses
- 4:17 Clad Resistor Derating and Cost
- 5:12 Saving the Output Transformer
- 7:30 Cooking With Tube Waste Heat
- 9:00 Griddle Element as Dummy Load
- 11:07 Testing on a Cheap Tube Amp
Full Transcript
Brandon: Can you hear me?
Claude: Loud and clear.
Brandon: So I'm thinking that I'm hungry and that I would like a grilled cheese sandwich, but the best way to cook it is probably to use a 1971 Marshall Plexi guitar amps output.
Claude: (laughs) A vintage Marshall as a sandwich press. (sarcastic) Collectors are going to LOVE this. Is the sandwich getting cooked by a resistive load on the speaker output, or are you just setting it on top of the power tubes?
Brandon: That's a great question because the power tubes could probably cook it pretty good. But no, I'm actually thinking a resistive load, like a power soak that also cooks food.
Claude: Okay, that's at least electrically sane. Is this the fifty watt or the hundred watt Plexi? Because a stove burner is around a kilowatt, and this thing is going to be pushing a tenth of that on a good day.
Brandon: Let's be positive and assume a good day, but it is the 100 watt version of the Plexi.
Claude: Good news, then. The sandwich itself only needs maybe twenty kilojoules, so a hundred watts is about three and a half minutes if you don't lose anything, which you will. What are you feeding the input to hold it at full power that long, a sine generator or you playing an A chord?
Brandon: This would be best if I was playing guitar, but we can feed a square wave. I just distort the hell out of it first with some kind of pedal or whatever. That's no big deal. So let's say a square wave input.
Claude: Square wave actually helps you. Same peak voltage, but RMS equals peak instead of point seven oh seven, so you might see a hundred forty watts before the supply sags. Heads up, that's brutal on the EL34 screens for minutes at a time. What's the cooking surface, a chassis-mount resistor bolted to a plate?
Brandon: Yeah, some kind of resistor that we could thermally couple to maybe an aluminum plate. I'm not sure.
Claude: Then watch the plate's thermal mass. A quarter inch thick, six inch square aluminum plate weighs about four hundred grams, and getting it to a hundred eighty C eats nearly sixty kilojoules, so seven minutes of preheat before the bread touches it. Are you cooking one side at a time, or a clamshell press with both sides heated?
Brandon: One side at a time, definitely.
Claude: Then convection is your real enemy. An open six inch plate at a hundred eighty C sheds around forty watts per face to still air, so if the underside isn't insulated, you're losing half your amp before the bread gets any. Were you planning to box in the bottom?
Brandon: Yeah, definitely. It needs to be insulated, maybe sealed in some kind of epoxy, maybe. Not sure.
Claude: Not epoxy. Most of it softens and starts smoking around one fifty C, right where you're cooking, and a thin coat barely insulates anyway. An inch of ceramic fiber board or calcium silicate underneath does the job. Would you put a lid over the sandwich while it cooks?
Brandon: Yeah, that's allowed.
Claude: Then this actually closes. Insulated bottom, a lid on top, losses drop to maybe fifteen or twenty watts, and you've got a hundred-plus watts going into food. Thinner plate, faster preheat. Is this a pure dummy load, or do you want some of the signal still going to the speaker cab?
Brandon: So I'm thinking I don't have to use a real speaker cab, but people need to be able to hear me play guitar. So maybe I would run the output to some kind of op amp circuit that lets me use that signal from the amp to hear it, but it wouldn't contain any power.
Claude: Easy part. You're swinging about thirty-three volts RMS across eight ohms, so a resistor divider knocks it down about thirty to one, an op amp buffers it, and that feeds a PA. But a raw amp output with no speaker sounds like an angry bee. Do you care if it sounds like a real four by twelve?
Brandon: I will run through an impulse, so that's not a problem.
Claude: Good. Then here's the one that bites. Those aluminum-clad chassis resistors are rated with a cool heatsink, and they derate to zero around two hundred C. You want the plate at one eighty. What surface temperature were you actually picturing for the bread?
Brandon: I have no idea what temperature that bread cooks.
Claude: Medium heat on a pan, so about a hundred fifty to a hundred seventy C at the surface. Browning gets going around a hundred forty. At a hundred fifty, a big three hundred watt clad resistor is only good for maybe seventy-five watts, so you'd need two or three ganged on the plate. Want to price that?
Brandon: Yeah, let's price that.
Claude: Three three-hundred-watt clad resistors, two point seven ohms each in series for an eight ohm load, run about fifty bucks apiece, so a hundred fifty. Aluminum plate, twenty. Ceramic fiber board and a lid, maybe forty. High temp wire, binding posts and a thermal cutoff, another thirty. The line out and op amp stuff is under twenty. Call it two hundred seventy dollars, and ten to fifteen hours of build and testing.
One real hazard. If a resistor cooks itself and fails open, that Plexi is suddenly driving no load at full output, and that can kill a fifty-year-old output transformer you can't replace with an original. Wire the thermal cutoff to kill the signal, not open the load.
And budget a matched quad of EL34s, roughly a hundred dollars, because you're making grilled cheese with them.
Brandon: okay so let's go back up just a second here so you're saying i need to cut off the input signal the actual guitar in the event that one of the power resistors goes open did i hear you correctly
Claude: Mostly, yes, but there are two separate things. The thermal cutoff is there to stop the resistors from ever getting hot enough to fail, and when it trips, it should mute the guitar signal going in, never disconnect the load. A disconnected load at full output is exactly the thing you're trying to avoid.
But if a resistor fails open anyway, muting the input after the fact might be too late. So the cheap insurance is a backup resistor wired straight across the output, something like a hundred ohms rated a few watts more than you'd think. Normally it barely carries anything. If the main load opens, the amp sees a hundred ohms instead of infinity, which keeps the flyback spikes from arcing the output transformer while you get your hand off the strings.
Brandon: Okay, so the big secret here is safety so I could protect this like $4,000 irreplaceable guitar amp I have.
Claude: (chuckles) Yeah, the sandwich is the cheap part. The engineering here is mostly about not turning a four thousand dollar amp into a lesson. Thermal cutoff, a backup load across the output, and keep melted butter off the resistor terminals, because a short isn't great for it either. Are you sure you don't want to test this on a cheaper amp first?
Brandon: I might test it on a cheap amp, but the same strategy should apply. So earlier you mentioned how hot this amp can get because of the tubes. Can we put this amp in some kind of enclosure to maybe help harness some of that heat and even use some for the sake of cooking the grilled cheese?
Claude: Tempting, because the amp's waste heat is actually bigger than its output. Four EL34s at full tilt dump maybe a hundred watts from the plates, another forty from the heaters, plus the transformers, so call it two hundred watts of heat.
But it's low grade heat. The air coming off the chassis is maybe fifty or sixty C, and you need a hundred fifty to brown bread. Heat only flows downhill, so it can warm the sandwich, it can't toast it.
And boxing the amp in is how you kill it. The filter caps lose roughly half their life for every ten degrees hotter, and the output transformer is already working hard. Seal it up, and you're cooking the amp, not the cheese.
The most I'd do is duct the exhaust air over the lid to keep the top warm. Free, harmless, and it'll look great on camera.
Brandon: It almost sounds like you're trying to talk me out of this. Why?
Claude: (laughs) No, I like this one. The resistor plate idea actually works, a hundred-plus watts into an insulated plate cooks a sandwich in about ten minutes. I'm only talking you out of the enclosure, because that's the one part where the physics says no and the amp pays for it.
Brandon: Yeah, we don't want the amp to pay for it. Hmm. So I wonder if there are different resistor choices that would be cheaper than what you mentioned already.
Claude: Yeah, and this one's kind of perfect. A cheap single burner hot plate or electric griddle. A fifteen hundred watt, hundred twenty volt element is about nine and a half ohms, which is close enough to an eight ohm tap. It's already nichrome, already rated to run glowing hot, already bonded to a cooking surface. Twenty to thirty bucks at a thrift store, and it replaces the clad resistors, the plate, and most of the insulation.
Two things to check. Measure the cold resistance, since nichrome barely drifts with temperature, so what you measure is what the amp sees. And check the inductance at a few kilohertz, because a coiled element looks a little reactive, and you don't want it rising too far above eight ohms in the upper harmonics of that square wave.
Obviously, rip out the thermostat and the mains cord first.
Brandon: Whoa! So we could basically just take a griddle, chop off the mains power cable, and put a speaker input jack and plug the amp into it. That might actually work.
Claude: It really might. But the thermostat is the part that'll bite you. It regulates temperature by opening the circuit, which means it would disconnect the load from your Plexi every time it cycles. Pull it out completely, along with any indicator light, and wire straight to the element. Then add the speaker jack and your backup resistor across it.
Brandon: So this is suddenly a fast project. I just have to manage the safety side to protect the guitar amp.
Claude: Pretty much. Now it's maybe a weekend afternoon, three to five hours, and around fifty bucks with the backup resistor, the jack, and some decent speaker cable. Most of the time goes to measuring the element and testing on the cheap amp first.
The safety list is short. Thermostat and pilot light gone, hundred ohm backup across the jack, and a cheap clip-on thermocouple so you can watch the plate temperature instead of guessing. The only thing left wearing out is the power tubes, and that's what they're for.
Brandon: So as far as testing this on a cheaper amp, I guess it does need to be a tube amp. I'm not real versed on tube amp electronics like I am solid state.
Claude: For the protection testing, yes, it should be tube. Solid state amps don't care about an open load; they hate shorts. Tube amps are the opposite, because the output transformer throws a big flyback spike when the load disappears, and that's the failure you're guarding against.
But you can do most of the work with gear you already know. Measure the element's impedance with a generator and a solid state power amp, check how hot it gets at a hundred watts, dial in the cutoff. Then the cheap tube amp only has to prove nothing weird happens.
One sentence of warning. A tube amp runs four or five hundred volts DC inside, and the filter caps hold it after you unplug, so don't go poking around in there.
Do you already have a cheap tube amp lying around?
Brandon: Yeah, I've got many tube amps laying around. I'm sure half of them are broken, which would give me a great chance to get shocked by those filter caps.
Claude: (laughs) Perfect, a whole pile of capacitor roulette. If you open any of them, bleed the caps through a resistor on clip leads and check with a meter before touching anything. Then grab one that actually works, ideally one around fifty watts or more, so the griddle's getting a real test.
Brandon: So it sounds like this project is super doable. It sounds like I really just need to plug a Marshall Plexi, which is a very, very well desired or highly desired amplifier. I need to plug it into a griffle, griffle, griddle. So that's the big question, boys and girls. Should I make this? And if you want me to make it, you got to leave a comment down below to tell me. And I will build that on my Lard Rocket channel. All right, let's wrap this up.