Trigger
An oscilloscope draws one line. Everything it is good for comes from deciding when to start drawing it.
New to oscilloscopes? Start here
An oscilloscope draws a picture by sweeping a spot from left to right across the screen while the signal moves it up and down. One sweep is one line, and a steady picture is many sweeps drawn on top of one another, fast enough that the eye holds them together.
So the picture only stands still if every sweep starts at the same place on the signal. Start them at random moments and each lands somewhere different, and the screen shows a smear. That is the whole problem, and the rest of this page is the answer to it.
A trigger is the rule that decides when to start: begin the sweep when the signal crosses a chosen voltage, the level, going the direction you chose, the edge. Hold-off is a dead time after each sweep, so the next cannot start partway through a shape that has not finished repeating.
The sweep
sweeps 50start phases 50phase spread 0.964trace spread 2.000
50 pictures on top of each other. The sweeps began at 50 different points in the signal, so the same shape is drawn 50 times in 50 different places.
1 Arrive: a signal that repeats, and a beam with nothing telling it when to start
2 Compare: a level you choose and an edge you choose, and nothing else
3 Sweep: that one comparison starts one traversal, and the trigger goes deaf until it ends
4 Overlay: fifty sweeps land on each other only because they began at the same phase
Checked when this page loaded: 8 combinations of signal, level and edge. Triggered, with hold-off at 0.50 periods, 8 of them put every sweep at one phase and the traces exactly on each other. Free-running, 8 of them smeared the traces by more than a quarter of the signal.
Two answers came first, and both need something you may not have
The oldest scopes had a sweep that ran continuously at a rate you set by hand. You turned the knob until it nearly matched whatever you were looking at and the picture crawled slowly instead of tumbling. It could not show you anything that failed to repeat, because there was nothing for the sweep to nearly match.
Wartime radar work produced the second answer. The sweep sat dormant with the beam cut off until a pulse arrived from outside, then ran once at a calibrated speed and stopped. A radar set has that pulse, because it is the thing sending it. The DuMont 248, sold in 1945, had this and it worked.
What it needs is a wire to whatever is generating the signal. Set that beside the ordinary situation of standing in front of a circuit with a probe in your hand and no idea what is making it do this.
The invention is one sentence long
A triggered sweep, in the standard description, has a circuit that develops the driven sweep's drive pulse from the input signal
. That is all of it. The signal generates its own start pulse, so the scope needs nothing but the probe.
What that circuit does is compare. You give it a level and a direction, and it reports the moment the signal crosses that level going that way. The moment is a phase, and a phase is arithmetic, which is why this page can measure the thing it is claiming instead of drawing a blurry picture and telling you why it is blurry.
Hold-off is the part that looks unnecessary
Comparing is not quite enough. A pattern that repeats can cross your level several times before it repeats, and each of those crossings is a perfectly good trigger. Take the word above: three of its transitions are rising, so a scope that re-arms as soon as a sweep ends will start the next one on whichever crossing comes round next, and you get three pictures of the same word stacked in three different places.
Hold-off is a deliberate deafness after the sweep. Set it long enough and the re-arm lands past the rest of the pattern, so every sweep starts on the same crossing. Turn the knob on the machine above and watch the count of start phases fall to one.
It does not fall and stay there, which is the interesting part. The settings that lock the picture are bands rather than a minimum: past the top of a band it comes apart again, and a longer hold-off is not a safer one. What matters is where the re-arm lands relative to the pattern, not how long it was. That is why the knob is turned by eye, and the page computes the bands and prints them rather than telling you to turn it.
What holds a picture still, and what does not
Sameness of start phase is the part the trigger is responsible for, and this page measures it. It is not the only thing involved: the phosphor holds a trace for a while after the beam has gone, and the signal has to repeat often enough that the next sweep arrives before the last has faded. Neither of those is here. A scope with a perfect trigger and a phosphor that decayed instantly would show you a picture flashing once per sweep, and that is a real instrument problem rather than a trigger problem.
1946 or 1947 depends on what you are dating
Howard Vollum and Jack Murdock are credited with the triggered-sweep oscilloscope in 1946. The instrument you could buy, the Tektronix 511, was introduced in June 1947 and made until 1953. Both years are correct about different things, which is why the date on this page is written as a pair.
The 511 sold for $795, its advertised price. The competitor it is usually set against is the DuMont Type 248, which really is two-piece: the scope weighs 60 lb and its separate power supply weighs 110 lb. The figure quoted for it is $1,800, and that number deserves to be read with its provenance attached, which is the note on the $1,800 further down.
The first version of the 511 could not show you the part of the waveform before the trigger. There was no delay line, so the edge you triggered on had already happened by the time the beam started moving, and that was one of the first things added.
These ran in this browser when the page loaded. Each claim, whether it held, and the number behind it.
| claim | held | measured |
|---|---|---|
| a sine crosses zero going up exactly 1 time per cycle, at phase 0.0000 | yes | found by search and interpolation rather than by knowing it is a sine; the same routine runs on the repeating word |
| over 50 sweeps the triggered scope starts at 1 distinct phase and the free-running one at 50 | yes | one phase is one trace drawn 50 times, which is a still picture; the free-running sweep draws 50 different ones on top of each other |
| the triggered starts span 0.0000 of a cycle against the free-running sweep's 0.96 | yes | the smallest arc containing every start; zero means they are identical, and the external-trigger scope also manages 0.0000 because it has a wire |
| the word has 3 rising edges and gives 3 overlaid pictures with no hold-off; 76 of 201 knob settings bring it to one | yes | the first is 0.23 periods. Hold-off is deafness with a duration, and it is tuned rather than turned up: at the maximum this signal is smeared again |
| at a level of 0.9 the sine still crosses 1 time, at phase 0.178; at 1.5 it crosses 0 times and there is nothing to trigger on | yes | moving the level moves where the trace starts, and past the peak the one-shot never fires; that is a real scope's behaviour and not a guard added here |
What is real here, and what is not
There is no beam and no phosphor
No electron optics, no deflection plates, no persistence, no bandwidth limit, no probe loading, no vertical amplifier and no noise. This studio does not simulate physics, and every one of those is physics. What is modelled is the decision about when a sweep starts, because that part is logic.
Traces that land on each other land exactly on each other, and real ones never do
When the model reports a trace spread of zero it means zero, because two sweeps that trigger on the same crossing are computed from the same number. A real instrument has trigger jitter, noise on the level, and a comparator with a finite slew rate, so its traces are a band rather than a line. The zero here is a property of a model with nothing in it to shake.
Fifty sweeps is a number chosen to be looked at
A scope draws thousands per second and the count is set by the timebase and the signal, not by anyone's preference. Fifty is enough that a smear looks like a smear and few enough that you can pick out individual traces in it.
The crossings are found by searching, not solved
The page samples one period at 24,000 points and interpolates inside whichever interval the crossing fell in. That is the only approximation in the model and it is good to about one part in 24,000. Solving the sine analytically would be exact for the sine and useless for the word.
The free-running sweep here is not tuned
Its period is 1.537 times the signal's, and it stays there. A real synchronised scope had a knob, and the skill was turning it until the picture crawled slowly enough to read. What that knob could never do is show you something that did not repeat.
Where the $1,800 comes from, and why it has not moved
Both prices on this page trace to one place. TekWiki’s 511 article gives the price as $795 and then, in the same sentence, “compared to $1,800 for a two-piece DuMont scope that was mostly pre-war technology”, attributing it to TekWeek of 23 March 1984. That is a Tektronix internal employee newsletter recalling a competitor’s price thirty-seven years after the fact, and TekWiki records in its own footnote that the same article gives the 511 as $595 against an advertised $795. A source that is wrong about its own company’s price is not a strong witness for a rival’s.
An outside reader put the DuMont nearer $1,870. Nothing here moved, because that figure has no source either: it appears in no catalogue, price list or period advertisement this page could find, and swapping one unsourced number for another is the same defect wearing a different number.
What was searched, so nobody repeats it: the Electronics Buyers’ Guide of June 1947, which turns out to be a directory of manufacturers and carries no prices at all; DuMont’s own Cathode-Ray Equipment catalogue, eighth edition 1951, which is technical and priceless in the literal sense; DuMont’s house journal The Oscillographer, whose archive begins in 1950, three years too late; the vintageTEK and Oscilloscope Museum pages for both instruments; and a general search for the $1,870 figure, which returns only restatements of TekWiki. The Type 248’s two-unit construction is documented and its price is not.
Sources
- TekWiki, 511, which is where the $795 and the $1,800 on this page both come from, and which attributes the comparison to TekWeek of 23 March 1984 while noting in a footnote that the same article says $595 for the 511. It is the origin of the sentence, not a second witness to it.
- Oscilloscope Museum, DuMont Type 248A, for the two-unit construction: a 60 lb scope and a 110 lb power supply joined by a multi-conductor cable. This is what “two-piece” means here, and it is the part of the comparison that is documented.
- Secondary, and used as such: Oscilloscope history, for driven sweep, the DuMont 248 and what triggering added
- TekWiki on the Tektronix 511, introduced June 1947 and produced until 1953
- The Tektronix museum on the 511, including what the first version could not do
- Secondary, and used as such: Murdock, Vollum and the founding of Tektronix
- No primary document is cited for the prices. The $795 for a 511 and the $1,800 for the DuMont come from the TekWiki entry above rather than from a catalogue or a price list. The honesty note on the page sets out the provenance and what was searched.