I've read in various places that an incorrect signal of the wrong refresh rate can cause physical damage to a CRT monitor, but I've never found an adequate explanation of why this is so. Is this true? And if it is, how?
1I am surprised no one mentioned the most famous example, the PET killer poke. 6502.org/users/andre/petindex/poke/index.html– Maury MarkowitzJun 8, 2018 at 11:49
Horizontal deflection circuit is intended to make linearly increasing current in the deflection coil. At the end of ray scanning from left to right, the deflection coil should be re-magnetized in a short time by applying higher reverse voltage to it. Typical schematics for the circuit looks like this: http://repairfaq.cis.upenn.edu/sam/deflfaq.htm#dsbdc .
L2 is the deflection yoke and T2 is the flyback transformer, that is also usually used to get high voltage for the CRT anode.
During the linear ray moving, the switch is open and the constant voltage supply causes the current to raise linearly in both L2 and T2. When the switch closes, the magnetizing energy in the inductance of T2 with the help of the capacitor C1 makes a short resonant high voltage spike, that remagnetizes deflection coil L2 to the initial ray-to-the-left position.
At the too low horizontal frequency, the period the switch is open becomes too long, that causes the following:
- The flyback transformer T2, that is usually made of ferrite core with a little gap, may get saturated. Once its ferrite core is saturated, it virtually disappears from the circuit, thus greatly reducing the inductance of the T2 winding. As a result, in a very short time the current gets unboundly high and it could damage the switch or the power supply.
- Even if T2 is not getting saturated, still the voltage spike during switch off time may get too much (as it is determined by the energy absorbed in the inductance of T2 during the switch on time), thus breaking down the switch, the capacitor C1 or the snubber diode.
- A third option is to saturate the switch (usually an NPN transistor), which would lead to emitter-collector voltage rise and then to excessive heat dissipation at the switch.
This all could happen if the controlling circuit inside the monitor allows it to operate at the inappropriate frequencies. This might happen on older (and simpler) CRT monitors.
1The circuit probably doesn't FEED the coil a current ramp. It feeds it constant voltage and has a current ramp developed through the inductance. That current can climb higher than tolerable if the ramp is allowed to build up too long, no core saturation needed... Jun 5, 2018 at 0:04
1@rackandboneman: It's possible that a drive transistor might saturate before the coil does, but in either case the collector voltage on Q1 would rise even though it's supposed to be "on". I wonder how much cost is saved by not including a circuit to detect that and force a retrace? Jun 5, 2018 at 15:28
Wow, thanks! Switch saturation is also an option! [improved the answer]– lvdJun 5, 2018 at 18:46
1@lvd: A follow-option to #3 is that the transistor may melt and fail shorted, resulting in coil saturation and burn-out. Jun 5, 2018 at 19:12
For the common switched power supplies used in all but (probably) very old CRT monitors,I don't think coil burnout is an option. Usually SPS have overcurrent/shortcircuit protection. If there is no such protection, output shortcircuiting leads to its own switch(es) burnout and then the fuse finally burns.– lvdJun 7, 2018 at 7:18
I've witnessed a monochrome monitor (actually, two of them) fried through software.
I worked with a friend who had a weak grasp of ethics. He worked on an IBM PC/AT clone running DOS. It had both a regular CGA card driving a color CRT, and a Hercules card driving a monochrome CRT. I no longer recall why he had an extra monitor; I can only guess that it was a useful setup for debugging, since he could have the program's output shown on one screen, and debugging output shown on the other.
My friend had written a small TSR (Terminate and Stay Resident) program that would, on demand, capture the contents of a text screen to memory, and then recall that screen using a hot key. This was like the many so-called "boss key" programs available at the time, but instead of displaying a spreadsheet or some other generic work-like content, it displayed whatever had been previously captured, and the captured screen could be freely changed at any time by pressing a hot key. You can see how this might be an advantage for a programmer, who doesn't generally use spreadsheets, and who wishes to sometimes not work while at work.
As I recall, when the "boss key" was pressed, the program needed to save the screen contents, which might be in a graphics mode, and replace them with the saved contents, which were in text mode. I think (this was a long time ago) the program made the mode switch without the assistance of the BIOS, by directly hitting the CGA card's I/O ports. I think those ports had the ability to control the monitor's scan rate, as evidenced by the destruction the TSR caused.
My friend found out the hard way that when the monochrome monitor attached to the Hercules card was in use, and the "boss key" TSR was triggered, it had a bug that could program the Hercules card with incorrect frequencies. The evidence of this was:
- Triggering the TSR by hitting its "boss key."
- The monochrome monitor emitting a very loud, very high-pitched squeal
- The monochrome monitor's display contracting to a point
- The monitor ceasing its squeal with a "pop"
- A puff of smelly magic smoke escaping the monitor
- The monitor's permanent removal from the set of working monitors
My friend tried to debug this, but after he fried the second such monitor, he decided it was too expensive a problem to work on and just stopped using the TSR.
My friend and I worked very closely together, so I personally witnessed both of these monitors being fried.
My friend had a lousy grasp of ethics in his youth. We stay in touch, and I can now say with certainty that he would never consider doing such a thing, and is ashamed of having once done it.
Magnetic coils (the coils that magnetically deflect the electron beam in a CRT) have an electric resistance which is a direct function of the frequency of the current flow. If that resistance is getting too low (because the coils and circuitry are operated at a different frequency they are intended for), either the circuitry that regulates the current through the coil or the coil itself can be damaged.
Great answer, I had always wondered "what" got damaged. This makes sense, it is the same principle as a induction heater. Jun 4, 2018 at 8:32
I worked on a variety of CRT terminals and monitors (and converted the former to the latter), and never had one fry, with two exceptions: (1) IBM-PC monitors, and (2) Exact clones of IBM-PC monitors. Everything else maybe made strange noises, but were never damaged. So it was certainly possible to design horizontal scan circuits that didn't burn out their drive transistor when given improper signals, and most were like that. Lots more fear, uncertainty and doubt than is really called for. Still, this is my empirical data rather than any industry-wide truth.
4Television receivers needed to be immune from damage caused by sync signals that were out of spec, since television sets would often receive signals of varying quality. Monitors whose designs were derived from those of television sets would inherit such immunity. If such immunity were not required, a slightly simpler design could be used, and in some cases was. Jun 4, 2018 at 14:49
4Agreed with @supercat; devices such as televisions that expect a potentially-noisy input apply an extra level of logic. There is a sync generator: it takes syncs spotted in the incoming signal as an input, and produces syncs as an output. It always produces them at a safe frequency, which will coincide with the input only if it's in the proper range. The curvy sine patterns you sometimes see when sync is bad are actually real-life simple harmonic motion because of the way that they attempt to track phase through adjustments in frequency (i.e. the derivative of phase).– TommyJun 4, 2018 at 18:40
@Tommy: Does a typical sync generator simply takes syncs as input and produce them as output? My impression was that a television set's sync circuit effectively adds the sync input (which is limited to be within a certain voltage range) to a rising signal derived from the horizontal coil current, and triggers a sync when the sum gets high enough. Even if there's no input, the horizontal coil current will eventually get high enough to trigger a sync by itself, and even if the input is at maximum it won't be able to trigger a sync until the coil current reaches a certain point. Jun 4, 2018 at 19:26
@supercat possibly we're talking about different implementations; I was thinking specifically of flywheel synchronisation as described e.g. at machineryequipmentonline.com/video-equipment/… : "The flywheel sync circuit maintains an average frequency of the sync pulses by monitoring and taking the average frequency of a number of incoming line pulses so that a random pulse will have very little effect on the frequency" etc.– TommyJun 4, 2018 at 19:51
@Tommy: An approach like you describe would probably work better at decoding dodgy signals than the approach I describe, at the expense of more circuitry. From a "safety" perspective, though, I would think forcing a retrace when the deflection coil current starts to get too high and disabling retrace when it's too low would be better than hoping the flywheel oscillator's capture range is suitable. Jun 4, 2018 at 21:14
I fried the CRT of a friend of mine trying to setup the X11 a long time ago :(. Just a startx fried it. That time I was using the Monkey Linux distribution.
1This answers the "is this true?" but not the "how?". It's an answer, but not very useful. Do you think you could provide some details? Brand of screen, configuration of X (if you know it), what exactly "fried" is etc..– wizzwizz4 ♦Aug 16, 2018 at 19:42
This is entirely plausible. Before monitors got the ability to communicate their specifics (such as horiz. and vertical sync rate) to the graphics card, you had to configure them manually in the X11 configuration file. Aug 18, 2018 at 4:02