How a plasma display panel (PDP) works

[tom66]

There seems to be a lot of misunderstanding about how plasma display panels work. I've learned a lot about them by reading patents, training manuals and service manuals.

Misconceptions about plasma displays
A plasma display does not have a backlight
The display technology is self-emissive. Each pixel emits its own light. Like a fluorescent bulb, which takes time to light, the pixels need to be primed before they are ready to emit light. The priming makes the pixels emit a small amount of unwanted light. If they are not adequately primed, they could fail to ignite when necessary. This causes underdiffusion -- random fields of dead pixels. If they are primed too much, they overdiffuse -- random fields of lit pixels. Newer plasma panels have very low black levels (priming light), almost invisible under even dark room lighting.

A plasma display has a fixed resolution
Unlike CRT displays, a plasma panel has a defined row/column matrix size.

Power consumption varies with brightness
Unlike LCD technology and most CRTs, the power consumption of a plasma display is variable with the amount of light emitted. Modern plasma displays typically use more energy than LED-backlit LCDs, and less than most CCFL-backlit LCDs for the same display content.

Reasons plasma displays still exist
The main reasons people like plasma displays are the higher contrast ratios, greater brightness, and overall better motion performance than LCDs -- less than 1ms delay -- great for gaming and fast-moving content. Plasma technology is far from dead and many manufacturers are getting back into re-badging plasma displays.

How a plasma display works
A plasma display is broken down into millions of individual cells, with each pixel broken down into three subpixels, one red, one green and one blue. Each cell is filled with neon or argon (depending on the display technology; newer displays use more exotic combinations too), mixed with a small amount nitrogen. The cells are pressed between two thick glass plates.

The display is not a vacuum, like a CRT is. However, the integrity of the display must be maintained. If the display is cracked, the gas will leak from the display. The evacuation of the gas (and replacing it with normal air) will cause a loud vibrating buzzing sound from the display surface due to the low dielectric constant of air (compared to noble gases.) No light is emitted due to the missing gas. The display is now considered a paperweight, because although it may be feasible to re-gas a panel (if you can get the right composition of gases) the gas will simply leak from the cracks. Even if you could repair the cracks, the electrode lines would likely be severed, leading to missing pixels or columns.

Each cell has three electrodes crossing it. I'll refer to them using my own terminology, and clear it up later. There are two Y-electrodes, the scanning electrode and the common electrode. There is one X-electrode, called the data electrode.

By applying a voltage, Vda, typically around 50~70V, to the data electrode, and connecting the appropriate scanning electrode to common, a small charge is built up on the cell. (The nitrogen helps here.) This charge can be stored for several minutes, called the wall charge. The voltage is too low to cause light emission from the panel at this time. This process is performed for cells which need to be lit, and isn't performed for the ones which don't need to be lit. This allows an image to be "written" to the panel. Note that the charge level is binary. We can only make cells be on or off, however, it will become clear later how greyscale is added. Each row is written individually; the Y-scan boards select the appropriate line to write data to.

Then, the scanning and common electrodes are used as part of what is known as the sustaining process. Since a wall charge has been built up on the cell it's now very conductive. By applying a voltage, Vsus, around 200V, to one end, and grounding the other end, the cell becomes positively charged, conducts current from scan to common, and a current flows through the cell, causing light to be emitted. Then, the directions are swapped. The scan electrode becomes grounded, and the common electrode becomes Vsus. The process occurs again. It is repeated as many times as necessary to produce the required light level. The current through the display peaks at over 100A! However, it is very brief. The average current with a 200V sustain voltage is typically 2 amps for a full white image.

The priming process occurs at the beginning of each sustain period, or on newer plasma displays only as necessary (to minimise light emission for maximum black level.) To prime the panel, before addressing occurs, a ramp waveform from Vsus to Vp is applied. Vp is typically around 180~200V higher than Vs. The ramp occurs over a few microseconds. A second ramp, from GND to Vn is applied; Vn is typically around -180~200V, relative to GND. This exploits a characteristic of the neon gas - the negative resistance means that the slow ramp frees the electrons from the gas, allowing it to more easily emit light. Sometimes the ramp is repeated a number of times, or in lesser incarnations, at different periods, to ensure the display properly sustains light.

Erasing the panel is accomplished in two ways. On some displays, the negative ramp erases the display. On others, a second voltage, called Ve, is used to reset the cells, which is usually applied on the common side. Sometimes a combination of technologies is used.

Adding greyscale levels
So far we've got a display that can very effectively and reliabily display a black and white image. Great -- but what do we do for greyscale? It turns out we can actually sustain and write the display very rapidly -- over 600 times per second for most typical displays. The image is displayed at 60 Hz (for a 600 Hz sub-field drive plasma display) by breaking it down into 10 individual sub-fields. Each sub-field is displayed for a different amount of time. Depending on the display technology, the fields will be in binary lengths (512, 256, 128, 64, 32, 16, 8, 4, 2, 1) which allows any cell to be varied from zero intensity to 1023 intensity (arbitrary figures.) Since the display updates 600 times a second, motion appears very much instantaneously, flicker is essentially unnoticeable.

1024 levels is great, but we can go bigger! To minimise the effect of false contours (the transition between 128 and 127 is noticeable, due to imperfections in the sustain process), odd numbered sub-fields are used, making the transitions less noticeable. A typical pattern might be 57, 31, 19, 17, 13, 8, 6, 4, 3, 2, 1. But this only gives us 160 levels -- we need at least 256, but ideally, we should have more. So, we can use adjacent pixels, and time itself, to control pixel brightness percepton; this is called dithering. One dithering technique used by Panasonic uses 10 odd sub-fields to get 6144 levels. An upgraded technlogy in some of their other models gives 24,576 levels. This is equivalent to 12.5 bit and 14.5 bit resolution. Conventional LCD is limited to 10 bit currently, with most panels only being 8 bit resolution. Generally you can't tell the difference beyond 10 bits (most people even stop at 8 bits), but this allows for precise calibration, brightness, contrast and similar controls without limiting the resolution available to express the image.

Energy recovery
A plasma panel is essentially a very large capacitor. Charging and discharging such a capacitor at a high frequency is extremely inefficient and leads to large losses in the sustain process. To avoid this problem, the energy that would otherwise be lost in plasma panel, which is stored on the panel capacitance, is discharged using an inductive resonant circuit. This relies on a few poly-film capacitors and large inductors, and is the major cause of buzzing emitted from a plasma display.

The energy recovery circuit reduces power losses by as much as 150W on a 42 inch, 720p display. Typically, the energy recovery circuit is found on both drivers. However, to lower costs, modern displays use one energy recovery circuit for both electrodes. The address electrode does not require an energy recovery circuit because the electrode is 90 degrees to the two scan and common electrodes, which means the capacitance it sees is much smaller.

Obviously the energy recovery circuit costs money to include, but it greatly lessens the demand on the power supply, reducing costs there.

Terminology
The scan electrode is the same as the Y-sustain (LG), Y-main (Samsung), SC (Panasonic), Y-sus (Hitachi-Fujitsu/Chungwha) or Y-drive (NEC/Pioneer)
The common electrode is the same as the Z-sustain (LG), X-main (Samsung), SS (Panasonic), X-sus (Hitachi-Fujitsu), X-bulk (Chunghwa) or X-drive (NEC/Pioneer)
The data electrode is the same as the address or X drive electrode.
The driver boards are typically referred to as sustains or drives.
The scan boards are typically referred to as buffer boards (Y-buffers or SU/SD/SM on Panasonics.)

[maybenever]

WOW All of this is going on, and I still love my Panasonic 65" plasma.
It's around 2 years old and has over 13,000 hours on it. Thats a lot of cycles.

[tom66]

Yes, a full HD plasma panel typically receives over 1.24 billion impulses per second.
Every single one must be perfect or the display will get dead pixels or other problems.

[ghiotom]

Originally Posted by maybenever

WOW All of this is going on, and I still love my Panasonic 65" plasma.
It's around 2 years old and has over 13,000 hours on it. Thats a lot of cycles.

Holey moley!! That's like 18 hours a day, every day!

[ben7]

AWESOME!

Thank you tom66! Sweet truck-load of info there. Will have to read a few times over

-Ben

[ratdude747]

Yay! A fresh thread in this sub-forum that actually belongs here.

Nicely written. Good job.

[tom66]

Triggering on sustain waveform:
http://www.youtube.com/watch?v=nGr5y...ature=youtu.be

[ohmmy]

So there are no small amounts of mercury in a plasma screen that produces light? Its done with neon gas instead?

[mockingbird]

Tom66 -

What do you think of the Panasonic VT50? It's supposed to be the best plasma out on the market today, perhaps considered even better than top of the line LCDs.

Also, why do pre-HD resolution plasmas have 1:1 native resolutions (Like 1024x1024) when the screens dimensions do not match?

[Topcat]

Good information, thanks for sharing that!

[tom66]

Originally Posted by ohmmy

So there are no small amounts of mercury in a plasma screen that produces light? Its done with neon gas instead?

Zero mercury in a plasma display panel. It's all neon, xenon and argon plus a small amount of nitrogen. Newer PDPs use more exotic combinations.

Originally Posted by mockingbird

What do you think of the Panasonic VT50? It's supposed to be the best plasma out on the market today, perhaps considered even better than top of the line LCDs.

It is the best PDP commercially available, though the VT60 is due to come out later this year (probably around March) with a better display filter -- but that's about it. However, it has nothing on the old 9.5G Pioneer Kuro. The 9.5G Kuro is the only flat-screen TV (aside from OLED, which is about 5-10 years away) that has achieved an infinite contrast ratio (i.e. zero black level.) Panasonic bought the tech, but hasn't used it because it's very expensive to implement; Panasonic achieve a real-world contrast closer to 30,000~50,000:1 typically. There also appears to be minimal difference between the GTxx, STxx and VTxx -- the actual performance difference is not due to the PDP or drive electronics (they're the same, with a few minor changes) but the main board, which has different software on it. So it may be possible to upgrade a GTxx into a VTxx -- someone on this forum recently installed a V20E main board in a G20B plasma, and it worked...

And any plasma will beat an LCD hands down on contrast ratio, colour depth, and motion performance. The penalty is slightly reduced brightness (though still about the same on normal viewing), glossy glass screen, and higher power consumption, though a modern plasma is less hungry than a CCFL backlit LCD.

Originally Posted by mockingbird

Also, why do pre-HD resolution plasmas have 1:1 native resolutions (Like 1024x1024) when the screens dimensions do not match?

They are ALiS displays (Fujitsu-Hitachi), the only interlaced plasmas. They don't make them any more, left the business in 2008 I think. The display is physically 1024x512, but each electrode is capable of sustaining in two halves, which effectively doubles the display resolution. Basically a way of getting HD with minimal extra cost (still using VGA-resolution panels.) They are HD capable -- as they can display 720p content, arguably better than most 720p plasmas, which are 1024x768 (for a 42".)

As to why they chose that odd resolution, I don't know. If they'd done 1024x768 or 1368x768 I could understand, and that'd be cheaper too, as it'd only use 384 vertical lines...

Originally Posted by Topcat

Good information, thanks for sharing that!

Couldn't do it without this great forum. So thank you!

[kc8adu]

great job!

[Bazz]

A most interest explination how a Plasma screen works. Thankyou.

[shovenose]

Thank you, I feel educated now and finally understand how a plasma TV works question: are LCD or plasma more reliable?

[tom66]

They have about equal reliability, but LCD failures can quite often be unrepairable (driver pack failure) whereas plasma panels are almost always repairable.

[akp]

Thanks for all the info!!

[munirutech]

great info thankz tom

[ScooterMcTavish]

Good info on this forum for a variety of applications.

Need some help myself, but have a feeling I'll be back here often between plasmas, audio gear, and pinball machines.

[JackMiles]

Hello Tom66.
In short time i'm planning to build my own pdp panel driver and really i'm out of knowledge. (very dissapointed about panasonic vt60, theres huge mistakes they didnt want to fix) The problem is: sub-fields and scan line+data line selection.
1)I don't ungerstand sub-fields at all) are dispaying subfield with "brighness 1" and then with "brightness 3" gives a brightness 4? How dithering works? Whats the order of subfields? One per line, all lines and again, or all per line and then next line?
2)About line selection: as i see it uses standart signals for making sustain/scan signals, but line selection and data lines for X board are lvds. What format, clock and how many bits it uses?
What oscillograph freq do i need to watch all signals in pdp, 100mhz is enough?
I'm very sorry for my bad english, but please help me if it is possible.

[tom66]

Originally Posted by JackMiles

Hello Tom66.
In short time i'm planning to build my own pdp panel driver and really i'm out of knowledge. (very dissapointed about panasonic vt60, theres huge mistakes they didnt want to fix) The problem is: sub-fields and scan line+data line selection.
1)I don't ungerstand sub-fields at all) are dispaying subfield with "brighness 1" and then with "brightness 3" gives a brightness 4? How dithering works? Whats the order of subfields? One per line, all lines and again, or all per line and then next line?
2)About line selection: as i see it uses standart signals for making sustain/scan signals, but line selection and data lines for X board are lvds. What format, clock and how many bits it uses?
What oscillograph freq do i need to watch all signals in pdp, 100mhz is enough?
I'm very sorry for my bad english, but please help me if it is possible.

It's an ambitious project, I'll give you that.

The subfields are basically just same brightness levels but displayed for a longer period of time, so for example a subfield of length 1 might be about 1% as bright as one which is displayed 100 times over, each sustain event emits about the same light.

The line selection is logic level from control board, ground referenced. At the buffers, it is VFG referenced, that is the sustain node (without the Vsc applied.) This requires optoisolation.

Driving IGBTs at the required rate (around 200kHz) is a high level power electronics project. I would not be surprised if the first few boards work for some small amount of time before catastrophically failing. For learning experience, I would try experimenting with a different panel... maybe one not so valuable

A common sustain driver (SS) is a lot less ambitious than the addressed sustain driver.

Scope: minimum 100MHz digital scope like Rigol DS1074Z, DS2072A, Agilent 2000X, etc. You need the high waveform update rate to see the most detail.