Wellll.... I was bored and I have a Panasonic ST60 plasma that sits next to my Sony A90K in my lab...
I use the Panasonic as part of pic quality evaul because it has nearly no video processing. Over a long time tho, i have clearly seen that from a suitable distance, the Plasma can hold its own, and sometimes best, the OLED. YES the plasma has a list of issues. Banding because of a lack of bits and just a list of technical issues. However... Its also very fast, scenes with a lot of motion look clearer on it even tho is 2K VS 4K. I liked the ST60 VS the VT. The ST60 just seemed like a less processed picture.
I am going to mod a 42" A90K in the weeks ahead as well. See what i can do with it.
ANYWAY... I pulled off the back cover and dove in.
First step, look it all over and decide what to do and how to do it.. COme up with a first pass and order parts. I have now done this.
Stuffing test equipment into the plasma was VERY interesting.
The waveforms that drive the panel on its 3 drive lines are super trippy.. And very high power. At a average of 2A with nanoseconds of 1000A in a scanning system at 218.6V its all VERY high power, wide bandwidth and high precision. Its incredibly analog and non-linear. Nobal gas mixtures and actual plasma, capacitance with the whole panel being one big cap, complex algorithms all creating wild high power dynamics. Its CRAZY stuff..
Its insane it works at all.
This guy had a good simple summary of how it all works are a refresher. https://www.badcaps.net/forum/welco...ng/23618-how-a-plasma-display-panel-pdp-works
My kinds of circuits. High power, wide bandwidth, high precision..
Takes me back to my CRT days.
FREAKY waveforms. I will take screen shots next time I am in it. Gaps with nothing, I assume "sustain" and then seriously dense bunches of pulses I assume are sub fields or maybe whole frames. Then big gap.. Then again.. Its one set of these with like 6 blocks of pulses but nearly all blank gap. The whole set of these then gets repeated. Each gap is a different length going from shortest to longest, but then getting shorter again all in a clearly non=linear way. The whole set of pulses and gaps has a arch to it and peaks in the center. The voltage of these 20nS pulses in a group seems to vary with picture content. There are not enough pulses to account for a line of video. I assume its addressing only the pixels it needs too, but, these groups are a fixed number of them. Overall the whole waveform varies in a slow rolling manner.. Impressive stuff for sure.
Its one of the most complex waveforms I have ever seen. The bandwidth is stunning. Square, clean, 10nS pulses all the way down to slow rolling near DC low freq. The precision is also striking. Considering all this is a 218.6V 2A avg 1000A peak signal, wowee..
The panasonic is easy to work on. The important boards are very high quality with great parts. Super impressive 14ga coils and huge film caps. PC board is impressive and very much designed for the wide bandwidth and precision needed.
Doing all this really right would be way too expensive for a consumer product. So they cut corners to make it affordable. Which is where i come in..
Looking at the power supply that supplies this insanity it was clear they spent time on getting it cheaper and yet still good enough. Still,, it required a bunch of big caps on the driver boards along with film caps. Also the power supply board has caps on the same rail. They also divided the power supply and drive boards into 2. No doubt this was easier to control. VSUS is the voltage that has the big power requirements and bandwidth. Its high precision at 218.6V quite exact and well regulated. However,, all that high current mess at high freq creates a 21V wideband pk-pk noise. This cant be addressed economically. This requires huge WIMA polypropylene caps. A bunch of them. Also they clearly cheaped out on the big electrolytics. There is even space to put 2 on the PC board, but, maybe the ST used the same boards at the more expensive one and these boards get populated for that unit.
The 120V line to DC switching supply needs to develop close to 400V. It uses voltage doubling on the line voltage it appears. This is a less then ideal was to do this, but, it is what it is. So the main power supply caps are SUPER critical. They are undersized for sure and I will quadruple the uF. Maybe more. This will provide a very stiff 400V for everything else and less ripple by a magnitude. The 120V AC > DC converter is clearly mid-fi at best. So I will look at that further later on.
Space is a issue. I have 28mm of space. BUT tons of room width and high wise. Just depth is limited. So I can make cap banks with 28mm dia caps or poly caps less then 28mm on a side. This will work.
As i mentioned the driver boards have unpopulated locations for power supply filtering. Not only the big caps but also locations of big polys, I will populate these and swap out existing with bigger and better parts.
SO.. I ordered up parts from mouser. They will be here next week.
I then started thinking if there was something I could do now and remembered i had 2 giant WIMA polypropylene caps in 250V. 22uF in fact. Which is a BIG WIMA. So I worked out how to place it in the circuits for best affect. 22uF across a 218 V rail that has 20V of really high freq noise would SERIOUSLY kill off high freq noise. Pulled the boards, which was easy, and did the work...
Fired it up and let it all settle.. The high freq noise was now down to 2V. A 10:1 reduction. But this also means that the accuracy of the waveform got better and it had lots of control at those high freq.
Sure enough, the pic clearly improved. Everything that required precision like color rendering was better. Also little spots of brightness, like a few pixels in the fireworks against a black sky got far more defined and had sparkle. Fine detail was better especially on things where there was a big differences in level per pixel. Fully lit colors with adjecent pixels being black was clearly better. The 3D illusion was better. Textures on things were clearer.
So great progress just from 2 caps. So I have no doubt there will be really obvious improvements to come.
The power supplies all sag with a full white screen.. I can really address this. I can make sure frame to frame and scene to scene the voltages dont vary. This should also improve the pic.
So this will be fun...
One of the driver boards..
I use the Panasonic as part of pic quality evaul because it has nearly no video processing. Over a long time tho, i have clearly seen that from a suitable distance, the Plasma can hold its own, and sometimes best, the OLED. YES the plasma has a list of issues. Banding because of a lack of bits and just a list of technical issues. However... Its also very fast, scenes with a lot of motion look clearer on it even tho is 2K VS 4K. I liked the ST60 VS the VT. The ST60 just seemed like a less processed picture.
I am going to mod a 42" A90K in the weeks ahead as well. See what i can do with it.
ANYWAY... I pulled off the back cover and dove in.
First step, look it all over and decide what to do and how to do it.. COme up with a first pass and order parts. I have now done this.
Stuffing test equipment into the plasma was VERY interesting.
The waveforms that drive the panel on its 3 drive lines are super trippy.. And very high power. At a average of 2A with nanoseconds of 1000A in a scanning system at 218.6V its all VERY high power, wide bandwidth and high precision. Its incredibly analog and non-linear. Nobal gas mixtures and actual plasma, capacitance with the whole panel being one big cap, complex algorithms all creating wild high power dynamics. Its CRAZY stuff..
Its insane it works at all.
This guy had a good simple summary of how it all works are a refresher. https://www.badcaps.net/forum/welco...ng/23618-how-a-plasma-display-panel-pdp-works
My kinds of circuits. High power, wide bandwidth, high precision..
Takes me back to my CRT days.FREAKY waveforms. I will take screen shots next time I am in it. Gaps with nothing, I assume "sustain" and then seriously dense bunches of pulses I assume are sub fields or maybe whole frames. Then big gap.. Then again.. Its one set of these with like 6 blocks of pulses but nearly all blank gap. The whole set of these then gets repeated. Each gap is a different length going from shortest to longest, but then getting shorter again all in a clearly non=linear way. The whole set of pulses and gaps has a arch to it and peaks in the center. The voltage of these 20nS pulses in a group seems to vary with picture content. There are not enough pulses to account for a line of video. I assume its addressing only the pixels it needs too, but, these groups are a fixed number of them. Overall the whole waveform varies in a slow rolling manner.. Impressive stuff for sure.
Its one of the most complex waveforms I have ever seen. The bandwidth is stunning. Square, clean, 10nS pulses all the way down to slow rolling near DC low freq. The precision is also striking. Considering all this is a 218.6V 2A avg 1000A peak signal, wowee..
The panasonic is easy to work on. The important boards are very high quality with great parts. Super impressive 14ga coils and huge film caps. PC board is impressive and very much designed for the wide bandwidth and precision needed.
Doing all this really right would be way too expensive for a consumer product. So they cut corners to make it affordable.
Which is where i come in..Looking at the power supply that supplies this insanity it was clear they spent time on getting it cheaper and yet still good enough. Still,, it required a bunch of big caps on the driver boards along with film caps. Also the power supply board has caps on the same rail. They also divided the power supply and drive boards into 2. No doubt this was easier to control. VSUS is the voltage that has the big power requirements and bandwidth. Its high precision at 218.6V quite exact and well regulated. However,, all that high current mess at high freq creates a 21V wideband pk-pk noise. This cant be addressed economically. This requires huge WIMA polypropylene caps. A bunch of them. Also they clearly cheaped out on the big electrolytics. There is even space to put 2 on the PC board, but, maybe the ST used the same boards at the more expensive one and these boards get populated for that unit.
The 120V line to DC switching supply needs to develop close to 400V. It uses voltage doubling on the line voltage it appears. This is a less then ideal was to do this, but, it is what it is. So the main power supply caps are SUPER critical. They are undersized for sure and I will quadruple the uF. Maybe more. This will provide a very stiff 400V for everything else and less ripple by a magnitude. The 120V AC > DC converter is clearly mid-fi at best. So I will look at that further later on.
Space is a issue. I have 28mm of space. BUT tons of room width and high wise. Just depth is limited. So I can make cap banks with 28mm dia caps or poly caps less then 28mm on a side. This will work.
As i mentioned the driver boards have unpopulated locations for power supply filtering. Not only the big caps but also locations of big polys, I will populate these and swap out existing with bigger and better parts.
SO.. I ordered up parts from mouser. They will be here next week.
I then started thinking if there was something I could do now and remembered i had 2 giant WIMA polypropylene caps in 250V. 22uF in fact. Which is a BIG WIMA. So I worked out how to place it in the circuits for best affect. 22uF across a 218 V rail that has 20V of really high freq noise would SERIOUSLY kill off high freq noise. Pulled the boards, which was easy, and did the work...
Fired it up and let it all settle.. The high freq noise was now down to 2V. A 10:1 reduction. But this also means that the accuracy of the waveform got better and it had lots of control at those high freq.
Sure enough, the pic clearly improved. Everything that required precision like color rendering was better. Also little spots of brightness, like a few pixels in the fireworks against a black sky got far more defined and had sparkle. Fine detail was better especially on things where there was a big differences in level per pixel. Fully lit colors with adjecent pixels being black was clearly better. The 3D illusion was better. Textures on things were clearer.
So great progress just from 2 caps. So I have no doubt there will be really obvious improvements to come.
The power supplies all sag with a full white screen.. I can really address this. I can make sure frame to frame and scene to scene the voltages dont vary. This should also improve the pic.
So this will be fun...
One of the driver boards..
