ElBartoME wrote:ElBartoME wrote:I am now leaning towards using the 2SC3782 NPN transistor
The 2SC3782 looks very interesting, but you linked the KSC3503DS instead?
ElBartoME wrote:since the 2SC3782 is sufficiently high voltage and high bandwidth
The bandwidth is very impressive but it seems the breakdown voltage is 200V. The transistors used in my E3431D have a breakdown voltage of 300V. I haven't measured any voltage near that but I assume there must be a reason why they chose that number. I assume it depends on G2. If I crank up G2 the voltage to achieve black levels change too. So I guess in my operation point I have at the moment I don't need that high voltage to achieve black. But still I would feel better if the breakdown voltage of a new transistor is the same. Do you have any suggestion?
Sorry, I copy pasted the wrong part number. I meant to say the 2SC3503, which is what I linked to on mouser.com. It definitely has a breakdown voltage of 300V, and should be a drop in replacement for most consumer grade CRT TV cathode amplifiers. BTW, the chroma signal path may have a bandwidth bottleneck at the pre-amp stage. If your oscope probing for the chroma signal right before the cathode amplifier shows the same high frequency distortion, then the pre-amp is also a bottleneck and would need to be upgraded. Pre-amplifiers are not very demanding as they deal with amplifying a roughly 1 volt peak to peak signal to at most 5 volts peak to peak, which is easy for transistor amps to achieve have a very high bandwidth. However, consumer TVs may have used very low grade pre-amps, so best to be sure and check with the oscope.
If the pre-amp signal is not distorted at high TVL, then just upgrade those cathode transistors!
I am not sure if the 2SC3503 can achieve the bandwidth target needed for high TVL, but since the part is a drop in replacement for most TVs and the part is still produced... it is compelling. You could try the much higher bandwidth 2SC3782, as it was used in the last generation of HDTV CRTs that were capable of outputting 1080i (which is 810 TVL), but as you mentioned, the breakdown voltage is only 200 volts, so you'd need to: (i) use a voltage divider to lower the amplifier circuit's input voltage rail, and (ii) make G1 voltage more negative to restore cutoff. Here again, we see the benefit of negatively charging the G1 anode, as it lets us operate the cathodes at a lower voltage, which reduces the burden on the cathode amps.
Looking at the E3431D's schematic, it looks like its neckboard uses a 4 transistor amplifier design for each color for a total of 12 transistors. For a single color there are 2 stages to the amplifier design. First there is a cascode amplifier input stage, which is then followed by a push-pull amplifier output stage. Please double check my work though because I am not familiar with the TV and only briefly looked at its schematic on Google just now.
Blacksheep wrote:At first, spot size/beam focus was considered the limitation for TVL, when LukeEvansSimon stepped on the stage with his confident demeanor. The OP now talks about other limitations, but all of that was edited in later.
What you are observing is real life engineering, as opposed to somebody that advertised a finished product. Nothing works perfectly the first time, and so two things are needed for success: (i) a vision grounded in research of the technology landscape, and (ii) persistence to not give up, because iterating on prototype revisions is always necessary before the vision is achieved. This thread started as a technical vision for a mod. Many naysayers said it was a bad idea and wouldn't work. Some people in this thread, such as
ElBartoME, decided to start experimenting variations of the idea, adding their own expertise and ideas in to refine the idea, and these are the people that have made real positive contributions and made the CRT gaming scene better because of their experimentation and their constructive improvements to the mod. As opposed to just saying it won't work, don't try it, these people try it, find issues, and propose solutions.
Blacksheep wrote:
Josh128 wrote:Vertical granularity however, is an entirely different story. Its very clear that the spot size has decreased and the the larger and more defined blank lines make it easy to see.
Correct, if anything, the proposed mod, when spot size is decreased, actually seems to be able to increase
vertical resolution of the CRT,
not horizontal resolution (TVL). Because on the horizontal axis, the beam is simply doing one continuous streak, so how should a decreased spot size even increase resolution there?
It is visually obvious that vertical resolution is improved. The reason why vertical resolution is improved and horizontal resolution is not, is due to a limitation in chroma signal amplifier rise and fall times (that is, amplifier bandwidth). The screen is drawn as a series of horizontal rows of varying color. The electron gun draws each row at the same speed, regardless of the TVL. So a higher TVL image needs to have the colors varied faster, and the speed that the colors can be varied is currently limiting the horizontal resolution improvement from being achieved by the mod.
ElBartoME's oscope images show exactly that occurring. The chroma signals for alternating vertical white and black lines are 3 square waves of around 1 volt peak to peak, corresponding to red, green, and blue alternating from black to peak color illumination. These square waves need to be amplified to around 200 volts peak to peak! The physics involved in swinging voltage up and down by 200 volts fast enough to draw, say 800 TVL, is a big challenge. Most amplifiers have rise and fall times that are too slow.
In this thread, we are engineering towards a vision. We have demonstrable progress towards realizing that vision, but we found an issue. Do we listen to the naysayers and give up, or do we determine the root cause that is limiting success and design the next prototype to test? Removing the chroma signal bandwidth bottleneck could be as simple as upgrading the cathode amplifiers to higher bandwidth parts such as the 2SC3503 or the even higher bandwidth 2SC3782. However, if the pre-amp stage is also a bottleneck, then that will have to be upgraded too. If the pre-amp stage occurs in the jungle chip, then upgrading pre-amp bandwidth will be challenging. If the pre-amp is made up of discrete NPN and PNP transistors, then common parts can be used to upgrade the bandwidth of the pre-amp.