CurtPalme.com Home Theater Forum

[redfox001]

Yes I found it will try that later

[the dude]

Also note this probe has 3pf. This is not well suited if you are going to measure the cathode. If you are about to buy such a probe only for cathode measurements you should be warned that even with such a probe this is complicated and a real bandwidth test at full constrast might even not be possible. I think its better to look on the screen for bandwidth measurements at the last stage.

[redfox001]

[redfox001]

By the way Gjaky has a switcher driver and neckboard from me. I think he will provide more professional measurements in time!

[the dude]

Just testet the unmodded boards.

Guess what? The results are very similar, maybe with my modded boards the 1:1 pattern is a few percent brighter. My modded ones use less peaking, so the mods are fine anyway.

But this proves something: There are no miracles can can be done by replacing just a op amp in the video chain. There are no ugly weak parts in the video chain that need to be replaced. I think less peaking is the only real advantage you get with such mods. Thats all voodoo (sorry Mike ). And it proves that we tend to be less objective when we replace a part (or pay for it), because we expect the picture to be better and so we see a better picture. Less noise and peaking is also a very important goal to archieve, so i don't want to call such mods useless. With my boards the picture for sure looks a little better. But they don't increase bandwidth significantly.

There is a real weak part for sure in every projector. And thats the cathode driver. Changing the neckboards in my 1209s really did the trick. But thats still the weakest part.

Its pure physics. You need a amplitude of 100Vpp at very high frequencies for a single ended driver. Think in rise time. Its a completly different story than a op amp with 2Vpp. Look at the sanyo video packs: The best can do 180Mhz, but only at 30V supply voltage. The highest voltage one (VPH08) can do only 85 Mhz, but has 200V supply voltage and can deliver 100Vpp. Its all about rise time. This are the limits we must defeat.

So the next project is developing a G1 driver for the 909 board. I will do that. Not today or tomorrow, but in the near future. It need not to be 100V, maybe a VPA15 would be good (150 Mhz with good rise time). Its not rocket science, its only electronics and we have schematics of working circuits.

Very interesting and educational discussion so far

[redfox001]

Yes I found that too testing Eisseman modfied boards and my own modified boards there was almost no difference in bandwidth. But Eisseman and Mike too also modify noise using small inductors to suppress noise on the rails and using better low esr caps sometimes adding some caps. Does it improve the image? It is difficult to measure but takes looking at the screen modding etc. I think there is some objective improvement to make.

But on a pure mathematical point of view why is real bandwidth better than compensated bandwidth? If all frequencies are the same amplitude fourier says there is no difference and the rise time of a transient should be identical? If not in practise please explain.

Also when you modded your boards the picture will be better that is a law even if you only modded the powersupply cord

[redfox001]

You say the modified boards where a little brighter but did you check on the scope? I found that a small visible difference was more visible on the scope and so it could do something.

[the dude]

On the scope it was not clear. There is still a little bit of ringing, so this would be difficult to judge. When there is a difference, its very small.

And for peaking (as far as i understand it): The frequency is one thing. The simple RC filters rolloff is always 6db/octave and never matches the amplifier roll off in a real world in a accurate manner. In fact the amplifier frequency limit is not a certain frequency, the limit is rise time. Thats why the cut off frequency depends on gain. You can see this figures in the datasheets. It says -3db at frequency X for gain Y. And some lower frequency on higher gain. So if we turn the constrast low, we need less voltage swing in the the same time and so we can archieve higher frequencies. A peaking network is something static.

[redfox001]

Ok but there are not much variable gain parts only ad835. In fact the -3db of a part depends mostly on internal design of the transistors an compensation, capacitance etc. Anyway I just try you allways need to compensate for cable capacitance and miller anyway.

In the marquee there are a number of motorolla transistors on the neckboard they all have a peaking network. Some happen to have the gain that matches the peaking and they are the high bandwidth types. In that design we se the internals of a chip.

[gjaky]

[redfox001]

If I use the same probe on the 300 MHz Tektronix would that really be better? If so I will buy one. I read the post from Craig and others again I would ned to use the coax trick I get it. Best to also get me a pattern generator. Ah no a good vga card also works!

[redfox001]

Better read this again. Dude do you see the edge enhancements on the hdfury? Your shots do have the brighter top and bottom where the moome has a small brighter top. What is causing that?

[redfox001]

If I could design a CRT for the 2020 th decade I would try something similar to digital audio amplifiers.

Switch a 10 to 100 trafo (or so) to amplify the samples that are on the digital source. So you get huge spikes on the high voltage side of the trafo. Put that on the G1. Might need to shape these pulses somewhat. The beam itself does a Gaussian low pass filtering and the result should be a smooth video signal not limited by bandwidth.

Think about it the vertical lines do the same. One line gets a pulse (at a certain pixel) the beam does the Gaussian shaping and the next line gets another pulse (at the pixel below). Works perfect.

I would call it the digital CRT.

[gjaky]

[the dude]

Thanks for explanations gjaky, thats quite interesting. I never thought of driving a probe as a load instead of the tube, but for sure the capacitance is quite in the right range. My probes can withstand 50V at 100 Mhz, so this would be quite useable.

The sony G70 uses a sanyo video pack for both cathode and G1, so i thougt that may be a good idea. But the G70 neckboards are known to run quite hot for sure. I once saw a machine with custom made water cooling. Not many parts really needed this, most would be happy with the slow fans that were still installed for basic cooling. The neckboards were one of the few parts that needed a water cooler!

Adapting the Electrohome neckboard would require some work, at least copying of the IK sens and sample/hold and the G2 regulator and a power supply but maybe it is worth the effort.

@Redfox: The advantage of the Tek scope isn't only the bandwidth, 300 Mhz is nice but thats not the point. It is analog, so the deflection speed affects the brightness and areas that are hit by the beam more than once will apear brighter also. So if for instance the area of the multiburst lines are uniform in brightness you know there is no ringing in the signal. It is easy to discover overlapping signals this way. Be sure to get a true analog scope. There are scopes with CRTs that are early DSOs, in this case the CRT is only a monitor.

If you are going to buy one of these late high BW Tek be aware of that horizontal deflection chip they use. It is custom made and inreplaceable, it serves several special functions like generation of readout cursors also. Unfortunally on some the internal chip bonding wires can lift up from the die, and if this happens the scope is garbage. You need some cooling for this, since heat up and cool down cyles are suspected to cause this due to termal stress. I made a heatsink for it so it runs quite cold.

And for the fury: I didn't test it at 75 Ohm, maybe i should get a T adapter and solder a resistor on it. Only this would give accurate results.

[gjaky]

[the dude]

Hmm, is there a special reason for using the cascode emitter to get the drive signal for g1? Use the lower cascode transistor as a driver for both stages and use less parts?

[gjaky]

[the dude]

Ok, that makes sense!

[redfox001]