CRT General Discussion.

[dominicbeesley]

One of the problems with the NBTV signal spec is that whilst the line sync/line blanking period is a modest 10 to 15% of total line time, the spec. makes no provision for a longer frame flyback period.

Too true, on the current waveforms (which I've now lost) the flyback spike was taking up a lot of the headroom in the amplifiers so I had to reduce the flyback speed a fair bit!

Also the "missing sync" is a pain in the unmentionables. I spent a fruitless few hours trying to implement a flywheel sync using a 4046 this weekend but can't work out how to mask off a sync hole on a CRT! (well not without lots of complications).

For my purposes I'll probably just "lose" the first line of the picture. In my analogue colour system this is used for the reference burst etc and coming soon teletext....maybe!

Cheers

Dom

[Steve Anderson]

...And that implies a very rapid flyback even at 12.5 Hz. Regards, Graham

One of the advantages of electrostatic CRT's is this absence of an inductive spike during flyback. One of the reasons I suggested the TIP 120-127 series of Darlington is that they have back-EMF diodes built-in.

The traces earlier in this thread I posted show a frame retrace time of around 10-15us, less than one pixel at 48kHz sampling. The traces were of the frame/field waveform, line even better.

In the current system under construction the line is either 50 pixels long with 48 for active video and two for flyback time at 36 lines, or 100pixels, 96 active video and four for flyback at 72 lines. This is for line timing.

Frame is a little more relaxed as after the frame reference burst there's a period of black where clamping is done and the frame retrace is done.

But in interlace mode there is still only two or four pixels to do the field retrace which is why the frame/field retrace time is so short. There is no frame/field sync in interlaced mode, just the frame, "The Brute" knows when the do a retrace in interlaced mode after either 18 or 36 lines.

So there are no half-lines and even number of lines can be used. One line in 36 or 72 is lost but the same can be said of FSTV.

Steve A.

[Steve Anderson]

Also the "missing sync" is a pain in the unmentionables. Cheers Dom

Cannot agree more, which is why I dumped it. Suggest everyone else does too.

OK, I lose one line in 32 (3.1%), one line in 36 (2,8%), one line in 48 (2%) or one line in 72 (1.4%)

Compare that with FSTV that loses 50 lines in 625 (8%). Not doing so bad after all methinks. Similar proportion at 525 lines. Not sure at 405 lines...Steve O?

Steve A.

[dominicbeesley]

Well here's the results so far, looks like I could do with a bit more oomph The vertical (12.5Hz) will go a lot bigger but the horizontal (400Hz) coils are a lot lower impedance and they don't look to be so easy to change to run in series...anyway it looks better than I was expecting so far!

This already has a sine-wave "spot wobble" applied as the lines are otherwise so sharp that it is difficult to make the picture out. Anyway looks like it should be a goer!

Flyback takes nearly two lines in this example, it should be possible to improve this but I'll leave it until I've got everything else working correctly.

Dom

[Steve Anderson]

Well here's the results so far....Dom

Well you've gotten far better results than I did 20-odd years ago, I got a fat nothing and gave up. Keep at it.

If the lines are that sharp at this reduced EHT I don't see any reason to increase it unless the intensity is a bit low.

Steve A.

[AncientBrit]

Dom,

Here's some circuitry I built a few years ago using a Leader LF X-Y scope as the basis of a x4 NBTV display.

I used a frame store to give me an output of x4 NBTV standards, ie 50Hz and 1.6kHz scans.

You'll note that I rotated the scan coils and used the old line scan as frame and vice versa. Also I rewired the H scan coils in parallel to get the drive I needed.

The forced linearity circuit works very well.

For the frame scan I used the Leader built in scan power amp, just about works with the addition of spot wobble.

I think you'll agree that 12.5 Hz direct display on a larger picture is very difficult to watch for long periods of time.

(added)

I see you were also asking about line accurate detection of the “missing pulse”.
This circuit also performs that function.

The 4520 is a 32 line counter which is synchronised late by the missing pulse.
The 4068 gate then pulls out an earlier pulse which is used to initiate frame flyback at the correct point.


Good luck,


Regards,


Graham

[Klaas Robers]

In most TV deflection coils the H-deflection consists of two coils, one at each side of the tube neck. Each of these coils is then wound with several wires in parallel to overcom problems with skin effect. The harmonics of the 16 kHz are needed to a few MHz, so the skin effect is becoming a problem.

But the two coils are connected in parallel to get a self inductance low enough for the deflection circuit. You might connect those two coils en series. In most cases this is not so difficult. Then you need half the current and twice the voltage.

[dominicbeesley]

Thanks All for the advice and encouragement, as always it is appreciated!

Steve,

thanks for the encouragement. The brightness is spot on, I can get far more by upping the g2 voltage without losing focus. It's cranked down at the moment though as my video amp only gives a few volts of drive could do with about a 30V supply for that part...(should have used split rails!). However its still perfectly viewable in a well lit room though. If I ever use this for 405/525/625 I may up the EHT however as the focus is not quite as tight as the original set's but more than enough to show up details at >16kHz in the picture.

Graham,

Thanks for the circuit plenty to think about there.

I can't say I fully understand the line drive circuit but will give it some more careful study later and may well be back with more questions later!

I'd thought about the idea of a counter but am trying to keep the parts count down at the moment. I may use the idea later though in an add-on sync stabaliser. Though I'll probably use a uController and try and do some fancy auto-detection of the standard being used and re-insert syncs, not entirely sure yet, might even try and make it a standards converter with frame store...

12.5Hz certainly is a bit flickery it has to be said, but for now I'll stick with the "standard" as I'm really trying to try out the spot-wobble ideas so its nice to be able to compare like with like.

What frequency did you use for the spot wobble? And did the square wave work or did it give the effect of double lines (thats what I get when I add a square wave wobble!). The main problem I've got with the spot wobble at the moment is that the error amplifier I've built doesn't pass high frequencies very well so I have to use a low frequency which means the wobble is visible (though it looks fine on photos!)

Klaas,

The tube I've got seems a bit odd in many respects not least that the frame coils were three thick wires in parallel giving about 3.4R whereas the line seems to be a single winding (unless there is something funny going on inside the yoke) with a lower resistance. Trouble is its all quite small and covered in lots of glue....I will probably try and undo some of this tonight and see what is actually going on. Also for some reason my picture is rotated and I can't quite work out why - the waveforms all look ok!

I will still try and put the windings in series if possible, though the trouble with that is I may want this to be a 405/525/625 monitor as well in which case the coils will need switching as well as everything else....I may need more wafers on my big standards switch!

Cheers

Dom

PS: I still can't work out how the little TV managed to make all this work with small frame drive transistos with no heatsinks - I'm using power ones with big washers as heatsinks and I still keep getting burnt fingers!

[AncientBrit]

Dominic,

I can't remember the exact freq. of the spot wobble. If you calc the tc of 50k and 220 pF that's about the middle of the range. It starts out as a square wave but by the time it's passed through the limited freq. response of the Leader scan amps its a sine wave so the result on screen is "tapered" edges, not the sharp left/right action of a square wave.

Regards,

Graham

[dominicbeesley]

Aha, thanks Graham, I'd not thought of that, so about 100kHz.

I'd like to go higher (6Mhz or so for 405 lines) if possible but suspect it might be difficult.....

And another quick update, Klaas was indeed correct about the H-coils being seperate. After some digging about with the soldering iron these turn out to consist of 4 seperate strands which I've quickly connected as 2x2 which gives double the scan width but a quite distorted output so more investigation needed there....

Dom

[Klaas Robers]

Yes of course I was right, hum hum.

The deflection coils always are in pairs, or two saddle coils laying against the tube and a ferrite yoke over them (the frame coils are always constructed like that), or two toroidal coils on a toroidal yoke around the neck of the tube, these coils were connected in parallel, but switched in opposite phase in order to get a magnetic field across the hole of the ring, through the tube neck.

So always two coils.....

[dominicbeesley]

As always, Klaas of course you were right!

I'm pretty comprehensively stuck now....

I've played around more and am now struggling to get the line flyback to be acceptably fast. The simplified circuit I have below gives good linearity over most of the line but not at the start as the flyback is far too slow. Any ideas how I can get the stored energy out of the line coils quickly during the flyback - or do I just need to use a bigger supply voltage?

(The problem seems to be that the driver transistor "bottoms out" (high current 0V at collector) during the flyback

Help?!?!

Dom

[Steve Anderson]

I'm pretty comprehensively stuck now....Help?!?! Dom

Dom I've made a start on this but time is getting late here, I'll do some more in (my) morning. What is the winding resistance of the coils? We know the inductance is 5mH.

Steve A.

[dominicbeesley]

Hi Steve,

The inductance measurement is not proving to be that easy to do with accuracy!

There are four windings on the H coils I currently have this as 2+2 in parrallel / series this actually gives about 2mH (well 3mH measured on bridge and 1.2mH measured on DMM!). With 3.8R series resistance (accurate to 1 dp on both valve voltmeter and DMM)

If they are all in wired in parallel then this changes to 1mH / 1.9R as expected.

The 5m figure was from trying all the windings in series (this proved to be a failure - not enough width!).

I've no idea which is the most correct. I think I need to do some calibration on my gear! (The inductance bridge I'm using I only just got working so it is probably a bit on the dodgy side).

If this needs to be measured with more accuracy to design the circuit is there something I could knock up with an AF sig gen and a scope to get the correct measurements?

Help, as always, appreciated....

Cheers

Dom

[dominicbeesley]

I've just rebuilt the circuit above from scratch to see if I'd done anything odd. I've now got better though by no means perfect results...so must have had something wired up wrong somewhere. I think there may have been an intermittent short somewhere in the H-coils. Probably caused by my mucking about with the soldering iron... Anyway that would explain the funny inductance readings, the component bridge must have put out a bigger voltage and overcome the insulation breakdown.

I also dispensed with C7 - can't remember why it was there in the first place - I think probably to slug a pole....

I now get 1mH for 3.8R series or 1/2mH for 1.9R for the H coils. [I can't work out why a short would have caused the odd measurements though!]

Anyway picture attached is showing some improvement, though could do better! [Lines are finer than shown in picture, slight ripple is caused by my crappy flywheel sync circuit which I still can't get working properly!]

I'd still be very grateful for any suggestions for speeding up the circuit though as the flyback is still a bit lazy. A similar circuit is in use for the frame and this has a similarly slow flyback [about 2 lines are lost in total].

Anyway is very later and I should be in bed....

Dom