Linear Sweep Using Tubes for 60-Line CRT Display

[cameramanlink]

This is my first time posting to this forum, and I have read some of the posts related to the topic of displaying NBTV signals on CRTs. I decided to join because I'm having difficulty with my display project and I'm hoping you can help me out.

Here's the story: I wanted to record video images onto audio cassettes and reel-to-reel tape, and I needed to build a device that would display the playback signal coming from the tape deck. I build exclusively with tubes (which you would know if you saw my YouTube channel) so I started with a converter box for displaying on a Heathkit IO-12 scope. After a modification to the front-end circuitry of my converter, I was able to get a pretty good image from my Akai reel-to-reel. I have no interest in 3 things: playback from the computer soundcard (defeats the purpose of displaying NBTV in my opinion - why watch low-definition unless there's a reason, like it was recorded on audio tape. Computers are used for too much stuff in my opinion), 32-line display (unusual aspect ratio and low resolution), and mechanical scanning (I'm only interested in electronics). Having said that, I'm having trouble with the electronic scanning specifically.

The 12.5 Hz vertical sweep is really giving all my oscillators a hard time with linearity. I've built 8 configurations of different oscillators in an attempt to fix the linearity issue, and none have come close to a good result. Here is what I've tried so far: positive-slope ramp mulitivibrator using 12AT7 with tied cathodes to a common resistor, "Heathkit" style negative-slope multivibrator with the R and C in the cathode circuit of the second triode (12AT7), standard thyratron circuit, Jean-luc's posted thyratron schematic for "Horisontal Timebase," a feeble attempt at a pentode plate load for the previous thyratron circuit (with plenty of exploded resistors and burned up pots!), a phantastron used in the "Toy CRO" project I found online (I changed some values around, but no luck), and even two solid state ramp generators using a 555 timer chip, one with a single transistor for charging the cap, and another using a "current mirror" using 2 transistors. I even tried using Audacity to generate a sawtooth tone at 12.5 Hz to see what it would look like. The waveform is perfectly linear on the computer's sample plot in the program window, but the soundcard's output very much follows the typical RC time curve when viewed on the scope.

After tweaking values in all the circuits, I've made the following observations: increasing the frequency will GREATLY improve the linearity, producing almost a perfect sweep in most of the circuits, especially the thyratron and "Heathkit" multivibrator. However, decreasing the frequency back to 12.5 Hz re-introduces non-linearity, even in the so called "constant current" circuits, such as the two solid state circuits. Opposite extremes of R and C to produce the same time constant have practically no effect on linearity.

Here's what I know from what I've read:
The initial portion of the RC time curve is more linear than the entire curve, so that part should be used. This makes sense, especially when increasing the frequency: it gives the thyratron less time to reach maximum charge on the cap and uses the initial part of the RC time curve. BUT, that also increases the frequency to too high a value. Here's a picture of one cycle produced by Jean-luc's thyratron circuit running at 12.5 Hz:

Notice that it starts to go flat, but is cut short by the thyratron firing at this point. This shows the concept, but I wish I could scoot that point back a little more to get rid of the "tail" on the curve. This is the most linear wave I've been able to produce.
The next photo shows a raster using this curve. The bunching up of the lines becomes very apparent at the top:

Note that the scope does not have an option to reverse polarity, so the raster is scanned upside down by this positive-sloping ramp. This forms the picture as played back from my Akai:

When dealing with such a low resolution, every line needs to be viewable, and a picture with this much distortion is just not watchable.
I also know that, in theory, a constant-current source for charging the cap *should* produce a *perfectly linear* ramp. Hmm...then why did the solid state circuits not do this using the two types of transistor circuits to regulate the cap's charging current? I just don't get it. This is one of the MANY reasons I don't believe in theory. I'll probably get in trouble for saying that, but it's been my experience that theory is basically worth nothing...

With this being my experience, here is my question: can someone please post a practical schematic of a thyratron relaxtion oscillator using a pentode as a constant-current source for charging the capacitor? Knowing the idea and theory hasn't helped me, so I'd like to see it as a circuit that actually WORKS the way it's supposed to!

Bryan

[Steve Anderson]

Hello Bryan and welcome to the forum,

You seem beset with linearity problems! First question...excuse me for asking this...but are you sure the scope you are using is linear itself? i.e. have you looked at the same waveforms using a different scope?

I am assuming that it's the same scope used for observing the waveforms and used as the display device...from the photos it appears that way. If the scope is the culprit it will affect all that's displayed, much in the manner shown.

Also, is the vertical input DC-coupled? I have seen effects like this whereby a 12.5Hz saw-tooth has been passed through one (or more) stages of AC coupling. This would possibly occur not only within an AC coupled scope, but also AC coupled sound-cards.

To illustrate this two simulations are attached, 1_6 Hz is a single RC circuit, 100nF & 1M, common AC input values for a scope....red trace is the correct sawtooth, the green the differentiated result. 3_2 Hz is the same, but with a 47nF capacitor. Don't be fooled by the -3db point! Which in the first instance is at 1.6Hz....almost eight times lower than the vertical rate/timebase....the -3db point is only for sinewaves, not complex waves like these...

The fact that faster saw-tooths appear OK points to this. (Should that be 'Saw-Teeth?)

Steve A.

[Steve Anderson]

Bryan,

I have just looked up the circuit diagram for the Heathkit IO-12 assuming that it's the device you're using.

It is an AC only scope and has FOUR AC coupled paths between the input and the CRT plates!! I think that's going to be your problem. Waveform or raster display alike.

Steve A.

[cameramanlink]

Steve:
Thanks for the fast reply and taking time to demonstrate your idea! At one point I thought it might be the scope, but I dismissed it because I had viewed the waveform on another "vintage" scope and the waveform actually appeared even less linear. The only 2 scopes I have are pre-solid-state and AC-coupled only. The Heathkit from the '60s is the better of the 2, with the second one I mentioned being a DeVry Technical Institute oscilloscope kit for a training course they offered during the '50s.

I really appreciate your answer as this is something I haven't investigated yet (although it seems I've tried everything else but the kitchens sink!). It would explain why ALL the circuits I tried produced the same results on the SAME scope. You are very insightful! The previously-mentioned DTI scope has unbalanced drive with deflection plates tied to ground and the opposite plates getting the signal. It has a quick disconnect point for the plates, so I will try connecting the thyratron's output directly to these plates with a large value cap, such as .47 or 1 microfarad.

I found the Heathkit's frequency range amazing, as it could reproduce the color burst from a color video signal with an amazing amount of clarity while preserving the perfectly straight steps of the "porches" test pattern. I never thought of the low-frequency end of the scope as being a problem with such a wide-band amplifier. But you're right: asking an amp to pass 12.5 Hz, especially sawtooth, is quite a tall order. This would also explain Jean-luc's use of large coupling caps and a SINGLE gain stage for driving the deflection plates.

Your answer has renewed my motivation as I had almost given up on this project! I will post once I have tried direct drive and take more pictures. Thanks!

Bryan

[Steve Anderson]

Bryan, no problem,

As most here are probably aware I'm somewhat of a 'Tube-Head' myself (but not exclusively), with an keen interest in electrostatic CRTs and 'scopes in general.

Looking at the IO-12 circuit there are actually FIVE AC-coupled points from Y-input to CRT plates, I missed one in my previous count-up. Two of these are worth modifying to improve this vertical linearity problem.

Biggest culprit is the 100uF/50V(?) capacitor connected to the 'hot' end of the Vert. Gain pot. First, if it's original it could well be below it's nominal value, a simple test is to tack a new one in parallel, and/or increase its value.

But better is if you could replace the Vert. Gain pot (2000 Ohms...why 2000? Doesn't need to be) with something like a 10k one. But this may not mechanically be so easy....see below...

Second culprit is the 100n that feeds the differential output stage. From pin 2 of V3 (12BY7)...also see below... trace back through the 100 Ohm resistor to a 100n cap, there's also a 3.3/3.5M resistor at this point. Put a 1uF/200V (non-electrolytic) cap in parallel with the 100n.

This should result in a noticeable improvement, but it still won't be perfect. At least it identifies the problem.

Steve A.

Added later...an error crept into the above, V3 is actually a 12BH7, a 12BY7 is a single Pentode. Not my error (for once), it's what it clearly says on the only copy of the IO-12 circuit I had....well up to a few minutes ago. Downloading a couple of others were noted with the far more suitable twin-triode 12BH7!

I mentioned replacing the Vert. Gain pot with a 10k one. This depends if you need to preserve the 5MHz quoted bandwidth for this scope. If you wish to retain the 5MHz bandwidth, leave it at 2k. It's that low to assist in circumventing the following stages' Miller capacitance effect. The bigger cap (470uF/50V) is a better solution.

Not only note Jean-Luc's use of large caps, but also large-value resistors! In one instance 2x6M8s in series, 13M6 in total. Though there is usually a maximum value of deflection circuit resistance specified. 5M in both the 5UP1 and the 5BP1.

The other illustration is to put a 12.5Hz square-wave into this scope, you'll see the classic operation of differentiation!

[cameramanlink]

Well, it turned out all along that the problem was my scope and not the circuitry connected to it. By connecting the output from the thyratron circuit directly to the vertical drive plate of the CRT in my DTI scope, I got this result:


Note that the line curves upward on the right slightly. This is because the DTI scope has its own nonlinearity at low frequencies, and scans left to right. This produces a slowing down of the trace that otherwise rises linearly. This photo shows how a sequence of lines looks when scanned with direct drive, and that I truly am getting a linear voltage rise:

Notice how the lines are uneven in places, but do not bunch up at the top like they did with the Heathkit scope. The unevenness is due to interference in the scope itself, which I think is caused by the power transformer. These photos were taken with the thyratron oscillating at a frequency lower than 12.5 Hz. I was even able to get a sweep so low that it would look like a horizontal line that would rise from the bottom evenly, then re-appear at the bottom!

Steve:
Thanks for the info on the Heathkit! Now all I have to do is build the output amp to drive the plates to make sure that will work properly. I'll try changing those parts you talked about in the Heathkit. I have the same schematic printed out that says there's a 12BY7 tube rather than 12BH7. But I have the original schematic from the assembly manual and that says 12BH7. I wonder why someone would change it...strange.

The Heathkit uses a combination of point-to-point wiring along with PCB construction, which is difficult to trace without flipping it back and forth from top to bottom. The big electrolytic cap should be easy to spot, though. I also think the vertical gain pot should be easy to change because it is next to the edge of the panel rather than the middle. I don't need the "5MHz" that it claims to have, as I don't look at any frequencies that high. I don't do much of anything in ways of radio or transmission: too much stuff that goes wrong, too many coils to find/make, and too much frustration trying to get the darned thing to work!

Next I'll post about the project I plan to build.

Bryan

[cameramanlink]

I plan to build the 60-Line monitor project into an old Heathkit scope chassis and cabinet. About 4 years ago, I scrapped one for a replacement frequency selector switch that is in my current Heathkit scope (IO-12, the switch had frozen). The project I had built into the chassis was called the "Simple Oscilloscope" which used a thyratron oscillator (how I have a thyratron tube in the first place) and unbalanced drive. There was no sync whatsoever, and the waveforms displayed would come out distorted and dim. Changing the brightness would de-center the trace! But anything less than full brightness was not viewable. This project was an experiment at building a REALLY simple scope for fun, but it turned out to be a waste of time. So now I plan to turn it into a 60-line monitor instead.
Here is a view of my test setup using the partially-scrapped chassis and the thyratron oscillator direct-driving the DTI scope's vertical:


I'll try building Jean-luc's circuit first, changing the horizontal to vertical, and vertical to horizontal and increasing the line rate. All knobs will come through the front panel as there are PLENTY of holes! The graticule will be removed so I don't see a grid over my picture. The only problem with this is that the screen has less contrast and needs a reduction in ambient light for better viewing. Any ideas for a green transparency like the graticule but without the lines? I'm thinking maybe a gel for theater lights.
No spot wobble will be built unless absolutely necessary and I feel brave. To me, that is a luxury at this point.

The audio amp will also not be built as mounting a speaker inside the Heathkit scope cabinet could be tricky, especially making a place for it where the magnetic field would not interfere with the CR tube. Then there's a matter of making holes for the sound to come out in the side of the metal cabinet and not having noise from the oscillators come through the speaker (12.5 and 750 Hz). Instead, there will be a pass-through jack for the audio, where the white and red stereo cables will connect to respective jacks on the front, and another red jack will be the "Audio Output," which loops the input around to this jack. This will allow for minimal interference and more monitoring options for high quality sound when desired. Video signal will be on the white (left) cable as standard.

The goal is for playback from 1/4" reel tape and compact cassette, but Jean-luc's design has plenty of gain to be driven from a computer soundcard's relatively low output. In previous experiments I had problems with overdrive from the tape deck because it has a lot stronger output than the soundcard. The use of a potentiometer before the first gain stage will ensure proper gain adjustment depending on the signal source.

Unlike regular video, horizontal sync can be tempermental, especially with tape playback as the amplitude of the signal varies over each frame and with changing scenes. Several adjustments will be made for sync, such as a sync amp control that will adjust plate voltage on the sync separator circuit. I found this especially helpful when using my experimental circuit for playback from tape. I got consistent h-sync without deforming the picture. It has been tricky at times, but not so frustrating as to the point of having to be constantly adjusting it. An adjustment for sync drive to the thyratron might also be helpful to deal with varying levels of sync.

Vertical sync has been the most difficult, as there is no vertical sync pulse, only a missing horizontal sync pulse. The frame rate is so low that there is only one retrace line - which makes sense because this originated as a mechanical system. Right now, I can get the vertical to stop rolling some of the time, but changes in the scene cause it to lose sync. Also, the picture will go out of phase randomly, with 2 frames displayed at once with the gap in the middle of the screen (top half of the picture on the bottom, bottom half on top). I think this is due to a lack of a definite vertical sync pulse. To get around it, I will "letterbox" my videos slightly before conversion. This will produce a point where no signal is present between frames and may give the thyratron a point to latch onto. This produced a better result in the past with letterboxed 16:9 videos.

I will post more as I make progress on this project. I'm sure I'll have plenty of questions!

[Steve Anderson]

Bryan,

All very interesting, I will certainly be keeping track of this thread as it develops...

I assume that not only the chassis and tube came from an old oscilloscope, but perhaps other parts too, e.g. power transformer. Is the 'donor' scope a Heathkit? If so, do you know the model number?

It's interesting to stab the calculator and derive a few figures for your system..

This I have gleaned thus far...please correct if wrong...

12.5Hz vertical timebase, top-to-bottom sweep,
750Hz horizontal timebase, left-to-right sweep, resulting in 60 scan lines,
A sync-pulse system very similar to the NBTV Club arrangement, with a 'missing' pulse for vertical retrace.
It's hard to tell, but maybe an aspect ratio of conventional TV, i.e. 4:3, slightly landscape.

Now given a degree of latitude, a guess at the required bandwidth is around 30kHz. With a high speed reel-to-reel you could achieve some 70% of that, but I doubt very much via a cassette.

The other thing, and I ask this on behalf of other members...where and how do you acquire your source material as shown in the videos and in the photos here?

Steve A.

[cameramanlink]

Thanks for your interest, Steve.

Yes, many of the parts are salvaged from the donor Heathkit scope I used for replacing the broken horizontal frequency selector in my IO-12. The donor scope, I believe, was a model O-10 or O-12. I think I had one of each, both of which I recycled. Now that I think about it, I don't remember how I ended up with two Heathkits of similar models in the first place. One of them was in pretty beat up condition and didn't work too well (the O-10 I believe - it was a few years ago and back then I didn't keep documentation of my projects). This may have been the scope for the donor switch. All I know is that this scope's chassis, cabinet, transformer, and CRT have been recycled into my "Scope-O-Vision" project, which is a circuit for standard definition display in "green and black." [I haven't posted that project to the forum because it's normal bandwidth TV, not NBTV). The O-12 is the chassis you see in the picture of my setup (updside down next to the DTI scope). This one was also called a "Laboratory Oscilloscope" and had a similar layout to the IO-12. The cabinet was in rough condition in terms of paint and appearances, so I spray painted it with a color I call "oscilloscope gray!" Heathkit scopes are great for recycling into projects like this and are plentiful on eBay. I found the peaking coils especially useful in the "Scope-O-Vision" project's video amp. It produces a sharp picture [relatively], even though the lines blur together a bit.

As for the parameters, you hit the nail on the head, including the 4:3 aspect. I find this shape more useful than the tall ratio used with 32 lines, especially as a videographer having worked with this ratio for as long as I can remember!

I knew the bandwidth would be pushing it with 60 lines. I don't plan to put much on cassettes; it's more for the "cool factor" of playing video from audio tape (why I started in the first place). My reel-to-reel unfortunately doesn't do 15 or 30 ips, which would be ideal. Perhaps if I used high quality tape at 7.5 ips? It's definitely better than cassette. I tried that once and things looked quite blurry. I might even be willing to go with CD or "DVD Audio," just to see how it works...just not into the computer playback thing as a habit, but for testing that's fine. I like putting something in a machine and pushing play, partly because I'm old fashioned. Quick thought: how about overcranking a cassette drive to double or triple the speed?

I'm not so much interested in preserving the entire bandwidth as having a recognizable image coming from tape/CD. If 70% is what I'm getting with my Akai (model GX-280D-SS) machine, that's fine by me. I'm not picky. But I do know that when I play back the raw 96kHz files from the computer, they look pretty nice! A studio tape machine would be best, but I can't afford one of those. Out of curiosity, how would I calculate the tape speed necessary for reproducing the 30kHz bandwidth? Helical scanning would be cool to do a freeze effect and would produce plenty of bandwidth...but I'm getting ahead of myself

What do you mean by "source material" shown in the photos? If you mean electronics parts, I buy online from Antique Electronic Supply. But if you mean the photos themselves, they are linked to my Photobucket account.

I'm going to start "hacking" into my Heathkit IO-12 now and replace some of those caps.

Bryan

[Steve Anderson]

What do you mean by "source material" shown in the photos? Bryan

I mean the source of the video signal. I realize that it's played back off tape or (heavens forbid) a PC sound-card. But how was the 60-line signal created? It appears to come from some type of camera in the first instance, not just a bunch of gates or software creating test-bars or similar.

Steve A.

[cameramanlink]

Thanks again, Steve. You've been a great help to me in this project! I just finished changing out the parts you suggested and it made a noticeable improvement. Here are some photos of the changes I made:

Original Input Section:



Replacing the Potentiometer with a 10K Linear:


Replacing the Electrolytic Capacitor:


Adding 1mfd Capacitor in Parallel with the 100n:



Improved Waveform with the Thyratron:


Raster with Thyratron:


Video Image with Converter Box (multivibrator):


Everything worked just the way you said. This made quite a noticeable improvement, but it is not as good as the direct drive test I did earlier today, which is to be expected. I am very happy with these results as it makes it more enjoyable to watch from the converter box.

To answer your question about the video source, I use the Video2NBTV application from this website http://users.tpg.com.au/users/gmillard/nbtv/nbtv.htm
It takes in a video clip or live feed (like a webcam), then either makes a wav file or outputs through the soundcard in real time. This is how I put the video on the tape. Yes, the video is recorded onto the tape from the soundcard.

This is totally off the subject, but I also wanted to build a NBTV camera using an iconoscope tube. I found a website that had plans for an iconoscope camera for the "home experimenter" in October 1940. The particular iconoscope tube used is electrostatically-deflected, and runs on the same sweep generators as the monitor, keeping it in sync with the monitor tube, which was similar to a 2AP1. It included circuits for generating sync and blanking signals, and mixing them together to form a video signal. I figured if I could construct the circuits shown in the diagrams, and change the sweep generators to produce the 60-line standard I'm using, I could make an all-electronic camera for NBTV. The tricky part is getting the 1940s iconoscope and some other parts mentioned. The scans are also in low quality and are difficult to read in places. If successful, I would have a *computerless* video recording and playback system
Just a thought, but I don't think I can do it. And if I think that's hard, what I would REALLY want to make is an [get this] all-tube, COLOR, 1080p, video camera, tv monitor, and VTR for recording and playing back full HD video in a completely analog system. Probably not possible, but I can dream on...

[Steve Anderson]

Thanks again, Steve. You've been a great help to me in this project! I just finished changing out the parts you suggested and it made a noticeable improvement..

To answer your question about the video source, I use the Video2NBTV application from this website http://users.tpg.com.au/users/gmillard/nbtv/nbtv.htm

An element of success, good. You could go through the other stages of this scope one-by-one making improvements, but if you intend to build a purpose-made monitor for 60-line NBTV, is it worth it? It's proven the point about the LF performance required from circuits handling NBTV, so I would tend to leave things as they are. But that's me, maybe not you.

GMillard is a member here often posting on various NBTV topics including his software when questions/suggestions pop up regarding it. You'll notice him often 'lurking' (meant nicely) by being listed as logged-in near the bottom of the Home Page here, his user-name is simply 'gary'.

GL, another member (user name 'Ancient Brit'), made a NBTV camera some time back using a Vidicon which scanned the target at NBTV rates, i.e. 32-lines vertical and so on. It wouldn't take much to modify it to the system you use. I agree, a 1940s Iconoscope is going to be a very rare beast indeed. If one could be found (that works) it will be horrendously expensive, but that's unlikely as their life was only a few dozen hours at first!

As for the 1080 system...I think you'll be dreaming for quite some time yet...but then again, I've often been proven wrong!

Steve A.

[cameramanlink]

I agree. The IO-12 is my general purpose scope, so I don't want to make lots of changes just to improve low frequency response. I've learned something new, and that's what matters for my next project. I would have thought after building 3 standard definition monitors successfully, NBTV display would be a piece of cake. I now know there's more to it than I thought, but I'm glad to always be learning. They don't teach you this stuff in school anymore!

Perhaps I will start a new thread for an all-tube camera, but I don't want to get ahead of myself; I haven't even built the monitor yet. The only reason I am afraid of normal types of camera tubes is the focusing and scanning coils: no idea what they look like, how they work, how to drive them, or how to adjust them once the circuits are working...scary! If you know of any newer all electrostatic camera tubes, I would feel a *little* more confident.

Now comes the long task of compilng a parts list, developing a layout and wiring diagram, and placing yet another order to AES. They're probably wondering what's happened to me and will be glad to know I'm still alive after over half a year of no orders [Yes, I build a lot of stuff and fast; but with long times in between "soldering binges"]

I'm going to start a new thread for the actual project construction, as the topic here is linearity, which has been solved!
Time to get moving...

[Harry Dalek]

Just a question as i have noticed now and again photos and such posted like here in the past have gone missing ? Which i think happened when the new look forum was put back up .
I am wondering if they are gone for good ?

[Steve Anderson]

Well, the board upgrade was just over a year ago so if photos, drawings etc. are absent I doubt they could be recovered. There isn't a way I can see that I could do it, but maybe Andrew could with his Administrator status rather than my own Moderator position. Somehow though I don't think so.

I guess it's best to download/copy stuff when it appears, nothing on the 'net is permanent. For a similar reason I don't like the concept of the 'cloud'.

Steve A.