CRT Clock

stan74

Senior Member
Looks retro,nice valves. Is that a ht diode with the cap on? The other's not a double triode. Electrostatic deflection on those or a yolk?
Very playaboutable.
 

techElder

Well-known member
You can see the trace lines as it moves from one image to the next, ...
This visibility of those parts of the trace are indicative of a slow edge on the blanking / intensity signal. You'll need a real scope to see that.
 

binary1248

Senior Member
Looks retro,nice valves. Is that a ht diode with the cap on? The other's not a double triode. Electrostatic deflection on those or a yolk?
Very playaboutable.
Yes, that is a HV tube, 1Z2, filament and plate only. The other is 12AX7, dual triode.
.
.
This visibility of those parts of the trace are indicative of a slow edge on the blanking / intensity signal. You'll need a real scope to see that.
There is no blanking,it simply steps to a new position and pauses for each dot in the display. If I turn down the intensity a small amount, the retrace lines disappear.
I am still working on the circuit (tracing out the board).
EDIT: I will scope the control signal on the cathode ckt and see if there is any blanking attempt.
My correction: I believe it is blanking between dots, but it will take me awhile to figure this out. I will post some scope images later of the control signal from the uChip and after the HV conversion. to see what distortion takes place in the HV circuit.
.
aaaaile0 - Copy.jpg
.
Well, from the attached waveform it is obvious (I think) that the circuit is very slow on turning off the beam, so between dots or groups of numbers, it just isn't fast enough. I think I know a way to fix that since it is just an open collector drive to the cathode (????). Maybe just some pullup resistor.
.
Well criminy, I need to figure out which way is on and which way is off. Be back later with an update.
 
Last edited:

binary1248

Senior Member
Well, this is strange, at lest for me. There is no pullup on the cathode NPN transistorcollector, it apparently relies on the cathode electron emissions alone. Yes I double and triple checked the circuit, see attached.
Moving the cathode voltage in the plus direction turns off the display, from minus 300V display on, to minus 280 volts display off.

.
IMG_0304.JPG
 

AllyCat

Senior Member
Hi,

Yes, electrons are being emitted by the Cathode but electrons have negative charge. So the "conventional current" flow is in the opposite direction, i.e. from the cathode and through the collector of the transistor.

Cheers, Alan.
 

binary1248

Senior Member
Yes Allycat, but my old school thinking still thinks of the electrons being boiled off the cathode and being attracted to the CRT aqudog coating (or positive plate voltage) which is at a high voltage plus potential. So my thinking is if the electrons are boiled off and/or emitted from the cathode and stream toward the positive plate (electron tube or valve). That's s my old fashion way of thinking which has gotten me thru years of electronics design. If I think of electron holes moving opposite of electron flow, it works for me.
 

techElder

Well-known member
My correction: I believe it is blanking between dots
Yep, and I believe the designer of that circuit couldn't figger out a way to do it, either! :D

Look at some other examples at other CRTs. I've seen them AC couple a blanking pulse into that circuit, but they weren't trying to display arbitrary dots.
 

AllyCat

Senior Member
Hi,

It doesn't matter how you conceptualise it (e.g. "holes" are the absence of electrons so travel in the opposite direction) but electrons have a negative charge so the Conventional current flows in the opposite direction to the electrons, i.e. from Anode/Collector/Drain (for "Tube"/NPN/N-channel) to Cathode/Emitter/Source.

Assuming that the CRT grid is correctly biassed (more negative than the Cathode, to repel electrons) and provided that the PNP (Q11) and perhaps NPN (Q10) are rated to 300+ volts, then that blanking circuit (above) should work (but perhaps only slowly). How could cathode current flow in Q10 if Q11 is fully "off"?. But fast, 300 volt transistors weren't cheap when CRT 'scopes were being developed, so ac coupling was an easier solution. It may well be adequate anyway, particularly if a (semiconductor) diode is used to "dc-restore" the blanking signal level (or the white level if you want to protect the screen's phosphor)..

With a 'scope, the (direct coupled) deflection bandwidth is usually (always?) the highest priority, so the Y (and X) deflection plates are biassed close to their amplifier's optimal output voltage (i.e. close to zero volts with transistors). That means the cathode and grid must be at a significantly negative voltage, but you could apply blanking to either (or both).

In CRTs (and some really old 'scope tubes) the aqua"dag" is connected only to the Extra High Tension supply. But better 'scope tubes use the resistance of the dag to act as a potential divider between EHT and "HT" rails to give the required potential gradient for "Post Deflection Acceleration".

Cheers, Alan.
 

binary1248

Senior Member
The scope trace above shows the digital signal (direct from microchip, 18f26K20) as 0 to 5V, the lower trace is on the cathode of the crt (2BP1) and goes from minus 300 to 0V (zero is at top). The transistors are 400V Vce rated. At minus 280 volts on cathode, the beam is cutoff enough the trace is no longer visible.
Keep in mind I didn't design this, it is a bought board and some of the things on it seem strange, but I expect to be making some mods to conform to my design requirements, also more voltage to handle larger electrostatic deflection tubes (if needed).
Thanks Allycat
Paul
 
Last edited:

binary1248

Senior Member
Here is a screen shot of the one of the deflection plates (X axis) which my bench scope is set for 100 volts/div, gnd being at the little red arrow at the bottom. The other plate is at a constant plus 250 volts, adjustable for position.
The other signal is the digital signal that turns on or off the dots, high being dot on. Blue arrow is gnd.

bbbFile0.jpg
 
Last edited:

binary1248

Senior Member
Now that I have the circuit board figured out, I am going to try substituting PICAXE chips. In the block diagram the micro. The attached block diagram shows the pinouts that are currently out of the microchip cpu's, with the substitution PICAXE chips shown below. First evaluation it seems they are pin compatible except power and ground. I will make adapter sockets so I don't have to cut any traces on the board.
I need to order 28x2 as I have none of these in my PICAXE inventory.
I do have some 20X2 so since I have to make a adaptor maybe I will try to use the 20X2. Only 20 pins were actually used on the U1 (18F26k20)

BlkDia.jpg
 
Last edited:

binary1248

Senior Member
Like so many projects I start, once figured out and tested I tend to lose interest. Just rec'vd my new logic analyzer (LA2016), so interest has been boosted. Here is a screen shot of the two programmable resistors (MCP4132 10K) being fed data from what will be a picaxe 08M2 eventually.
.
pstnshftr2.jpg
(Click on image for larger view.) Trace 1 is the 16 bit clocking, trace 2 is the data stream. The first 8 bits are always zero (normally used for special configurations) the 2nd group of 8 bits are the position data sent to the programmable pot every 5 or 10 seconds.
Trace 3 & 4 are the same but for the other programmable pot.

.
Doing the image shifting with external devices doesn't seem necessary but simplified the main code (not mine) that generates the clock images, since it could be done in the main code that drives the deflection plates for image generation.
 
Last edited:

binary1248

Senior Member
Well, I am going to skip the shifter circuit, very simple, will come back to later.
Now I want to examine the character formation provided by the dual DAC signals.
Here is a capture of the wr, a/b select, and a few data bits. AD write occurs going low.
Click on icon to expand
.
ADclkData.jpg
The write to the two DACs occur rather quickly, in about 2 us, but the update between new dot positions is about 30 us. So I have to think or do some testing to see if the 2X2 or 40X2 picaxe is going to be fast enough.
 
Last edited:

binary1248

Senior Member
Here are unedit shots of the write cycle between AD which is very fast, about 1.8 us. But dot data is updated about every 30 us.
First screen capture shows write time between two AD is 1.8 us.v
Second screen shows dot position update is about 30 us.
I am guessing the write time between the A B A to D selection is not that critical, so a few uS should still work.

ADwritecycle.jpg
ADcycle.png
 
Last edited:

binary1248

Senior Member
Well, looks like I have to abandon this crt clock using picaxe. Just too slow, even with 20x2 overclocked at 64mhz.
Code:
        ;------------------------------------
	inc b1					;+
	readtable b1,var_out			;+-- This code segment takes 100 us to execute
	;send byte0 to b.0 thru b.7		;+    with 20x2 #freq m64
	let pinsb=var_out 			;+ to fetch and send two bytes to the dual DAC
	;------------------------------------
 

Buzby

Senior Member
Well, looks like I have to abandon this crt clock using picaxe. Just too slow, even with 20x2 overclocked at 64mhz.
Don't give up yet, you've got a great testbed for ideas there.

This is a picture using 128 step sine waves at about 100uS per step. The overall update is about 65Hz, so that's over 16K dots per second.

MS215_Circlea.jpg

I'm sure this is fast enough for a simple clock display.

I've not had chance to do much PICAXE work lately, my sample&hold vector generator is not ready yet, so that display is just dots.

Cheers,

Buzby
 
Last edited:

binary1248

Senior Member
Thanks Buzzby, I needed that. I will continue and see how it goes.
First test will to generate a single small character. I thought I would use an 8X8 grid array for character layout, same as I used when writing programs for a ticket printer.c
 

Buzby

Senior Member
[video]https://youtu.be/cBj6rggZA6A[/video]

This is a PICAXE 28X2 running at 64MHz, driving two discrete R-2R DACs, fed into a vintage CRT oscilloscope.

The lines are made of pixels, not vectors.

I'm sure it could make a clock, but clocks are so old hat, I want Battlezone !.

It will have to wait for the sample&hold for that though.

Cheers,

Buzby
 

binary1248

Senior Member
Thanks Buzby,
That is just what I needed to encourage me to continue,
Waiting on some headers so I can wire in my 20X2 into the exsisng soket on the board I have.
Paul
 

Buzby

Senior Member
That looks interesting. How did you get the voltage levels for the dacs, from a table?
Hi Stan,

I wrote a bit of VBA code in Excel to make a list of pixels for a specified line, then copied them into a few macros in PE.
A couple of nested 'do loops', and I've very nearly got a clock !

line.JPG

The code was based on Bresenham's, but the next task is to make circles. There is a Bresenham Circle algorithm, but it might be quicker to code if I just use Excel's trig ability.

Cheers,

Buzby
 

stan74

Senior Member
Ah yes,I was wondering if you'd used line algo. Mid point circle algo..no trig integer math. I got 2 12c 8pin a-d to try something along this thread. There's code for rotating 3d cubes etc. and you tube vids of pic-32 animation on lcd. What's simpler is the audio to graphics that some guys done.
What happens if you just give a start and end x/y in a loop,doesn't pov make a line?
 
Top