Crt tube blurring.

[dr pepper]

Finally got time to work on my oscilloscope tube clock, well radar tube.
The image was very blurred, turned out the deflection (elctrostatic) amp was getting noise over the supply from the switching converter I put together that generates the voltages for the anode, focus, brightness grid and cathode.
I got this to improve filtering the supply, however now I have this odd blurry image on one side of the tube, clear on the other, reversing the deflection plate wires inverts the image but doesnt invert the blurred side of the tube, moving the image around with the centering pots brings the blurred part clear if I shift the image all the way across, so what was the blurred side of the image is on the other side of the screen.
Have I got a dead tube?, or is my fairly small amount of experience with crts biting me again.
The clock uses vector graphics from an esp32 dac outputs.

 

[unclejed613]

you could still have some deflection voltage leaking into the focus voltage. are the voltages developed on a voltage divider? you should really have power supplies for the static tube voltages separate from the power supply for the deflection amps. normally the only voltages derived from a single voltage divider are the anode, focus, and brightness.

another cause, could be something off with the tube geometry, but that's pretty rare unless you bought the tube from ebay and the tube was a factory reject.

 

[rjenkinsgb]

Do you have a transformer or anything with a permanent magnet anywhere near the tube? A stray magnetic field can easily mess up a CRT display.

 

[dr pepper]

I'll see if I'm getting any noise on the focus grid.
All the high voltages come from a uc3842 flyback power supply driven off 12v, all of the voltages are tapped from one winding, with their own rectifier and cap, of course only one has feeback, and that is the deflection amp power.

If I rotate the tube on the spot the blurred side doesnt change, wouldnt it do so if it was an external magnet?

I'm using an Lf358 as a follower buffer, its non inverting input is biased via a 220k resistor to 1/2 vcc, then the input is applied through a cap.
220k seemed high so I lowered it right down to 10k, the image overall is now clearer, but still out of focus on the clock No.8.
I have 2 tubes, the other one is more or less the same.

 

[rjenkinsgb]

If the blur rotates with the tube, it must be internal to that, or from some form of interaction between the signals to the tube.

You could try coupling a trace of one of the deflection plate voltages to the focus to see if you can null it out?

Another thought - are you driving the deflection plates symmetrically or single-ended.
A single ended drive would vary the average electrode voltage and could affect other things?

 

[dr pepper]

Yes maybe trim some of the interfereing voltage back to the focus grid,if it is that.
The deflection is symetrical, I'm using a long tail pair made from ztx458', one tranny is biased with a pot which controls image posiiton, and the other has the dac signal from the esp applied, I wonder if theres an impedance thing, the pot is 50k, the signal side comes direct from an op amp so will be very low imp, might try a 47k resistor in series.

 

[dr pepper]

I removed the op amp altogther, and connected the dac from the Esp direct to the deflection amp via a 5k pot 'volume' control.
It worked, number 8 ever so slightly out of focus, but acceptable.
Then over time it blurred a little, still not so bad, but the image is now clipped, seems like the o/p of the esp isnt happy, its the same connected to my 'scope in xy mode, without this circuit connected.
My original Esp32 module literraly blew up I accidently connected 400v to the dac o/p, oops, as a temp measure I'm using a Esp32 cam module, which doesnt have gpio25 & 26 (dac out) brought out to pins, so I soldered wires direct to the chip, I'm now thinking there might be some analogue stuff on the dev kit module that I'm missing, will have to wait till the couple I oredered show up.

 

[rjenkinsgb]

It's interesting that the blurring coincides with the areas that are clipping - it looks as if something in the drive circuit may be struggling to produce a stable voltage with the deflection one polarity?

 

[dr pepper]

Yep I think your right.
The schematic for the Esp32 dev module doesnt show any hardware on the analogue outs, but I'm thinking there is.

 

[rjenkinsgb]

I've not used any of those, so no idea on the hardware, sorry.

Fantastic project, by the way!

 

[dr pepper]

In the post box when I got home, 2 new Esp 32 dev modules.
They are different, the image is much better, however I tweeked the circuit a little, I put a non inverting unity op amp buffer with a op amp virtual ground inline with the signals just after they come from the Esp.
The image is now probably as good as the 1947 therminoic picture tube gets, the retrace lines dont show with the naked eye, and the face isnt quite round, I'm having issues with trim pots shoved in breadboard.

 

[unclejed613]

you can probably eliminate the retrace lines by adding some negative bias to G1. that's the thing about vector graphics, you have to drive G1 into cutoff pretty hard to eliminate those retrace lines

 

[dr pepper]

Well you cant see them, just the camera does so I'm not that bothered.
I was originally thinking of adding a blanking routine in the code , and either modulate G1 or the cathode itself with a transistor.
The datasheet for this tube wants you to run the cathode at 200 volts negative, 300 volts (already used for G1) would probably turn the beam off.
Might do that after I've run it a while and the retrace appears.
Believe it or not that tube is well used, theres a failry large burn blob in the centre, the second hand shows it on the pic, again in real life you have to squint to see it.
Fairly pleased with this project, I'm going to dig round for some old ply and bakelite knobs to make a stand with.