NBTV without any moving parts. THE NIPTRIX

Forum for discussion of narrow-bandwidth mechanical television

Moderators: Dave Moll, Andrew Davie, Steve Anderson

Postby Viewmaster » Tue Jan 29, 2013 7:29 pm

Steve Anderson wrote:Albert, are you saying that three complete lines are missing? All 48 pixels dead? i.e. you only have 29 lines. Or three pixels dead, only 45 pixels in each line but 32 lines?

Whichever way, is it consistent? Lines 1,2,3 or 30,31,32 or pixels 1,2,3, or pixels 46,47,48?
Steve A.


Sorry Steve for describing this fault badly.

It is 3 horizontal 'lines' missing, not verticals on the 32 lines.
(Its the 2nd, 4th horizontal pixels line from the bottom and also near the top..........see piccy.

It is being caused by low o/p from 3 transistors in the 48 line clocking, so that at each pixel at that horizontal point clocked is lowish o/p.
I changed the relevant 4017 and transistors but no cure yet. I haven't devoted much time to this but when I do will let you know more about what I find.
Attachments
three lines missing.JPG
Three horizontal lines missing
three lines missing.JPG (11.06 KiB) Viewed 16212 times
“One small step for a man,"......because he has Arthritis.
Albert.
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Steve Anderson » Tue Jan 29, 2013 8:12 pm

Albert, although I know your sketch isn't drawn to scale, something funny is happening with pixels 2, 4 and what appear to be either 31 or 32, binary numbers. why this isn't occurring on 8 and 16 too is a mystery.

I don't know how you've done the cascading of the 4017s but I would assume the same method would have been applied to the line counter (which seems OK) and the pixel counter. Are you getting a healthy pulse out of the appropriate 4017 pins? If so (or not) it's a 'silly' somewhere...it happens.

The 4017 has a rather strange arrangement for cascading as shown in the attached datasheet (Page 11, Section 12). But seeing however you've done it for one axis should work for the other.

Steve A.
Attachments
HEF4017B.pdf
(122.44 KiB) Downloaded 644 times
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Postby Viewmaster » Tue Jan 29, 2013 8:33 pm

Steve Anderson wrote:Albert, although I know your sketch isn't drawn to scale, something funny is happening with pixels 2, 4 and what appear to be either 31 or 32, binary numbers. why this isn't occurring on 8 and 16 too is a mystery.

I don't know how you've done the cascading of the 4017s but I would assume the same method would have been applied to the line counter (which seems OK) and the pixel counter. Are you getting a healthy pulse out of the appropriate 4017 pins? If so (or not) it's a 'silly' somewhere...it happens.

Steve A.


There are std methods of cascading, one of which you showed.
Mine is quite std with a small mod to reset everything, which is working OK and would not be the casue of this problem

It is a 'silly' somewhere as the relevant 4017 was changed and the 2N3904s too. 4.7k base Rs were also checked OK. There was less volts on the tranny base, seen on a scope, on the low o/p pixels as I recall. It was a few weeks ago when I first investigated this so I will have to definately recap on all this.

BTW, when standing straight in front of th Niptrix the picture is too dazzlingly bright with the under run LEDs, so I tried some club diffuser sheet in front (this was suggested by Harry, I seem to recall, some time ago).
This also allows one to view the picture at greater angles so I may make that a permanent feature.........or have half at full dazzle and other half diffused :shock:

Good Gamma correction to be experimented with too, soon.
“One small step for a man,"......because he has Arthritis.
Albert.
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Steve Anderson » Wed Jan 30, 2013 5:01 pm

Viewmaster wrote:BTW, when standing straight in front of th Niptrix the picture is too dazzlingly bright with the under run LEDs...

This was one of my worries as I expressed earlier in this thread, but it would appear unfounded. I guess I should have taken note that at least two other displays of this nature have been done by others who don't mention anything about the display being dim.

"Anderson. Pay attention!"

Steve A.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Postby AncientBrit » Wed Jan 30, 2013 6:28 pm

Albert,

What is the duty cycle for each LED?

The reason I ask is that you mentioned you suffered 'infant mortality' of a few LEDs.

I presume there must be an upper safe limit to how much current you can pulse through each LED even if the average current is well within the specified steady state limit.

Regards,


Graham
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Postby Viewmaster » Wed Jan 30, 2013 7:59 pm

AncientBrit wrote:Albert,

What is the duty cycle for each LED?

The reason I ask is that you mentioned you suffered 'infant mortality' of a few LEDs.

I presume there must be an upper safe limit to how much current you can pulse through each LED even if the average current is well within the specified steady state limit.
Graham


Graham, each LED is clocked at 19.2kHz so each LED is switched on for approx 50u sec....but bear in mind that any particular LED may not be on full or even on at all, depending on the video content at that time.

They are 3v LEDs.
They now have 800 ohm in series the 12v supply so you can see that they are well under run. Additionally each LED is in series too with 2 off 2N3094 trannies so more effective R there as well....see sketch.

Steve......those other displays you mentioned both used coloured LEDs and were well over run, one at 80m/a and the other at 120 m/a.

added......sketch shows wrong collector/emitter arrows on one tranny.
They are BOTH NPN 2N3094's
Attachments
Basic cct of LED..JPG
Basic cct of an LED
Basic cct of LED..JPG (23.02 KiB) Viewed 16155 times
“One small step for a man,"......because he has Arthritis.
Albert.
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby AncientBrit » Wed Jan 30, 2013 11:32 pm

Thanks Albert.

The LED currents seem conservative so it shouldn't be a problem.

Cheers,

Graham
AncientBrit
Green padded cells are quite homely.
 
Posts: 858
Joined: Mon Mar 26, 2007 10:15 pm
Location: Billericay, UK

Postby Viewmaster » Fri Feb 01, 2013 12:39 am

The handbook shows the gamma correction cct for a 100 m/a LED current .......see piccy.
Would I be right in assuming that for just a 10m/a LED current then the R values could just be increased by a factor of 100/10 , in order to maintain the 1 volt peak as shown on the cct ?
So the 33 ohm becomes 330, the 15 becomes 150 and the 12 becomes 120.
Attachments
Gamma correction.jpg
Gamma correction for 100 m/a
Gamma correction.jpg (12.64 KiB) Viewed 16104 times
“One small step for a man,"......because he has Arthritis.
Albert.
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby gary » Sun Feb 03, 2013 3:30 pm

According to my calculations that is correct Albert - in fact due to the lower current the forward voltage (according to data sheet) of the Schottky should vary just enough to make the instantaneous current almost exactly 10mA at 1V, whereas at the original values I would expect it to be around 90 mA assuming perfect resistors.
gary
 

Postby Viewmaster » Sun Feb 03, 2013 6:15 pm

Thanks for that Gary.
In fact, as is usual with me, I left out a critical short sentence.....
"Would it still work OK at these elevated values of R?
I tried it out and although there was some better gamma correction it is still not good, so I am going to up the R values even more.
“One small step for a man,"......because he has Arthritis.
Albert.
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Steve Anderson » Sun Feb 03, 2013 6:42 pm

Albert, I haven't jumped in here as I believe this was originally done by Klaas. I don't know how much research he did in terms of different peak currents and he may pop up in the near future to offer some advice.

I have done some Gamma-correction circuits but they operate as part of the signal path not within a driver such as this...they're also much more complex. Simplicity being the elegance of design I would stick with this and somehow fudge the values to suit.

Perhaps changing to smaller (signal type) diodes may be a step in the right direction, say a 1N60 in place of the 1N5817 and a 1N4148 in place of the 1N4001...just a thought...also perhaps a smaller MOSFET? (2N7000). But these are just ideas thrown into the air...the resistors would of course need scaling, but perhaps not an exact 10:1 ratio...a real 'suck it and see' exercise...

Steve A.

Interesting to note that some call the 1N60 a Germanium device, whereas others as a Schottky diode, either way it has a lower forward voltage than a similar Silicon version. Cricklewood UK have then in stock...but not cheap at £1.50.
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Postby gary » Sun Feb 03, 2013 8:56 pm

Viewmaster wrote:Thanks for that Gary.
In fact, as is usual with me, I left out a critical short sentance.....
"Would it still work OK at these elevated values of R?
I tried it out and although there was some better gamma correction it is still not good, so I am going to up the R values even more.


Hi Albert, well that's what I thought you meant.

What I am saying is those values will give an identical instantaneous gamma curve to the original - all other things being equal.

The caveat I gave is that the forward voltage drop of the Schottky varies a little with current so there may be some variation - but I doubt that it would be significant and if my interpolation is correct it should improve things if anything.

Now IF you meant will the gamma work (as in giving a better picture) well you are in virgin territory my good man and your guess is as good as anybody else's I would venture to say.

Personally I am not much of a gamma man and although the theory is perfectly sound, to my eye there never seems to be enough improvement to warrant it's use. OTOH others swear by it.

I have spreadsheet calculations and normalised curves if you want.

I haven't taken into account the effects of the PCM nature of your LED modulation and have assumed linearity.

I am sure I am teaching grandma to suck eggs here but for some it may not be obvious that the diodes in this circuit act as break points in the voltage-current curve and thereby approximating a curve by 3 straight lines.

See newsletter 26 volume one for Klaas's documentation.
gary
 

Postby Steve Anderson » Mon Feb 04, 2013 6:30 pm

Albert, I did a simulation as per the handbook (see .gif attached) and then simply inreased the three source resistors by tenfold. The green trace is as per handbook, the brown trace with the source resistors increased x10. The load resistor was also increased x10 to allow for the lower current.

The curve is negative as this stage inverts the signal.

I was surprised to find them very similar, just a slight reduction in the input drive and they match very well. So forget about changing diodes as I suggested.

Steve A.
Attachments
Gamma 1.gif
Gamma 1.gif (10.93 KiB) Viewed 16017 times
User avatar
Steve Anderson
"Fester! Don't do that to 'Thing'"
 
Posts: 5360
Joined: Fri Mar 30, 2007 10:54 pm
Location: Bangkok, Thailand

Postby Viewmaster » Mon Feb 04, 2013 11:30 pm

Once again thanks to the two NBTV stalwarts, Gary and Steve for suggestions etc re gamma, and also the graph, Steve.
I shall still experiment (that's more mucking about for me :) ) with values.

There is one more outstanding item I am still working on to complete
the Niptrix..........
A plus/mius one pixel incorrect clocking showing up as a kinda slight variable ghosting above and below full white areas going into full black areas.

The start of each 48 pixel line must vary slightly for a small part of the time every frame. The small part causing the ghosting effect. It doesn't show up on, say a portrait, but does on some test tracks of the club's CD.
Maybe insufficient decoupling somewhere.

It must be very slight erratic 48 clocking somewhere, so more mucking about for me! :lol:
“One small step for a man,"......because he has Arthritis.
Albert.
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

Postby Viewmaster » Tue Feb 12, 2013 12:58 am

Nearly there !
Just an interesting fault now that I am assembling it finally in its cabinet for filming it all.

At some pixel places, clocked at 19.2kHz a very slight brightening of a horizontal line shows. When the relevant driving transistor is disconnected from the board and wired seperately the fault goes.

It NEVER has happened to the 400 Hz clocked transistors.

My very, very tentative theory is that at the 19.2kHz freq there is more chance of some kinda capacitive/resistance leakage across to the neighbouring transistors which are very close together, so the faulty one is slightly brighter.
At 400 hZ leakage is much less and never shows on the 32 lines.

I have had to float 2 transistors from the board to avoid this fault, so hopefully that will be a permanent cure, no matter what the problem really is.
“One small step for a man,"......because he has Arthritis.
Albert.
User avatar
Viewmaster
Frankenstein was my uncle.
 
Posts: 1306
Joined: Fri Apr 06, 2007 4:50 am
Location: UK Midlands

PreviousNext

Return to Mechanical NBTV

Who is online

Users browsing this forum: No registered users and 19 guests