Construction Diary -- Part 1, The LED matrix/display

[Andrew Davie]

One expected a certain amount of standardisation. Now I know I need to check these things. How inconvenient -- and why on earth don't they actually mark the pinout on the part itself? Backwards it is. I'll reverse it tonight, and try once again

[Steve Anderson]

Andrew,

Just a suggestion...if you really want to make a start in learning electronics may I suggest a book written by Horowitz & Hill, called 'The Art Of Electronics'. ISBN 0-521-37095-7 (Hard-back version)

It's not a dry tome and is actually quite readable. Although the last edition was published in 1990 (I think) the vast majority of the content is still applicable today, just the sections relating to microprocessors and recent digital developments are well and truely out of date.

It's certainly more than enough to get to grips with the electronics used in NBTV and the like. Even today I often use it as a reference. They've kept the maths to a minimum although some is essential.

I hope it helps, and I hope you can get a copy. I wish it were around when I was studying!

Steve A.

[Steve Anderson]

Andrew,

Here's a link to the book...

http://www.artofelectronics.com/

Steve A.

[Andrew Davie]

Just a suggestion...if you really want to make a start in learning electronics may I suggest a book written by Horowitz & Hill, called 'The Art Of Electronics'. ISBN 0-521-37095-7 (Hard-back version)

OK, Steve, I'll chase that one up. As good a time as any to mention the website www.abebooks.com -- a very handy worldwide interface to second hand booksellers worldwide. I've bought quite a few books through this site, and have been very happy every time. Highly recommended.

I note that I can get a new copy from a seller in the USA for US$31 or so. I might have one sent to my in-laws to bring over next time they visit. The cheapest copy in Australia is an absurd US$141

[Steve Anderson]

Andrew,

Yes, US$141 is absurd, but even at that price it is really well and truly worth it! You'll have no need for any other reference material except for recent developments since the last publication. (pdf files from the component suppliers cover that).

If anyone is wondering, no, I have no connection with Cambridge Unversity Press or the authors.

Steve A.

[Andrew Davie]

Andrew,

Yes, US$141 is absurd, but even at that price it is really well and truly worth it! You'll have no need for any other reference material except for recent developments since the last publication. (pdf files from the component suppliers cover that).

If anyone is wondering, no, I have no connection with Cambridge Unversity Press or the authors.

Steve A.

I've been doing a bit of searching, and note that the softcover version of this book (2nd edition) is available on eBay from India for 1350 rupees, including shipping. That's roughly US$32. The downsides are that the printing is black and white, and the cover is softcover. Neither of those really bother me. Other than that, it's supposed to be word for word the same.

Thought I'd throw up this information for some opinions, before purchasing.

[Andrew Davie]

One expected a certain amount of standardisation. Now I know I need to check these things. How inconvenient -- and why on earth don't they actually mark the pinout on the part itself? Backwards it is. I'll reverse it tonight, and try once again

I switched around the transistor, and the LEDs *still* remain fully lit (though there is a very slight variation when playing with the brightness knob). The waveform across the LED array looks basically the same... I'm really starting to get a bit frustrated with this lack of progress. There must be something obviously wrong but I can't see it. My first real question, though, is that given there's a (rough) sine wave showing at the LED... where the hell is this coming from?!! The power input is a ripple-free DC 12V, rectified at both ends of the circuits (that is, the power to the LEDs is rectified 11.8V ripple-free) and the power to the NBTV circuit boards is a ripply 17V, but internally rectified to 12V on those boards.

I suspect some subtle part substition again. Looks like it's going to be a difficult debugging time ahead

[gary]

One expected a certain amount of standardisation. Now I know I need to check these things. How inconvenient -- and why on earth don't they actually mark the pinout on the part itself? Backwards it is. I'll reverse it tonight, and try once again

I switched around the transistor, and the LEDs *still* remain fully lit (though there is a very slight variation when playing with the brightness knob). The waveform across the LED array looks basically the same... I'm really starting to get a bit frustrated with this lack of progress. There must be something obviously wrong but I can't see it. My first real question, though, is that given there's a (rough) sine wave showing at the LED... where the hell is this coming from?!! The power input is a ripple-free DC 12V, rectified at both ends of the circuits (that is, the power to the LEDs is rectified 11.8V ripple-free) and the power to the NBTV circuit boards is a ripply 17V, but internally rectified to 12V on those boards.

I suspect some subtle part substition again. Looks like it's going to be a difficult debugging time ahead

Andrew,
Can you tell us what the magnitude and frequency of the wave form across the LED array is please.

Can you also measure (this is important) what the signal at the base pin of the transistor is (magnitude and frequency if any).

Is there any input to the LED driver when you see this? If so what is it?.

This could be mains hum being picked up and amplified through grounding errors so that is worth checking.

The other possibility is that one of the op-amps is oscillating so also check on the output of each op-amp.

Welcome to the wonderful world of electronics! At least you will have learnt how much there is to learn at the end of it.

[Andrew Davie]

Can you tell us what the magnitude and frequency of the wave form across the LED array is please.

Can you also measure (this is important) what the signal at the base pin of the transistor is (magnitude and frequency if any).

This could be mains hum being picked up and amplified through grounding errors so that is worth checking.

The other possibility is that one of the op-amps is oscillating so also check on the output of each op-amp.

Welcome to the wonderful world of electronics! At least you will have learnt how much there is to learn at the end of it.

After my last post, I disconnected all wires, examined all joints carefully, noticed a couple that appeared to be less than desirable, and re-did those ones. Then I reconnected everything, and proceeded to do the measurements. Things have changed, and the LEDs are (mostly) off -- though I don't really know that this is a good thing.

I did try NOT connecting the power to the NBTV boards, so the LED matrix was going through the NBTV boards' ground -- and lighting up. But I didn't do the waveform on this as it was before your question. NOw I can't get the LED to light up at all.

Looking at the input signal, this still looks OK at the pot -- it increases as I rotate the knob. It is recognisable NBTV type waveforms.

But what is very very strange is when I look at this signal as it gets to the LED driver board. Where the "video 1.4V in" wire comes from the sync board, I connected the red lead of the oscilloscope. I connected the black to a nearby "ground" pin on the same board. And the video signal has "shifted" by +6V. That is, I see a NBTV signal, but it's way "up" on my screen. I don't know what is happening here.

When I measured same pins with multimeter, I get 6V too, so I know it's nothing to do with the oscilloscope settings. Very strange.

I am wondering why all of the grounds on the circuit diagram are marked as "ground", except for one which is marked "power ground". I have assumed there is no difference.

I've not yet got around to measuring voltage at the transistor, because my LEDs are no longer operational and I have a 6V offset of my video signal.

I'm almost tempted to order a new set of NBTV boards and start completely from scratch. Frustrating.

One new spanner in the works is an oddity I'm seeing on the oscilloscope signal when i have nothing at all connected -- just plain oscilloscope leads unconnected. I see a very small sine wave -- about 0.1V @ 20ms (=50Hz, our power frequency). If I ground the leads across me, the signal jumps to about 10x size. I'm a living amplifier.

If I bring the leads nearish to the powered-up circuits, it also increases in size. It's quite bizarre, I can't figure out where/why this is happening. Sometimes I notice it, sometimes I don't. Sometimes I see a definite sine-wave ripple in the NBTV signal I'm measuring, sometimes not.

One other thing I noticed -- I measured the AC voltage across the 15V tag of the transformer itself. It wasn't a sine wave, but only a semi-reasonable approximation of one. Flat top, sloped (but not curved) sides. More like a kind of triangle wave with the tops cut off.

[gary]

After my last post, I disconnected all wires, examined all joints carefully, noticed a couple that appeared to be less than desirable, and re-did those ones. Then I reconnected everything, and proceeded to do the measurements. Things have changed, and the LEDs are (mostly) off -- though I don't really know that this is a good thing.

Well it *should* be. That is what you would expect if there is no signal.

I did try NOT connecting the power to the NBTV boards, so the LED matrix was going through the NBTV boards' ground -- and lighting up. But I didn't do the waveform on this as it was before your question. NOw I can't get the LED to light up at all.

Well, if I understand your statement correctly they shouldn't be lit as the transistor is turned off due to there being no bias at the base.

Looking at the input signal, this still looks OK at the pot -- it increases as I rotate the knob. It is recognisable NBTV type waveforms.

But what is very very strange is when I look at this signal as it gets to the LED driver board. Where the "video 1.4V in" wire comes from the sync board, I connected the red lead of the oscilloscope. I connected the black to a nearby "ground" pin on the same board. And the video signal has "shifted" by +6V. That is, I see a NBTV signal, but it's way "up" on my screen. I don't know what is happening here.

I assume when you talk about the 'pot' means you are referring to the sync separator board (contrast)? That appears to be correct as there is a 6V bias on the op-amp. because the input to the LED driver board is AC coupled this DC offset is removed before it reaches the second op-amp of the LED driver board.

Here are the important places to take a measurement:

Pin 3 of the 2nd CA3140 op-amp on the LED driver board.

Pin 6 of the 2nd CA3140 op-amp on the LED driver board (or the base pin of the transistor).

If there is not a reasonable signal at these locations we can then start to work backwards.

When I measured same pins with multimeter, I get 6V too, so I know it's nothing to do with the oscilloscope settings. Very strange.

Not at all that is just the voltage you would expect to get due to the voltage divider effect of the two 22k resistors at the +ve input of the first op-amp.

I am wondering why all of the grounds on the circuit diagram are marked as "ground", except for one which is marked "power ground". I have assumed there is no difference.

That is just a convenient place to connect your supply on the board. If you look at the tracks you see that the video and power grounds are connected, so your assumption is correct.

I've not yet got around to measuring voltage at the transistor, because my LEDs are no longer operational and I have a 6V offset of my video signal.

We still need that measurement as well as the other I mentioned above.

I'm almost tempted to order a new set of NBTV boards and start completely from scratch. Frustrating.

Don't rush into that. I believe we are very close.

One new spanner in the works is an oddity I'm seeing on the oscilloscope signal when i have nothing at all connected -- just plain oscilloscope leads unconnected. I see a very small sine wave -- about 0.1V @ 20ms (=50Hz, our power frequency). If I ground the leads across me, the signal jumps to about 10x size. I'm a living amplifier.

You have just discovered one of the electromagnetic fields we walk through every day of our lives. You are NOT an living amplifier, you are a living antenna! This is the mains hum I referred to in a previous post and we may have to get back to it when we get the rest of the circuit working. It is normal.

[Andrew Davie]

I assume when you talk about the 'pot' means you are referring to the sync separator board (contrast)? That appears to be correct as there is a 6V bias on the op-amp. because the input to the LED driver board is AC coupled this DC offset is removed before it reaches the second op-amp of the LED driver board.

I don't know what "AC coupled" means, at this stage. There is no AC connection anywhere, other than the rectifiers, which convert to DC before anything gets to the NBTV boards...

Here are the important places to take a measurement:

Pin 3 of the 2nd CA3140 op-amp on the LED driver board.

Pin 6 of the 2nd CA3140 op-amp on the LED driver board (or the base pin of the transistor).

If there is not a reasonable signal at these locations we can then start to work backwards.

It is nigh-impossible for me to get the probe's alligator clips attached to the CA3140 pins.

When I use the multimeter to measure the voltage from the pins of the transistor to ground, I get 12V for all three. From Base to Emitter is 0V, from Base to Collector 0V and from Emitter to collector, also 0V.

By the 'second' CA3140, I assume you mean the one closest to the transistor. Nonetheless, I've measured both... with the input being track 2 of the club CD (black screen), track 5 (white screen) and no input...


in order... CA3140 #1 pin 3, pin 6 CA3140 #2, pin 3, pin 6
black screen 0.1mV, 4.8mV 1.7mV, 7.8mV
white screen 0.1mV, 4.8mV 3.4mV, 7.8mV
no input 0V, 3.5mV 11.7V(!!), 12V

After this, I couldn't repeat the above observations. EVERYTHING I measured was 12V. That is, I re-measured black screen input across pins 3, 6 etc... getting 12V or thereabouts. I have a loose connection somewhere...?

VEERRRRY interesting. I re-did the ground connector just under the HEF4053 on the LED board. It seemed to be slightly wobbly (I could only find these really bad pin-type connectors for the board interconnects -- Jaycar had nothing else, could suggest nothing). They're a pain to put in, and they tend to pull out the circuit traces -- I have several wire 'bridges' where traces have come off due to excessive heat.

Anyway, as I wrote, I re-did one of them and re-measured...

Now I get...

no signal #1(pin3=0.8V, pin6=1.1v) #2(pin3=0.8V, pin6=8mV)
track2(black) #1(pin3=0.8V,pin6=1.1V) #2(pin3=0.8, pin6=8mV)
track5(white) same as above

One step back, two steps forward?

I have now hooked back in the LED array. And the LEDs are lit up. Bah.

Here are some measurements...

across the LED +/- 7.2V
transistor base-collector 0V
base-emitter 0
emitter-collector 0V

from LED input on board to ground on same board -- 0.8V
from LED input (ie: neg of led matrix) to ground on rectifier board 0.8V
at this point I note that only about 75% of the LEDs are actually lit.
Some, it seems, have a very low voltage to light up.

re-doing the measurements with the LED board installed

no signal #1(pin3=385mV, pin6=0.65v) #2(pin3=385mV, pin6=0.86V)
track2(black) same as above
track5(white) same as above

I'm making NO sense of this.

Now I've removed the LED matrix (just disconnected the neg input from the matrix to the LED driver board)

no signal #1(pin3=385mV, pin6=0.65v) #2(pin3=385mV, pin6=0.86V)
track2(black) same as above
track5(white) same as above

ie: no change.

I'm too tired to do any more. Values and behaviour appear to be changing as the night goes on. Many thanks for your help so far, Gary.

[gary]

Ok I thought the "AC coupled" would throw you, I had meant to put in an explanation. When we say a signal is AC coupled we are normally talking about the signal being connected to the rest of the circuit through a capacitor. An arbitrary signal is composed of a DC component and an AC component, for example you measured the NBTV signal on the input to the LED driver and saw that it was offset by 6V. This 6V is the DC component and the NBTV signal is the AC component. The characteristics of a capacitor is to only allow the AC component through so if you measured the signal described above on the other side of the capacitor you would see that the 6V offset is removed and the NBTV signal is now averaged above and below 0V. This is also referred to as indirect (AC) as opposed to a straight through direct (DC) coupling. In the case of the LED driver the signal has the 6V DC component removed and then the clamping circuit (the first op-amp) causes the negative-most part of the signal to be raised by another DC offset such that the sync 'tips' are at a known and (relatively) constant level. This is the 'real' video signal we want to see and it is imperative that we have a look at this signal with the CRO, looking at the signal with the multimeter is not of much use at all since the signal is alternating in an arbitrary fashion.

Now, the CRO should have come with a probe that has a kind of a hook thing on the end which is at the end of a sleeve which is sprung to allow more or less of the hook to be exposed and also to hold the hook in place when wrapped around a wire or pin or whatever you are measuring. This sleeve should be removable by simply pulling it off and this will expose a new probe tip which is a sharp pin-like spike. This is ideal for measuring the op-amp pins. If you don't have such a probe just use the alligator clip with a short piece of straight copper wire, or pin, or needle, or even a nail, it doesn't matter much as long as it has a thin or pointy end that can reach down onto the op-amp lead.

Do you know how to determine which op-amp pin is which? If not let me know and we will go through it.

At the two points I have measured you should see a signal very much like the signals shown on the simplified diagram in the documentation. If you don't then we can start worrying about the other measurements.

Ok, so when you have the right probe arrangement, measure the signal at:
a) the input pin of the LED driver (you have already done so but do it again as a sanity check against the next two measurements.

b) at pin 6 of the second op-amp (yes you normally read a schematic from left to right, it would normally be identified as say IC2 but in this case it isn't). This is the signal that has passed through the capacitor to have the 6V offset removed and then 'clamped' to have the synch tips raised to a known point and should look like the first signal shown on the simplified diagram.

c) at pin 3 of the same op-amp. This should look like the second signal shown on the simplified diagram and is what is fed into the driver transistor to modulate the LEDs.

Any deviation of these signals from that shown on the simplified diagram should give us a good indication of what is wrong.

BTW debugging a circuit is not that much different (in principle) from debugging software. You need to know where to set your break points (measuring points) and you need to know what variables should have what values (the signal). You know that in debugging you don't want too many variables and you certainly don't want to make too many changes at once because you won't know what is changing what - the same goes for circuit debugging.

I attach a picture of a typical CRO probe (disassembled). If you have one great, if not improvise - but get those measurements!.

I'm tired now too so good night and good luck!

[Andrew Davie]

Well, very interesting. I did build myself a little homebrew probe tip that would allow me to measure the waveforms at the pins suggested. I will just let the coments on each picture speak for themselves... NOTE: pictures are in reverse-order -- please read/review from bottom up!! Also, I see that comments with images are truncated. I hope it's all clear enough, though.

A couple of things I've learned... firstly, playing an all-black screen (track 2,3,4) is NOT a good idea when really trying to see what's coming out the other end. Using track 5 (a white screen) is a much better idea!

It looks to me like we're getting valid signals all the way to the transistor (AND to the LED matrix, too!) the only issue is that "extra" signal sitting in there buggering everything. This is possibly one of the major problems all along -- when I'm seeing the LED 'on' I'm really seeing them pulsing to this half-sine-wave. Maybe.

I had surmised in the picture comments (truncated, alas) that the 'hum' - that is, the 50Hz sine wave overlaid on the video image -- was coming in somehow because I have the LED 12V rectifying circuit connected to the same transformer and tags (15V/0V) that I have the other 17V rectifying circuit connected to. I suspect the solution to this problem is either going to be separating these two circuits or perhaps putting in a magic capacitor somewhere to 'dampen' the hum.


Anyway, this is really a big step forward for me, and I"m quite chuffed.

[AncientBrit]

Steve,

Don't forget that your scope Y amps will have the facility for both AC and DC coupling on their inputs

When in AC coupled mode any DC (or steady vert offset) is blocked.

When in DC mode the whole signal (both AC and DC components) is displayed.

Why provide this function?

Well some AC signals "sit" on a DC component that is much larger than themselves.

If you crank up the gain to get a good view of the AC part the signal shoots off the screen at the top(or bottom) due to the DC part.

So we then switch to AC mode to get a good view just of the AC part of the signal.

So remember to stay in DC mode if you need to get the overall picture.

(Re-reading your posts you probably have found this out anyway)


Regards,

GL

[gary]

Andrew, have you looked at your 'clean' 12V unconnected to the LEDs with the 'scope? If not then it is possible that this supply is not so clean after all so have a look at it.

As a sanity check, if you have a couple of 9V batteries handy you could connect them in series to provide a guaranteed clean DC supply to the LEDs.

At first glance the level of the unwanted signal seems to be too high to be just 'hum'. Its almost as though the rectifier is halfwave rather than fullwave although if this were true you would still expect it to be much smoother due to the smoothing capacitor.

It is important now to identify the signal at the emitter of the transistor, I am pretty sure that the intention of the circuit to not have any syncs at this point (truncated by the .7V drop across the transistor base-emitter). If they are there then the DC level of the input to the transistor base is not right (this could be just a matter of setting the brightness).

The images of your scope seem to indicate that it is out of focus or maybe that is just an artifact of the snapshot itself.