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

[gary]

OK you *definitely* don't want to do that either, the ground on the 'scope is the same as the black lead on your probe and so you are shorting out your secondary through one of the diodes when you do that. I.e. you are shorting the -ve terminal of the bridge to the bottom of the secondary winding (0V) thereby shorting out the reverse biased diode. Phffft!

[Andrew Davie]

OK you *definitely* don't want to do that either, the ground on the 'scope is the same as the black lead on your probe and so you are shorting out your secondary through one of the diodes when you do that. I.e. you are shorting the -ve terminal of the bridge to the bottom of the secondary winding (0V) thereby shorting out the reverse biased diode. Phffft!

Fortunately the magic smoke didn't escape, so it still works. What I don't understand, though, is why I get 17.1V when measuring with my multimeter (DC), but when I measure with the oscilloscope, I get a flat line (where I expected ripple) and according to my readings of the scales, it's only showing about 6V. Something odd is happening (which is why I was playing around with that ground wire in the first place).

Cheers
A

[gary]

Well, that could be lots of things, some that spring to mind:

Have you calibrated the 'scope? (see documentation).
Have you check the gain of the probe?
Have you checked against a known benchmark like a 9V battery?

What timebase are you using? (note ripple period = 10mS)

[Klaas Robers]

Long time that I was not on the Forum. Gary will learn you to use the oscilloscope. Good thing that you bought one. It will open your eyes.

One important thing: The BDX54 that the electronics shop sold you is a transistor of the wrong type. Transistors are made in two variations, called NPN and PNP. The NPN-type has an arrow pointing outwards and are ment for "positive supply voltage" circuits. PNP's have an arrow pointing inwards and are for "negative supply voltage circuits". The fact that you have got a PNP transistor explains why the LEDs are just on and no more.

The BUT11 is an NPN transistor, in fact a high voltage one that can withstand some hundreds volts. I had one lying around and had a power supply of 300V for a real neon lamp. So I needed this high voltage transistor.

For low voltages and LEDs you may take a simpler transistor, e.g. BD139. Try to get a non-darlington transistor (Darlingtons are BDX-transistors, the X indicate "darlington".) You can use other types as well, as log as it is NPN, and a Ic (collector current) of about 1A.

[Klaas Robers]

Andrew,

the LED-driver is laid out for 50mA max current. So with your LEDs you might make a double chain, two chains of LED in parallel. I should look back to see how many LEDs you have, but that number will define your needed voltage.

The driving voltage might be higher than minimally needed. This is because the TRANSISTOR swallows the remaining voltage. See the transistor as an automatically variating series resistor (resistor to ground). The value of that "resistor" defines the current through the LEDs.

The magnifiscent thing of the transistor is that the current is NOT dependant of the voltage accross the transistor. So if the voltage on the top end (+) of the diode chains is varying somewhat this is NOT visible in the CURRENT through the LEDs. So the + supply for the LEDs might be an unstabilised voltage that is just "high enough". When things work I can advise you to look around for a small transformer of e.g. 55V or 2 x 24V, rectify that and have a smoothing capacitor, and that is it for the LED-chain. Then you will get about 70 volts.

In this case there is no need to recalculate the values of the "gamma resistors".

[Klaas Robers]

Hmm. I see 4 x 6 LEDs.
That might be 2 chains of 12 LED's
4 volts per LED = 48 volts across the LEDs at full brightness.
Transformer 2 x 24 V will give about 65 volts rectified and smoothed. Smoothing capacitor 220 uF at 100 V.

Then something else. I see you giving extremely precise values of resistances, capacitances, voltages etc. In electronics we are never so precise. A deviation of 10% is seldom a problem and is mostly not observed. So these deviations from the given values are in most cases permitted. This also implies that values are having just 2 digits and the place of the decimal comma (as we use it) of dot (for you anglicists).
A capacitor of 1,8 uF may vary from 1,6 to 2 uF. Some capacitors (electrolytic) might have a value of -20% tot +50%. No problem in almost all cases.

Your problem is that you have a digital meter. They are giving you all these digits. The oscilloscope will learn you the coarse way of viewing.

By the way, have you solved the AC measuring mystery already? I saw on your photographs that your meter's switch has a setting ~V. Wouldn't that be for measuring AC?

The output waveform of the CD-player as I saw it from your oscilloscope looks fine, at least the polarity is correct, negative syns pulses. Track 2 is black (look on the frontside of the box) The negative sync pulses should have a "depth" of 0.4 volts. You measured something similar with the oscilloscope. When you go to track 5 you will see that you have pulses of 1,4 volt, i.e. 1 volt video (white) + 0,4 volt for sync. Browse through the disc and try to understand what you see. And remember: deviations of 10% are allowed.

[Klaas Robers]

Oh Andrew,
your oscilloscope: Both attenuators for the sensitivety (channel 1 and channel 2) have a red continuous control on top, marked "var". This should be in the "click" position most of the time. Then a little lamp, marked "uncal" goes off. Only than you can read voltages off the screen.

Once used to an oscilloscope you will use it more than a voltmeter, even to check if the 12V supply is there.

[Andrew Davie]

Well, that could be lots of things, some that spring to mind:

Have you calibrated the 'scope? (see documentation).
Have you check the gain of the probe?
Have you checked against a known benchmark like a 9V battery?

What timebase are you using? (note ripple period = 10mS)

I have just found the cable to connect the square-wave calibration square-wave to the input channel 1, and when I do this I do indeed see a square wave but instead of the 1V that it is supposed to show, I read just 0.3V via the scales. So that is where the problem lies. I will learn how to calibrate the 'scope today. Since the square wave is showing just 1/3 of the voltage, this ties in well with my estimate of seeing only 6V when I was expecting 17+.

The timebase I was using -- not sure, will reanalyse this after I get the calibration done.

Cheers
A

[gary]

Hi Klaas,

It's always great to hear from the designer of the circuit!

I can't believe I missed that the BDX54C is a PNP - it explains everything. I looked up the BDX54C on the internet (see my attachment a few posts back) and saw it was a Darlington but missed the fact that the BDX54C is the complement to the BDX53C NPN version - damn! See Andrew, even a (retired) professional can get it wrong. However I will use it to reinforce my comments about "the man in the shop".

Good point about using a BD139 - these are readily available here and I have used them for this purpose.

[Klaas Robers]

Andrew,

the square wave has another purpose as well. When you use a probe that attenuates by a factor of 10 (it has a built in resistor of 9 megaohm, with the 1 megaohm this gives a attenuation of 10 x) Should be adjusted for the high frequencies. They have some little screw in the probe itself or in a small box in the connector, that you can turn on.

When you do this while connecting it to the calibration output of the oscilloscope, you will see that the steep edges on the screen change in length. The horizontal part then gets a curved beginning, curved down or up. The correct setting is: no curve at all, so straight horizontal lines.

This is not needed when you use just a piece of coaxial cable or a probe of the type 1x.

[Klaas Robers]

Gary,

I have a simple DOS-based program "Torbase". it gives me short information about lots of transistors. Unhappily for you it is in Dutch, but all numbers and pinnings should be readable still. I will send it to you by e-mail. It goes much faster than googleing.

[gary]

Thanks Klaas, sounds very useful to me and others.

[Andrew Davie]

Today I went out and bought a BD139, snipped out the incorrect transistor and soldered in the replacement. If you'd ask me what was the absolute LAST thing I would expect to happen, it would be that the LEDs would still light up...

The LEDs still light up! I can't see a single thing different in how the circuitry "works" compared to the other transistor. Time to start suspecting my soldering joints, no?

I calibrated my oscilloscope using the .1V/1V square wave today. The gain was way out of whack on both channels. Somebody had a play with the adjustments, I guess. Could have been me... don't remember which knobs I played with when it first arrived. In any case, now that the gain is correctly set, I'm getting "good" readings (ie: they correspond to the multimeter readings) on the oscilloscope for the rectifier circuits.

Now that I've had a very good play looking at the signals from the club CD, I can just about recognise from the oscilloscope trace what image is actually being displayed. Most of the waveforms are very distinctive.

When I was on the hunt for a CD player with a line-out, I actually found two. I found out today that they have different 'polarity' -- that is, the line-out signal is inverted (wires are switched) compared to the other. This is kind of handy for me, but in any case I can always hook up the 'scope to make sure I have things right.

The first thing I learned is that the variable potentiometers were connected 'backwards'. That is, of the three tags on the potentiometer, i was previously unsure which was which -- were the circuit diagrams showing the pot "up" or "down"? The answer is, "down". From the diagram point of view, we're looking at the bottom of the pots. I discovered this by watching the waveform after it went through the pot, and noting that as I turned the contrast pot to the right (clockwise), the voltage range of the signal reduced. That's not what I wanted, so I switched things around so it looked OK.

Then I did some measurements at various places in the circuit.

The really interesting bit for me was checking the line pulse and frame pulse on the sync board. This was kind of tricky to get the oscilloscope to display (because, I think, of the very low frequency of these pulses). In any case, though I was not able to measure the actual rates (I didn't really feel like trying to count very close together dots), I could confirm that there were regular pulses on these two and they seemed to be in roughly the correct ratio.

I had a look at the video signal as it got to the video in of the LED driver board. Looked OK at that point. I noted that as I adjusted the contrast pot, this signal reduced/expanded. I also noted that when the pot was at 0, the line/frame pulse also stopped.

The rectified power on the sync board was 12.06V (sorry Klaas -- I really do know the difference between accuracy and precision and the lunacy of too many significant digits). I'll get used to saying "12V" eventually.

I measured the range of voltage across the brightness pot with my multimeter. It went from 0V to 2.6V. This, too, seems a very correct figure to me.

But my LEDs were still on. I measured 7.2V across the whole array. Adjusting the brightness had no effect on the LED array at all.

I tried with the select connected to ground, and to 12V -- no difference for either of these.

I measured and photographed the waveform across the LEDs. This appears to be a corrupted sine wave, no idea what is going on here.

Finally, I measured the voltage across the C/B/E of the transistor. Did I put it in backwards, perhaps? I put the writing such that it was facing me with the left pin closest to the LED matrix input.

From the "left" to "middle" measured 1.47V with no input.
With input from CD (ie: signal) this lifted to 1.52V
from the "left" to "right" measured 1.4V/1.52
from the "right" to "middle" measured 0V

So, that's it for another night. A few steps forward, some steps backwards. I'm learning a bit every day, but it's clear to me now that these boards and debugging them is not a learner's job. I'll get there in the end

[DrZarkov]

About a cheap, small and very good CD: Take an old CD-rom-drive from a computer! On the backside, where you connect the soundcard, you have line-out. The other problem is power, usually they only need +12 Volt (the yellow wire, black is earth), the IDE or SCSI-cable is of course not necessary for us. If you take a flat one from a notebook, you maybe can built it into your NBTV-monitor.

BTW: Today I've got from the club-shop the same circuit-boards and my Nipkow-disc (good quality, interesting material. I've already cleaned the holes and cut it out). I will go back in this thread, and then we will see when I do arrive at the same problems. I'm living a bit too far away to borrow your osciloscope, but luckily Klaas is a "buurman" (neighbour) to me, as I live only 3 km from the dutch border.

[gary]

Today I went out and bought a BD139, snipped out the incorrect transistor and soldered in the replacement. If you'd ask me what was the absolute LAST thing I would expect to happen, it would be that the LEDs would still light up...

The LEDs still light up! I can't see a single thing different in how the circuitry "works" compared to the other transistor. Time to start suspecting my soldering joints, no?

...

Finally, I measured the voltage across the C/B/E of the transistor. Did I put it in backwards, perhaps? I put the writing such that it was facing me with the left pin closest to the LED matrix input.

From the "left" to "middle" measured 1.47V with no input.
With input from CD (ie: signal) this lifted to 1.52V
from the "left" to "right" measured 1.4V/1.52
from the "right" to "middle" measured 0V

So, that's it for another night. A few steps forward, some steps backwards. I'm learning a bit every day, but it's clear to me now that these boards and debugging them is not a learner's job. I'll get there in the end

Andrew, it *sounds* as if you have it in backwards: