Power Supply

[Steve Anderson]

That's a new device to me, but browsing through the datasheet it seems quite tame, shouldn't be any issues with it. Just remember to keep the input and output caps close by.

The only other issue is ripple. For this 12V regulator you're inputting a certain voltage as measured by your DC multimeter. What is next is to calculate the size of the capacitor in the raw supply. Any idea how much current you'll be drawing at 12V? The bottom of the ripple (as shown on an oscilloscope) should never go less than the regulator output voltage plus its drop-out, in this case around 12.8V.

In other words, if your multimeter shows 14V DC (on load) and the scope shows 2v p-p of ripple the rule of thumb is that the input varies at 100Hz from 13V to 15V - which for this device is OK. Just.

It could well do with a small heatsink as it will be dissipating around 2W with a 1A load.

Steve A.

In a way you're doing this backwards. Most design the apparatus first, measure the current required, then design the power supplies required. This is why most will have 'bench power supplies' which can be varied on output voltage and usually have current limiting of some form. You can then slowly wind up the voltage and switch off if there's an obvious problem.

[Andrew Davie]

That's a new device to me, but browsing through the datasheet it seems quite tame, shouldn't be any issues with it. Just remember to keep the input and output caps close by.

The only other issue is ripple. For this 12V regulator you're inputting a certain voltage as measured by your DC multimeter. What is next is to calculate the size of the capacitor in the raw supply. Any idea how much current you'll be drawing at 12V? The bottom of the ripple (as shown on an oscilloscope) should never go less than the regulator output voltage plus its drop-out, in this case around 12.8V.

In other words, if your multimeter shows 14V DC (on load) and the scope shows 2v p-p of ripple the rule of thumb is that the input varies at 100Hz from 13V to 15V - which for this device is OK. Just.

It could well do with a small heatsink as it will be dissipating around 2W.

Steve A.

Understood. Will keep all in mind.
I have zero concept of how much current will be drawn by the NBTV circuitry.

As to the current through the LED grid, the last post about this in my previous build was from Klaas who said I could increase the brightness by 4x - it "currently" (see what I did there?) had 40mA and thus each LED string had just 5mA through it. And from the specs the LEDs could go 7x brighter (35 mA). Amazing, but I didn't want to use high voltages, and the LEDs were already completely overkill as far as brightness goes.

But anyway, that would be 40mA, I guess. It's limited by the gamma resistors, which Klaas said were hard to change.


I could just junk this idea and use the two LM2596 adjustable output DC-DC converters I have in my parts drawer. That is, feed the ripply 13.7V into the LM2596. The forward dropoff voltage on this seems to be quite low, too. More complex solution, though.

My current order of preference...

1) Use a stock-standard 12V regulator, and the lowest suitable input AC voltage - one which never has the rectified DC voltage drop below (12V + regulator forward voltage dropoff) when under load. That would be *possibly* using the 9V AC but most likely the 15V AC. And put on a heatsink on the regulator.
2) Get a ultra-low dropoff regulator and try to use the ~9V AC -> DC supply.
3) look at something like the LM2596

I think with 1) the regulator is going to get very hot, but that might be OK.

[Steve Anderson]

I could just junk this idea and use the two LM2596 adjustable output DC-DC converters I have in my parts drawer.

I don't recommend switching regulators as 1) they are more complex and 2) they put out quite a lot of garbage on the supply. Usually OK for logic, but you could well have problems with analogue linear circuits, especially at low signal levels.

With all the televisors built thus far the vast majority have used linear regulators where required with no real issues regarding heat dissipation in the power supplies.

If you're planning to use one supply unregulated (+22V) for the LEDs there's no issue with regulator dissipation. The other supplying +12V regulated to all the other electronics I would be surprised if you got anywhere near 100mA, one tenth of an Amp. The drive motor may be a different issue though.

Steve A.

[Klaas Robers]

Andrew, integrated voltage regulators have an automatic high temperature switch off. When the circuit runs too hot, it switches off. Then it cools down and switches on again. Be aware of this behaviour.! If you see that your monitor switches off now and then a regulator may become too hot. So it needs a larger heat sink, or even better, split up the circuitry and power them by separate regulator ICs.

It is even permitted to design two, or more, regulators in parallel. I once built a 12 volts power supply for 5A with six regulators of the type 7812 in parallel. What then happens during time I don't know, but it worked and works very well. Because of the very low price of the 7812 chips, this was cheaper than using a thick parallel transistor with the circuitry around it. And you can spread the heat development.

It is indeed not needed to use a stabilized voltage for the LED cluster. They are current controlled by the video output transistor, so the supply voltage is not affecting the light output (= current) as long as the unstabilized voltage stays high enough. You can check that by measuring (oscilloscope?) the voltage at the collector (drain) of the video output transistor while displaying a white field. It should stay a few volts higher than the voltage on the base of that transistor.

[Andrew Davie]

I don't recommend switching regulators as 1) they are more complex and 2) they put out quite a lot of garbage on the supply. Usually OK for logic, but you could well have problems with analogue linear circuits, especially at low signal levels.

With all the televisors built thus far the vast majority have used linear regulators where required with no real issues regarding heat dissipation in the power supplies.

If you're planning to use one supply unregulated (+22V) for the LEDs there's no issue with regulator dissipation. The other supplying +12V regulated to all the other electronics I would be surprised if you got anywhere near 100mA, one tenth of an Amp. The drive motor may be a different issue though.

Steve A.

Noted. Thanks!

[Andrew Davie]

Andrew, integrated voltage regulators have an automatic high temperature switch off. When the circuit runs too hot, it switches off. Then it cools down and switches on again. Be aware of this behaviour.! If you see that your monitor switches off now and then a regulator may become too hot. So it needs a larger heat sink, or even better, split up the circuitry and power them by separate regulator ICs.

Pleased to be learning this from you. Thanks! My #1 televisor's regulator was getting hugely hot - I had a fair sized heatsink on it, but I don't think it ever switched off. It was converting from 22V down to 12V, so given my new setup where I'll be feeding "just" 17V (or maybe even 13.7V) I'm hoping it will be OK. But I'll know if things switch off/on that I should check the regulator. Understood regarding the LED current/voltage explanation. Thanks!

[Andrew Davie]

One of the two transformers I have was getting very hot. Burning smell, and when I did some unrelated work, the transformer no longer seemed to be... right. Couldn't pin it down, but this was a transformer that I bought unboxed from the store for a discount. It seemed too dodgy so I went and bought a replacement - which is now working very well without overheating issues. I wonder if I damaged the original by trying the two power outs from the one transformer (remember, I was getting weird results when I did the common ground). Anyway, happy to not have the burning smell and the really hot almost melting plastics on the transformer

[Steve Anderson]

Somehow, somewhere, I suspect you created a short, or a least a low resistance path which stressed the transformer as you say. If all is OK now best leave it as is.

The old adage, "If it ain't broke don't fix it."

Transformers are usually very tolerant to overloads - of a reasonable duration. If you get excessive heat and/or that characteristic burning/melting smell you've probably gone past that stage.

Steve A.

[Andrew Davie]

Somehow, somewhere, I suspect you created a short, or a least a low resistance path which stressed the transformer as you say. If all is OK now best leave it as is.

The old adage, "If it ain't broke don't fix it."

Transformers are usually very tolerant to overloads - of a reasonable duration. If you get excessive heat and/or that characteristic burning/melting smell you've probably gone past that stage.

Steve A.

Not sure yet about the new one.
I have it driving the LED array, and I'm seeing streaking and noise in the resulting image. I haven't changed anything else, so I suspect the new PS.
Will have to look at the waveform. Easiest perhaps to swap the two around and see if I still get the image artefacting.

[Klaas Robers]

Andrew, scope your rectified voltages. If one of the diodes in the bridge rectifier is "blown up" due to previous experiments, you are working on half wave rectifying. That works, but your transformer is loaded with a DC current, which it doesn't like at all. It runs hot and more.

If the double wave rectifying is working properly you see a ripple in the rectified, not yet stabilized voltage of 100 Hz (10 msec). When not it is 50 Hz (20 msec) and you should replace the rectifier bridge.

[Andrew Davie]

Andrew, scope your rectified voltages. If one of the diodes in the bridge rectifier is "blown up" due to previous experiments, you are working on half wave rectifying. That works, but your transformer is loaded with a DC current, which it doesn't like at all. It runs hot and more.

If the double wave rectifying is working properly you see a ripple in the rectified, not yet stabilized voltage of 100 Hz (10 msec). When not it is 50 Hz (20 msec) and you should replace the rectifier bridge.

OK, will do. I'll report back tomorrow.