Power Supply

[Andrew Davie]

Had a nice chat with one of my electronic-whiz workmates about the power system on my original televisor. I wanted to understand why the (how embarassing)... component... on my circuit board with the whacking great heatsink just after the capacitor which smooths the voltage from the rectifier gets so hot that I can feel the heat radiating from centimeters away. I'll clarify this all when I get home.

He advised that I might consider some LM2596 on the tail-end of the power chain, and gave me a couple. Capable of 3A apparently. I'm pretty sure I have the "adjustable output" version.

What I plan to do is start with the power supply side of things, and get suitable separate power for the motor and the lights/circuitry.

I *think* I will need to supply 15V and 12V, but I'm not sure about that - perhaps I can do just 12V. I have a lot to remember and re-learn.
In any case, currently thinking 240V -> 15V AC transformer feeds into two rectifiers, giving me 15V and 12V - with attached capacitors of appropriate values to then smooth out the rectifier output, followed by the LM2596 on each output and I should have good 3A-capable 12V and 15V supplies.

Edit: Correction for the above red section - the input is 12V AC and/or 15V AC which is rectified to ~17V DC and ~21V DC. The LM2596 is pointless - that's just a DC to DC converter, and I don't think I need to be guaranteeing a set DC voltage at the power stage. So, no LM2596 for now.


I *know* I'm back to novice/amateur status, but hey... I'm giving it a go

[Andrew Davie]

Have been reviewing my original build and construction diary. In actuality I have a 12V supply and a 15V supply.
It's the 12V supply with the 4700 microFarad capacitor and it's the close-by [..?.. MOSFET ??? ]that's getting REALLY hot.
That's what I want to fix/avoid this time around.

Edit: Not "12V either". A couple of things I'll note

1) I really wish I had kept good documentation in terms of accurate circuit diagrams, and
2) I wish I had NOT used perfboard - there are a couple problems - all the traces are on the non-component side, and I can't easily see which components connect to which traces. Sometimes nearly impossible! And of course everything flush with the "top" there are no pins or leads to place probes onto. I have to flip things over and then I've lost context. Perfboard is a pain in the butt to diagnose/debug. Never again!

I now think the "12V" side (as marked by my scribblings from a decade ago on the perfboard" is actually putting out some 20+ volts. I can't read the markings on the rectifier, but from GND to + on it, looks a tad over 20V.

Edit: This doesn't make sense - if it's 12V AC being rectified then we would expect about 17V DC. However, there's that giant capacitor there - its intention was to smooth the ripple, but is it actually increasing the measured voltage, too?

[Andrew Davie]

Another bit of information I had forgotten - a 15V AC supply is actually measuring the RMS voltage. In fact AC goes from +15 to -15V
Still getting my head around how a rectifier can make 20+ volts from 15V AC though. Is there 30V "potential" there...? More studying/revising to do.

Edit: The above is WRONG. RMS voltage of 15V AC implies a waveform from -(15*sqrt(2)) to +15*sqrt(2)
That is, from -21.2V to +21.2V as explained by gary, below. I have left my errors in for posterity.

[gary]

Another bit of information I had forgotten - a 15V AC supply is actually measuring the RMS voltage. In fact AC goes from +15 to -15V
Still getting my head around how a rectifier can make 20+ volts from 15V AC though. Is there 30V "potential" there...? More studying/revising to do.

I think you have made a typo there Andrew, a 15V AC (RMS) goes (peak to peak) from (+15 x (the square root of 2)) to (-15 x (the square root of 2)) - that is (+21.2 to -21.2V).

The rectifier (diodes) *doesn't* make 20+ volts, it just makes the -21.2V part of the AC voltage +21.2V (full wave). The things that makes the DC 20+ volts are the whopping big capacitors that hold the peak voltage between peaks. Notice this results in "ripple" and that's where the voltage regulator comes in.

ACDC.JPG (13.94 KiB) Viewed 16177 times

[Andrew Davie]

Another bit of information I had forgotten - a 15V AC supply is actually measuring the RMS voltage. In fact AC goes from +15 to -15V
Still getting my head around how a rectifier can make 20+ volts from 15V AC though. Is there 30V "potential" there...? More studying/revising to do.

I think you have made a typo there Andrew, a 15V AC (RMS) goes (peak to peak) from (+15 x (the square root of 2)) to (-15 x (the square root of 2)) - that is (+21.2 to -21.2V).

The rectifier (diodes) *doesn't* make 20+ volts, it just makes the -21.2V part of the AC voltage +21.2V (full wave). The things that makes the DC 20+ volts are the whopping big capacitors that hold the peak voltage between peaks. Notice this results in "ripple" and that's where the voltage regulator comes in.

ACDC.JPG

Hi Gary thank you for your posting/clarification.
My understanding is good on all of it, except for perhaps the terminology of 15V RMS.
In the way I explained it, -15V to +15V AC would actually be 15/sqrt(2) = 10.6V RMS. But yes, I got that bit wrong.
I was assuming 15V AC had a range of -15 to +15 when in fact as you say it's -21.2 to +21.2 and the 15V "value" is the RMS calculated from those.
I'm sure I understand now. Already understood how the rectifier changed the negative part, and we had a series of positive hills, and how the capacitor "fills in" the dips to give a more even waveform, but with ripple. Appreciate your clarification! It's been 10 years and I'm re-learning all of this, but THIS time I'm going for *understanding* not just a working thing made by following advice. Cheers!

[gary]

Another bit of information I had forgotten - a 15V AC supply is actually measuring the RMS voltage. In fact AC goes from +15 to -15V
Still getting my head around how a rectifier can make 20+ volts from 15V AC though. Is there 30V "potential" there...? More studying/revising to do.

I think you have made a typo there Andrew, a 15V AC (RMS) goes (peak to peak) from (+15 x (the square root of 2)) to (-15 x (the square root of 2)) - that is (+21.2 to -21.2V).

The rectifier (diodes) *doesn't* make 20+ volts, it just makes the -21.2V part of the AC voltage +21.2V (full wave). The things that makes the DC 20+ volts are the whopping big capacitors that hold the peak voltage between peaks. Notice this results in "ripple" and that's where the voltage regulator comes in.

ACDC.JPG

Hi Gary thank you for your posting/clarification.
My understanding is good on all of it, except for perhaps the terminology of 15V RMS.
In the way I explained it, -15V to +15V AC would actually be 15/sqrt(2) = 10.6V RMS. But yes, I got that bit wrong.
I was assuming 15V AC had a range of -15 to +15 when in fact as you say it's -21.2 to +21.2 and the 15V "value" is the RMS calculated from those.
I'm sure I understand now. Already understood how the rectifier changed the negative part, and we had a series of positive hills, and how the capacitor "fills in" the dips to give a more even waveform, but with ripple. Appreciate your clarification! It's been 10 years and I'm re-learning all of this, but THIS time I'm going for *understanding* not just a working thing made by following advice. Cheers!

Yep whenever we talk of AC we generally (almost exclusively) talk in RMS - e.g. 240V mains is 240V RMS, etc etc. If we are talking peak to peak we should always specify peak to peak.

I suppose, after rectification, but before smoothing by capacitor, we have DC with 100% ripple, but I never hear of anyone putting it that way LOL.

We use RMS because, roughly speaking, RMS (Root Mean Square) is the AC equivalent, in terms of delivering power, of the same value of DC - note that seems counter intuitive with regards to what we have been talking about but that's only because we are talking about power not voltage. We may get a higher voltage, but the power remains the same, therefore max current is reduced.



An easy way of thinking about it is

[Steve Anderson]

Yes, it's one of my pet hates, when people say a certain voltage should be such-and-such AC. Do they mean RMS (say 10V), peak (about 14.14V) or peak-to-peak (28.28V)? Be specific - we're talking about an almost 3:1 difference here. Those with a bit of experience can often work matters backwards and work out what the author meant - but sometimes it's completely lost. Particularly with regard to audio levels. Analogue video is always (NO exceptions) specified as p-p. (Peak to peak). i.e. you need an oscilloscope to measure it.

Though they may exist I have never seen a multimeter that reads either peak or p-p voltages. Most are calibrated to read the average AC voltage on the assumption that the source is a sine-wave - usually of a fairly low frequency, 50/60Hz. Above 1kHz most roll over and die. Again a 'scope is the answer. Of course there are exceptions, if you have the budget for a Fluke then the chances are it will read True RMS and have a much wider bandwidth. But it's still no use in measuring video levels.

Any 'scope, perhaps even these $15 specials, are invaluable. Just keep in mind its limitations, which equally applies to a multimeter.

Steve A.

Picking up Gary's thoughts above, 240V AC is about 340V peak (240 * 1.414) or 680V p-p (twice the previous).

Google RMS if you want some bedtime reading...

In statistics and its applications, the root mean square (abbreviated RMS or rms) is defined as the square root of mean square (the arithmetic mean of the squares of a set of numbers).[1] The RMS is also known as the quadratic mean and is a particular case of the generalized mean with exponent 2. RMS can also be defined for a continuously varying function in terms of an integral of the squares of the instantaneous values during a cycle.
For a cyclically alternating electric current, RMS is equal to the value of the direct current that would produce the same average power dissipation in a resistive load.[1] In Estimation theory the root mean square error of an estimator is a measure of the imperfection of the fit of the estimator to the data.

..told you...

[Andrew Davie]

One of the problems with my #1 televisor was the power conversion area, where I had two rectifiers for a "15V" and a "12V" (AC) source. What I seem to have goofed on is at some point when I realised I wasn't supplying enough voltage I switched to the 15V AC as input to *both* of those conversion blocks. In other words, rather than supplying 12V AC to the one marked "12V", rectifying and de-rippling that, and using the resultant ~17V DC, I seem to have fed 15V to that, too, and consequently it gets VERY hot. More specifically, the "thing" with the giant heatsink on it is practically melting. That's the supply that has the 4700nF smoothing capacitor on it. I really don't quite remember/understand why the "15VAC" section has a smaller (1000nF) smoothing capacitor and the "12VAC" section has the whopper and is so hot. Furthermore, the "12VAC" section has a small capacitor that seems to short the 17VDC to GND - I seem to recall this is to safely discharge the 4700nF capacitor over time. I'll have to review the circuit as it stands, and the original build thread.

OK, so now I need to understand some stuff.

1) I'm going to be connecting the two power sections to DIFFERENT AC inputs this time. One will have 12VAC RMS supply and the other 15VAC.
2) The 15VAC is converted to ~21.2V DC and that is power for the LED array. Given LED voltage approx 4V then each LED string could be a maximum of 5 long (better, 4). I'm going to make a much smaller LED array this time - perhaps 5 strings of 4 rather than the original 8 strings of 5. (!!)
3) The 17VDC is going to be driving the main electronics and the motor... ? It doesn't seem to be unhappy, so keeping the original design seems OK.

The confusion is because I'm using just a single 15VAC input to both rectifiers, and although they are labeled "12V" and "15V" on my original board, they are both, of course, rectifying to 21.2V DC. Furthermore, the 1000nF capacitor on one, and the 4700nF capacitor on the other are both suppressing ripple - but don't really have anything to do with output voltage other than that (I think!). If I supply different input voltages (12V AC and 15V AC) to the appropriate rectifying circuits, then I end up with a 17V DC source of power, and a 21.2V DC source of power. The 21.2 DC source is sufficient to power the LED array, and the 17V DC is sufficient to power the motor and main electronics.

And by pumping just 17V DC through the (what I assume is a voltage limiter with a huge heatsink) instead of 21.2V DC, it's going to run a LOT cooler.
Which makes me think - maybe if I put 9V AC through that power section, I'd get 12.7V DC which may be enough and things will be cooler still.

Hopefully I'm starting to understand this better.

[gary]

Yes, I think you have it Andrew - the voltage limiter is actually a voltage "regulator" and the extra voltage is dropped across that regulator as heat.

[Andrew Davie]

I'm wondering how the capacitance of the ripple-fixing capacitors are chosen.
The 17V side has a "massive" 4700nF - and this came from a circuit diagram found/supplied during my original build.
The 21V side has a much smaller 1000nF one. Seems to me that there's little rhyme or reason here.

[gary]

I'm wondering how the capacitance of the ripple-fixing capacitors are chosen.
The 17V side has a "massive" 4700nF - and this came from a circuit diagram found/supplied during my original build.
The 21V side has a much smaller 1000nF one. Seems to me that there's little rhyme or reason here.

It's fairly simple, it depends on the load, if it's a low impedance load, that is, it draws a high current, the capacitors need to be heftier to maintain the voltage.

C = I / (f * V) (for 50hz supply f = 2 * 50 = 100)

so for 1 amp C = 1 /(100 * 12) = 833 uF or nearest available above that.

[Andrew Davie]

I have decided to go for modular design instead of monolithic. My original televisor had everything transplanted onto a single perfboard, and was practically impossible to trace and measure things, and when I made a mistake, that was difficult to fix. So I've started with a small perfboard - the one with the "railings" of holes connected together. Until last night I thought these were the stupidest things ever, because you couldn't place components wherever you wanted. But once I got out of that mindset and realised there was a major "+" railing and a major "-" railing, so to speak, then I could use that to minimise the number of joining wires I had to add - and with a bit of forethought it made the building very neat and simple.

I also bought some screw-type headers, so connecting wires to and from my modules won't be soldering, it will be through the headers. This will make things much less prone to damage as I move things around, and also make it easier to remove/fix modules, and wire it all up.

With that in mind, I built a simple "power" module - just a rectifier and a capacitor. I chose 4700nF basically because that was handy and I think hefty enough for any likely load. When I powered it off the transformer's 9V AC, I got 13.3V DC. So that's looking pretty good, much as expected. I hooked up the oscilloscope to the output and saw no sign of ripple at all - even at 10us timebase. Just a flat line, basically, within the noise limitations of the scope. Next I tried the 6V AC input and saw... 13.3V DC. Uh. Think, think.... and of course it's because the capacitor being rather large is slowly discharging from the previous test. So I disconnected input power, held a 360 ohm resistor across the output terminals and watched the voltage drop and when near 0 tried the 6V AC input again. Something like 8.3V DC so yep, totally understand this part now.

IMG_20170218_160528.jpg (377.94 KiB) Viewed 16152 times

I did briefly consider putting a bleeder resistor in to always drain the capacitor but given the low voltages involved, and also that this being connected to the other circuitry anyway down the track, the capacitor will discharge pretty quickly anyway. But I do understand the operation/reason for bleeder resistors, and what's happening, so all good.

Next step is to build another power unit and have one running off 9V AC input, and the other off 15V AC input, with the OUTPUT ground wires connected together (to give the common ground). I would then expect to have a fairly ripple-free 13V DC supply (which will drive the motor and main circuit) and 21V DC supply (which will drive the LEDs), and that will be my power side done. If the 13V is not enough I can go to 12VAC input ( --> 17V DC).


I realise this is all VERY simple stuff, but it's good to go back to basics and make sure I understand it all properly. Also I would like these posts to help others learn the basics and see the mistakes I'm making. It's amazing really that my original televisor didn't just go up in smoke - I don't think I quite understood WHY things were working, and my goal this time is to get the understanding, rather than just have something that works in a semi-magical sort of way.

[gary]

I hooked up the oscilloscope to the output and saw no sign of ripple at all - even at 10us timebase.

Good stuff Andrew, nothing like putting something together and testing it to get a feel for what's happening.

Just one thing, yes you have a ripple free output when you have no load on it, but when you put a load on it things will be different because the capacitor will be discharging between cycles - check the ripple with a dummy load that will be drawing a current at least as much as the real load with be and test the ripple then. I think it will be ok because that's a fairly good sized cap and I doubt you will have too much of a heavy load.

If you are going to have a regulator you don't want the ripple dipping below the regulator output voltage + drop out voltage of the regulator (~1.5V from memory)

[Andrew Davie]

Thanks Gary. Understood about the load.

I did build the 2nd board and cross-connected the ground on the two.
I fed one of them 9V AC and the other 15V AC.

When they are not connected together, they work OK: 13.3V DC and 22.1V DC give or take.
When I connect the ground outputs together (i.e, my concept of "common ground") then they BOTH read 22.1V DC.

So I'm NOT UNDERSTANDING this concept of common ground.

[Andrew Davie]

If you are going to have a regulator you don't want the ripple dipping below the regulator output voltage + drop out voltage of the regulator (~1.5V from memory)

Which is good to know - I had actually thought of this scenario and was wondering what the regulator would do/require. Given the 12V regulator (from memory) in the NBTV circuit, then 13.3V (from 9VAC) will probably not be enough DC input and I'll have to step up to the 12VAC (17VDC) input, which will then have sufficient for the regulator's voltage drop. That's easy; just a shame there's all that "wasted power" going to heat