I've bought a broken scope haven't I ?

[aneng]

I can see I'm going to need to learn more about negative voltages and grounds etc because I'm hitting a 'layman's logic block' here !  e.g. If I connect two psu's in series, say, 15V each, then I'm going to get 0V at the -ve terminal of the first, 15V at the +ve terminal, and, as it's connected here, at the -ve terminal of the second, then 30V at the +ve terminal of the second.  Therefore, my '0V' is really +15V isn't it ?

If I were to connect this to a circuit that also had an earth connection, that would cause current to flow from the circuit '0V' line to earth (the 'real' 0V).

I know there's a hole in my knowledge here!  Where can I go to grasp this concept properly ?  Is it in the Forest Mims book ?  (I'll look now...).

[Ian.M]

There is no such thing as an absolute 0V.  *ALL* voltage measurement is differential.  Picking a circuit node, calling it 0V and measuring all voltages with respect to that point is merely a convenience.   Sometimes the node is picked for you, e.g. Earth ground is commonly taken as the 0V reference point, and  oscilloscopes usually have the BNC shell grounded so forcing you to ground the 0V node of your circuit.    Thefore tpyr two series connected floating 15V PSUs cold be considered as a 30V PSU (+ve or -ve, tapped at 15V) or as a +/-15V supply, and if your circuit includes level-shifting between sections, what you consider the PSU to be can vary between section.   e.g. it is sometimes advantageous to make the 5V logic supply the same as the 12V power rail  and provide a logic Vee/Vss rail at 7V that acts as 0V for the logic section - because it makes it easy to control P channel MOSFETs as high side switches.   If one of the supply rails is connected to the case, chassis or earth, it is generally foolish NOT to use that as the 0V reference point.

[Brumby]

I've used a couple of 9v batteries for this illustration, but you can look at each of them being a floating power supply...

Question: What's the difference between these 3 circuits?



Answer: Where you put the 'common' (black lead) of your multimeter.  The circuits themselves are identical and any other circuitry off to the right will not care where you put your multimeter leads.

[Brumby]

While there are situations where a few more details might be relevant, the following has been kept really simple - to demonstrate the basics....


These batteries are just like 'floating' power supplies (usually a pair of red and black terminals) - they aren't connected to anything.

The (usually green) earth terminal IS connected - but only to the earthing pin that goes back through your power point.  It is not connected to any of the red or black terminals ... unless you make a connection.

Now here comes the head spin....

With the circuit above applied to two floating supplies, you can connect the earth to any one of the three measuring points without affecting anything.  FURTHER, if you keep the common lead of your multimeter connected to whichever of the three measuring points you used before, the readings will not change.


There are two simple rules here:
 1. That all voltages are measured from a specific reference point.
and
 2. A voltage measurement is only valid if the reference point and the measurement point are electrically connected through a circuit of some kind.  (Unconnected windings on a transformer don't count as they are magnetically coupled.)

[aneng]

Thanks for that guys...  it's a lot clearer now.  SO, taking my function generator as an example, if I were to connect two 15V supplies as per the diagrams above to give me VRAW (+15V), 0V, -VRAW (-15V), what would the scope do if I measured the output of the unit (say a sine wave) with the ground clip of the probe connected to this arbitrary '0V' point - given that the shield of the BNC connector is likely connnected to mains earth ?
(actual circuit notwithstanding - I'm assuming the output would centre around the 0V line for the sake of this argument).

Would it show as a +15V offset or not ?  ...and, presuming the power supplies were isolated from any earth, would any current flow if I connected the +ve terminal to mains earth ?  If I've understood your posts correctly, I'm assuming no current would flow. ?

[MarkL]

Would it show as a +15V offset or not ?  ...and, presuming the power supplies were isolated from any earth, would any current flow if I connected the +ve terminal to mains earth ?  If I've understood your posts correctly, I'm assuming no current would flow. ?

Your assumption is correct.  It would not have an offset, assuming you asked the function generator to output a waveform with 0V offset.

If you connected +ve to mains earth no current would flow, BUT as soon you connected your probe ground clip to the 0V point on the circuit, you have now created a short between +ve and 0V since they are both at earth potential.  The shorting path is through the earth ground in the scope and power supply line cords.

[aneng]

Ooh... now.... err.....

Right, so it wouldn't have an offset because, although the waveform would be centred around +15V relative to the -ve lead of the first power supply (i.e. the middle point in your middle example pic), because there is no circuit between the function generator and mains earth, there is no potential difference between circuit 0v (i.e. the centre of the waveform) and the mains-earth-connected ground clip of the scope ?

My brain is aching at this point..

"you have now created a short between +ve and 0V since they are both at earth potential."

The... what... but....... eh ?

Oh... there it goes...  it's imploded. 

[Brumby]

Breathe .... relax ....  You are not far from understanding!

I refer you to this part of my previous diagram - as this is the more typical way of expressing a split power supply...

If you connected +ve to mains earth no current would flow,

Restated: If you connected the +ve terminal (the one at the top) to the earth terminal on the power supply, then nothing nasty happens.  The +ve terminal is now directly connected to mains earth, through the power supply cord going back to the power point. No current would flow.

You then bring out your oscilloscope, plug it into another power point, where it is earthed as well.  For several reasons, oscilloscopes have this earth firmly connected throughout, including to the 'earth' connection on your probe.
 

BUT as soon you connected your probe ground clip to the 0V point on the circuit,

Connecting the probe's ground clip to the 0V point (the one in the middle)...
Do you see the problem...?

you have now created a short between +ve and 0V since they are both at earth potential.  The shorting path is through the earth ground in the scope and power supply line cords.

This is something about which you should be very careful - as you can start blowing up things!  So much so, that Dave has done a video on just this issue:

[KJDS]

Have a look through the PSUs I have listed on ebay. PM me if any take your fancy and I'll do a good deal on it.

[aneng]

Right... I get it now - thanks Brumby (and others !). !

@KJDS Thanks for the offer - I'll have a look.

As for getting a bench PSU, I think I'm going to need one regardless.  I would prefer to buy a commercially-made one I think as they look SO much better than the DIY ones!  Not to mention it would take me weeks to build it.

So, in Dave's video about setting up a lab, he suggests building a dual-tracking psu and says there is loads of info on the forum.  However, I've searched and can't find any threads that have designs for them.  Am I not looking in the right places or what ?

I don't really want to spend more money than I can afford, but....  if I'm not going to save a great deal by building one (i.e. at least 50%), then I may as well just go straight for a manufactured one.  I've seen this one on eBay, but it is a bit more than I wanted to spend at 108 GBP, however...  it looks like it would last me a while... 

[aneng]

Naah...  changed my mind.  On closer examination, you need to move a slider on the display to switch between volts and amps.  That's a bit 'meh'.

[nfmax]

This one, maybe? http://www.ebay.co.uk/itm/Farnell-L30BT-/151951331949?hash=item236101526d:g:hqAAAOSwa-dWmM7B With luck, it may even need repair!

[aneng]

Thanks nfmax, that would have been a bargain if it had been a bit nearer to me.  It's collection only, and unfortunately, a 270 mile round trip.

Tell me guys, I can see that having a dual tracking PSU is the most desirable type of bench supply to have.  Why is it then that 90% of the PSUs on eBay are single output and I can't find a design on the forum to build my own 

Do most jobs require only a single output ?

[MarkL]

Naah...  changed my mind.  On closer examination, you need to move a slider on the display to switch between volts and amps.  That's a bit 'meh'.

Here's one that displays volts and current for both channels at the same time.  And there's a third output with 5V @ 3A.  The third output has no readout and is not adjustable, but that's fairly common and could be used for powering digital circuits.

  http://www.ebay.co.uk/itm/261646567155

The specs are ok if you can trust them, and 5A is more than you'll ever need to power small projects.  I've never heard of MLINK, but the design looks a lot like other makes.  No doubt a Chinese OEM.

Do most jobs require only a single output ?

Depends what you're doing.  If working on digital circuitry, then usually one output is fine.  If you do a lot of work with analog and op-amps a dual supply can be a big help instead of having to derive your own 0V reference.  Hybrid analog/digital often will need a dual or triple supply, and sometimes more.

You'll never regret buying more outputs than you think you need.

[nfmax]

Dual tracking supplies can be useful, but they almost invariably have a common 0V output, that is, one output gives you plus some adjustable voltage, while the other gives you minus that voltage (or some adjustable fraction of it). You can't connect the two in parallel if you want more current, they are always in series. On the other hand, at least in good quality supplies, they will turn on and off together - i.e. you never get, even for a moment, one supply much higher voltage than the other. This can matter with sensitive circuits.

Good power supplies also never overshoot when turned on, the voltage rises steadily from zero to the set value, never going above it. Good supplies also have low levels of noise on their output, they have accurate meters, and either multi turn controls or separate coarse and fine controls so the output voltage is easy to set. They are also reliable, and naturally cost more. The advantage is they go on for donkey's years so a 40 year old supply like that Farnell is quie a feasible proposition.

For someone like yourself, just starting out, I would recommend a dual independent (isolated) supply, covering up to about 15 or 20 volts at about 1 amp, with either digital or good quality analogue meters, and adjustable current limiting. Like that Farnell, in fact. You don't really need simultaneous measurement of voltage and current for both supplies (though it is nice) because you generally use the voltage meter only to set the supply output, then measure current while your test circuit is operating. On the other hand, an adjustable current limit is very useful. This allows you to set a maximum current that your circuit can draw from the supply: if it tries to take more, by some mischance, the supply will automatically reduce its output voltage to keep the current at the limit value. This can help prevent tears!

Designing a good quality (bench) power supply is harder than it looks. eBay is full of the offerings of people who have yet to learn this lesson. You will probably do better with a second hand unit from one of the leading manufacturers - HP (Agilent), Thurlby-Thandar, Tektronix, Keithly, Farnell, or Hameg are likely candidates, but there are many others.

[aneng]

Thanks for all that info both.  Is the 'tracking' bit just a convenient way to match the two outputs then ?  i.e. if I got a two (independent) output supply that didn't have tracking, I presume I could just manually set the voltage and current limit for each output separately, achieving the same result (with the exception you mentioned that one rail could be different or on/off compared to the other) ?

That one you linked to looks alright Mark.  Shipping isn't too bad considering it's in Barcelona !  If what I've surmised above is correct, then I may go for an old-school unit for a budget price - unless of course I see something that looks too sexy to resist !  I'm a sucker for good looking equipment !

[Brumby]

Is the 'tracking' bit just a convenient way to match the two outputs then ?  i.e. if I got a two (independent) output supply that didn't have tracking, I presume I could just manually set the voltage and current limit for each output separately, achieving the same result (with the exception you mentioned that one rail could be different or on/off compared to the other) ?

    10/10

[aneng]

[MarkL]

That one you linked to looks alright Mark.  Shipping isn't too bad considering it's in Barcelona !  If what I've surmised above is correct, then I may go for an old-school unit for a budget price - unless of course I see something that looks too sexy to resist !  I'm a sucker for good looking equipment !

That one looked liked a reasonable new unit, but old school would be great and, in my opinion, actually preferable if you can find one in your price range and from a known reputable manufacturer.  HP/Agilent, Lambda, Sorensen, TTI, Power Designs are the ones I would seek.  Leader, Tektronix, Xantrex, Kikusui are also decent.  Leader makes the LPS-151 and LPS-152 which are priced well and I think would make a nice "old school" first power supply, although it doesn't have all-way metering.

I would also try to find a linear supply for your first supply.  They have less ripple and noise than switching supplies.  Many listings don't say, but you can usually tell because they are heavy and have cooling fins sticking out the back.  Most old supplies will be linear.

[nfmax]

Note:

• TTI = Thurlby-Thandar Instruments
• Lambda, Sorenson & Power Designs are more common in the USA than in the UK

I forgot to mention the linear/switching difference & I agree with Mark you want a linear supply. You should also not bother with a (computer) programmable/remote controlled supply at this stage.

[aneng]

I've found an old school one that appears to do 50V and 0.5A that I might get for a bargain.  I know that Dave suggests 1A in his "Setting up a lab" video, but I'm hoping 500mA will suffice for now?

[MarkL]

500mA is a little lean, but it would certainly get you through the current project of repairing the function generator.  I agree with Dave and would opt for something with 1A or more given the choice.  I guess it depends how much of a bargain it is.

But no matter, if you stick with doing electronics projects I can guarantee it won't be your last power supply.  There are some people on this forum who can never have enough of them.  (You know who you are...)

[aneng]

I'll know how much of a bargain it is come Sunday...  All will be revealed then (hopefully).

[Brumby]

There are some people on this forum who can never have enough of them.  (You know who you are...)

Does the old laptop power unit I keep on hand for 12v up to 4A count?

[aneng]

Well, the 500mA old-school one I was watching went for stupid money for such an old device, so I let that one go.  There's a few more coming up this week so everything is on hold really until I get a dual-output PSU.  I must say, I'm a sucker for the LED displays, especially if they are different colours like this one...