Bipolar Power Supply Questions

[Jdansti]

Here's a few questions for the power supply experts. I just built the bipolar power supply below for use as a bench supply. It seems to me that the positive and negative supplies are somewhat independent, but:

1) Would it damage either half of the supply to use just one side?  For example, if I'm testing a circuit and I just need +9v, I might connect to the positive side and the common ground, and not connect anything to the negative side. Should I set one side to +4.5v and the other to -4.5v and connect across the positive and negative sides, or would either way be ok?

2) Would it damage either half of the supply if they were unbalanced. For example, the positive side set at +10v and the negative side at -5v?  I don't think I would ever do this, but I'm just curious.

3) The schematic shows a transformer with the center tap connected to ground. I'm using a plug in transformer with no center tap. I have about 22vac across the transformer output going to the rectifier. Is there any danger to the supply or to me due to the lack of a grounded center tap?

[Nasse]

I think no no yes. Get another transformer

[slacker]

The two regulated voltages are totally separate so you can just use one of them or have them at different voltages. If you're powering it with an AC wallwart or a transformer without a centre tap you can use the rectifier arrangement shown here.

http://www.musicfromouterspace.com/index.php?MAINTAB=SYNTHDIY&VPW=1075&VPH=517

[Seljer]

It doesn't work without the center tap if you want it to be full wave rectified bipolar power supply.

You can however pull off a half wave rectified bipolar supply with such a transformer, and providing you don't need truckloads of current for whatever device you're powering, you can put in some bigger filter caps and get away with it

edit: here is an example http://metroamp.com/wiki/index.php/Half_Wave_Dual_Polarity_Rectifier (its been used in a few AC powered pedals to get bipolar power)

[artifus]

slacker's link: http://www.musicfromouterspace.com/analogsynth_new/WALLWARTSUPPLY/schematicpage1_assembly.pdf

idiot question: when is 0v not 0v? is it simply a reference bias voltage in the middle of + and - or should it actually be 0v? what are the pro's and con's of a simple resistor voltage divider in this regard? and is the answer 'it depends' ?



*edit* 0v with reference to what? i s'pose ...

[Seljer]

slacker's link: http://www.musicfromouterspace.com/analogsynth_new/WALLWARTSUPPLY/schematicpage1_assembly.pdf

idiot question: when is 0v not 0v? is it simply a reference bias voltage in the middle of + and - or should it actually be 0v? what are the pro's and con's of a simple resistor voltage divider in this regard? and is the answer 'it depends' ?

You can use a resistive divider to generate a bipolar supply, having an isolated supply (say a battery) and connecting the middle point of the divider to ground. The issue is that its prone to becoming unbalanced as soon as one side drags more current than the other. If your circuit will exhibit such behavior, this may be an issue depending on how much headroom and/or voltage stability you need on both rails. You can make the two resistors smaller which improves on this problem but also increases your total current consumption.

In pedals, with the Vcc/2 generated like this its usually not an issue since its only a reference voltage, with barely any current being drawn (most likely only hooked up to the high impedance inputs of an opamp)

If you need bipolar supply capable of delivering slightly more power you can however make an active buffer for the 'virtual ground' instead of the resistve divider which improves this issue greatly.

[artifus]

cool. think i'm getting it now. i'm guessing more and bigger caps of varying values help with the balancing of current consumption, to an extent, across all frequencies on both sides. my confusion lies in seeing 0v and gnd listed, is this just short hand for bias or half supply? i've seen opamps used as such and even a 386, but the question remains - when is 0v not 0v? can it be 4.5 @ 9v and 6 @ 12, etc., or must it be 0v? again, with reference to what? - i feel i'm missing something stupidly simple here...

[Seljer]

cool. think i'm getting it now. i'm guessing more and bigger caps of varying values help with the balancing of current consumption, to an extent, across all frequencies on both sides. my confusion lies in seeing 0v and gnd listed, is this just short hand for bias or half supply? i've seen opamps used as such and even a 386, but the question remains - when is 0v not 0v? can it be 4.5 @ 9v and 6 @ 12, etc., or must it be 0v? again, with reference to what? - i feel i'm missing something stupidly simple here...

Caps only help against noise. The posts above are about the issue when you split a single supply in half to generate a true bipolar supply by generating a half-way point and connecting it to ground. 0V is 0V, a predetermined starting point from which you base the rest of your voltages. Kind of like the way you can hook either the + end or the - end of a 9V battery to ground to use it as either a +9V supply or a -9V supply.

We make opamps work on a 'virtual ground' by using coupling capacitor to reject DC and keep only the AC signal. With a true bipolar supply you can have a dc coupled opamp circuit.

[artifus]

0V is 0V, a predetermined starting point from which you base the rest of your voltages. Kind of like the way you can hook either the + end or the - end of a 9V battery to ground to use it as either a +9V supply or a -9V supply.

We make opamps work on a 'virtual ground' by using coupling capacitor to reject DC and keep only the AC signal. With a true bipolar supply you can have a dc coupled opamp circuit.

a predetermined starting point could be 4.5 or 6 or 9v, if working from 9, 12 or 18v, or whatever, just as 0v could be - the only disadvantage being less headroom, or not so much of an ac swing as far as i can currently see - what am i missing?

[Gurner]

cool. think i'm getting it now. i'm guessing more and bigger caps of varying values help with the balancing of current consumption, to an extent, across all frequencies on both sides. my confusion lies in seeing 0v and gnd listed, is this just short hand for bias or half supply? i've seen opamps used as such and even a 386, but the question remains - when is 0v not 0v? can it be 4.5 @ 9v and 6 @ 12, etc., or must it be 0v? again, with reference to what? - i feel i'm missing something stupidly simple here...

When is 0V actually 0V?...when the point you are speaking of, is physically grounded.

For example, for a a battery .....is the black wire 0V & the red wire 9V?  Nope, the black wire is most certainly not 0V   ...it's simply at a potential difference of -9V in relation to the other (red) wire. If you attach a battery black wire to a true ground, then the battery black wire becomes 0V. If you attach the battery black wire to a voltage sitting at say 20V above true ground, then the black wire is +20V & the red wire is 29V wrt true ground.

Voltage is a relative thing... e.g. are you referencing a measurement to ground or are you referencing to another potential (like my battery black wire example) etc.

A virtual ground is just a convenient way of biasing up single supply circuit (to save having to go the bipolar supply route) ....a virtual ground is not true ground (cos it's not 0V!) ...hence the virtual bit in the name....when working with virtual grounds you start then thinking of the AC signal swing in relation to the virtual ground voltage (vs. true ground). For most 9V circuits, a virtual ground would be 4.5V....this is because you want the AC signal to swing the maximum amount to the positive & negative side...the only way of achieving that, is by having the AC signal 'sit' on a DC voltage exactly half way between the 9V supply voltage.

[artifus]

potential difference - that was the phrase that was on the tip of my tongue - thank you gurner. which brings us to potential divider, or voltage divider as mentioned earlier. ok, forgive me - not much sleep these past few days. next question, down to nuts and bolts - can a device be damaged by sending it half supply, a virtual ground, rather than referencing it to a true 0v rail? is it merely a reference for it to work from or must it be an actual 0v (if such a thing exists in reality outside of text books)? sorry for being so slow - may as well get this cleared up and unstupided now if i can - thanks guys.

*and apologies to jdansti for the brief thread derailment!*

[Gurner]

- can a device be damaged by sending it half supply, a virtual ground, rather than referencing it to a true 0v rail? is it merely a reference for it to work from or must it be an actual 0v (if such a thing exists in reality outside of text books)? sorry for being so slow - may as well get this cleared up and unstupided now if i can - thanks guys.

You'd need to narrow down what you mean by 'device' (also when speaking of virtual grounds, we aren't really 'sending' anything).

this is both a difficult one to explain (& if my own personal experience is anything to go with....difficult to grasp!)...that's becuase there are many terminolgies & ways of referencing all of this (& a lot of people using the terms wrong too!). The virtual ground is just a way of shifting the DC level in relation to cater for the incoming AC signal you want to handle. For an opamp, if you've an AC signal swinging ±4V around true ground (0V) & if you don't use a bipolar psu, then the negative swings won't get through (you end up with a half wave rectified signal)....only the AC signal swings above 0V will get through the ..so the trick is to offset/shift the DC level upon which the AC signal swings on...therefore if you create a DC level of 4.5V,both the +ve & -ve swings of the signal will get through the device. So your signal then becomes referenced to the 4.5V virtual ground you've created.

[artifus]

yes, it is the use of language which is confusing me somewhat, we are indeed referencing and not sending. ground, 0v, vref, etc. and other such pedantic semantics.

i s'pose the question is bipolar versus virtual ground... i'm re-reading all your replies and it's slowly sinking in, or some of it at least, i'm sure it'll all make a lot more sense after a good sleep. by device i meant chip, so yes, transistors and transistor arrays in opamps, etc. *and by damage i mean destroy or otherwise cause irreversible physical damage rather than just work wonky.*

i guess i'm struggling with the idea of a true bipolar/dual supply of say +4.5v, 0v and -4.5v versus a virtual ground circuit using 9v, 4.5v and 0v. i'm not sure i see a *(potential - zing!)* difference if it is all reference.

[Gurner]

i guess i'm struggling with the idea of a true bipolar/dual supply of say +4.5v, 0v and -4.5v versus a virtual ground circuit using 9v, 4.5v and 0v. i'm not sure i see a *(potential - zing!)* difference if it is all reference.

the 'virtual' ground is with reference to the AC signal swing....after all we are dealing with an AC signal & we want that to flow through our circuits without bashing into the rails (clippage) .....so, if you've only got a single 9V supply circuit then the AC signal needs to be give a leg up & sit on a DC level at to half the supply (4.5V) else the negative swings won't get through (they''ll be clipped)

If you then measure the AC swing referenced to your circuit's 'ground' the AC signal will be seen swinging 0 to 9V, however if you measure the AC swing referenced to your virtual ground (which is sitting at 4.5V), the AC swing will be +4.5V & -4.5V ...like I say, it's a relative thing.

[artifus]

so it's all reference and doesn't really matter* - relativity. got it. nice one, gurner, thanks again.

*within the datasheet min & max ratings and the circuit at hand

[Seljer]

i guess i'm struggling with the idea of a true bipolar/dual supply of say +4.5v, 0v and -4.5v versus a virtual ground circuit using 9v, 4.5v and 0v. i'm not sure i see a *(potential - zing!)* difference if it is all reference.

Thats the point, an opamp works in both situations (usually), you just have to adapt the rest of the circuit around it . The feedback does it's thing, the opamp tries to make the + is at the same potential the - input. Potential is always relative

Using a bipolar supply has benefits in many situation. Sometimes you don't want to use large (de)coupling capacitors (i.e. a power amp or any low impedance load for that matter where you need big capacitors to not have low end loss). Sometimes you need to amplify a DC voltage (say with a force/temperature/pressure sensor that only puts out a couple of millivolts). Theres also noise issues and ground current issues (i.e. if the load is balanced on both supply lines, theres no current flowing to ground which leads to less noise)

[artifus]

The feedback does it's thing, the opamp tries to make the + is at the same potential the - input. Potential is always relative

and the key fits the lock, slowly turns and the door is opened - now for the big push... thank you very much, seljer. i'll try not to be blinded by the oncoming light!  

*edit* i should have taken notes - http://www.diystompboxes.com/smfforum/index.php?topic=98499.0

[Jdansti]

The two regulated voltages are totally separate so you can just use one of them or have them at different voltages. If you're powering it with an AC wallwart or a transformer without a centre tap you can use the rectifier arrangement shown here.

http://www.musicfromouterspace.com/index.php?MAINTAB=SYNTHDIY&VPW=1075&VPH=517

Thanks to everyone for the info.

Slacker, I think that this is the circuit is the one you are referring to.  Yes?

[slacker]

Yeah, that's the one, sorry I didn't realise my link didn't go straight to it.

[Jdansti]

Yeah, that's the one, sorry I didn't realise my link didn't go straight to it.

No problem!  Thanks for the help!