Single transformer for positive and negative ground circuits

Started by DimebuGG, December 30, 2017, 09:09:24 AM

Previous topic - Next topic

DimebuGG

I'm not sure if the title was correct but here's how:

I was planning to build a +12V distortion circuit and a -15V DMM delay both in a 1590DD box. Initially thought of putting two dc jacks for each circuit but I came up with the idea of using a single transformer to power the two circuits. Basically, an 18VAC xformer/half wave rectifier but using 7812 on the positive side and 7915 on the negative side, having GND in the middle. To be clear, it will be similar to this http://www.generalguitargadgets.com/wp-content/uploads/ggg_bipolar_ps.pdf.

I think it will work :icon_biggrin: but I'm not sure though if there would be issues due to uneven voltage or maybe the current draw of the two circuits. Any thoughts?.

PRR

I think it will work.

The unbalanced current is not kind to the transformer when used this way. Use a plenty-big iron.
  • SUPPORTER

Rob Strand

Using two of these in opposite polarity is a bit more friendlier to the transformer but requires two extra caps.
Send:     . .- .-. - .... / - --- / --. --- .-. -

DimebuGG

Thanks for the heads up guys. I think the DMM draws about 50mA current and the distortion circuit probably draws around 10-15mA max.

Meanwhile, I found this:

DimebuGG

Quote from: Rob Strand on December 30, 2017, 11:32:44 PM
Using two of these in opposite polarity is a bit more friendlier to the transformer but requires two extra caps.

Hey Rob, you mean like tapping the same AC lines but with all components in reverse/opposite polarity for the negative output?

Rob Strand

QuoteHey Rob, you mean like tapping the same AC lines but with all components in reverse/opposite polarity for the negative output?

Yes.

So you would mirror that circuit downwards, then change all the part polarities.  The top tap on the transformer connects to two caps.  The bottom tap on the transformer is the common and becomes 0V.

Don't forget there's a voltage doubling on *both* outputs.

[FYI:  On that Springer circuit  I'm not sure C1 and C2 are the right way around.]
Send:     . .- .-. - .... / - --- / --. --- .-. -

DimebuGG


R.G.

If I understand transformers correctly, it is not possible to saturate the core by loading conditions on the secondary. Otherwise, half-wave rectifiers would be impractical; yet they are used effectively in many applications.

The circuits making bipolar power from a single secondary amount to two half wave rectifiers, each working on alternating half cycles, so regardless of different loadings on the opposite polarities, the transformer is >better< balanced by any loads whatsoever on the two outputs. The worst cales of imbalance, no load on one polarity, contracts down to a simple half wave rectifier, which is known to work.

The advice to use a bigger transformer, i.e. more iron, may be wise; half wave rectifiers are not efficient users of transformer iron. But again, if the transformer works for the more heavily loaded half of this circuit, it likely works for the bipolar version. This is because loading the unused half cycle uses the "unused" half of the AC cycle, using up some of the thermal capability that a simple half wave doesn't use. Using a larger transformer improves the thermal ability of the tranny, and that's always good if you can afford the size, weight, and expense. Transformers in DIY equipment are rarely designed right up to the edge of the transformers' capability anyway.

The real issue in using dual half-wave rectifiers to make bipolar is really the capacitance needed. You need twice the capacitance value for filter caps because these caps have to support the load current for twice as long between refills from the rectifers. Looked at another way, for the same capacitance used as a filter, the ripple voltage is approximately twice as big.

By the way, it's useful to read "Power Supplies Basics" at geofex.com
(http://www.geofex.com/Article_Folders/Power-supplies/powersup.htm)
for some understanding about power supply rectification and filtering.
R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.

Rob Strand

QuoteIf I understand transformers correctly, it is not possible to saturate the core by loading conditions on the secondary. Otherwise, half-wave rectifiers would be impractical; yet they are used effectively in many applications.

Every time the half-wave core saturation thing comes up it makes my head hurt .   I know where you are coming from: the core regulates the B field regardless of load.  However, the whole situation is very subtle, it does actually saturate!

If you load-up a transformer with a half-wave rectifier you will notice is groans quite loudly.  It's bad enough that you can't help thinking more it going on.

The simplest way I can explain it is like this:   For a concrete example, take a simple situation where you have a positive half wave rectifier with a resistive load and no filter caps.  Consider the case where the transformer primary has a series resistance Rp.  Suppose the input current is Iin.

On the positive cycle the *internal*  primary voltage of the transformer is,

Vtp   =   Vin - Rp * Iin

On the negative cycle the *internal*  primary voltage of the transformer is,

Vtn   =   Vin

Here, Vtp and Vtn are half sinusoids and Vtn is swinging negative.

The core flux is related to the *internal*  primary voltage by,

Vt  = N1 (d phi  / dt)   ; N1 = primary turns;  phi =core flux

If we integrate this to find phi we get,

phi =   (1/N1) time integral of (Vt)

     =    (1/N1) time integral of (Vtp)   +  (1/N1)  time integral of (Vtn)

The positive Vin part in Vtp cancels out the negative Vin part of Vtn.

However we are left with a small voltage due to the voltage drop under load on the positive cycle,

phi  = (1/N1) time integral of (- Rp * Iin)

As we integrate over through more and more cycles this part grows so the core flux keeps growing.
It grows until the negative flux swing gets near saturation.   At this point the magnetization current rises on the negative polarity. This ends up causing an IR voltage drop on the negative input cycle Vtn which balances out the IR voltage drop on the positive side Vtp.   The higher the load current the deeper it has to get into saturation to balance things out - that's why the transformer groans under high loads.
Send:     . .- .-. - .... / - --- / --. --- .-. -

PRR

> if the transformer works for the more heavily loaded half of this circuit, it likely works for the bipolar version.

Agree. I was in my dog-porch mode, not aerospace mode. If I calculate my dog-porch will need 0.87"x1.9" joists, and find 1.5"x2.5" wood cheap, I'm happy to have extra stuff to cover my slop or possible dog weight-gain.

The 50mA one side will cause a DC flux in the core, increased peak flux and also higher reactive current. Small transformers already have high reactive current, and run warm from that. However their surface/volume ratio allows high loss without high heat. And half-wave is common in small supplies.

And as you note, the 15mA on the other side offsets the first side.

EDIT-- I bet Rob covered it better while I was typing.

High capacitance.... a poor rule of thumb suggests 50uFd. At this size and economic, 500ufd is perfectly affordable. And there's regulators to take much of the curse off. A million-unit production run would need a sharper pencil, but a one-off should be fine simply over-built.
  • SUPPORTER

R.G.

Quote from: PRR on December 31, 2017, 09:23:34 PM
The 50mA one side will cause a DC flux in the core, increased peak flux and also higher reactive current.
Best I can tell, no, it won't cause a unidirectional flux in the core. At least not appreciably.

If I have this right:
> The reactive current in the primary energizes the core.
> The M field in the core causes a voltage to be induced into the coils of wire that surround the core - the same coils that constitute the primary inductance that the magnetizing current flows through that make the M field.
> Current in the primary increases with the applied incoming primary voltage, but the back EMF is in the direction to oppose the primary voltage.
> It is the difference between the applied primary voltage and back EMF that let primary current flow, and enough current flows to make an M field that generates just enough back EMF to let that current flow.

If this sounds like an over-thought way of thinking about minutiae about inductors, well, yes, it probably is. The difference comes when you let some of the M-field enenrgy out through a secondary. The secondary lets some of that core energy stored in the M field leak out. The lessened field supports a lower back EMF, and that provides a bigger difference in applied voltage and back EMF to let more current in the primary, restoring the balance of the M field.

The secondary can be thought of as a spigot letting water out of a tank with an automatic float level. Water flows into the tank, and water flows out the "secondary" pipe, but the level of the water in the tank changes hardly at all, only the tiny differential needed to open the float valve.

Letting energy out through a secondary is instantly balanced by increased current into the primary, and the change in the core M field is essentially zero. So there is no net change in the conditions in the M field in the core; only the minute change needed to let through enough energy to balance what's going out the secondary. It's that "let through" as opposed to "provide" that means that you can't substantially affect the flux in an AC sine wave excited power transformer by drawing on the secondary. Note that this is NOT true for switching type transformers that are fed various duty cycles of square(ish) waves. In that instance, you do have to do the math to figure out the energy balance for the core from moment to moment.

What's confusing is that you can overheat and burn out a transformer by sucking too much current through the secondary, and that seems like the same thing that happens with DC primary voltage. But it's not. Just a similar smell in the combustion products.

Another indicator of this is that the phase of the AC line current changes as secondary current increases. With no secondary loading, the primary current is essentially all reactive, 90 degrees (about) to the primary voltage. As you load the secondary, you can watch the phase change to match the sum of the reactive magnetizing current and the reflected secondary current, whatever that phase is. Resistive loading is particularly easy to see the phase change with.

I spent about two hours arguing this with the professor that convinced me of its truth. I told him it was patently obvious that secondary current could indeed saturate the core, and we had at it in his office for a while. He was kind enough not to throw me out of his office, but argued with me until I got it. I finally agreed with him, apologized for the argument. He told me kindly that he was glad I was willing to put in the effort to support my opinion and understanding well, even if I was wrong. He also told me I had an A for course and didn't need to take the final.

THAT lesson stuck.
R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.


R.G.

This is a particularly difficult facet of transformers to understand, and I have run into other electronics professionals who would argue with me loud and long over the point, just as I did with my professor. And I have not personally dug deeply enough into transformers to completely understand them, which is why my comments were salted with words like "can't substantially affect", allowing for some perverse situations.

But in general, I think it is substantially impossible to saturated a sine-wave driven transformer by loading the secondary. Note that this excludes using the power from the secondary to feed back and offset the primary voltage - that's cheating, and of course you can figure out ways to cheat. :)   

Use half wave rectifiers freely. I have for decades. I've never had a situation where any failure can be attributed to half wave loading walking the core flux, although I have killed them other ways.    :icon_eek: 
R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.

amz-fx

I've used this general setup many times...  adapted from a Paia design:



Or you can use a dual secondary transformer:


Rob Strand

#14
QuoteThis is a particularly difficult facet of transformers to understand, and I have run into other electronics professionals who would argue with me loud and long over the point, just as I did with my professor.
I've argued with myself many hours (and many times).  I believe I got to a point where I can explain all the behaviour.  Using a non-linear magnetics model in Pspice I can see exactly what is predicted.  I've even put in a soft-start up to eliminate all the complexities of inrush current and transients (which are made worse by magnetics saturation; this  saturation is not from the half-wave, but in general, as a transformer is normally designed with flux swings close to saturation.)

A hint that saturation can occur is the transformer's primary current waveform should only have cycles of one polarity as the load is only present on one half cycle.   That implies there is a mean DC current going into the winding.

QuoteI think it is substantially impossible to saturated a sine-wave driven transformer by loading the secondary.
It's not a full saturation, the tips of one side of the swing rides the edge of saturation.  They are pushed progressively deeper into saturation with increasing load.

The saturation current at the tips creates a primary current waveform of opposite polarity and removes the mean DC current from the primary.
The biggest eye opener from all this is the transformer primary is *forced* to dissipate power in the primary on half-cycles where the load is disconnected.

QuoteUse half wave rectifiers freely. I have for decades. I've never had a situation where any failure can be attributed to half wave loading walking the core flux, although I have killed them other ways.
Me too.  From outside, it's hard to know the evil going on inside!   There is some hint in the current waveform.  I built a PSU using a 50VA transformer with the rectifier in amz-fx's post.  When operating at highish currents with a one-sided load it made a loud noise and got hot.  The noise was not present when operated symmetrically.   After spending a lot of time looking at it I realized some of the evils of half-wave rectifiers.

Send:     . .- .-. - .... / - --- / --. --- .-. -

R.G.

Quote from: Rob Strand on January 01, 2018, 05:05:23 PM
A hint that saturation can occur is the transformer's primary current waveform should only have cycles of one polarity as the load is only present on one half cycle.   That implies there is a mean DC current going into the winding.
Yes. There is a mean DC current going through the primary. And there is a mean DC current going through the secondary. The phasing is such that these two cancel in terms of the effect on the magnetic field in the core.

Quote
It's not a full saturation, the tips of one side of the swing rides the edge of saturation.  They are pushed progressively deeper into saturation with increasing load.
I've never found any indication of this, including with some very abusive half wave rectifiers.

QuoteThe saturation current at the tips creates a primary current waveform of opposite polarity and removes the mean DC current from the primary.
This makes me very curious. The current out of a half-wave rectifier, as reflected to the primary, can easily be vastly larger than the magnetizing current. A decent AC mains transformer design will have magnetizing current well under maximum primary current - generally under 5%, often under 2%. It is easy to load a half wave rectified secondary to a substantial fraction of the full current for a resistive load. This is arguably in the range of half the full resistive load, and therefore about half the primary current. The "offset current" in the primary is then many times the magnetizing current, so the thing ought to be driven dramatically into saturation if the offset primary current was not substantially cancelled. That way of looking at it seems to indicate that the one-directional primary current isn't pushing the magnetics into saturation. I think.

QuoteThe biggest eye opener from all this is the transformer primary is *forced* to dissipate power in the primary on half-cycles where the load is disconnected.
Of course it is. The question is then - power from where? It's clear that the primary wires heat up more on the half cycle of higher current, and this heat has to be dissipated over the next cycle, but this isn't a statement about the M-field. It is true that the RMS value of the current in a half wave rectifier is different from the RMS value of the current in a full wave, and that must be taken into account in transformer heating.

Or am I missing your point entirely?

QuoteI built a PSU using a 50VA transformer with the rectifier in amz-fx's post.
I must have built fifty similar ones. That schematic is one example of the vast general case. There are probably many, many more of them on the net, in varying degrees of specificity.

QuoteWhen operating at highish currents with a one-sided load it made a loud noise and got hot.  The noise was not present when operated symmetrically.
I have never run into that situation where I didn't find that I had made some other mistake. Not that it's impossible, but all of the situations I've found were explicable by other means than core flux walking.
R.G.

In response to the questions in the forum - PCB Layout for Musical Effects is available from The Book Patch. Search "PCB Layout" and it ought to appear.

Rob Strand

#16
QuoteYes. There is a mean DC current going through the primary. And there is a mean DC current going through the secondary. The phasing is such that these two cancel in terms of the effect on the magnetic field in the core.
That's exactly right.  Sorry you caught me out thinking in the "old way".   When the secondary conducts, the field due to the primary and secondary always cancel.  Also when there's no primary resistance, the flux in the core *always* agrees the input voltage (in the sense of E = N1 d phi /dt).  It doesn't matter what we do this will always hold true.

It's because that view makes so much sense it hard to get off the trail - I know because I kept getting stuck.  The flux creep problem only shows up when you have a primary resistance.   Flux creep relates to voltage imbalance not current.  So you have to think in terms of primary voltage.  With current you get stuck in a self consistent loop.  However because of the primary resistance,  voltage and current gets entangled through IR drops.

QuoteI've never found any indication of this, including with some very abusive half wave rectifiers.
I never thought much of it myself until I started looking into it.   However, when I think back I remember when I was a kid we had a car battery charger (maybe 30W to 50W).  It had a "trickle charge setting" which was basically a half-wave rectifier.   On half-wave mode it used to make the same angry buzz on my PSU, and it got worse with load.

QuoteThis makes me very curious. The current out of a half-wave rectifier, as reflected to the primary, can easily be vastly larger than the magnetizing current. A decent AC mains transformer design will have magnetizing current well under maximum primary current... The "offset current" in the primary is then many times the magnetizing current, so the thing ought to be driven dramatically into saturation if the offset primary current was not substantially cancelled.
Yes, that's exactly how I see it too.  (As a passing observation only, notice the primary and secondary currents are in phase with the input voltage but the magnetizing flux is actually 90 out of phase with the input voltage.  The primary and secondary flux components, which cancel, are therefore out of phase with the magnetizing flux.)

Another weird thing is if the core had infinite saturation, we would still see the same magnetization current and the transformer would still work perfectly normal.  From outside all is normal. 

The invisible process going on is the continual increase in the core's B-field.  The key step is to realize that the (internal) primary voltage on the phase when the load is present is a bit less than the (internal) primary voltage on the phase when the load is disconnected.   The difference voltage is just the I*Rprimary drop due to the load.  And this voltage effectively looks like half-sine pulses of only one polarity being applied to the transformer (it connects to the ideal transformer part which is surrounded by the winding resistances.)  This source has an average DC voltage.  The average DC voltage causes the flux to ramp up each cycle.  It has to, it is just like connecting DC to an inductor.

It's important to see how this DC drives the flux up because it is not possible to understand what goes on without it.  That *is* the main idea.

QuoteOf course it is. The question is then - power from where? It's clear that the primary wires heat up more on the half cycle of higher current, and this heat has to be dissipated over the next cycle, but this isn't a statement about the M-field.
Or am I missing your point entirely?

Yes, that's right.  At first everything looks normal, so we can't see "a problem".   There's no mechanisms to cause any magic current on the unloaded cycle.

At this point, I'm not sure if you believe that the flux has a creep mechanism yet?

Suppose the flux has shifted in one direction.  The flux due the normal magnetizing current will add to the flux offset.   A some point one side of the total flux swing will hit the core saturation point.  It doesn't need much shift because a transformer normally operates with magnetizing flux swings not far from saturation.

When that occurs the magnetization current increases because magnetizing inductance drops.  In addition the magnetization current becomes asymmetrical.

So far we still haven't removed the voltage imbalance that caused the flux creep.   The flux will keep creeping until there is a voltage balance at the (internal) transformer input terminal.  The IR drop associated with the increased magnetizing current, which is now asymmetrical, creates the voltage balance.   It's not hard to understand that the large magnetizing current will eventually create an substantial IR drop.   (In reality it's more painful than that because the magnetizing current peaks are near the input voltage zero crossing so we need quite large current to create enough skew.)

Anyway that is the mechanism which creates current draw when then half-wave load isn't conducting.

QuoteI must have built fifty similar ones. That schematic is one example of the vast general case. There are probably many, many more of them on the net, in varying degrees of specificity.
I've built many as well, maybe not 50.     The main point is the point is the nature of the saturation isn't catastrophic like you get with a switch-mode power supply.   It's more like clipping which self regulates the amount of DC flux.  Under normal conditions everything remains working and, provided the load isn't too heavy, the transformer is still happy.  From outside we only see some extra heating and perhaps a funky primary current.

It's only when you start pushing things that you see some ill effects and perhaps excessive primary current and heating.     (The angry buzz is fairly objectionable.  Different parts of the core saturate slightly differently which creates field gradients and field gradients lead to high forces which cause the buzz -> I'm not sure about this now.)
Send:     . .- .-. - .... / - --- / --. --- .-. -

amz-fx

Another method:



The Alesis 3630 compressor power input looks like the above drawing, with linear regulators following on the V+ and V- rails.

Since we are only taking about 65 ma maximum for this project, it should be fairly easy to acquire a 100ma or larger transformer that is not stressed by the current requirements.

regards, Jack

DimebuGG


imJonWain

I nominate this for thread of the year, the transformer discussion was really interesting.
  • SUPPORTER
TFRelectronics