"Fake" Split PSU

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
janneman said:
This is generally possible for low current loads, like a few mA for a preamp or filter. From your numbers I assume you want to use it for a power amp? If so, it won't work I'm afraid.

Yes, it will work just fine for an audio power amplifier (I assume from the name of this board that that's what he's building :D). I've posted this before, and I'm only too happy to post it again...

It works because audio is mostly symetrical, hence the average earth currents don't amount to much. I can't see any real problem with the circuit, and it's certainly better for the planet than throwing away a perfectly good transformer just because it's not centre-tapped. I don't know why more people don't use it :)

Mark
 

Attachments

  • quadpsu.jpg
    quadpsu.jpg
    18.8 KB · Views: 793
mhennessy said:


Yes, it will work just fine for an audio power amplifier (I assume from the name of this board that that's what he's building :D). I've posted this before, and I'm only too happy to post it again...

It works because audio is mostly symetrical, hence the average earth currents don't amount to much. I can't see any real problem with the circuit, and it's certainly better for the planet than throwing away a perfectly good transformer just because it's not centre-tapped. I don't know why more people don't use it :)

Mark
OK for the principle, but in practice I'd use something much heftier: real amplifiers are not specially designed with quiescent current balance in mind, and if the artificial ground begins to drift, there might be some regenerative mechanism driving the whole amplifier into latch-up, with everything, including the output stuck to one of the supply rails. We certainly don't want this to happen. I'd divide all resistor values by ten and use driver-size transistors at least.
LV
 
Elvee said:

OK for the principle, but in practice I'd use something much heftier: real amplifiers are not specially designed with quiescent current balance in mind, and if the artificial ground begins to drift, there might be some regenerative mechanism driving the whole amplifier into latch-up, with everything, including the output stuck to one of the supply rails. We certainly don't want this to happen. I'd divide all resistor values by ten and use driver-size transistors at least.
LV

Of course, like any idea which we "borrow", you have to engineer it. In case you didn't follow the link I gave, this does indeed work in practice - this is an extract from a Quad 306 schematic.

Can you find me an example amplifier that has quiescent earth currents of more than a few milliamps? I can't think of many, if any at all, apart from perhaps some esoteric designs. The DC earth currents normally consists of biasing for current sources and similar. And these can alway be re-routed to the other rail - good practice anyway - earth should be about signal only.

Needless to say, the quiescent current in the OPS doesn't flow in the ground unless there's a DC offset, and even then the currents are small (50mV into 8R is 6mA)

Fair point about latchup, that again needs to be tested. However, having ground stuck at a rail probably won't damage anything: in fact, this circuit acually offers DC failure protection, precisely because it's not able to supply much current (as I mentioned in the original post link to above). But each capacitor must be rated at the total supply voltage.

If you want something more complex, have a look at this. It takes 36V from another regulator, and makes +/-18V for all the analogue stages in a Calrec broadcast sounddesk. Exactly the same principles as above, but using feedback to more precisely regulate the centre. Of course, the op-amp sets the upper limit of input voltage to 36V.

Mark
 

Attachments

  • calrec.jpg
    calrec.jpg
    39.4 KB · Views: 673
Here is an example of a not too untypical amplifier drawing more current on the positive rail than on the negative, and also susceptible to latch-up if the ground is too loose:
http://sound.westhost.com/project27.htm
The current supplement is caused by D1, and the latch-up happens in the following way: when the ground becomes positive enough to cause a decrease in the tail current of Q1/Q2, Q3 receives less bias, sending the output (and consequently the ground) further towards the positive; at this stage, the process becomes regenerative and drives the amplifier into latch-up, with output and ground at the positive.
It is of course also possible to find numerous examples of amplifiers behaving perfectly with a very light ground, but if you want to be on the safe side with any amplifier, it is preferable to overdesign somewhat the circuit.
LV
 
Hi again LV,

I certainly see 6ma in D1, but that isn't enough to upset even the simple Quad circuit, apart from maybe sweating one of the 2k2 resistors which should of course be sized according to the application (both value and power rating).

I'm less sure about the latchup because of the initial statement "when the ground becomes positive". The big question here is what is going to cause it to go positive? Presumably a fault condition - certainly not normal operation with AC signals. So in that case, it's protecting the speaker against DC - perfect :)

BTW, you missed a rather more serious issue - the BC559's are only rated at 30V IIRC. That's a good example of why this isn't newbie territory ;) (I wonder if this might be why you're taking a more conservative view?)

Just to be clear - I'm generally a big fan of over-engineering things, and I admit that when I initially saw that Quad circuit, I had all the same thoughts that you did. But the more I think about it, the neater and elegant it seems. Hence my evangelism!

What I'm trying to say to everyone is:
  1. It works, and it works better than you'd think.
  2. Please consider trying it, BUT, you have to engineer it! It won't work with every amp, but if you're designing an amp yourself, then you can optimise the design (like Quad would have done) at no real cost.[/list=1]
    If you get it right, the benefits are:
    1. DC offset protection for free
    2. The opportunity to use a transformer that isn't centre-tapped (and that's a bit "plus" to someone that happens to have a large transformer itching for an application. The environmental benefits of this speak for themselves.)[/list=1]
      Downsides:
      1. You have to THINK about it. This isn't a bolt-on for newbies to use with any old amplifier kit. Both the design of the amplifier and the centring circuit need thought.
      2. You have to rate the two smoothing capacitors at > the full DC voltage.
      3. The centring components might run warm with a non-ideal amp. But attention to point 1 takes care of this.[/list=1]
        The reason I'm so passionate about this simple circuit is that so many people dismiss it out of hand, and until now every argument I've heard has been pretty spurious. It's nice to discuss it properly at last :)

        Regards,

        Mark
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.