Dual Power Supply (e.g. 15V and 3V) and general voltage dropping…!

[jBam]

Hi all,

Been a while… I've been scooting around doing my usual "too many hobbys" trick; but still have been working on my weird gadget I was discussing earlier in the year… I'm now at the point where I've modeled a good working prototype in LTspice; but I'm missing a lot of the peripherals.

I have a few possibly basic questions regarding power supplies AND stuff to do with DC voltage sources for other parts of my circuit…

Power Supplies:

I've designed up a lot of variants of my schem, and have found superior performance (in sim) when using some comparators that run on a different V source (significant enough to consider them strongly for use in a prototype).

So most of my circuit runs on 15V; but the comparators want around 3V.  Even if this changes, it's got me thinking about "what is the best way to generate a dual power supply"...

Is the best solution here to use a power transformer with dual output options to get a separate 15V and a separate 3V?  The stupid part of my question is:  can I just burn off 12V from the 15V supplies; or is this just, as mentioned, stupid?!  i.e. inefficient or dodgy?!  If I went, say 500 series, could I still consider 3V supplies (as the Xfmr situation is less attractive here)...

I've searched around, but can't find any discussion specific to what I'm wondering (which might imply that it's not good practice to burn up the 12V!!)...

DC Questions:

Within my circuit I'll be using a few steady DC sources.  Similar to above, just wondering what good practice is here, as these will be typically around or less than 1V…

So if I DID use a 15V supply, would I rectify this, and then burn off 14V somewhere?  seems inefficient!  Would this burn amps; or do I just scale up resistors to be low current (I'm still getting a little confused here, as I read that high ohm is effectively low current, but can't quite understand how you can drop voltage through a divider without wasting electrons?!).

OR is there another way that I can syphon off 1V efficiently...

What's best practices here?
-----

As always, I really appreciate any help here - sorry if this is 101 basics again!  I've searched around, but can't get any solid advice on the do's and don'ts…

Jonny

 

[PRR]

> the comparators want around 3V.

Ask "why??"

Comparators are normally designed to ignore supply voltage. At least between some low wake-up voltage and some high blow-up voltage.

If SPICE is telling you different, something aint right.

(Yes, too-crude comparators will be supply critical, but nobody makes chips like that.)

As for your other questions....

How much power you talking? 12V at 40 AMPs to power a boom-car amplifier? 12V at 4 milli-Amps to feed a comparator? And this 1V reference.... is it delivering more than stray bias?

 

[jBam]

Hi mate!  Yup - me again ...

Thanks for the response...  The comparator I'm simulating is an LT1721, and I'm using it to trigger a CMOS gate to open and close (swaps between one signal and another once the signal passes a threshold).  The datasheet suggests abs max supply of 7V... I say 3V... but really, it's recommended up to around 6V... just not "15V")....

I've A-B'd nearly every LT comparator that meets appropriate specs, and this one is really catching things the best, with minimal signal over threshold before the switch.  Yes - it's crazy fast (insanely fast), but it's working tandem with slower processes (gate opening times etc.) and so far seems to produce the best results...  It's the difference between around 0.7mV (negligible) overs and 10-15mV or more (noticeable) overs with fastest options that can run on 15V supplies...

Let me know if this still doesn't sound right...

Like I said; I may end up using a different arrangement (different comparator) though - possibly different manufacturer; I note that I'm kind of sticking with LT components at the moment due to ease within LTspice.  But the question got me thinking about dual supply in general too...

As for the second question - it'll be driving Opamps, comparators and some THAT VCAs (218x)... so quite the latter!!  Low amps is assumed, but still don't want to sit around burning through electrons if it's generally poor practice!

The "1V DC" is a reference V to trigger the comparator swap.  It'll act as a kind of threshold (well actually - there's a few of them creating windows for comparator swaps when a signal passes this "threshold"), and they're going to be adjustable (e.g. between 0V and maybe up to 1.5V).  So it's not exactly a bias voltage (through my understanding of bias voltage use).

Thanks again...

 

[Monte McGuire]

The LT1721 draws about 4mA and it can run on a 5V supply. It seems simplest, if you're only running one or two comparators, to just drive a 5V regulator off of the 15V regulated supply. That allows you to simplify the basic power transformer arrangement, and also end up driving the 5V regulator with a clean 15V supply. It shouldn't really improve things much to have pre-regulated power, but IMHO, the extra complexity of providing a specific transformer winding, rectifiers and filter caps just for this supply is a lot of extra 'stuff' for little reason.

In this way, the 5V regulator will drop 10V, and dissipate 40mW per comparator driven.  This is a pretty small dissipation, so it should be easy to handle, even with a small regulator package and a bit of care laying out the PCB.

Best of luck!

 

[jBam]

In this way, the 5V regulator will drop 10V, and dissipate 40mW per comparator driven.  This is a pretty small dissipation, so it should be easy to handle, even with a small regulator package and a bit of care laying out the PCB.

Best of luck!

Perfect... Thanks Monte.... That's sort of where my head was at, but I'm also pretty damn junior when it comes to electronics experience!... so thought it best to ask...

I'll look into 5V regulators and get them into my schem setup... And I'll post any further questions I may have here

 

[Andy Peters]

Is the best solution here to use a power transformer with dual output options to get a separate 15V and a separate 3V? 

Those are impossible to buy in DIY (onesie/twosie) quantities.

The stupid part of my question is:  can I just burn off 12V from the 15V supplies; or is this just, as mentioned, stupid?!  i.e. inefficient or dodgy?!  If I went, say 500 series, could I still consider 3V supplies (as the Xfmr situation is less attractive here)...

I've searched around, but can't find any discussion specific to what I'm wondering (which might imply that it's not good practice to burn up the 12V!!)...

Dropping 12 V in a regulator isn't the worst thing you could do.

Later in the thread, you say,

The comparator I'm simulating is an LT1721, and I'm using it to trigger a CMOS gate to open and close (swaps between one signal and another once the signal passes a threshold).  The datasheet suggests abs max supply of 7V... I say 3V... but really, it's recommended up to around 6V... just not "15V")....

Which leads me to ask, why the LT1721 and not an LM339 or similar? You say a CMOS gate is the load, but what kinds of signals are you switching that require "Crazy-fast" comparator response times? With a 339, you power it with your usual ±15 V supplies. The trick is that it has an open-collector output, so the output switches between 0 V and whatever voltage you pull up to. So what is your actual logic supply voltage? Is it 3 V, or is it something else?

Obviously the switching threshold is set by a voltage divider or whatever at either the comparator's inverting or non-inverting input.

DC Questions:

Within my circuit I'll be using a few steady DC sources.  Similar to above, just wondering what good practice is here, as these will be typically around or less than 1V…

So if I DID use a 15V supply, would I rectify this, and then burn off 14V somewhere?  seems inefficient!  Would this burn amps; or do I just scale up resistors to be low current (I'm still getting a little confused here, as I read that high ohm is effectively low current, but can't quite understand how you can drop voltage through a divider without wasting electrons?!).

OR is there another way that I can syphon off 1V efficiently...

These "steady DC sources," what is the expected load current on each, and what is the required accuracy? It may be good enough to use a voltage divider between +15 V to ground, with the divider output buffered by an op-amp. A voltage reference is required if the divider tolerance, or that of the +15 V supply, isn't good enough. Some references can source 10 or 20 mA.

Lots of questions here!

 

[jBam]

Hi Andy - thanks for the detailed response and questions…

Xfrmr response - great - thanks… This and your next comment on regulator usage really helps steer me in the right direction if I stick with the lower supply comparator…

LT1721 vs LM339:

I might still continue mucking around with comparators AND other components, because - as you allude to - something in the "crazy fast" category seems overkill for what I'm doing: triggering at crossing points (e.g. zero crossing) and opening and closing a switch (currently cmos) based on an audio signal...  Think "audio rectifier", which I AM using it for, but this same scheme is then also used in the circuit for slightly more complex things too… Being audio rate, I'm a bit suspicious as to why I'm getting such superior results with something flipping over in nS, when surely uS would be fine (within audio range). BUT - it is happening, and I've worked toward refining the performance using different comparators and then BAM ---> the 1721 is nailing it.

Basically, I'm working with a whole range of compounded time related processes here - I've got minor group delay in opamps, propagation delay and slew rates in comparators, on / off delays etc etc… All of these things combined are resulting in a delay from when the signal crosses a ref voltage (e.g. 0V) and when a switch actually opens or closes… further, it's not so much a time issue in terms of audio speeds, but more a voltage (volume) issue (i.e. it's the loud ANDfast crossing that are hardest to trap / switch; and it's the higher V overs that I'm trying to avoid… With these, the 1721 simply appears to be superior (considering it's seemingly impossible to actually completely catch everything, "as close as possible" is what I'm aiming for)…

I'll run through the design again, and look to make sure that the issue isn't a byproduct of something else...

On one hand, this is a functional intent that I'm trying to design toward… On the other, I'm still just learning through this whole thing, and therefore slowly learning about the can's and can'ts - I may later realize that the functional benefits are negligible between a 1721 and something slower!  I'll get a model of the LM339 into LT too, and see how it performs… would certainly be a great outcome if a 15V supply comparator had similar results!

CMOS gate / switches:

I mentioned "gate" a but in my comments… essentially I'm meaning CMOS swiches… This is all part of a side chain (not final audio), BUT should I be considering a different analogue switch setup?  I'm after: Very High Speed; low through distortion… ideally quite a lot of headroom OR overload protection for the through signal (will likely be around 1.5V typical max, but will be fed by a 15V setup at the end of the day… so I guess it could end up with something smacked out at the rail V unless I build in additional protection?).

At the moment I've loosely modeled a standard switch based on a CMOS that I think goes on at around 3V… might need to double check that!  I've also realized earlier today that I AM running the comp at 5V … oops… 3V was an older version …  but same same for discussions sake!

Steady DC Sources:

My understanding is that these will have a negligible load - I'll Buffer it; split it and send one version to a comparator input, and the other heads off into another opamp input via a switch (very crude explanation that doesn't really make sense of the intent(!), but that's the end points of the DC source).  Would I be right in assuming there is not likely a loading issue here, particularly once buffered?

I'd also like to aim toward "very stable DC" for now… any suggestions to get high performance here?

---------

Sorry if any of the above doesn't make sense… please feel free to ask for further clarification.  I'll be happy to post up some LTspice outputs or the schem details if you like…

 

[JohnRoberts]

As has already been shared if the low voltage current draw is modest, it is no big deal to scrub off a few volts and allow the pass element to dissipate the wasted power. For high current and low voltage a switcher may be the best approach.

You mention in passing zero crossing synchronization for gating audio.. This sounds a little like one of my early technician jobs in the 70's when we were splicing together pitch shifted audio samples. (Performing the splices at zero crossings reduces perturbations like clicks, thumps, burps).

More recently zero crossing synchronization has been used with digital pots to reduce zipper noise from gain changes. I don't have a link handy but I recall some DPOT app notes publishing pro forma zero cross detector circuits. Some adjustable gain mic preamps have it built in so you can't see precisely how they did it.

In general just detecting any zero crossing is difficult as DC offests become important, and there will be circuit lags meaning you are always late, unless you start early. Often you end up using a window detector, and perhaps more sophisticated first derivative (slope direction) detection.

Good luck

JR

PS: I am tempted to say just use a microprocessor with A/Ds and use that to detect, but when I keep saying that it irritates my analog dog friends. I was once an analog only dog, so try to understand, but it can be hard.

 

[jBam]

Hey John... Thanks for the comment...

I was wondering about introducing small DC offsets in the signal to fake a zero crossing --> slope detection would be the key though hey! I'll read into it...  In terms of a theoretical solution, I'd love to see how far I can push it - like I said, I might realise that the functional benefits aren't justified, but learning about it all has value none the less ...

This part of my circuit is critical; and it seems to be working "acceptably" for the purpose BUT: there's a whole range of cool options beyond what I'm currently doing... Love ur audio wave swapper - a great example ... Also recall reading a comment from Jakob a while back talking about zero crossing triggered vca changes for a compressor to avoid distorting waveforms which seems interesting... Ironically it seems a lot happens at "zero"