SMPS for small signal analog circuits

[teleman]

I've not seen any small signal analog circuits like preamps, guitar effects etc driven by a SMPS units! Why is that? Are there any special reasons for this?
Is this safe & viable alternative to linear supplies, especially running these small analog audio units?

Thanks

 

[Michael Bean]

For low wattage applications, most of the advantages such as size and efficiency that switch mode supplies have over linear supplies pretty much disappear, especially if you need a dual polarity source for opamps. For hobbyists the added complexity and expense generally aren't worth it.

Mike

 

[teleman]

Thank you! I was thinking in terms of using a 24v/1.2A SMPS by splitting the output using BUF364 or similar voltage splitter to obtain +12v-0 -12v virtual earth supply! The maximum current draw of the circuit would be around 100mV/ rail! A dumb idea, is it not?

 

[Michael Bean]

A dumb idea, is it not?

Who am I to judge?

If you already have the switch mode supply in hand and everything else you need, it might work OK. But in my opinion, it's usually better to use a "real" split supply where possible rather than one derived from a single supply.

Mike

 

[Calvin]

Hi,

have You thought about a single lowvoltage DC supply that feeds into PointOfLoad (POL) DC-DC converter modules?
Its generally good to make the supply lines short.
Isolated converter modules allow fast and short, galvanically isolated supplies in close proximity to the Point of Load, hence where the power is needed.

There are numerous modules in wattage classes from 0.5W to >>100W from many manufacturers and in different makes.
For manufacturers see CUI-inc, TDK Lambda, Wuerth, TI, Maxim, XP Power etc. etc.
Modules with single or Dual oututs, unregulated or regulated, isolated or non isolated, with reduced transformer coupling capacitance, shielded casings et al.
For small signal circuits 1-5W is often fully sufficient.
See for example the TI DCP02 series, or CUI-Inc's PBQ3 and other P series.
Such modules need only a few external parts, mostly for pre and post filtering.
Its absolutely easy for Hobbyists to work with and any of the well known distributors offer those at reasonable cost.
As main DC supply You may use any lowvoltage type of sufficiently high wattage.
A external wallwart supply also spares You all the hassle of having to deal with high voltage power line voltages and associated safety issues.

jauu
Calvin

 

[teleman]

Who am I to judge?

If you already have the switch mode supply in hand and everything else you need, it might work OK. But in my opinion, it's usually better to use a "real" split supply where possible rather than one derived from a single supply.

Mike

Thank you Calvin,
Ha ha..I wish there were more of you in this world...! Seriously, you've said it more eloquently in a few words than what most of these experts in most"commercial" websites have been trying convince! I suspect they have vested interests in promoting their own products, ie:selling SMPS?

Have You thought about a single lowvoltage DC supply that feeds into PointOfLoad (POL) DC-DC converter modules Its generally good to make the supply lines short.

Yes, I've actually tried using a single supply with a DC-DC converter/ splitter, but for some strange reason when the load was connected the voltage dropped to very low level! I in fact have built a few good linear dual supplies which work very well & are almost completely noise free, but this "eureka" moment came to me w.r.t using SMPS instead as it seemed to offer a cheaper alternative! I'do have all the parts neccessary to try this.....however it's not a life & death issue right now.

My linear psu is based on the drawing below using LM317T/337T regulators. So i thought I'd omit the main stage & do it this way to minimize the high output ripple smps have! It's ready, but I'm a bit worried connecting a load to it!! black charred face? Lol!

I think I got carried away by this article!

AC DC Power Supplies, DC DC Converters, Wall Adaptors, External Power Supply - Future Electronics

 

[marce]

SMPS for lower power circuitry like this is more prevalent in commercial designs, where either battery operation, low power requirement, space limitations and cost are factors.
I second what Calvin has said, and most controllers from the likes of Texas have layouts that you copy exactly, so very DIYable.
If laid out and designed correctly SMPS's do =not have to have high output ripple.....

 

[teleman]

Thanks Marce,
I may have not explained myself properly....I'm not attempting to DIY a SMPS as it is way above my capabilities & knowledge! I already have a SMPS 24v/1.2A DC supply unit which I bought on e-Bay, hence thought of splitting it to give a virtual earth supply.

W.r.t SMPS not having high ripple, again I may have read or understood it wrongly. In any case I will refer to the Texas implementation notes. Thanks for the lead on this.

Cheers

 

[marce]

One problem I have seen with SMPS's and analogue circuitry is power matching, using an over powered SMPS can cause the supply to go into burst more when there is not enough power drawn. This quite often causes the noise spectrum to drop dramatically in frequency often into the audible range. So it is best to match the power output to the circuit requirements.

 

[teleman]

Really?...thank you indeed for your timely warning! I'm so glad that you mentioned it! Ok, so this is what I want to do.

SMPS 1.2 A.......to split supply......powering a circuit drawing < than 100mA/ rail

Do you then reckon that this may cause the problem you've mentioned?
What would you recommend then; 250mA/400mA/500mA?

Thanks

 

[marce]

Git caught out with this on a project I worked on, the engineer couldn't understand where the 8kHz noise was coming from (quite intrusive in a headset) at first, then realised that the power requirements were so low the SMPS was always going into burst mode.

 

[teleman]

Seems like I'd probably solve one problem only to discover many others! I think this requires some more investigation/ research!

Cheers

 

[marce]

As I said earlier there are numerous example of point of load use of SMPSs but these are usually designed from scratch for a particular task.
keep us informed of how you get on

 

[jackinnj]

i am thinking of having some boards burned for the Linear Tech LT3439 "Quiet Switcher" -- i posted a negative version #6 on this thread:
http://www.diyaudio.com/forums/analogue-source/256062-equal-opportunity-mm-pre.html

SY's latest invention requires negative bias for a JFET in the grounded grid input circuit.

obviously it can be modified to provide plus and minus regulated outputs.

The device is only available in SMT. Unlike other switch controllers, it rounds off the edges of the switching transient, so it is much quieter than a TL494 or SG3525 etc.

 

[Calvin]

Hi,

but for some strange reason when the load was connected the voltage dropped to very low level!

Some SMPS and DC-DC converters feature a overcurrent protection with restart, called Hickup.
If for example too much capacitance is connected at their output, the high inrush charging current of the caps triggers the hickup mode.
In effect the supply doesn't really start up at all.
Typically a maximum capacitance load is specified in the Datasheet of such supplies.

jauu
Calvin

 

[teleman]

I get it! I will try again & report back. Thanks for this very useful tip!

 

[sesebe]

For low noise you need to chose a resonant SMPS. It is ideal fitted for you need.
The usual problems with noise from SMPS it is almost not visible because the current and voltage on output are sinusoidal and because you do not have large variable power request, the usually limitation of stabilization range are not a problem.

 

[teleman]

I connected the SMPS to the BUF634 circuit & it gives +12v/-12.04 volts without any load connected. So this means there's no "hickup" mode so far even though it's rated at 1.2A! Is this promising? I will now try with the LM 317/377 regulator circuit shown & see what happens next!

sesbe, is "resonant" SMPS available off the shelf? I'll google & see if I can get one.I wonder if going for a lesser rated, say between 8-10 VA(watts) SMPS would be even better than the one (1.2A) I have a.t.m? I need only around 150-200mA/ rail split supply! My prime concern is NOISE!

This is very promising & I'm so grateful with all you guys giving me so much invaluable advice! Viva DIY!

 

[marce]

A lot of SMPSs can be solved by correct layout. They also do spread spectrum controllers that help minimise noise levels.
There are also quasi-resonant designs, again often with spread spectrum clock to distribute the noise instead of one sharp peak.

 

[AndrewT]

CRC are very good at attenuating unwanted HF from a power supply.
CLrC are even better. The r is the parasitic resistance of the air-cored inductor.

BUT
can a low loading on an SMPS, that already has an LC filter inside, benefit from some added external CRC or CLrC ?

Or is there a minimum loading that has to draw a defined current to get the attenuation to work well?