Smart Circuits to learn from

[JohnRoberts]

As I think I mentioned earlier start-up is a bit of a trapeze act.  but it works at least for the original application.

To make start up a little more reliable, and the pre-regualtor more efficient I would try substituting a mosfet for T2 so the pull-up resistor could be increased. to reduce the current. "and" drop a 5V zener to ground in place of the emitter connection of T1 to the VR output. This should make it independent of load on the regulator. To keep the pre-regulation knee sharp you might want to add a pull up resistor to the 5V zener to keep it turned on.

Dumping the base drive current into the 5V rail was just to recapture that current into the 5V load. Making that current less with a mosfet makes capturing it less important too.

Or you could just use a switcher.. ;D ;D

JR

 

[Rochey]

Or you could just use a switcher.. ;D ;D

I'll need 5V AND 3.3 to make this work. A switcher is looking more and more appealing!

 

[moamps]

Some interesting things came in my sim of that circuit (if anyone is following out of interest).

There are a tight relations between values of R4, C3 and load to get stability. It remains me to early SMPS designs  .
This circuit is a dead end. Use a small SMPS as JR suggested. 

 

[JohnRoberts]

Some interesting things came in my sim of that circuit (if anyone is following out of interest).

There are a tight relations between values of R4, C3 and load to get stability. It remains me to early SMPS designs  .
This circuit is a dead end. Use a small SMPS as JR suggested.

I feel a little defensive about this circuit, there are thousands out in world, that worked fine and were stable as designed. But yes it is not trivial to tweak for other voltages, and I am not recommending wide use of this.

For a mic preamp I'd be tempted to use a switcher to generate multiple voltages. While I recall working on one trick DC to DC that was supposed to regulate two different output voltages and it's HF transformer was mechanically noisy, so I disabled the lower voltage and added a pass regulator. 

FWIW another obscure PS trick I did years ago to generate a low current 48V supply from +/- 15V rails where there was already a DC to DC switcher in use to generate a high current 5v rail, was to hang a cap doubler/tripler off the high frequency switching waveform.  With the high switching frequency the doubler/tripler caps could be tiny (0.1uF) SMD puppies.  Crude un-regulated feeding a 48v pass regulator.

For only one or two mics if you already have a switcher going, it may be worth looking at. Same as low frequency doubler/tripler, but you can use smaller boost caps and steering diodes. You still need decent sizer reservoir cap.

JR

 

[Rochey]

There are a tight relations between values of R4, C3 and load to get stability. It remains me to early SMPS designs  .
This circuit is a dead end. Use a small SMPS as JR suggested.

Your forgetting how low cost this design could be.

a pair on npn's at $0.01 each. and a 7805 ($0.10).

A full on SMPS also has a terrible reputation with Audio DIY'ers. The rest of the world got over it and did good layout and filtering, but audio guys are stubborns with beliefs. This circuit runs a little slower, little easer. Less dumping into GND etc.

 

[JohnRoberts]

There are a tight relations between values of R4, C3 and load to get stability. It remains me to early SMPS designs  .
This circuit is a dead end. Use a small SMPS as JR suggested.

Your forgetting how low cost this design could be.

a pair on npn's at $0.01 each. and a 7805 ($0.10).

A full on SMPS also has a terrible reputation with Audio DIY'ers. The rest of the world got over it and did good layout and filtering, but audio guys are stubborns with beliefs. This circuit runs a little slower, little easer. Less dumping into GND etc.

Those component prices sound Peavey cheap, not DIY cheap...These days the cost to pop a part is more than the part especially for small components, and PCB real estate matters too..

Of course I like the general approach but the exact design might be different for 3.3V or 5V LDO rail. If you also need a low current 48V rail I lean toward the switcher for 5V  with cheap doubler/tripler for 48V.

JR

 

[moamps]

Your forgetting how low cost this design could be.
a pair on npn's at $0.01 each. and a 7805 ($0.10).
A full on SMPS also has a terrible reputation with Audio DIY'ers.

The right question for me is:
Does the SMPS implementation in an audio device make it better or only cheaper and lighter?
Look into this UA preamp-compressor. 

 

[My3gger]

The right question for me is:
Does the SMPS implementation in an audio device make it better or only cheaper and lighter?
Look into this UA preamp-compressor. 

The same question for me; haven't seen much use in tube pro audio, although i know about few examples from the last few years when people get them working well. That is for tranistor's low voltages, seen one kit company offering SMPS supplies for tube preamps. People had all sorts of problems before, that seems to be fine now with newer ICs.
If we are talking about diy or small series, aren't linear supplies with TL783, LM317 regs, bypassed zeners, capacitance multipliers, or even simple RC filters more durable, predictable? It takes transformer with a few taps which can be cheap, the rest is very simple and needs only few parts.
Power One still uses open frame linear supplies for low and high voltages, some people here use them a lot. I wonder why, is it because of UL/CE, or there aren't reliable prebuilt SMPS modules like this?

I don't like how this UA compressor is built, it is getting harder and harder to service things like these. Still think there will always be people who like and respect bullet proof mechanical designs.

 

[Andy Peters]

There are a tight relations between values of R4, C3 and load to get stability. It remains me to early SMPS designs  .
This circuit is a dead end. Use a small SMPS as JR suggested.

Your forgetting how low cost this design could be.

a pair on npn's at $0.01 each. and a 7805 ($0.10).

A full on SMPS also has a terrible reputation with Audio DIY'ers. The rest of the world got over it and did good layout and filtering, but audio guys are stubborns with beliefs. This circuit runs a little slower, little easer. Less dumping into GND etc.

Those component prices sound Peavey cheap, not DIY cheap...These days the cost to pop a part is more than the part especially for small components, and PCB real estate matters too..

Of course I like the general approach but the exact design might be different for 3.3V or 5V LDO rail. If you also need a low current 48V rail I lean toward the switcher for 5V  with cheap doubler/tripler for 48V.

He left out the mains transformer, which is not exactly free!

I built up a little power-supply module that takes in 9 VDC and gives +/- 15 V, 5 V and 3.3 V. 

The +/-15 V comes from an LT1371 switcher, which is set to do +/- 18V out using a VERSA-PAC transformer. The +/-18 V are followed by LM317 and LM337 regulators for the +/-15 V.  Max out is about 400 mA per rail. I need to do some real-world noise testing to see whether the post-regulators are really necessary; I can change the switcher output voltage with just one resistor.

+5 V comes from a TI TPS54329 buck regulator hanging on the +9 V input. The +3.3 V comes from a Micrel MIC37100 LDO hanging off of the +5 V. If I use this design with something that doesn't need the +5 V output, I can choose to not stuff the LDO and change the feedback resistor so the 54329 does a +3.3 V output instead of +5 V.

Every time I look at using a mains transformer/rectifier/regulator set-up, I start to look for a transformer with a separate lower-voltage winding for the digital stuff, and it seems like those things aren't available through the usual distributors.

 

[PRR]

> to simulate.

Not sure what you get, or what you expect.

If this follows-on from before, you need a non-ideal transformer (add a few Ohms) and a real load.

I do not see what D3 D4 C1 do for you here (they served another purpose in the original).

The essential action is that T2 turns-on at the *falling* slope of the sine just as it crosses the desired (not peak) voltage. When it does, the load at C3 (load you omitted) causes the (omitted) resistance of TRCT1 to sag. While I have not thought it through, JR's text suggests positive feedback from the low transformer resistance through a high-gain amplifier causes a snap-off of conduction. If there is no load, and no source resistance, there's no snap. (Also no dissipation, so JR's design does not "fail" if the load falls off.)

IMHO, the source could be reduced to a simple sine generator plus a resistance. The CT winding is primarily for +/-DC (not simmed here) and increased ripple frequency (cheaper caps). It should work fine half-wave, one AC source.

Not that "extra parts" matters for this simple circuit when you probably have a Sexium 6666MHz CPU. However I been simming since 8087 days, can still chug a 3GHz Pentium, and I pre-look for the simplest sim which does capture "all" the features of the circuit. I'll shamelessly throw-in a 999V source and large resistor, because real-world "current sources" are current limiters which is a complex sim (and SPICE's current source is a real SOURCE which will gladly blow-up to >1EE32V and blow-up the math).

 

[JohnRoberts]

> to simulate.

Not sure what you get, or what you expect.

If this follows-on from before, you need a non-ideal transformer (add a few Ohms) and a real load.

I do not see what D3 D4 C1 do for you here (they served another purpose in the original).

D3, D4, and C1 create the +unregulated rail that supplies the base drive for T1

The essential action is that T2 turns-on at the *falling* slope of the sine just as it crosses the desired (not peak) voltage. When it does, the load at C3 (load you omitted) causes the (omitted) resistance of TRCT1 to sag. While I have not thought it through, JR's text suggests positive feedback from the low transformer resistance through a high-gain amplifier causes a snap-off of conduction. If there is no load, and no source resistance, there's no snap. (Also no dissipation, so JR's design does not "fail" if the load falls off.)

IMHO, the source could be reduced to a simple sine generator plus a resistance. The CT winding is primarily for +/-DC (not simmed here) and increased ripple frequency (cheaper caps). It should work fine half-wave, one AC source.

Not that "extra parts" matters for this simple circuit when you probably have a Sexium 6666MHz CPU. However I been simming since 8087 days, can still chug a 3GHz Pentium, and I pre-look for the simplest sim which does capture "all" the features of the circuit. I'll shamelessly throw-in a 999V source and large resistor, because real-world "current sources" are current limiters which is a complex sim (and SPICE's current source is a real SOURCE which will gladly blow-up to >1EE32V and blow-up the math).

 

[Rochey]

He left out the mains transformer, which is not exactly free!

It's free in that it's a cost that has already been accounted for in a current build. I'm looking at the methods that I can add digital control to already existing systems. That means that I suddenly need a supply to drive microcontrollers, LED's and Relays. So far, on our Eden Mic Pre, we've used a small USB power brick, taken in parallel to the usual toroidal etc.
We're fortunate in the Eden build that the signals are completely isolated from each other. Even grounds are separated - but I suspect I won't be so fortunate in the future.

There are a few of us here that could easily justify, understand and accept a 12VDC wall wart and a smart SMPS to create the old school voltages that we like in Pro Audio, but for the most part, most folks go numb at the thought of an SMPS polluting their precious audio signal.

I do have a small 5V to ±15V power supply that I had PCB's made for to drive the mic pre's from a USB Battery block - but that's a very different end of the market. That used a Nat-Semi / TI switcher and some other Unitrode tricks.

/R

 

[JohnRoberts]

So do it...

I might even help.... 8)

JR