NE555 or MAX1044/LT1054 for 18v supply inside the enclosure of a buffer?

[jul059]

Hello everyone,

I am designing a simple high quality opamp buffer. I want it to be as versatile as possible, and therefore I want it to operate at a high voltage to accept high input signals. Ideally, I want it to be powered by a simple 9V supple, therefore I need a converter inside the enclosure. I have designed a switching supply with a simple NE555 which gives me ~24V, which is easily regulated back to 18V with a LM317, or perhaps a shunt regulator, I'm not sure yet. Here's the circuit :



If my memory serves me right, switching frequency is between 35 and 40 kHz. It can be adjusted by changing the value of the 470 pF capacitor. Of course there would be a RC filter on the power input.

My problem with this circuit is the very high (1A peak and more) transient currents through all the switching capacitors (C1, C2, C3), the snubbing capacitors (C5, C13) and the OUT pin on the NE555. I am afraid that even if the circuit ground is properly laid out and "isolated" from the rest, there will be strong magnetic fields inducing noise in my buffering circuit.

My question is this : would packaged charge pumps (MAX1044, LT1054, etc.) introduce less noise through the emitted magnetic field? Is the packaging also used to block those fields? Also, does anyone have any ideas or recommendations about the above circuit or how it could be be improved?

I was thinking about designing a multi-layer PCB with the power supply parts and their traces on one side, then a ground plane, then the traces of the buffer circuit and finally the parts of the buffer on the other side. The board would be made as large as possible to fill the entire enclosure. Would that block out most interference?

Thanks for any input!

[amptramp]

I can see serious amounts of noise for either approach.  The NE555 has massive current spikes on the order of 1 amp just switching on its own without the connection to a capacitor.  I would eliminate 9 volt operation and use a 24 volt wall wart (I have three of them that I use in series for powering old battery radios, so they do exist).  By their nature, charge pumps take large current spikes.  If you are wedded to 9 volts, a boost converter may make more sense since the efficiency is better and the current used to supply the converter sees an inductor as a load and the current does not change as suddenly when it is switched on although the turn off is a step function from high to nominally zero.

By its nature, a buffer usually works with small, high-impedance signals and charge pumps are not really compatible with low signal levels.

[jul059]

I agree that the best option is to simply use a higher voltage source. However, I have seen that the 1044 is used in the Klon Centaur for example, and although I have never listened to this pedal, I heard good things.

I'll have a look into boost converters then.

[JerS]

I have used the LT1054 for this very purpose - and it worked very quietly. If memory serves, I used it to create a +- 9V swing around the opamp giving you approx 18V peak to peak.

Good luck!

[TejfolvonDanone]

I've built a valve tremolo which worked with 36V/45V on the valve. I used a simple DC-DC boost converter with a 555 with an external switching transistor running in open loop. The switching frequency was around 100khz. It worked well without any noise. The switch transistor needed a small heatsink because i just threw the first power transistor I've found.

[Transmogrifox]

For low power a charge pump is generally more efficient. 

Something is very wrong if you need a heat sink on a power transistor when your switcher is delivering < 1Watt.

If using inductor buck converter then higher switching frequency (>100 kHz) helps because you can use a smaller inductor and smaller caps to filter input and output.  There is no reason you would need a linear regulator off a well-designed buck converter.  Any high frequency spikes can be rejected with a modest RC filter.  The same goes for charge pump circuits.

If you want headroom and you're already boosting to 24V, then why not use 24V as your VCC?  I know the thinking is to reject switching noise but you can probably get the same effect in the same amount of board space with larger low ESR caps and a resistor (normally would be an inductor filter in higher power applications).  Using a resistor instead of inductor makes the control loop stability problem easier to address.

Here is a simple self-oscillating buck converter circuit I have played with (using a variant of it in a 320V B+ for tube preamp).  Below is a variant that will get you close to where you want to be.


Just about any general-purpose NPN and PNP transistors will work.  A little bit of tweaking needed to get it where you like, definitely suggest simulating it.  Here's an online calculator I made for it:
http://cackleberrypines.net/transmogrifox/BoostConverter/boost.cgi?Fsw=190.00&Vin=9&Vout=18&Iout=8&SWloss=0.75

[Unlikekurt]

Couldn't you do something similar to the Klon.  Granted in that pedal the higher headroom is applied at a later stage, but you could go +18v / - 9V from the charge pump.  Then Vbias is 4.5v which is sensible for the rest of the circuit when operating at +9v. 
You'd get 27v swing there.

[jul059]

Couldn't you do something similar to the Klon.  Granted in that pedal the higher headroom is applied at a later stage, but you could go +18v / - 9V from the charge pump.  Then Vbias is 4.5v which is sensible for the rest of the circuit when operating at +9v. 
You'd get 27v swing there.

Indeed, I think that's what I'm going to do. Much lower parts count too. Using 1n5817 schottkys, biasing with resistors 10k and 9.1k should put me almost right in the middle of the voltage range. Thanks for the idea.

Last question, do you guys think it would be a good idea to put a RLC filter between the 9V supply and the charge pump to try to isolate it? Something like this 82 uH/1.62 ohms choke http://pdf.datasheet.online/datasheets-1/3l_electronic/EC36-820K-U.pdf followed by a 220 uF cap to ensure overdamping. If it fits on the board, of course. It should give at least -50dB at the switching frequency. This way, the biasing voltage should be clean.