Need a DC +12v to -15v converter good for at least 50mA. Any ideas?

[anotherjim]

Ok, I could get one of those little isolating boost converter modules for not much money, but I'd rather use commonly available parts.

I'm thinking of a 555 driving a diode multiplier, but I'm struggling to find the right multiplier network, bearing in mind the -15V output must have common 0v ground with the +12v supply?

Any suggestions/pointers are welcome. The basic negative converter is well covered on the web, but a multiplying negative voltage with a 0v common is, it seems, more elusive.

Stop press. I just found this...

With an input cap to make the oscillator AC, reversing those caps and diodes and 3x +6 is -18? We'll see. "To the breadboard, Tonto"


Cheers
Jim

[Marcos - Munky]

There's this one: http://tagboardeffects.blogspot.com/2014/01/bipolar-voltage-converter.html

[anotherjim]

Well, Tonto said no Kemosabe. I can barely get -10v out of that!

Thanks, Marcos - It would have to wait from to get some of those chips.

[Marcos - Munky]

Probably a ICL7660 or a MAX1044 will do the job, just connect pins 1 and 8.

[antonis]

@Jim:IMHO, you'll have to use HUGE caps or you'll result in "not acceptable" ripple for your current need..

[italianguy63]

Maybe a charge pump doubler and inverter... then trim it down to 15V with a regulator?

MC

[EBK]

Edit:I forgot to ask what you are planning on using this for.  The application could affect what would be considered practical.  I don't think my suggestions below are necessary great....


Have you considered buying a larger power supply from a surplus supplier and dividing and regulating to get +12V and -15V? 

For example, allelectronics.com currently has a 48VDC 0.38A power supply for $3.50.  Just another option, unless you are absolutely against a solution that requires an additional wall outlet.

https://www.allelectronics.com/item/ps-480/48-vdc-0.38a-power-supply/1.html

They also have a 29.4VDC 2A power supply ($9 though).

[anotherjim]

It's for an old drum machine with a PMOS rhythm chip and it's only that that needs -15v. However, I've measured the current at 40mA. The rest of it needs +12v at around 400mA. For all I know, the PMOS might be happy with only -12v?

The power brick I'm going to use is unregulated 13.8v DC at 1.4A.
It would be possible to change the brick output to AC and put the rectifier inside the unit if that made any difference to converter possibilities.

I've just had the idea to work out the diode pumps myself - the 555 driving 2 negative pumps, the second fed in parallel but the output stacked on the first one's output and using Schottky diodes. This gets -21.5v off load but the output swing of the 555 drops too much under load. This suggests it is workable, but I may have to buffer the 555 with a push-pull transistor pair.

[anotherjim]

As it happens, the diode pump I made is same as a published diagram. It just looked different the way I drew it. It is a Quadrupler, or would be if the input drive wasn't single ended.
I've boosted the 555 output with a 2N3904/3906 pair, optimized here and there and get -14.4v at 40mA load. That should be enough. I've been testing with 12V supply, so the unregulated DC supply will be a bit higher than that. A 15v zener at the output will clamp against overvoltage.

The 3904/3906 are 200mA which is the same current drive that the 555 should manage. I'm getting higher peak-peak drive from the transistors, maybe because there isn't the extra current limiting circuitry as the chip has, but they do get warm.

[Rob Strand]

The 3904/3906 are 200mA which is the same current drive that the 555 should manage. I'm getting higher peak-peak drive from the transistors, maybe because there isn't the extra current limiting circuitry as the chip has, but they do get warm.

Generally you will find there a right duty and frequency for those doublers/triplers.  if you are way off the performance suffers.   IIRC rankot had a thread around Christmas.

The high level drive on the 555 isn't that strong.  The higher output resistance when the 555 o/p is high means you need a minimum on-time in order for the charge to transfer.

[anotherjim]

I messed around with frequency and duty cycle, but also found the coupling "pump" capacitors quite critical. Currently at 30Khz 50%. A change from 1uF electro's to 680nF film got an extra volt! Lower ESR in the film caps?

[Rob Strand]

I messed around with frequency and duty cycle, but also found the coupling "pump" capacitors quite critical. Currently at 30Khz 50%. A change from 1uF electro's to 680nF film got an extra volt! Lower ESR in the film caps?

That's actually what I'd expect. 

Yes the lower ESR caps help a lot.  Standard electro's don't work so well.  If you want to use them you have to do a lot of playing around to find a sweet spot.  Even then it's a compromise between droop and ripple (and ripple frequency).  Low ESR electro's should work fine.  In the old days I even use tantalums.   

Oh a cap across the input rail can help too.  It stops the input supply sagging when the (slightly discharged) output caps get switched across the rail.

There's few tips in the datasheets for charge pump chips.   The 555 designs pretty much always have more droop than the purpose built charge-pumps.

[diffeq]

I messed around with frequency and duty cycle, but also found the coupling "pump" capacitors quite critical. Currently at 30Khz 50%. A change from 1uF electro's to 680nF film got an extra volt! Lower ESR in the film caps?

Or lower capacitance makes for faster (full swing) discharge at this frequency. Or both?

I had the same experience after replacing 22uF electro to 1uF film when I've tried an unbuffered 555 inverter some time ago. It would be interesting to see an efficient pump inverter made with discrete components.

Few thoughts:
1. Would using low Rds-on MOSFET push-pull drive make it better than BJT?
2. Another "trick" comes to mind: use charge pump inverting output as 555 ground pin, so that pumping action increase its own voltage swing. It'll need some regulation (LM7905?) as not to exceed the 555 18V maximum power rating.

[anotherjim]

I suspect in some ways, a little resistance loss is helpful, even though it's costing some output. Off-load it gives -22.5v, which isn't far off what you would expect. It won't be very efficient, not ever I suspect, since half a cycle is throwing away charge and there are 2 pumps doing that.
It is asking to switch full swing thru a cap and diode, which at some instants in each cycle is almost a short circuit. So stronger drive is going to be felt in the power supply.

Bootstrapping the 555 & pump -ve supply voltage using the output is a sneaky and clever idea. My suspicion is it would only drain itself of the charge it's trying to make.

As there are two pump capacitors, it occurs to me that I might drive one direct from the 555 and the other from the push-pull (which is inverting). In theory, it might improve a bit since there will always be a pump cycle happening.

[anotherjim]

Bi-phase drive works, but not as well because the 555 direct still lacks swing. Would need a 2nd transistor pair, but that's starting to get out of hand.
Tried metal can 2N2222 and 2N2907 pair. Loaded output shot up to -16V, but then started to drop towards -12v. Transistors getting very hot!
Then I wonder if they aren't switching off fast enough and there's shoot through. Perhaps add base diodes and pull off resistors?

[bean]

One of these could work:

https://www.mouser.com/Murata-Power-Solutions/Power/DC-DC-Converters/Isolated-DC-DC-Converters/_/N-brwkv?P=1yxt7dzZ1z0ztj0Z1z0k0eiZ1z0wavnZ1z0k0dsZ1z0wsm3Z1z0j54c

[Rob Strand]

Tried metal can 2N2222 and 2N2907 pair. Loaded output shot up to -16V, but then started to drop towards -12v. Transistors getting very hot!
Then I wonder if they aren't switching off fast enough and there's shoot through. Perhaps add base diodes and pull off resistors?

Shoot-through for sure.  It's virtually impossible to remove without more complex drive circuits.   You can minimize it sometimes by putting in base-emitter resistors to shift when the transistors turn on.  The fact the 555 high side doesn't technically hit the rails means to the top transistor's turn off is a little indeterminate.

Here's some ideas to improve the positive swing.  You will need to choose appropriate transistor and base resistors.  IMHO you might be better off with MOSFETs.   Obviously the 555 no longer supplies current to the load.



Honestly, by the time you add all the junk you are better off with a switcher or a purpose built charge-pump IC. 
If you use a 555 you just have to accept the limitations - if it's not good enough then move on!.

[diffeq]

I suspect in some ways, a little resistance loss is helpful, even though it's costing some output. Off-load it gives -22.5v, which isn't far off what you would expect. It won't be very efficient, not ever I suspect, since half a cycle is throwing away charge and there are 2 pumps doing that.
It is asking to switch full swing thru a cap and diode, which at some instants in each cycle is almost a short circuit. So stronger drive is going to be felt in the power supply.

Could this be remedied by a resistor+electrolytic de-coupling on the high side of the drive?
My understanding is that MOSFETs are better suited for power switching and at those speeds, 555 should be able to drive them without much loss (if any at all). A lot of them also come in TO220 package, which will help with heat dissipation. 

Bootstrapping the 555 & pump -ve supply voltage using the output is a sneaky and clever idea. My suspicion is it would only drain itself of the charge it's trying to make.

Even if the driving stage gives out more current than oscillator needs? And again, some sort of buffering/de-coupling comes to mind.  It doesn't have to be fully bootstrapped to -Ve, several volts should suffice to obtain the "efficient enough" range. A resistor divider 10R/100R from ground/-Ve? Diodes? This will require some thinking.

Honestly, by the time you add all the junk you are better off with a switcher or a purpose built charge-pump IC. 
If you use a 555 you just have to accept the limitations - if it's not good enough then move on!.

As far as board space is concerned, absolutely. Even if SMT packages are used for all of the drive and pump components, it'll still be bigger than a single 8-PDIP IC. In terms of economical gain, it's still "feasible", soft of. A decent charge-pump IC that works above 12V, LTC1144 (which still drops 2 volts under 40mA load...), costs about $4.95 in a single quantity. NE555 with a complementary pair of MOSFETs, Schottky's and caps cost slightly above $2. And then there is an educational value. Throwing parts until it "just works" can be satisfying too. 

[anotherjim]

Yeh, well, any fool could just go out and buy the right part in the first place 
I'm not just any old fool though, I'm a special fool
https://www.rapidonline.com/tracopower-tme-1215s-1w-dc-dc-converter-10-12v-dc-in-15v-dc-65ma-out-57-4976
I said they were cheap, although everywhere I look here in the UK, availability is poor. Of course, I'd buy two because I'd need a spare.

And then there is an educational value. Throwing parts until it "just works" can be satisfying too. 

Absolutely, and there the "we need this now... is there something you can make with what we have that does it?" aspect.

[italianguy63]

I'll make another stupid suggestion...

What about a cheap DC to DC Buck Converter?

MC

Edit-- $1.26 shipped on FleaBay

https://www.ebay.com/itm/2A-DC-DC-Boost-Buck-Converter-Step-up-Voltage-Module-Adjustable-3V-5V-9V-12V-24V/323013308636?hash=item4b3517f0dc:g:-OUAAOSwEIJbNKc5