High Current Pump Charge Converters

[TheWinterSnow]

I have built a circuit that is using the Max1044 chip which I fried because I miscalculated the current draw of the circuit.  The Max660 is much closer being that it can handle 100mA and 730mW.  What I need is something slightly higher, I calculated an average of at give or take 120mA at 1.05W (that is peak watts not RMS, same with the current).  Now I know that 8-pin dip chips can only really handle around 730mW can I make the assumption that 1.05Wp or 742mWrms would be okay for a 730mW chip or is it still going to fry?  Are there any pump charge converters that can handle 120mA to 150mA or can a 100mA chip handle being pushed up to 120mA to 130mA.  Obviously that is much better than the 1044s maximum of 10mA.  I fried about $16 worth of OPA2134s malfunctioning the converter, hard lesson learded to really calculate your load.

[Seljer]

If you want more power discard the charge pump and use an inductor based voltage converter. The simple ones come in 8 pin packages and you just need to add a diode, a little inductor and some capacitors. Browse over the Linear Technology and Texas Instruments websites.


Otherwise, maybe give heatsinks a shot.

[TheWinterSnow]

Thanks man, I kind of wanted to stay away from the inductor based voltage converter but 1.5A with an input voltage up to 15v in an 8pdip is perfect.  Still going to be slightly over dissipating but like you mentioned, I think a heatsink should take care of that.  I might just get the small adhesive sinks that you typically see for aftermarket air cooled graphics cards memory modules.  I have had good luck keeping DDR3 and DDR5 chips cool with those little devices.

I am going throw in another question.  I am looking at the TI MC33063A and given the max power equation given, at a 25C temperature, the max dissipation is 1.2W.  Now I figured, being in a black box in the direct sunlight type of situation, think like live outside shows in the middle of a hot summer, I expect the temperature in side to get upwards of 65.5C, that would limit the maximum dissipation to 876mW.  Since I expect the total power of the circuit to get up to 900mW, which is a nice rounded up value of the power figures, that would put the heat of the device over the maximum.  Obviously I would realistically need to use heat sinks, so how much could I expect a small heatsink, small enough to fit on an 8pdip to keep the heat down off the device?

[Cliff Schecht]

Before I comment on anything else, the term "RMS power" is a misnomer. Power is a term that is specified by the rms voltage squared divided by a known resistance or multiplied by a known rms current; thus power is already an RMS term. Multiplying a known instantaneous voltage by an instantaneous current will give you the peak power dissipated at any given time, however, and this is a common metric to use if it is specified that you are talking about peak and not RMS.

As far as your power solution goes, it sounds like you are trying to do too much with the charge pumps. They really are pretty crappy power supplies and are not meant to continuously supply very much current. If you need a burst of 100mA for a short time or you are only pushing maybe 75-80% of what the max specified current is then they are definitely usable, but you have to know that the output voltage will jump all over the place as current draw varies. This is why charge pumps are not a great idea for any sort of OTA-based solutions where the output voltage is dependent on a bias current. If you are running more than a few OTA's off of one charge pump, the output voltage will vary and thus cause the current to vary and, for a voltage-controlled oscillator, you will hear the voltage drop as a drift frequency. I tend to avoid charge pumps as a rule unless I need a compact solution to develop -3V or -5V at a pretty constant current draw (TI makes a part for this that works well in my experience). I also don't like my switching supplies to operate anywhere near the audio range. 100kHz is fine but 25-30kHz is not that great or desirable (you deal with larger magnetics as well).

Now, as far as the MC33063 goes, I really dig this chip for a cheap power solution. I've used them for inverting switchers in the past and was very successful. The really nice thing about this part is that you can go open up most any inexpensive car or wall cell phone charger and find one. They can supply ample current and come in 8-pin DIP packages which is good for heat dissipation in the IC as well. If you need more current, add an external switch and a slightly larger inductor. When all is said and done I can make one of these on a board that is smaller than 1" x 1" and get enough power to drive any analog pedal you'd want.

An even sexier solution would be a flyback converter but these typically require custom magnetics which is not an easy endeavor for a first power supply design.

[TheWinterSnow]

The MC33063 max current draw is beyond overkill, my only concern was overdissipation.  Under typical operating conditions the total power should be 20-30% of what it can handle, but if some idiot were to short circuit all the outputs and run it full volume in the sun on a hot summer day in a black box, it would be slightly overdissipating.  I did find a an 8 pid dip clip on heat sink and I realized my short circuit current limiting loads on the amps are actually too low and need to be at least doubled, which means the dissipation figures in even the worst case scenario of user error still won't fry the chip.  I am overly paranoid and thorough on ruggedness on all my builds.  Regardless if some idiot decides to short circuit all the outputs somehow and fry the pump charge, its a cheap chip and I use IC sockets so even user installation of a $0.70 chip is not that hard.

I estimated under typical operating conditions that the whole circuit would be pulling 30-40mA at 9v so about 400mW, well within the dissipation of the chip and considering the pump charge is only pulling current when the negative portion of the signal is being amplified, the actual dissipation figures in my circuit will be less than if it was pulling current all the time, so figure even half of that will actually go to dissipation in the chip.