Charge Pump: a bit of confusion

[JackSkellington]

Hi, everybody!

I'm going to build the Amptweaker Tight Metal, but the layout I found doesn't have the charge pump, so I have to build a daugherboard.
I bought some ICL7660SCPA, I want to build something based on it.

I never used a charge pump, but I always thought the following veroboard layout was simple and good:



Of course, I compared it with some ICL7660S layout. Some of them include some extra parts like a 100uF and/or a 1N400X diode, as filter in the power supply. But the layout of the Tight Metal I have to use already has this accessories parts, and I guess I don't need of them.

And now my doubt about the circuit. I see in the charge pump layout that the pin 5 goes nowhere, so it's isolated. But in some other layout, as some schematic, pin 5 is connected (with the pin 3) to the ground. Why in this layout it isn't so?

Here's the schematic of the Tight Metal by Grind Customs with a charge pump section similar to the veroboard layout, except where pin 5 and pin 3 are connected together, and both go to the ground:



Should I consider to put a jumper to connect the ICL7660S pin 5 to the ground like the schematic?

[alanp]

When in doubt, always look up the datasheet.

https://www.intersil.com/content/dam/intersil/documents/icl7/icl7660s-a.pdf

We're using the chip in it's voltage doubling application, which is on Page 10 on the second column. Pins 3 and 5 are indeed grounded. Pin 1 is the 'Boost' pin, which boosts the switching freq above audio range (always a good idea in an audio circuit!), which is why we have that pin attached to V+ when it is floating in the example circuit.

[JackSkellington]

Thanks for the datasheet.
That kind of connection causes the doubt. Why in that layout the pin 5 are isolated? And that layout is verified and used.

[alanp]

Pin 5 in the GCFX schematic you've shown is also attached to ground -- if you look, it's wired to the same GND point as Pin 3.

[reddesert]

Page 8 of the data sheet, left column: explanation of how the circuit works to use switches on a high-frequency cycle to charge an external capacitor and prevent it from discharging, and discussion of technical considerations of how the switching transistors are biased to avoid latching up. Page 8 right column do's and don'ts, point 7: "User should ensure that the output (pin 5) does not go more positive than GND (pin 3). Device latch-up will occur under these conditions."

In the voltage doubler application, we aren't using the V_out (pin 5) that is used for the negative voltage output. The do's and don'ts clearly suggest that it should be tied to ground. If it floats and is not tied to ground, the circuit may work, until something unexpected happens and it latches up, and then it won't work (and the user will probably say "My charge pump IC is bad," mis-diagnosing the problem).

[JackSkellington]

Pin 5 in the GCFX schematic you've shown is also attached to ground -- if you look, it's wired to the same GND point as Pin 3.

Yeah, I meant right this. The veroboard layout has the pin 5 isolated, in the Aeon (Tight Metal) schematic the pin 5 is grounded.

Page 8 of the data sheet, left column: explanation of how the circuit works to use switches on a high-frequency cycle to charge an external capacitor and prevent it from discharging, and discussion of technical considerations of how the switching transistors are biased to avoid latching up. Page 8 right column do's and don'ts, point 7: "User should ensure that the output (pin 5) does not go more positive than GND (pin 3). Device latch-up will occur under these conditions."

In the voltage doubler application, we aren't using the V_out (pin 5) that is used for the negative voltage output. The do's and don'ts clearly suggest that it should be tied to ground. If it floats and is not tied to ground, the circuit may work, until something unexpected happens and it latches up, and then it won't work (and the user will probably say "My charge pump IC is bad," mis-diagnosing the problem).

«If it floats and is not tied to ground, the circuit may work». That MAY worry me a bit.

There's even Sabrotone layout:

But this HAS the pin 5 grounded.

It's slightly bigger than the other, and it has the 100uF and 1N4001 extra parts. I assume I can omit them, right? The Tight metal layout already has them.

But I guess I can settle with the tagboard layout and to add just a jumper from the pin 5 and the ground stripe, a shortest jumper in the column between the diode and the cap.
Is that correct?

[alanp]

But I guess I can settle with the tagboard layout and to add just a jumper from the pin 5 and the ground stripe, a shortest jumper in the column between the diode and the cap.
Is that correct?

Worth a try If it doesn't work, you can always cut the wire out afterwards.

[JackSkellington]

Here's the modded layout with a jumper makes even the pin 5 grounded.



Thanks!

[JackSkellington]

Just because I got a piece of stripboard slighly smaller than what I need for this layout I drawn the it again.
I moved the jumper that make to go the pin 3 to the ground and I moved a cut.

None of these two layouts are verified, but they shoud be ok because the original one is really similar.

[JackSkellington]

A quickie question:
What if the layout includes a 1N4001 and a 100uF electrolytic cap from the 9v and the ground though the effect has already them on board?
The caps probably will be 100uF + 100uF = 200uF, but it or could be better or it doesn't matter.
What about to have two 1N4001s? Or this 1N4001 and, on the effect board, a 100R resistor in line and a 1N4148 to the ground?
I wouldn't a cut of voltage or frequencies.

Thanks!

[mjg]

If you have the 1n4001 diode on the charge pump board, there is really no reason to have one on the effect board as well.  They are just there to try to protect against someone plugging in the wrong polarity supply.  If your charge pump board protects against that, then the effect doesn't need to do it as well, because you will have them wired up correctly inside the box. 

For the protection diode, I'd consider using a 1n5817 inline with your +9v input, rather than the 1n4001 from 9v to ground.  That seems to be what people are doing these days, and it makes more sense to me.  If you plug in a backwards supply, it will not allow voltage to pass.  With the 1n4001 setup, if you plug in a backwards supply, it will create a short, which might protect your pedal, but seems like it could screw your power supply to me.  Something to think about.  :-)

I don't think it will hurt to have the extra capacitor... it will make sure your power supply is stable closer to the effect.  It might not matter if you leave one out, but I don't think it hurts to have both. 

[JackSkellington]

Thanks!

In my case, today, the charge pump will be switchable, and the effect has 100R + 1N4148 in line (should be better a 1N5817 diode? )

The effect has a 47uF as filter cap, and I don't expect more noise with the charge pump (maybe less noise because it will louder? ), I could use anyway the 100uF extra cap in case of noise.