2
\$\begingroup\$

A picture is worth a thousand words, and I tend to make a lot of mistakes in my words, so, first picture, the circuit in question:

Dickson Charge Pump Circuit

I know the Dickson charge pump looks a little different but it is essentially the same.

EDIT: We can basically forget about the charge pump type whether it is MOSFET or diode or if the MOSFETs are from drain-source, etc. The problem remains the same.

Below is my pulse engine which does a fantastic job of providing me with two sets of pulses - until I hook them up as in the next two pictures.

Picture 1. The pulses are not hooked up to the negative side of the capacitors and

Picture 2. The pulses are then attached and the pulses are not at all the way I want them, that is, they do not alternate and one of the two no longer returns to ground.

Picture 1: Great Pulses

Picture 2: Terrible Pulses

I've used op amps to create the pulses and inverted pulses and bipolars - but that's not really the problem. It's that there needs to be an outlet or something for the capacitors.

I've tried to put PNP transistors that fire when the alternating pulse drops to zero. I've used resistors and also a lot of other embarrassing things I won't talk about.

I've tried to put high pass or band pass filters on each pulse line but that does not work at all.

The question, again, is: How to keep the pulses alternating and returning to ground after connecting them to the negative half of the capacitors in a topology like this?

\$\endgroup\$
7
  • \$\begingroup\$ @Andyaka Well, in picture 1 they are nicely alternating pulses and returning to ground appropriately. In picture 2, once they are attached to the capacitors, they are not alternating and one of the pulses has an increasing dc bias. I want picture 1 after attaching to the capacitors \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 18:00
  • \$\begingroup\$ Here is a nice pic and link to research gate, but lots of ways of doing it: researchgate.net/figure/… \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 18:15
  • 1
    \$\begingroup\$ @Andyaka The reason to use Mosfets is because when done right you can get rid of most of the voltage drop associated with the diodes - even schottky. I used 1N5818's in previous charge pumps. But, I'm trying to achieve a buffer amplifier as well as a charge pump. \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 18:37
  • \$\begingroup\$ @Andyaka "so what was yours based on?" sorry for the long delay in responding. I've researched these over two months.. so many designs. Here's one complete circuit: circuitlab.com/circuit/5k5gmr/dickson-charge-pump \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 18:54
  • \$\begingroup\$ @Andyaka You are always super helpful to me Andy. I know my topology is different. I've worked it with diodes, and with mosfets both ways. In the circuitlab link, he used a sine wave and two not gates. how can I achieve THAT circuit with discretes or mosfets or op amps regardless of the charge pump topology? \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 19:23

2 Answers 2

2
\$\begingroup\$

I agree with jp314, that the MOSFETS are connected the wrong way and conduct in both directions. The difference in forward and reverse voltage across the FETs produces a small pump effect of 1-2V per FET.

The circuit you referred to with the researchgate link uses FETs as diodes because they can be produced straightforward inside a CMOS chip, there is no other advantage.

The other chargepump link works identical in the simulator if you just remove all MOSFETs, so this is not well understood either.

To drive a dual phase charge pump, you need two strong push pull driver stages, that can source and sink all the current for the charge transfers of all capacitors at once. It is a symmetrical requirement. Your driver stages may be capable to sink 1A or more, but can only source less than 100uA.

You can use H-bridge DC motor drivers or very powerful gate drivers (dual low side) between your pulse generator and the pump chain. R13, R16 and R17 should be <= 10kohm anyway.

\$\endgroup\$
1
  • \$\begingroup\$ Thank you for that answer. The H-Bridge DC Motor Drivers can be built using mosfets or discretes - I just couldn't figure out how to get my circuit to sink and source current. The rest.. it's.. well, what I'm trying to achieve isn't up for debate in this question ;) Thank very much for your help!! Great answer \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 23:21
1
\$\begingroup\$

Your MOSFETs are connected the wrong way. G-D means it conducts with a forward V of about Vthreshold. However the drain-bill body diode means it also conducts in the reverse direction.

Flip the FETs and connect gate-source. Not it will be off in reverse bias and in forward the FET and/or body diode will conduct. Note that if this is in an IC you probably don't want the drain-body diode to conduct...

\$\endgroup\$
3
  • \$\begingroup\$ Right, I've definitely seen that topology. I know that topology works as well. However that's not the question. In either case - that topology or this one - the pulse lines to the neg portion of the capacitors does not work for me. Do you know how to make those work? \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 18:45
  • 1
    \$\begingroup\$ What you have won't work -- it shorts the pulses to +/- 0.7 V from the supply. It''s not a topology choice -- it needs to be done that way. That's why your pulses are not appearing correctly. \$\endgroup\$
    – jp314
    Commented Jul 19, 2022 at 23:20
  • \$\begingroup\$ Thanks for your helpful thoughts. I did flip the mosfets around and add the body diode, but alas the pulses acted exactly the same way. Jens answer is perfect. \$\endgroup\$
    – RobMcN
    Commented Jul 19, 2022 at 23:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.