What's wrong with my voltage doubler?

Thread Starter

wayneh

Joined Sep 9, 2010
17,496
This is an inverter I'm preparing to use with EL tape. For now the load on the inverter is just a small speaker, R1. This will be replaced with a step-up transformer, a wall wart run backwards with an impedance of just 1-2Ω.

I wanted to use only N-channel MOSFETs for the switching, so I put a voltage doubler into the circuit to give me a "reservoir" of 2Vcc to drive the high-side gate. The high side gate is held at the doubled voltage except when Q1 turns on and pulls it low whenever the low-side switch is turned on.

In the schematic, U1 is an LM358 op-amp making two square waves. U1a sets the variable inverter frequency in the 1kHz range. U1b is a higher frequency (20kHz? I forget what I chose) and drives the voltage doubler. The sim shows 1N5818 diodes but I'm using similar SB140 Schottky 1A diodes. [correction - they are 1N5817]

Everything works fine in simulation. The voltage on C5 builds to 2Vcc in a few milliseconds and never drops more than a fraction. I've built the circuit on a breadboard and it "works" in the sense that the speaker buzzes at a frequency that varies with VR1. But I wanted to measure some voltages and be sure things are working as expected before adding the step-up transformer. Here are some observations with my cheapo Cen-Tech multimeter:
• The DC (AC) voltage out of U1b is 6V (12V)
• The DC (AC) voltage out of C4 is 12V (24V)
• The DC (AC) voltage on C5 is 12V (24V)
• The speaker buzzing stops if R9 is removed, but C5 voltage is unaffected
• Changing or removing C5 has no effect on the buzzing
• Attaching the LED indicator to the source of the high-side makes it glow less brightly than Vcc.
• Both diodes test OK; open in one direction and 145mV in the other.
My conclusion from these observations is that something is wrong with the voltage doubler. The DC voltage is not building on C5. I speculate that, when turned on, the high-side switch is pulsing on very briefly with each pulse of the doubler. I can't hear the high frequency of the doubler, so I'm not completely sure.

How can there be such a high AC voltage, compared to the DC voltage, appearing on the top of C5? Is this all just an artifact of my meter? Before you say yes, remember that the buzzer sounds identical if C5 is removed.

Screen Shot 2019-01-18 at 1.31.45 PM.png
 

Attachments

Last edited:

DickCappels

Joined Aug 21, 2008
10,153
It looks like it is only switching the "origianal" 12 volts on the output. The voltage on C5 is only used for gate drive to M1 -I don't see any way that voltage can make it to the output.
 

AlbertHall

Joined Jun 4, 2014
12,343
How can there be such a high AC voltage, compared to the DC voltage, appearing on the top of C5? Is this all just an artifact of my meter?
Try connecting your meter to the 12V DC supply and see what readings you get for DC and AC. If you get a reading on the AC range then your meter is not capacitor coupled and will give misleading results if DC is present as well as the AC.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,496
Try connecting your meter to the 12V DC supply and see what readings you get for DC and AC. If you get a reading on the AC range then your meter is not capacitor coupled and will give misleading results if DC is present as well as the AC.
Yeah, it reads 24V AC for the DC supply. But, why isn’t my doubler putting more voltage onto C5?
 

Sensacell

Joined Jun 19, 2012
3,432
The output impedance of the charge pump will be very high, all the current needs to be delivered in the very short pump switching transient.
The 4.7K resistor loads it down substantially.

I have employed similar tricks, the current you can obtain is always a disappointment.
Whatever drives the pump needs to have a really low output impedance and a high switching frequency.
 

AlbertHall

Joined Jun 4, 2014
12,343
The voltage on C5 is unaffected if R9, the 4.7k, is removed so loading does not seem to be the problem.
Check the connections to C4 and if it's electrolytic the positive side should connect to the two diodes.
If all seems correct there then try replacing C4.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,496
The voltage on C5 is unaffected if R9, the 4.7k, is removed so loading does not seem to be the problem.
Check the connections to C4 and if it's electrolytic the positive side should connect to the two diodes.
If all seems correct there then try replacing C4.
C4 is one of these, an MLCC 2.2µF. I'll try something else. I'll also have to take a hard look at that oscillator. Sure wish I had a genuine oscilloscope. :(

Huh, looks like the doubler frequency is set to only about 6kHz. I should be able to hear that!
 
Last edited:

Thread Starter

wayneh

Joined Sep 9, 2010
17,496
I've been probing around with some headphones, using a capacitor to couple the signal. I can tell there's a lot of power supply noise throughout my circuit, so I'll be trying a battery soon. I've been using an old hard drive PSU and it's pretty good but clearly noisy.

I didn't hear much of anything else besides the PS noise at C4, so I'm wondering if the LM358 can do the job I need at 6kHz. I used a tone generator on my phone to make sure I can hear 6kHz, and I sure can. It's ear splitting! Despite the PS noise I should hear a much louder shrill tone and C4 and I didn't.

So I've got some things to chase down.
 

LesJones

Joined Jan 8, 2017
4,174
I think your multimeter may be responding to DC voltages when set to an AC range. Try putting a capacitor (0.1 to 1 uF) in series with one of the test leads to block ant DC component.

Les.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,496
My op-amp may be blown. I'll test another soon.

This morning I used a battery to eliminate PS noise. The oscillator U1a works fine and a strong buzz is produced at the load. When I probed oscillator U1b with my headphones, all I could hear was U1a. After disabling that side, I couldn't hear anything coming out of U1b.
 

Thread Starter

wayneh

Joined Sep 9, 2010
17,496
What impedance headphones? They may be loading the circuit too much.
Could be. Apple headphones are said to be 23Ω. I'm using a 100µF cap in series to block DC.

But, I can hear the U1a oscillator plenty loud in the U1b output unless I disable it. And U1a is deafening when I probe it's output.
 
Last edited:

Thread Starter

wayneh

Joined Sep 9, 2010
17,496
Problem solved! I swapped out the LM358 for a fresh one and it made no difference. Checked the resistors making up U1b and all showed resistance in range. There was nothing left but C3 and sure enough, swapping in a 0.1µF gave a deafening tone.

Turns out I had mis-ordered from Mouser years ago. I wanted to order 0.01µF 103 caps but mistakenly ordered 100pF 101 caps. I never noticed the error - until just now.

Now the doubler works! With a battery supply of ~6V, I see ~10V on C5, which makes sense for the output range of the LM358 (Vcc - 1.5).
 
Top