King of Tone Charge Pump problem

[lars-musik]

Hi,

I just built a King of Tone and thought it might be a good idea to have a switchable voltage doubler at hand for an extended headroom at cleanish boost settings. As I sometimes give pedals away to people with a need to plug in obscure PSUs, I ordered a handful of SMDs for RG Keens good and cheap polarity protector and implemented that as well.

However, when I switch on the voltage doubler, the sound shortly fades away then comes back but distorts unpleasantly when I strike the strings hard (quite the opposite of what it is supposed to do). When I play on, the volume drops occasionally and some rumbling and noise takes place.

Would you have a look at the schematic what is amiss?

The polarity protection spits out 9.2 V at the "9V-prot-out" point, at the 18V point I get a fine 17.8V .

Complete schematic here: King's Drive Voltage Doubled



Thanks and all the best, Lars

[MrStab]

what kinda op-amps are you using? is everything else rated for at least 25V?

[lars-musik]

I am using standard JRC4558s and yes, I hope everything (caps, that is) is rated high enough. It was a problem, though and I had to salvage some old non-functional builds for the electrolytics (I am so much into the 1590a builds that most of the caps I got were only 16V). I also had to use some non-descript 10nF SMD caps (0805). They might be the culprits.

If there's no mistake in the schematic then that might just be it.

Thanks!

[MrStab]

ah okay, so not the op-amps, at least. i'd definitely check the caps.

FWIW, i built something ages ago with a switchable 7660S multiplier, and although there's a bit of noise upon switching, the circuit continues to function even if switched while powered. so either i got lucky or you can rule that out.

no probs! see if it's the caps & let us know how it goes

[Groovenut]

This may not be or solve the current issue, but I would recommend using a resistive divider on the 18V supply for your 9V instead of taking the 9V from the input supply. When you disconnect your 18V supply to go to 9V, you have no load on the 7660. When you switch to 18V, you suddenly have a load and the caps take time to stabilize. This can cause drop outs in audio or distortion as the voltage stabilizes. It may even diminish the popping of the switch at voltage change.

Sorry for hacking up your photo

The cap I added to the resistive divider is probably unnecessary because of C11 but I thought I'd throw it in there for good measure.

I used the 10ks but those can be adjusted for less current draw if it becomes an issue.

Again just IMO, allegedly.....

[MrStab]

I would recommend using a resistive divider on the 18V supply for your 9V instead of taking the 9V from the input supply.

i only roughly skimmed the schematic tbh, but +1 on this big time. in addition to Lawrence's point, fluctuations in the positive rail and bias point will be more relative with a divider. also, if using the 9V input, the diode drop from the multiplier stage would mean the bias point is slightly higher than centre (though most opamps are a bit uneven in their distance from each rail). maybe the reason described in the previous post is causing some sorta latchup upon switching?

[duck_arse]

if it is a bias resettle problems, maybe if you add another electro across R12, equal to the one across R13, the capacitor divider will divide somewhat faster.

(I don't know where the electrons will go, though ......)

[MrStab]

i really should've checked the diagram, now i realise what Lawrence actually meant lol.

as i mentioned, my own half-hearted attempt at this involved straight-up switching without a divider and all seemed to work fine within a second or less of switching. not noiseless iirc, though.

just a weird thought that popped into my head: the circuit i did this with had all non-inverting opamps, no straight connections from Vb to opamp inputs without a biasing resistor. maybe the voltage change is more likely to cause problems with inputs directly shorted to Vb? pure conjecture.

maybe when switching the opamp sees the vbias as the same or above the positive rail all of a sudden, latching it up. maybe Vcc goes to 9V first but for a split second the Vb is also at 9V because R12 & C10 slow down the change for it to go down to 4.5

[lars-musik]

Thank you all for your time, ideas and suggestions. There are some remarks I do not fully (or even half) understand, but the idea of getting to 9V per voltage divider from the 18V line makes sense to me.

if it is a bias resettle problems, maybe if you add another electro across R12, equal to the one across R13, the capacitor divider will divide somewhat faster.

Would you care to elaborate? Unfortunately you lost me already at "bias resettle" and I don't even know  where I would place a capacitor that should do its magic across R12 (from one R12 junction to ... ground ... VC ... the other junction?)
Dividing faster would be (like in a math test) ... good?

[Groovenut]

Given the bias network values on the schem, it will take ~3 seconds for the 100uF bias cap to discharge from 9V to 4.5V after the switch is thrown.

[MrStab]

it will take ~3 seconds for the 100uF bias cap to discharge from 9V to 4.5V

in the first half-second or so, that'll be above the positive swing of most op-amps, while the V+ change is seemingly instant.

maybe a buffered Vref on a spare op-amp would be another way to make it more reliable.

sorry for all the technobabble, Lars. basically, it seems for a moment after switching, the voltage going into the inputs of your op-amps is higher than the voltage they're supplied with, which is known to cause the chips to latch up.

[lars-musik]

Electrobabble is fine. That's what I am here for.

If I understand the matter correctly, the initial problem is that the 7660S is without load if the the circuit is at 9V. Now you established some alternative  ways to get the 9V supply (voltage divider) that in turn might cause the op-amp to latch up (bad again).

Now, what argues against just using a DPDT to take the 7600S completely out when switching to 9V?
That would even keep the part count lower than with voltage divider resistors and more caps?
Like that:

[duck_arse]

lars - if your problem stems from the time it takes for Vb to settle to a usable V/2 with the switch from/to more/less supply volts, a capacitive divider will help speed the transition (I only know this from reading a thread a few months ago).

so, you have the R12/R13 divider providing Vb, with C10 (? I'm squinting here) bypassing Vb. maybe swing the ground end of C11 across to the Vb point - it is now parallel with R12. now you have a pair of caps voltage dividing and AC bypassing the two divider resistors. you can add another cap from VC to ground to replace, but you can probably see the change I mean.

the two divider caps need to be the same value, as they divide in proportion, just like the resistors. [one of the experts will be along soon to make the necessary corrections.]

[Groovenut]

Also, just as a thought for future builds, if half/double switchable voltage is desired, it may be easier to implement a +9/-9 dual rail with ground as the reference voltage. Switch in a resistive divider or voltage regulator (LM317 adj to 4.5V) at the input of the charge pump. This yields ~4.5/-4.5 and the reference voltage never moves. It's easier/simpler/less parts to implement than a doubler setup, it only uses one switch and there are no bias voltage fluctuation issues. 

[MrStab]

Electrobabble is fine. That's what I am here for.
Now you established some alternative  ways to get the 9V supply (voltage divider) that in turn might cause the op-amp to latch up

i think the bias latching the inputs was the entire problem to begin with tbh, not just a problem with the divider approach. one way to narrow that down would be to see how the unit performs when switched between different voltages with the power off (ie. no switching when powered)

[lars-musik]

I tried the DPDT version last night and it seems to work. When switching from 9V to 18V it takes about two seconds to stabilise, then it works fine. The other way round (18V to 9V), there's half a of second squealing (some kind of discharge?) then everything is good. The switch is not supposed to be flipped in mid-play, so I guess I am fine. If I find the time, I'll build one of the charge pumps with the voltage divider (or I'll even give my breadboard another go) to see if it works better.
Best, Lars

[MrStab]

glad it works, Lars, and sorry for all the confusion. there's still potential for long-term damage to your op-amps by doing this, so you may wanna keep an eye on it or refine it.

http://www.geofex.com/circuits/when_good_opamps_go_bad.htm