Rangemaster with charge pump and Millenium bypass

[Paul Fawcett]

Hi guys,

First post here, so I'll just introduce myself briefly by saying that I'm a relatively experienced tube amp builder, but am just getting started with effects.  Some of you I probably know from other forums. For the rest - Hi! I'm Paul, you'll be seeing more of me.

To get my feet wet with PCB layout, I'm planning on building a Rangemaster variant that incorporates RG's charge pump and original JFET based Millennium bypass.  You can find a proposed PCB layout and schematic on my amp blog: http://wombatamps.blogspot.com/2012/01/dipping-toe-into-pedal-world.html

Anyway, the main question unresolved question I have before diving in surrounds the appropriate ground scheme that I need for this beastie.

As you can see on the blog, the schematic shows both C6 and C4, which uncouple the -Ve 9VDC to the charge pump ground and the effects circuit ground respectively. Only C4 is currently on the PCB.

Do you think I should have both of these capacitors, or just one or the other? C4 was from the original Rangemaster, whereas C6 is from the implementation of RG's charge pump on tonepad: http://www.tonepad.com/getFileInfo.asp?id=118

Normally, I don't think it would be a big deal, but evidently there are potential noise issues with the charge pump. Maybe I'm slightly confused about where RG is saying the big charging currents are coming from? http://www.geofex.com/circuits/+9_to_-9.htm
 

[garcho]

Amps are effects, especially guitar amps! Wise project to choose, but I'll let a qualified poster answer your question. Welcome to the forum, glad to have you.

[Morocotopo]

 Hi Paul, your name rings a bell, but not sure where from...AX84? I´m pretty sure you´ve answered some of my dumb questions somewhere! I´m Mostro on other forums.
On to the subject, think of those caps as power supply filter caps on an amp, they are there to eliminate ripple and/or switching noises, to get a steady DC. I´m currently prototyping a Fuzz Face with a charge pump, and using that same schem for it. Not a single noise other than the usual in an OD pedal. I´d try both options, but the caps are essentially in parallel and work as a single one. If noise worries you, you might try a 0,1 uF ceramic in parallell, but I don´t think it will make a difference. More important is, to use a charge pump IC that has a osc boost feature, like the 7660S or, I think, the LT1054 also, so that the osc doesn´t operate in the audio frequencies. By the way, i think the LT is overkill, since that has a max current of 100mA if I recall correctly, you might try a 7660S, 10 mA max. But both will work OK.
Hope this helps.
Disclaimer: I´m no expert, this are just my experiences.

[Paul Fawcett]

Hi Morocotopo,

Yes, I recognize your handle from AX84 as well, that is my usual hangout - I think you go by Morocotopo there too.

Thanks for your response.  However, I'm already perfectly aware of what these caps are for in terms of their role in the circuit.  Apologies if I didn't make what I'm after sufficiently clear - let me try again, perhaps I can express myself more clearly.  What I'd like are opinions on whether it is preferable to decouple supply noise on the -9V rail through the charge pump ground (i.e. C6), through the circuit ground (i.e. C4) or through both (albeit at different points in the ground path).  Note that according to the way I have drawn the schemo and the PCB, these are both independently star-grounded back to the input jack shield. So this is really a ground scheme question, the goal being to minimize audible noise attributable to the charging current in the IC.  Moreover, I'm also seeking more general comments about the sufficiency of the PCB layout I'm proposing,

Cheers and thank you!

Paul   

[Morocotopo]

Oh, I see Paul. To be honest, I -think- that it doesn´t matter much, since the involved currents are small. I´ve used the charge pumps in various schems with PCB´s designed by me, and never really paid too much attentiont to that. The only thing I´ve taken care when designing PCB´s is to not make a loop with the ground track. Also, when prototyping in breadboard, I just grounded the different grounds to the nearest ground available, and never got extra noises from that. Of course I might have been just lucky... Why don´t you try the different possibilities on breadboard? It might be important if you want to try to get the absolutely best performance possible. Who knows?
But let´s wait for the really knowledgeable people to state their opinions, I´m just a hacker.
Good luck with the project!

[Govmnt_Lacky]

OK, I have been looking into this topic (Charge pumps to generate -9V) for a while and I am hoping someone can drop some knowledge on me:

1) Why do all of the designs I see have the +9V input voltage going through a transistor before going into the charge pump? What purpose does that serve? Why cant the input voltage go straight into the charge pump?

2) Grounding...... Does the charge pump ground need to be seperated from the circuit/signal ground? Would there be a reference problem if they are not seperated?

Sorry if these are simple(ton) questions but, I am just getting into charge pump integration 

[Morocotopo]

The transistor is to avoid noises getting into the audio. See the Geofex page linked by Paul.
As for the other question, hmm, that´s what Paul is asking... and what I tried to answer according to my experience.

[Govmnt_Lacky]

The transistor is to avoid noises getting into the audio. See the Geofex page linked by Paul.
As for the other question, hmm, that´s what Paul is asking... and what I tried to answer according to my experience.

OK, questions about the link.

1) Did RG make an error with his labeling? It looks like he is showing the mono plug TIP going to the 6.8K resistor and the RING is labeled as the audio input signal 
2) If they are reversed... when you connect a mono plug into the stereo jack, aren't you STILL connecting the Ring to Sleeve?

[Morocotopo]

Yes, if I remember right, the ring and tip are reversed in R.G´s drawing.
And yes, you are still connecting ring to sleeve with a mono plug, but the ring connection now only serves to control the transistor´s base, it doesn´t carry audio signal, so the audio ground (that is also the power ground) doesn´t get noise into it.
At least that´s my view of it, maybe someone can explain better.

[Govmnt_Lacky]

And yes, you are still connecting ring to sleeve with a mono plug, but the ring connection now only serves to control the transistor´s base, it doesn´t carry audio signal, so the audio ground (that is also the power ground) doesn´t get noise into it.

OK, so if I use "star" grounding... what is keeping the signal and power grounds isolated from the transistor's base control?

Is it the transistor itself? Or is it the associated resistors in the control circuit (a.k.a. 6.8K/1K)

[PRR]

> +9V input voltage going through a transistor

That's just a clever on/off switch.

> Why do all of the designs I see have the +9V input voltage going through a transistor

Because R.G. sketched that back in the previous century and everybody else copies R.G.

[Paul Fawcett]

> +9V input voltage going through a transistor

That's just a clever on/off switch.

> Why do all of the designs I see have the +9V input voltage going through a transistor

Because R.G. sketched that back in the previous century and everybody else copies R.G.

Actually, R.G. makes a very specific claim about the switching with the transistor being necessary to keep the charging current from the charge pump out of the signal ground:

"You may want to use the stereo-jack trick to make power switching easier on your stompbox pedals; you'll find that this is not a good idea with a charge pump converter running your -9V supply. This is because in the standard stompbox setup, that wire is the circuit's power *and* signal ground. The current pulses from the switching  converter run on that wire, and will cause the signal ground of your effects circuit to hop up and down at the switching frequency - and you'll hear it!"

Personally, I'm probably misunderstanding something or missing the point, as it seems like the negative terminal of the battery would probably be grounded to the input jack shield in either case -- it's gotta go somewhere, after all... But perhaps you understand the issues more clearly and could address the specific merits of his claim?

[davent]

Hi Paul,
I've been playing around with this circuit, using a MAX1044 power supply, on the breadboard for the last couple weeks so just some observations.

With ear up to speaker, guitar volume at zero, pulling or reinstalling C4 (47uF) changed nothing but i'm using a very well filtered power supply and that may not always be the case so it stays in the final version.

Germanium transistors vary greatly so breadboarding your transistor is essential for fine tuning your component values.

The Geofex document states the circuit bias was usually set between -7.2v and -6.8v but at the suggestion of another forumite i've got mine set at around -4.0v and will end up setting it there in my build. So when you breadboard the circuit stick a 250k trimmer in your R5 spot and don't be afraid to wander outside the prescribed bias points and go with what your ears tell you, then pull that trimmer, measure the value and replace with the nearest value of fixed resistor.

Added the tone control from Joe Gagan's Easy Face and that turns it from a one trick pony into a much more versatile little beast. I've got a 5n and a 1uF cap in the tone control but may go a bit smaller at the 1uF end of things, i'll play with that further.

Output cap is 330n, just what was handy at that moment, will audition other values there. (Last time i looked, your schematic showed a 1n(c5), i'm guessing a typo.)

I'm curious about the MAX1044 switching as well and have it on my PCB but not sure whether i've set it up correctly so interested and hearing back as well.

(ANother typo, R3 from your schematic is a 6k8 in the Geofex document.)

Be prepared for a rocky road with toner transfer, have never gotten satisfactory results, others get great results but i'm sure it's taken a major effort to get to that level of proficiency. Photo resist is a breeze and to me the only way to go.

PCB's done with photo resist, Rangemaster bottom right.



Take care!
dave

[Paul Fawcett]

Hi Dave,

Thanks for the hints.

The 1nF output cap was indeed a typo -- according to the Geofex documents, the original value was 10nF.  Note that assuming the worst case 10K output impedance, and an input impedance of the circuit being driven of 1Meg, a 10nF cap is already sufficient to give you a corner frequency of approximately 16Hz (rather lower than you need for guitar), so there isn't a whole lot of benefit going larger unless you anticipate driving a circuit with substantially lower input impedance (which I suppose could be the case with some effects).   But 330nF will give you a corner frequency substantially below 1Hz, which won't really accomplish much except perhaps increase the chance that the input tube stage on the amp might experience blocking distortion.  In any case, even surprisingly aggressive attenuation of bass is rarely a bad thing when driving tube circuits, so I personally tend to keep such coupling caps relatively small.  Probably auditioning different values for the input cap would be more fruitful in terms of tuning the frequency response to something you might find desirable in typical use.

Thanks also for catching the 68K typo on the base of the input transistor. I managed to get the correct 6.8K on the PCB, but dropped the decimal I guess on the schemo.  One final not-quite a typo is R8, which can be adjusted to ensure that the LED is completely off when it's supposed to be off. Exact value will depend on the JFET.  RGs default is 4.7K there but can vary widely.

I was aware of the biasing issue with the germanium transistors, the values for R5 and R1 in my schematic are just default values. I will have to try biasing at the higher voltage as you suggest, just to check out the effect. 

[PRR]

Keep power-garbage grounds segregated from audio grounds. Power-garbage common should all come together so that the huge current-spikes are localized. Then the integrated power common goes to your "Star" via one path.

C4 and C6 are redundant, yet dump spiky ground current through different paths in ratio 10:47, the 47 through audio ground path.

C4 is the gross filtering, C6 may be the fine filtering, with a resistor in the power rail so each knows its job. Trace from C4 to 470 radiates hash so should be short and away from audio.

LED is very much a power load and will click. It should source from bulk-filter C4 and return to the power-section common, NOT through audio common.

There probably should be a cap AT the input to the power converter, especially if fed through contacts or transistor. The datasheet might have some thoughts. I'd certainly put pads for a 10uFd cap.

[Paul Fawcett]

Thank you Paul, some good constructive advice there!  I'm not sure why I couldn't see it myself, but it all makes perfect sense now.

Let's see if I can summarize this to confirm my understanding:

The key insight here is that the local star for the power ground should NOT be the input shield jack, but rather a local power section star should just reference back there as shown.   Now it's clear why the transistor switching helps.... when the star ground was at the input jack sleeve as shown originally, I don't think it would help much, because the power ground would have a shared impedance with the audio ground at that point regardless of whether transistor switching was used.   But with a separate power star the transistor switching ensures that there is no longer a shared impedance between the audio current and the full (noisy) power currents -- only between the audio current and the base current of the transistor (which, of necessity, must be routed through the ring/shield of the jack for the scheme to work).

I also hadn't considered that the LED would be a problem. I'm assuming that there is only an issue with a click when switching, as I can't imagine a noise issue with just a steady state current through the LED.   I wonder...rather than completely redesigning the board, if it might be possible to ameliorate LED switching noise by bypassing the LED with a few tens of nanofarads of capacitance?

As for the 470R resistor, I'm assuming its job being just to put an impedance in the path to isolate the grounds...  But wouldn't almost any value would do, even only a few tens of ohms? If the germanium transistor is biased for 1mA of current (I haven't calculated this, just a top of the head SWAG), you're looking at an almost half volt drop across a 470R part - seems like rather more than would be desirable when you're only starting with 9V to begin with?

Thanks,

Paul

[PRR]

> LED ... only an issue with a click when switching

Yes. But see the many-many threads about popping. Not all are due to LED. But the LED current is sometimes 10X the audio-systems current. And off/on instead of steady.

I just noticed... WHY is the LED fed from the power converter?? It will light the same off the raw +9V input supply. Inverting is not just a waste of converter capacity, it probably increases the hash out of the converter. Yes, the-other-way means inverting the polarity-sense of your LED switching, but it sure seems cleaner.

> completely redesigning the board

IMHO, un-tried circuits should be prototyped on something accomodating. Perf, etc.

I also have opinion about modularity and appropriate techniques. The basic Ge-amp stage could be wired on a 5-lug strip; many-many were back in the day. There's not much to the LED system. The converter does work better on perf-board. But the connections between the three sub-systems... the "cost" of running wires is small compared to the "cost" of PCB and the real cost of possibly re-working the PCB. But to each his own.

> the 470R resistor... wouldn't almost any value would do, even only a few tens of ohms?

If a 9V battery circuit is upset by a 0.5V drop, it's much too fussy for a real world. That's why there's twice as many resistors as absolutely necessary.

While the circuit is supposed to be self-stable bias, musicians typically tinker the bias; any small supply change would be included in this tweak.

Yah, I suppose 10r/47uFd is 300Hz so would have "large" impedance at converter frequency. If converter hash is a problem, I'd be inclined to increase this resistor (to cut hash current) until some other parameter (bias, headroom) got "bad", then split the difference. Of course you may tack-in various values and see what you hear.

[Paul Fawcett]

OK thanks, I've reworked the PCB to reflect most of your advice. Feeding the LED from the raw Vin+ makes a lot of sense, but I think I'll try and get away with it here, since the circuit otherwise draws so little current.  I'll do it as you suggest next time around.  Tomorrow I'll try and etch this board:





Incidentally, this whole project is intended as an excuse to gain experience with PCB layout and etching, so I'm not particularly concerned if it doesn't come out quite right the first time.  I guess these things are indeed a matter of perspective. Compared to my typical tube amp project the time commitment seems small, and the components seem practically free by comparison... and so far lots of fun! Thanks again.   I;ll let you know how it goes.

[Paul Fawcett]

Thanks everyone for the help on this one!

I finally had a chance to complete this today, and it works great.  No hum, no buzz, and the millennium bypass circuit works really well with the 2N5460 - there is no perceptible delay in the LED lighting.  Sounds good too.

I'll post a wrap-up when I get a chance, but in the meantime, please note that in the images shown in the post above, the 2N5087 on the ring circuit is shown backwards, as is the 10uF cap directly above the Max1044.  Finally, the OC44 (or PNP equivalent in a TO-1 can) should be oriented with the protruding base leg closest to the right-hand side of the PCB drawing.

My first fully completed pedal, thanks guys!  Now just waiting on the 3PDT switch to arrive so I can the the ABY box going properly.

What next, I wonder?

[boogietone]

...

C4 is the gross filtering, C6 may be the fine filtering, with a resistor in the power rail so each knows its job. Trace from C4 to 470 radiates hash so should be short and away from audio.

...

Could you explain this a bit more, in particular the point about "radiating hash."

Thanks.