Darth-Fader

[barleymow74]

I'm looking to build something similar to a GigRig Wetter Box for my pedal board. I've assembled a simple buffered splitter (FuzzDog Spluffer) which will send the signal to two different pre-amps. I then want to create a pedal that allows me to select between Pre-amp A or Pre-Amp B or a blend of the two (100:0% to 0:100%), which can be adjusted with my feet. I've done a bit of research and put together the following schematic which contains the following stages:
1.   A buffered input with  a relatively small gain so that I can adjust/balance the level of each channel. I've used a schematic of a micro amp as a starting point but made a few tweaks.
2.   A phase inverter on one channel based on a RG Keen circuit on Geofex.
3.   A cross fader (or Panner) again based on a RG Keen article on Geofex. It's also used in my amp on the effect return input to combine the wet and dry signals.
4.   A summing amplifier to blend the two signals.
5.   Two footswitches, one that selects blend or single pre-amp, if the latter than the second footswitch selects which pre-amp, A or B.

I've worked through the calculations provided in the above resources to hopefully come up with some suitable values for resistors and capacitors, but I'm no electrical engineer and have only a basic understanding of the sizing and location of different capacitors in particular. As I understand it, the input stage will provide a gain of ~1.75 to 8. The x-fader should hopefully reduce this by a factor of 2 with the resistances selected (R13-R17).

I've breadboarded stages 1 to 3 for a single input and it seems to be doing something sensible but I haven't done the full schematic. Before I commit to building the whole thing, I thought I'd see if anyone was willing to point out any issues/blunders or propose any mods that might make it more robust, stable etc.

Any suggestions would be appreciated, but this is my first post so please be gentle!

If you need any further information just let me know.

Thanks!

[PRR]

Welcome.

I don't see a problem. (Yet?)

[barleymow74]

Thanks Paul!

I've finally managed to complete the circuit and it is largely working well (after a fair amount of debugging  ).

I've been testing it with an oscilloscope and the only issue I haven't been able to resolve relates to the polarity/phase inverting part of the circuit. I've attached a few screenshots below to illustrate the problem I'm having.

Screenshot 1 shows the circuit working well with the phase switch closed. The blue trace is Input A, a 1 kHz sine with an amplitude of 500mV. Green is the output (with the gain control set to match the input). Red is the output of the phase inverting op amp after the decoupling capacitor (C4_8).


Screenshot 2 shows the issue when the phase switch is opened.


Looking more closely at the phase inverting op amp, Screenshot 3 shows the non-inverting input (blue), the inverting input (red) and the output (green).


The inverting input seems to deviate from a pure sine in the troughs. I'm presuming that it is this difference between the two inputs that then gets exaggerated at the output.

The DC offset and peak-to-peak values are also shown at the bottom. Shouldn't the DC offset be ~4.5v rather than 2.7V?

I've attached the latest iteration of the circuit below. Any thoughts on possible causes or solutions?!

Thanks!

[PRR]

Shouldn't the DC offset be ~4.5v rather than 2.7V?

I would think so. What is the actual voltage on the Zener? How teeny is the Zener current with 500k drop resistor?

Seems like the Zener wants more than a milliAmp to get happy.

What if you made the drop resistor 5k, or 2k?

Or lose the Zener and use two reasonable resistors as a divider. You don't want 4.50000V, you want about "half supply". When your battery sags to 6V, or runs wild at 18V, the divider is more right than a Zener.

Even so, your waves are >3V peak to peak, which is over 1 Volt of audio, which is BIG in guitar world.

[ElectricDruid]

+1 what Paul said. Take the 500K and Zener out and put a divider with a couple of 10Ks and a 10/22/47u across the lower one in instead.

In general the whole thing looks well-designed, like someone thought about it (because you did!). My only other comment would be that I'd take a factor of ten out of the two non-inverting "Return" op-amps. Values that large will start to get noisy. Going down to 47K, 10K, with 100K Rev-log pot and 100p/1n caps will do better.

[marcelomd]

Looks neat.

I think you need a current limiting resistor for the LEDs, no?

On a purely stylistic suggestion, I'd use multiple ground symbols, instead of a ground line snaking around the schematic. Same for vref.

[barleymow74]

Finally, FINALLY, cracked it!

I went back to the breadboard and played around with a few components. The issue raised previously appears to have been caused by the 1M resistor on the input to the phase inverter section. Reducing this to 100k seems to have solved the issue.

I thought I was there and boxed it all up, only to plug it in to my amp and discover major footswitch popping noise. Back to the breadboard I went and discovered some significant DC levels through the fader section. Reducing the 1u capacitor at the summing point to 470n dealt with the DC offset and with it the popping sound.

I've also added in the resistors for the LEDs (thanks Marcelomd), and hopefully simplified the layout style. Final schematic below along with a signal analysis for anyone that's interested.




Thanks for the earlier feedback. Against all advice I stuck with the Zener diode! One of the many, many rabbit holes I disappeared down during this project was due to concerns that fluctuations in the power supply were affecting my signal. I found this application note on 'Biasing and Decoupling Op Amps in Single Supply Applications' very helpful and concluded that a Zener was a more robust, stable solution (providing an appropriate current limiting resistor is chosen!). In the end I went with 330 ohms which should limit the current to 13 mA. The Zener I used was rated at 4.7V @ 5mA, leaving 8 mA of load current, which I'm hoping should be plenty. Even if the load current is a fraction of this, and more goes through the Zener, it can cope with up to 100 mA so well below its limit. I know its not conventional but as a newbie, I'm a bit surprised why Zener's aren't more widely used?

With regards to the high resistances I chose, the article above also advised on resistor sizing to minimise input bias current errors. Figure 4 states that R2 should equal Rin for a non-inverting setup. In my circuit I wanted Rin to be 1M so that I had a high input impedance, so went large on the feedback resistor too, meaning I then needed a big pot to get the gain I wanted. Fortunately I'm not getting any noise issues so have left it but again I appreciate it isn't the wisest choice.

A couple of minor niggles still but I can live with them. There's a pop when powering on and the LED bar display doesn't perfectly align i.e. it is above 50% when the two signals are balanced. I'm presuming this is to do with some asymmetry in the resistance and maybe flatspots at the start or end of the rotation?

Below are a couple of photos of the finished build. I think its fair to say I got a bit carried away with the Darth Vader theme! Its looking good on my pedal board though and working well.

[ElectricDruid]

Lol, looks fantastic  Nice work.

A couple of comments just to explain a bit of the background:

What you have read about input bias errors is all true. The question is whether it matters. Your op-amp input stages are followed immediately by a 1u cap (C7 / C9) so any DC offset at the op-amp output never reaches the next stage. So unless the offset is big enough to heavily affect the headroom, there's no real effect. Consequently, audio designs often prioritise noise performance over offset performance. Other applications like sensor reading where you have small DC voltage inputs would take the alternate view.

The zener bias is going to reliably give you the stated voltage until the supply drops so low that it can no longer do that. So it's "stable" in that sense. But if the supply voltage changes away from 9V, it is not longer a midpoint voltage, so the biasing gets *worse*. The use of a voltage divider in pedals probably stems from the days when battery power was common, and a pedal needed to work on a fresh PP3's voltage of over 9V down to whatever point the circuit dies, say 6 or 7V. The advantage of the divider is it continues to provide a midpoint voltage across that whole range. Indeed, with many op-amp-based pedal designs, you can change the supply voltage from 9V to 12V to 18V with no modifications at all and the thing works perfectly because the biasing scales to the supply.

Of course, we're not using old blue Everready's to power our pedals any more, so this original reason is far less relevant than it once was.

[duck_arse]

I'm a bit surprised why Zener's aren't more widely used?

calculations are required. who wants to do that?

Darth's head - do you step on it? do you need to use much force?


in the old days when zeners were a thing, we used to put loops in the leads, neatly, with pliers, close to the body, as additional heatsinking and thermal stress relief. but - look the interwires for a picture - 'zener diode lead loops' ---- crickets chirrping. maybe that radio museum site, but I forgot to look before posting.

[PRR]

'zener diode lead loops'

AI has several wrong answers:

Zener diodes are commonly designed with axial leads, which are straight, cylindrical wires used to connect the diode to the circuit. These leads are often bent or looped to facilitate soldering and mounting on a circuit board or in a circuit. The purpose of looping the leads is to provide mechanical stability and prevent accidental damage to the diode during handling and soldering.
Here's a more detailed explanation: [1, 2] 

• Axial Leads: Zener diodes with axial leads have two wires, one for the anode and one for the cathode, extending from the diode's body.
• Looping for Stability: When the diode is mounted on a circuit board, the leads are often bent or looped to create a stable connection.
• Mechanical Protection: The loops can help protect the diode from damage or accidental movement, especially during soldering or if the circuit is subjected to vibrations.
• Soldering: Looping the leads allows for easier soldering and ensures a good electrical connection.
• Mounting: The loops can be used to mount the diode to a chassis or panel using screws or other fasteners.

Generative AI is experimental.

EEVblog has a long discussion:
https://www.eevblog.com/forum/vintage-computing/was-this-a-thing-looping-components-on-vintage-assembly/

[marcelomd]

Finally, FINALLY, cracked it!

...

Wow, it looks so cool!

[barleymow74]

Lol, looks fantastic  Nice work.

Thanks! It definitely turned out better than I expected.

So unless the offset is big enough to heavily affect the headroom, there's no real effect. Consequently, audio designs often prioritise noise performance over offset performance.

Makes sense. I was battling with DC offsets for a long time causing  my opamp outputs to max out so I followed any tips I could find. In the end, better choice of capacitors probably sorted the problem out so I could look to reduce the values again should noise creep in.

The use of a voltage divider in pedals probably stems from the days when battery power was common, and a pedal needed to work on a fresh PP3's voltage of over 9V down to whatever point the circuit dies, say 6 or 7V. The advantage of the divider is it continues to provide a midpoint voltage across that whole range.

I think I get what you are saying. I guess, for me, running it off a battery was never a consideration and the stable voltage was much more appealing.

[barleymow74]

Darth's head - do you step on it? do you need to use much force?

You can operate it in two ways:
1. Put your foot on the top of the dial and rotate it. This is the easiest and gives more control but is better for finer adjustments.
2. Slide the side of your foot down the side of the dial. The red silicon is pretty grippy but it takes a bit of practice and is better with shoes on! You can make bigger/faster changes this way.

It doesn't require much force but I was concerned that I might apply more than a pot could cope with if I mounted the dial directly on it. Therefore I mounted the dial to a shaft with a bearing to support it and the rotation is transferred to the pot using pulley wheels and a timing belt. It also meant I could reduce the amount of rotation required by going from 20T to 16T on the pulley wheels.

in the old days when zeners were a thing, we used to put loops in the leads, neatly, with pliers, close to the body, as additional heatsinking and thermal stress relief

I've never seen this before but I might try it next time! Looks cool and could be practical. I also like the idea from the EEVblog of using it where you might want to attach a probe.

[duck_arse]

EEVblog has a long discussion:

hmmm. pity the poor modern engineer, some of those answers [and anything from ai has no idea, evidently]. well, obvs, no leads on modern parts.

here, see, an ex member build with twin loops both ends. but on a 1N34, so I don't think he knows right either.