ABY switching in a Fender 5E3

[Stephen1966]

ABY switching in a Fender 5E3

I LOVE the Tweedle Dee, and the amp that has evolved out of it for me, something I am now calling The Victor. I now have a great sounding bright channel and a great sounding normal channel and there's an elegant (mechanical) simplicity and logic to Fender's 4 jack input but how many times have we heard Deluxe players saying "I prefer the Bright channel" or something along those lines? Maybe if it wasn't such a **** **** to switch channels in the middle of a performance we wouldn't so often hear the impression that it is just a one trick pony.

What if, there was a way to switch between the channels, and a way to mix them all without having to drop everything, go over to the amp and step back into a '50s telephone exchange with physical jumpering. An ABY channel switcher of course, but even the idea of having two outbound cables from a floor mounted switch is not good enough for me and so to keep the signal in the amp where it belongs I took a leaf out of Dumble's book and came up with an internal, relay driven ABY switching based on a regular passive ABY pedal design.

DipTrace Schematic - ABY relay circuit.jpg

I haven't tried building this yet but the logic is okay I think. There's a cap straight after the input because this is part of a preliminary design for a circuit that utilises VVR and the cap is to block DC. There's also a pair of 22M resistors on the separate A and B channels to suppress pops like MrD did with the PAB relay. The Hi/Lo attenuators right before the grids (of V1a and V1b) are identical to the Fender jack switching but using manual DPDT switches. These are unlikely to be switched during a performance I think because it depends on the output of the guitar - I think we only need them when we switch the guitar but you nerds will straightway see that you can control these from relays as well.

The coil circuit is likely going to require a separate 12V supply and it borrows it's topology from the ODS etc. Using a BJT transistor net on the "anode side" of the coil's diode is a way to suppress switching noise and a relative no-brainer but a problem I want to address straight off from the ground up, is switching noise on the audio channels. On the one hand, placing the switching net early in the signal chain is a good thing because the currents are so small, but on the other hand, every small signal artifact is going to be amplified by the subsequent amplification stages so even tiny glitches, pops and crackles are going to become apparent.

The 22M resistors are likely sufficient for the switching of individual A and B audio channels but I have my doubts about the A+B switch when the signal gets summed in the parallel triodes, then mixed in the next gain stage, V2a. Merlin is my go to guy for these kind of things so I am curious what you think about series-shunt switching?

Optocouplers on the COM1 and COM2 lines out of the A+B (Y) relay might be the answer even though Merlin tends to avoid them, voltages rarely surpass 1V and current is very small. Meanwhile, solid state relays such as the PVT412APBF (https://cz.mouser.com/ProductDetail/Inf ... fL5Q%3D%3D) might be used - in conjunction with a 3PDT relay for shunt switching on the line outs from the plates (but after the coupling caps) of V1.

Or perhaps JFETS as switches for the audio, offer a more controllable offness with the ability to control the timing of the fast fade?

I am truly curious to hear what you consider the optimum solution for this kind of context.

[cdemike]

Interesting idea! Seems to me that the switching arrangement you posted should work fine. Since there's good ground referencing throughout, I doubt switching noise/pop would be too bad. My big question actually would have less to do with the actual switching arrangement and more to do with ensuring choices A, B, and C sound good (or are at least adjustable). I can't speak for everyone, but I'd rather see more panel space dedicated to that over high/low input switches, especially since I can essentially get the same effect as a low input by rolling down my guitar's volume. Because the interactivity is sort of a zero-sum thing in terms of EQ (for example, adjustments making one channel sound brighter when not jumpered generally will make the other channel darker), I'd more likely be interested in switching between tone stacks that individually tune-able. You could get a similar effect on gain as jumpering by making the first gain stage a switchable single or parallel triode, but I can't say I'd have interest in more than two settings on a traditional two-volume setup: In most cases, I generally would only use the bright channel on its own and sometimes the bright channel with a small amount of normal added.

I'm building a similar amp now basically on that principle (switchable 6G2 and 6G4 tone stacks with separate volume controls), so it's possible I'm just projecting...

[Stephen1966]

Interesting idea! Seems to me that the switching arrangement you posted should work fine. Since there's good ground referencing throughout, I doubt switching noise/pop would be too bad. My big question actually would have less to do with the actual switching arrangement and more to do with ensuring choices A, B, and C sound good (or are at least adjustable). I can't speak for everyone, but I'd rather see more panel space dedicated to that over high/low input switches, especially since I can essentially get the same effect as a low input by rolling down my guitar's volume. Because the interactivity is sort of a zero-sum thing in terms of EQ (for example, adjustments making one channel sound brighter when not jumpered generally will make the other channel darker), I'd more likely be interested in switching between tone stacks that individually tune-able. You could get a similar effect on gain as jumpering by making the first gain stage a switchable single or parallel triode, but I can't say I'd have interest in more than two settings on a traditional two-volume setup: In most cases, I generally would only use the bright channel on its own and sometimes the bright channel with a small amount of normal added.

I'm building a similar amp now basically on that principle (switchable 6G2 and 6G4 tone stacks with separate volume controls), so it's possible I'm just projecting...

Thanks. I try to look at these things from a practical viewpoint. The manner and methods of switching are what are keeping me up at night at the moment but I hear your other points too.

The 5E3 is an odd beast with its bipolar tone control. Sluckey did a great job of explaining 'The 5E3 Tone Stack' over on the Hoffmann Amp forum https://el34world.com/Forum/index.php?P ... 2#msg95202. It does give you a kind of compromise when switching channels and some may be more than happy with that because the tone control on my amp at least, lives almost permanently at the 12 o'clock position. Bright to dark is not always a bad thing and there are always the volume controls which can offer a lot of control over the amount of dirt you get in the standard configuration. I too, like to control those with the guitar volume, it's really a great way to exploit a lot of different voicings but I have met players who don't know what the guitar volume is for and leave it cranked. Fair enough I suppose, we can lead the horse to water and all that...

The ABY switching in the schematic above is probably not going to be built because I don't like the idea of the signal swirling around a pcb on its path to the grids. The point where the signal penetrates the amp is the most sensitive to noise, cable runs need to be short and as direct as possible. The Hi/Lo switching isn't actually going to take up that much real estate. The switch can be a simple SPST push button that drives a solid state relay placed with the grid stoppers, near the grids. Again, it's about keeping the signal path as short as possible and this is just one way to do it. Of course, a mechanical DPDT switch is a simpler option. The schematic above provides the ABY switching with series switching, interrupting the signal flow and increasing the risk of loud transients switching from one channel to the next, as well as forming an RC filter which doesn't do our tone any good. So now, I am rethinking the design with parallel shunt switching in mind sending the signal to ground before the grid stopper and after the coupling cap off the plate at the same time. I trust Merlin's logic that the signal to ground before the grid sends the bulk of the signal out of circuit and the one after the coupling cap sends anything that remains, resulting in a much cleaner state of off (offness) with no distortion or crosstalk. I think this is a better way to ensure the very minimum of Johnson noise as well.

The principle of bisecting a portion of the circuit at its top and tail might also be implemented with different tonestacks. Something, very much like the switching of the boost channel seen in the Marsall, 4210. Earlier models used BJTs (BC184), later an op-amp (CA3046) - if I'm not mistaken it's a two-state solution.

[Stephen1966]

There are no authorities, at best there are only experts.

From the Baloney Detector Toolkit, Carl Sagan.

As much as I respect Merlin Blencowe's work and opinions, it's important not to become blinkered to the universe of alternate possibilities and opinions out there.

I am still processing a lot of this but in the meantime, here's an interesting site with informative articles: https://sound-au.com/articles/muting.html

[Stephen1966]

Something along the lines of a feasibility study, I drew up the schematic for an onboard solid state ABY mixer for the 5E3/Tweedle Dee. I don't have anything against solid state per se but it's easy to suck the tone right out of an amp circuit with it and it needs careful handling.

The ABY circuit I've pinned here meets all criteria for a two-channel, high and low 4-input amp using a single jack, though there are still two separate jacks on the BRIGHT and NORMAL channels so that they can be operated independently. Plugging in the MIX jack puts voltage into the external footswitch circuit, activating it; plug into a channel jack and the footswitch is bypassed, obviating the need for manual switches on the back of the chassis. Most of the lines/traces you see here are part of the voltage control, switching circuit, the actual signal comes in at the jack and passes through the first optocoupler (the series switch), the tail end of the signal and any small artifacts of the signal that remain are then further attenuated by the second optocoupler (the shunt switch). This is series-shunt switching. Older circuits used the Vactrol™ VTL-5C4 optocouplers/LDRs and appear harder to obtain these days but the NSL-32SR3s are very close and are in current production.

The LBB120s are 2 Form B (DPST-NC) switches and control the channel switching, while the LCC110 are 1 Form C (SPDT-NO, NC) offering the antiphase switching that ESPs circuit using BJTs offered in a single DIP-8 package. The SSRs are not as power hungry as electromagnetic relays, 10mA versus 40mA on average, so I haven't worked out how much all these relays cost in terms of their current requirements yet, but it will be low and achievable with a modest PCB transformer sufficing for a separate 12V supply.

Project X-5_Page_1.jpg

The preamp section is very close to the Tweedle Dee circuit with split cathodes and smaller coupling caps in the parallel triodes. Ingenious readers with a passing knowledge of Marshalls and Hiwatts similar to my own will see straightaway, there are other ways to configure this ABY circuit. The electromagnetic relays here are for the cleanest possible signal. SSRs have unacceptable levels of capacitance across their switch input-output. Contact capacitance in the AZ850 is just 0.4pF while in the LBB120 it can be as much as 50pF, that makes the latter unsuitable for audio signals because of the filtering they introduce. Optocouplers also offer distortion free rise and fall times though so briefly they are hardly likely to be noticed.

It should be noted, the Hi/Lo switching is meant to be controlled with a separate, independent switch on the front panel (not the footswitch) because in my experience the Hi (unattenuated) or Lo (attenuated) signal is not a selection we generally make on the fly even if it were possible. There may be a "jumper" setting that you know is good or there may be an advantage at setting it for guitars with different pickups but these are factors which are often determined before the song starts. Front panel control for this is good enough! Plenty good enough.

An area of investigation that will be explored is the benefit of using the BJTs in the SSR relay circuit. My current level of thinking is that the relay circuit is better with them because of the fraction of current the BJTs need to drive them into their closed, saturated state. It isn't clear yet how much noise we can expect from a jack with a powered up switch inside it but my instinct here is that the BJT attenuates that noise to a greater extent. When I breadboard this circuit, I will put the scope on this and see how it looks. Anything that minimises the interaction between the switching DC signals and signal strength AC signals sounds like a winner to me.

Project X-5_Page_2.jpg

The CMOS flip flop IC at the heart of the footswitch is the same one mentioned in Merlin's book and I've expanded on the circuit to include the three SSRs. These act as gates to the COM line. That is distinct from the other GND lines and all the COM lines lead back to the secondary base of the 12V transformer. They are NOT the same as GND or Earth. The flip flop provides ample output current to drive the switching of the BJT's which in turn, close the circuit on the SSRs which drives the SSR as well as their indicator LEDs and thus open the gate for the channel control voltage to COM, closing their circuit and activating the channel in the amp.

I like this footswitch for a couple of reasons: first, the momentary switches are lighter of touch than mechanical, latching switches and virtually noise free. Second they offer a single switch operation, that is, when you flip, you also flop! Switch to channel A, all other channels are deactivated, switch to channel B, all other channels are deactivated, switch A+B both channels are activated. You don't need to deactivate one channel before switching into the next. Many pedals have been doing this for years, it isn't unreasonable to expect the same intuitive, ease of use from our own projects.

Project X-5_Page_3.jpg

A nice summer project for me now is to breadboard all of this and see how the theoretical performs in actuality. I am already thinking however, that the footswitch is something with real potential for a number of other pedal applications and might even be combined with the existing old-school tech for an enhanced design. It doesn't need an onboard DC supply so it might easily be incorporated into a number of ODS projects.

[Stephen1966]

I slipped up with the form B relays in the previous post, the circuit should use form A (normally open until energised)... moving on.

Again, the channel control circuit has very little interaction with the audio signal which uses optocouplers with a low on-resistance. The optocouplers themselves require a low voltage but a relatively high current, about 25mA which far exceeds the control current the flip-flop can deliver. The simple answer is to use BJTs as switches. Average BJTs like the 2N2222A are more than adequate for the task. The textbook approach is figure out how much collector current they are going to need to handle and then calculate the base current and base resistor value to drive the transistor into saturation. The calculations are not critical, and we don't have to be too precise with the base resistor calculation. The minimum current required is typically calculated as the Collector current divided by the beta, with the beta being synonymous with the hfe. If we say most npn transistors in this class have a hfe of 100, the minimum current required of the base for saturation is 100th of the collector current it is expected to serve. That's a minimum though, and to ensure a fully saturated and pardon the expression, a proper hard "on", the rule of thumb is to divide the collector current by 10 and call that your required base current. Some say you can add 30-40% more base current to that for certainty but it starts to get very wasteful on power and there is an increased risk it will fry the transistor. I've used a beta of 10 in this test, which I've written as beta prime.

ABY transistor calculations.jpg

In order to verify the circuit I built a couple of breadboards with simple LEDs representing the loads.
This works fine when the circuit is configured either as a Common Collector

IMG_20240705_091950.jpg

Or configured as a Common Emitter circuit

IMG_20240705_102938.jpg

The schematic showing the discrete components and connections shows R1 as 4.7k which is a wildcard value thrown in there to offer a little ballast. If we wanted to be precise about it, we might make it the equivalent of the two base resistors R6 and R7 in parallel, but again, it works fine in this configuration demonstrating that you don't need to too picky about the component values.

Schematic.jpg

Q1 (Channel A), Q2 (Channel B) and Q3 (both channels) control the activation of the leds, applying 12VDC to Q1 and Q2 activates their respective leds (channels) separately whereas apply the voltage to Q3 and both channels open. The current being pulled is only a couple of milliamps in this test, but when the circuit is refined for the logic level control it's relatively trivial to adjust the values so that they don't pull more than 1mA.

The common emitter circuit is the configuration that works best between switching individually or together with no visible dimming or brightening of the leds across the change. The same cannot be said for the common collector circuit which did exhibit dimming and where the voltage across the collector-emitter is significantly higher than the common emitter configuration which at close to zero volts represents an improved state of on-ness, or saturation.