ESP Logo
 Elliott Sound ProductsProject 219 

Speaker Switch For Valve Amplifiers

Copyright © October 2021, Rod Elliott


There often a requirement by guitarists to be able to switch their amp from one speaker to another, most often the internal speaker of a combo amp to an external speaker box.  This is dead easy with a transistor amp, because they don't care at all if the output is open circuit, but valve (vacuum tube) amps are placed at some risk.  This is because no switch or relay offers a 'make-before-break' function.  Relays used to be available with this function, but they're now obsolete.

With any switch or relay, it takes up to 6ms (milliseconds) before the moving contact has changed from one fixed contact to the other, and I verified this with a number of different relays, as well as a standard push-on, push-off footswitch.  6ms isn't very long, but if an amp is being pushed hard (well into distortion) when the switch is activated, the output is completely disconnected during that time.  Provided you never operate the switch while playing a note or chord, no harm will be done, but everyone knows that at some point it will happen.

When the output of a valve amp is open-circuited, very high voltages can be generated by the output transformer, which can cause 'flash-over' between valve base pins, or at worst, a damaged output transformer.  Damage isn't guaranteed of course, so you might find that a simple footswitch works fine ... until it doesn't.  Many valve guitar amps can produce spike voltages of anything up to about 3kV (3,000V, sometimes more) with an open-circuit output condition, and if you think that can't be good, you're quite right!  Some valve amps have protection diodes to minimise the risk of damage, but a great many do not.

fig 1
Figure 1 - Test Circuit To Measure 'All Contacts Open' Time

The circuit shown was my test setup, and while the moving contact (connected to the relay's armature) traverses from the NC to the NO contacts (and vice versa), there can be no output across R1.  A captured waveform is shown next, and it's unambiguous - there is an easily measured time when both contacts are open-circuit.  The 'oscillations' you can see in the trace are caused by contact bounce, and all relays and switches are similarly afflicted.  The waveform changes with every operation, but the time between a 'solid' connection for one output or the other is generally quite consistent.  As you can see in the next image, the 'all-off' condition lasts for 4-5 milliseconds.  Mouse over the scope trace images to see the full resolution.

fig 2a
fig 2b
Figure 2 - Switching Relay - 'All Contacts Open' (Activate Left, Deactivate Right)

The above shows the period during the switching interval.  A make-before-break relay would show a straight line, but it's anything but.  This is the 'danger zone' for a valve amp, where the speaker load is disconnected (then connected, then disconnected again) as the relay changes from one set of contacts to the other.  The traces are shown above for reference, and after testing many times (and with a number of relays, plus a push-on, push-off footswitch) similar results are apparent with all.  The 'break' time as the moving contact changes from one set of fixed contact to the other is surprisingly consistent, at around 4ms (4 milliseconds), but other relays will be different.  However, it's not just about the 'contacts-open' time, because all relays take at least 2ms to activate, and up to 5ms (sometimes more) to deactivate, depending on the magnetic circuit.

fig 2a
fig 2b
Figure 3 - 'Normally Open' Contact Operation Time (Activate Left, Deactivate Right)

The relay used for this was the one shown below, and these were the most definitive traces I obtained.  These are not 'worst-case' results though, but are 'typical'.  The time delay between application/ removal of DC to the coil is also of interest, and these are shown above.  The violet trace is the coil voltage, and the relay release time is almost 7ms.  This is due (almost) entirely to the diode in parallel with the coil.  There is little consistency with contact bounce (rapid opening and closing of the contacts).  Every test I did showed different results, particularly as the relay deactivates.

fig 4
Figure 4 - Switching Relay

The relay used for testing was the same as that shown above.  This is a very common relay, and is the same one used in the Project 39 soft-start unit.  The contacts are rated for a maximum of 10A at 250V AC.  You may be mystified by the rating being reduced to 3A at 240V COSΦ0.4.  This shows that the relay must be derated when the load has a poor power factor (in this case, either capacitive or reactive).  While a loudspeaker is reactive at some frequencies, because the voltage is relatively low no issues will be encountered (40V RMS at most, and that's for 400W into 4Ω).  If you happen to want to know more about power factor, see Power Factor - The Reality (Or What Is Power Factor And Why Is It Important?).

This style of relay is low-cost (less than AU$3.00 each) and readily available from most suppliers.  The contact arrangement is SPDT - single-pole, double-throw, meaning it has a changeover function (aka 1-Form-C, one set of contacts, changeover).  The coil resistance varies from around 200 to 270Ω for 12V versions, so the maximum coil current is 60mA.  Since two are used, the total current is up to 120mA which is easily handled by simple circuitry.

To eliminate the 'both contacts open' condition, we'll wire a pair of these relays to create a make-before-break circuit.  It should come as no surprise that this will involve some electronics.  That's what this project is about - providing a circuit that will ensure that there is no period when the amp's output is disconnected (when set for 'Valve' operation).

Circuit Description

To obtain a true 'make-before-break' switch we need two relays.  One operates immediately, and the other is timed so that it won't change state until the first has established a solid connection.  It's not just the contacts, as relays also have the turn-on, turn-off delay, and for an acceptable safety margin my suggestion is nothing less than 25ms.  At varying times during this period both speaker cabinets will be in parallel, but that won't cause any issues with a valve guitar amp.  However, if you use a transistor amp, that may cause the amp some stress.  Consequently, the project design includes a switch that lets you change from 'Valve' to 'Transistor' operation.  The switch has to be able to carry the full output current (up to 5A for a 100W/ 4Ω amplifier).

In addition, there's provision for a remote footswitch (push-on, push-off) so you don't need speaker cables all over the floor in front of you while playing.  The unit needs a 12V power source, which is most easily obtained with a 'plug-pack/ wall-wart' style power supply.  The current drain is modest (a maximum of perhaps 250mA), so the wall supply doesn't need to be large or expensive.  The circuit will default to the 'Main' output if there's no power available (or it gets disconnected by accident).

The circuit is straightforward, with one relay operated immediately, and the other is delayed for about 60-70ms.  During this transition period, there will be times when both speakers are connected in parallel, but this doesn't cause a problem for a valve amp.  The output valves are overloaded for ~50ms, but that does them no harm.  When set for 'Solid-State' operation, the delayed relay is disconnected, as transistor amps are more likely to be damaged if the load impedance is too low.  Damage can occur in a couple of milliseconds (or even less) in some cases, but an open-circuit will not cause damage.  Of course, any amp can be switched if there's no signal, but as noted above, at some point it will happen (Murphy's law can never be underestimated).

fig 5
Figure 5 - Circuit For Speaker Switch (Valve/ Transistor Version)

The circuit uses a small-signal MOSFET (Q2) to provide the delay.  R4 and R5 are used to get roughly equal timing when voltage is applied or removed from the control line, with each optimised by a diode.  The control voltage is also isolated by a diode to drive the 'instant' relay, because some voltage persists across the relay coil after power is removed.  The added delay is small but unpredictable, as it depends on the relay itself.  With the component values shown, the delay before RL2 operates is about 70ms, as it needs to be long enough to ensure that RL1 has activated (or de-activated) before RL2 activates (or de-activates).  During this overlap period, both speakers are connected, but the time is short enough that even the most stressed valve circuit won't be damaged.

You don't need to be too 'precious' with the values for R4 and R5.  Although I aimed for a delay of about 60-70ms, anything greater than 30ms should be perfectly alright.  It doesn't matter if the timing is different either, provided it's long enough (in both directions) to ensure that there's enough overlap - the period when the two relays are in opposite states.  When I ran workshop tests, I just grabbed a couple of resistors that were 'about right', and was never able to trick the circuit into an 'all contacts open' condition (and I did try hard to make it misbehave, but it refused).  If you can hear the relays activating and deactivating with a definite 'click-click' it should be ok.  If it sounds like they operate at the same time, the timing is too short.

note Important:  When using the switch unit, always ensure that both speakers are connected.  Verify that both work by testing the switch box at low volume (as in very low) to prevent any possible amplifier damage.  This is essential every time you set up using the switch box, and all the precautions taken by the switching circuits to avoid an open-circuit output are for naught if you fail to verify that everything is working properly before using it at high levels.  If one of your speakers (or speaker leads) is not connected or faulty, damage is very likely with a valve amp.

The remote footswitch requires some extra switching to allow the use of a standard guitar lead.  A 'simpler' method would end up being more complex, because you'd need to use an isolated jack socket, which is a nuisance, and it's more trouble mechanically (even if simpler electrically).  An isolated socket could also cause issues if the remote plug (and/ or box) were to come into contact with other pedals.  The on-board switch must be open, and shorting the remote connection (R1) to ground turns on Q1 which bypasses the on-board switch.  Everything else works in exactly the same way, and it's only the provision of power to the relays that changes.  When using the remote, just make sure that the 'Auxiliary' LED is off before connecting the remote cable.

Operation is simple.  When the footswitch is operated, RL1 acts (more-or-less) instantly, and connects the amp's output to the second (Auxiliary) speaker.  The first (Main) speaker is still connected to the amp at this instant.  Around 70ms later, the second relay (RL2) activates, and also connects the amp to the 'Auxiliary' speaker, at the same time disconnecting the 'Main' speaker.  When the footswitch is operated next time, the same process occurs, but the amp ends up being connected only to the Main speaker.  There will never be a period when the amp's output isn't connected to anything, provided the delay is set correctly.  The minimum possible delay isn't recommended, because it's essential to ensure that there's enough overlap to ensure that the amp is always connected to a load.  A delay of up to 100ms is perfectly reasonable if you wanted to go that far (increase the value of R4 and R5 to extend the delay).

The critical factor is the delay.  It must always be long enough for both 'activate' and 'de-activate' operations to ensure that the 'instant' relay has fully settled to its new state before the other relay operates.  This is why I elected to use a relatively long delay time, and two different time-constants are necessary (R2 and R3) because the MOSFET is not symmetrical, due mainly to the 12V used for switching.  Note that if you can't get hold of the 2N7000, almost any MOSFET can be used.  You may have something far bigger in your junk box that will do nicely, as it's not at all critical.  Don't omit the zener diode though - it's essential to protect the sensitive gate of any MOSFET, as a static charge may cause irreparable damage - a microsecond is more than enough!

fig 6
Figure 6 - High Accuracy Circuit For Speaker Switch (Valve/ Transistor Version)

If you think that high accuracy is worth the extra effort, then Figure 4 is ideal.  Using one half of an LM358 dual opamp, the on-off timing is 100% predictable as it doesn't rely on the MOSFET's gate threshold voltage.  A ½ supply voltage is created by R6 and R7, and with the values shown, the delay is 65ms for turn-on and turn-off.  R4 can be changed to get longer or shorter delays (adjusting R4 sets the timing for both turn-on and turn-off).  It's probably over the top, but it shows how a consistent delay can be achieved very easily.  The second half of the LM358 is not used, but do not substitute the opamp.  The LM358 was selected because its output can go to ground, and most other opamps cannot do this.  Pins 6 & 7 should be joined, and pin 5 connected to pin 2.

fig 7
Figure 7 - Circuit For Speaker Switch (Transistor Amp Only)

If you only use a transistor amp, then the Figure 7 circuit is the simplest.  It still uses a relay so the footswitch can be remote, connected with a normal guitar lead.  The remote switching isn't changed, as the requirements are the same.  It could be simplified if you don't need the LED to indicate the 'Auxiliary' speaker is selected, but I suspect that most players would need that.

The box sits on top of the amp, or behind it, whichever is most convenient.  If it's always going to be used with a remote footswitch, then the internal switch can be omitted (this also applies to the Figure 3 & 4 versions).  I strongly recommend using a relay for switching, because that prevents a mess of speaker cables from being underfoot when playing.

Please note that to allow this switch-box to be suitable for use with valve and transistor amplifiers, the speaker output sockets are not shorting types.  Most valve amps use a shorting socket, because a short-circuit is always preferable to a disconnected speaker, but using them may destroy many transistor amps that aren't fully protected.  Even those that do have protection (such as the Project 27 guitar amp) usually cannot withstand a shorted output for extended periods (usually limited to a few seconds at most).  In use, always test that both speakers are functional at low volume before cranking the amp to '11' and expecting it to work.

Note that one omission is a remote LED.  It's easy enough to do though.  Simply reduce the value of R1 to 1k5, and wire the LED in series with the remote footswitch.  Check polarity to make sure it's the right way around - nothing will work if you get it wrong and the LED may be damaged.  When the switch is on, so is the LED, so it indicates that you're switched to the 'Auxiliary' speaker.


This simple circuit will allow owners of valve or transistor guitar amps to switch speaker cabinets at the press of a footswitch.  While it doesn't appear to be a particularly common requirement, several companies sell devices for the same purpose, and they are not inexpensive.  This DIY version includes the facility to use an external footswitch, and it can be optimised for valve or transistor amplifiers.  The two amp types have completely opposite requirements, so making it adaptable to either means that it becomes 'universal'.  If you only have one amp, then just select the circuit for your needs.

The circuit described has just one job - to switch from one speaker to another without damaging the amplifier.  You can switch over in the middle of a full-power chord if you wanted to, although I can't imagine why anyone would do that.  The important part is that you can do it, and it will happen sooner or later anyway, whether by accident or otherwise.  It's not designed to switch two amps into a single speaker, and doing so is likely to cause serious damage.  It's not recommended that you even try it, and it doesn't have the facility to switch the inputs anyway.  To do that requires a significantly more complex switching arrangement, and would normally only ever be done when the amps are idle (no input signal).

Similar switch-boxes are available commercially, but from what I could find, they are rather expensive.  No-one else appears to have published a (workable) circuit to provide the make-before-break function reliably, so this seems to be another ESP 'First'.


There are no references, because no circuitry could be found anywhere to achieve the results achieved by the circuits shown.  I did see one example, but it used logic gates (two separate packages with most of the circuitry unused) two MOSFETs (and the relays), and it won't work well anyway.  Einstein is claimed to have said that "everything should be as simple as possible, but no simpler".  The circuit I found violated this principle.


HomeMain IndexarticlesProjects Index
Copyright Notice. This article, including but not limited to all text and diagrams, is the intellectual property of Rod Elliott, and is © 2021.  Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro-mechanical, is strictly prohibited under International Copyright laws.  The author (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference.  Commercial use is prohibited without express written authorisation from Rod Elliott.
Change Log:  Page published October 2021