Looking for the absolute best pro grade channel switching solution.

[SolidStateCircuitReject]

Hi Folks.

I have spent literally months designing and working on an amp... lots of testing of different components and values etc. For the sake of argument and future reference, that the amps layout is just like an old 2 channel Dual Rec, where the two channels split after v1b for separate gain pots, rejoin at v2a, then again separate at v3 cath follower to two separate EQ sections and controls, then again combine at the PI input.

So to date I have tried a pair of relay switches, and perhaps this is not the best solution, or perhaps someone can point out where I have gone wrong. But the issue is that I get a sound that is similar to losing tons of gain... moving around wires can sometimes fix the issue temporarily, other times moving around wires creates a nasty crossover sounding distortion. Obviously suggests poor soldering but once I take the pair of relays and replace them with two DPDT switches, everything works as expected... After that again I tried to wire up relays, I get the behavior described above. So perhaps I am powering them incorrectly.

I will attach a scheme of the way I have them set up for power below.  I have studied lots of schemes to try and figure this out... Mesa, Fender, peavy etc most seem to use mechanical relays except older Mesa and Soldano who used optocouplers. Optocouplers, handing never used one, seem odd to me. It seems odd to connect the full circuit of all channels only suppressing certain parts with tons of resistance, but can't help but wonder if it's a better solution than what I am doing. (Any takers on that?)

I had a custom power supply built for my amp that contains a 3A 6.3VAC coil to power what will eventually be at least 2 relays, maybe more in the future. The power circuit for the relays was lifted from a Silver Jubilee schematic - I tried this one first because this solution had worked for me in the past.

Relays: Finder 30 Series, 6V - https://www.finder-relais.net/en/finder-relays-series-30.pdf
I live in Belgium, so these are readily available.

Power scheme attached.

Can anyone offer a better solution, cost is of no matter, within reason, but I need something solid and pro grade.

Tank you guys for your input!!!

[tubenit]

IF you didn't know, there is ALOT of information on relay switching, schematics, layouts, etc .........  on relays in the ARCHIVES:

http://el34world.com/Forum/index.php?topic=14348.0

Maybe you can find your solution there?   I am attaching two drawings but I hope you will look thru the entire ARCHIVES thread. 

I use Hoffman's relay boards and relays and they work flawlessly for me. They are silent in switching. Switch instantly. And add no noise to the signal level.

With respect, Tubenit

[sluckey]

I think the problem is your power supply and control circuit. That LED can't possibly light up as drawn. Follow the link tubenit posted and build a simple bridge power supply.

[tubenit]

I edited two of the relay pictures to try and illustrate what I am thinking you are needing.

With respect, Tubenit

[92Volts]

Some sound (mostly treble) will bleed through switches/relays. That's why you use 2 relays so only the active channel gets signal, instead of letting both operate and just "listening" to the active channel. Some sound (not but the right amount/tone) is consistent with the relays being wired backwards (sending sound to channel 1, listening to channel 2), OR one switches and the other doesn't. As slucky notes your power supply is weird. Could you test with a 6v wall wart, 4xAA battery holder, or other known 6v source?

[SolidStateCircuitReject]

Thanks all for the input.  Rather than hastily replying yesterday I am still trying to further educate myself to make more intelligent decisions and replies.

Firstly, I had schemed the resistors/led wrong. Should be as attached.

However, while Tubenit's diagrams help, and I had previously read the referred thread last week (and will do so again), I see no way other than having two relays trigger from the same "event" to a/b between the two channels.  BUT, I have decided I want to do a three channel amp, so I need a way to run three channels now. So that:

Channel 1 will split at the input and go to v1a and it's own gain pot, rejoin into v2a, split out after v3b for it's eq section, rejoin at v4a (PI input)
Channel 2 will run to v1b, split to it's own gain pot, rejoin at v2a, split out after v3b for it's eq section, rejoin at v4a (PI input)
Channel 3 will run to v1b, split to it's own gain pot, rejoin at v2a, split out after v3b for it's eq section, rejoin again at v4a (PI input)

So this becomes tricky in my mind since there needs to be a relay to split the input for ch 1 or ch 2 and ch 3.  Then a relay to split ch 2 and ch 3 at v1. Then some type of way to combine all three into v2a. Then split it three ways for the eq section.

With two channels it could have all been done with 2 dpdt relays... with three it seems a magnitude more complex in my mind.

And, how is the DC conversion reservoir capacitor value found? Would it be the same formula as you could use for finding your power supply reservoir cap? According to Merlin Blencowe (to whom I hold to a pretty high standard) is:
C = I / (2 f Vripple)

Thanks!!

[silverfox]

You may want to look at how Peavey incorporates LDR switching. While they do leak a little, that can be overcome by using a drain resistor to kill the unwanted channel sound.

silverfox.

[PRR]

> Channel 1 will split at the input and go to v1a and it's own gain pot, rejoin into v2a, split out after v3b for it's eq section, rejoin at v4a (PI input) - Channel 2 will run to v1b, split to it's own gain pot, rejoin at v2a, split out after v3b for it's eq section, rejoin at v4a (PI input) - Channel 3 will run to v1b, split to it's own gain pot, rejoin at v2a, split out after v3b for it's eq section, rejoin again at v4a (PI input)
> So this becomes tricky in my mind

Tricky? No fake, jake! This is more complicated than directions around Maine! "Take a right at the red rock. Unless it is summer, then take the logging road or you never get past the jam at the lobster pound. But if a ship is in port, you better go all the way out to Sullivan and sneak back through the quarry..."

OK, I found a thing with rectifiers and *with the plan in hand* your scheme makes some sense. (Wish you had posted that, not relay specs. Relays is relays.) I marked it up Red Orange and New Yellow setting.

One source can feed more than one load; input switching may not be needed. I'm sure the guitar can drive two grids. I'm less sure about a bunch of tonestacks. Leo improvised that cathode follower when he reduced the slope resistor and discovered the plate impedance limited the result. His values just-barely do the job for one stack. Here we have two (now three) stacks and Master pots after each (so the CF may be working harder). I dunno.

I would say that Mesa's switching is as "Pro" as it gets. It does use Lead photo-cells which are getting harder to get. Being used On/Off, they do not have to be finely-specced cells, the stuff from China should be fine. PbS photo-cells are still legal for "repair", and a small postal package is unlikely to be searched. I would look to using Mesa's plan.

But it is already awful complex. And going 2 to 3 doubles the complexity. There is no 3-throw relay (in modern life). You want two 2-throw relays. One A/B, one over-rides to C. With switching around both ends of a HIGH gain chain, you surely want two clumps of relays so those nodes are not all together.

[SolidStateCircuitReject]

> Channel 1 will split at the input and go to v1a and it's own gain pot, rejoin into v2a, split out after v3b for it's eq section, rejoin at v4a (PI input) - Channel 2 will run to v1b, split to it's own gain pot, rejoin at v2a, split out after v3b for it's eq section, rejoin at v4a (PI input) - Channel 3 will run to v1b, split to it's own gain pot, rejoin at v2a, split out after v3b for it's eq section, rejoin again at v4a (PI input)
> So this becomes tricky in my mind

Tricky? No fake, jake! This is more complicated than directions around Maine! "Take a right at the red rock. Unless it is summer, then take the logging road or you never get past the jam at the lobster pound. But if a ship is in port, you better go all the way out to Sullivan and sneak back through the quarry..."

OK, I found a thing with rectifiers and *with the plan in hand* your scheme makes some sense. (Wish you had posted that, not relay specs. Relays is relays.) I marked it up Red Orange and New Yellow setting.

One source can feed more than one load; input switching may not be needed. I'm sure the guitar can drive two grids. I'm less sure about a bunch of tonestacks. Leo improvised that cathode follower when he reduced the slope resistor and discovered the plate impedance limited the result. His values just-barely do the job for one stack. Here we have two (now three) stacks and Master pots after each (so the CF may be working harder). I dunno.

I would say that Mesa's switching is as "Pro" as it gets. It does use Lead photo-cells which are getting harder to get. Being used On/Off, they do not have to be finely-specced cells, the stuff from China should be fine. PbS photo-cells are still legal for "repair", and a small postal package is unlikely to be searched. I would look to using Mesa's plan.

But it is already awful complex. And going 2 to 3 doubles the complexity. There is no 3-throw relay (in modern life). You want two 2-throw relays. One A/B, one over-rides to C. With switching around both ends of a HIGH gain chain, you surely want two clumps of relays so those nodes are not all together.

BINGO.

First, being a coder, your directions through Maine make some sense ;-)

I have attached the 3 channel triple rec block diagram... basically what you laid out above but channel 1 has an early tone stack, thus driving the CF twice instead of three times....

But according to the schem, I see that these are real relays, not opto's... because of the lead this I assume? They are not RoHS compliant... I am living in Europe now for almost a decade... there's all kinds of restrictions on this and that... not to mention INSANE taxes on everything.

Next step... figuring out how to trigger all of these relays in a proper order. Wondering if it cam all be done by 5 pin.

Thank you

[SolidStateCircuitReject]

To give an even better idea, here is my block diagram....  just whipped it together.

Contains a total of 5 relays. Is this possible? I have no real idea...  I see plenty of amps running tens of relays, so my logic tells me it is a possibility... but admittedly I need to learn more. I study for hours a day, read Merlin's books, study schematics and more importantly safely test plenty of stuff myself having some three or fours testing amps in my "lab" :-)

But I have yet to complete something this ambitious. But regardless which way it needs to happen, I will finish this.

I had a PT made that puts out 720VAC (360-0-360) (500VDC after rectification), 6.3V 7.5A, 6.3V 3A, 5V 3A, and 60V bias tap. According to datasheets, unless I am mis-understanding, relays will consume typically 33mA each.

Thanks :-)

[92Volts]

A conceptually easy way to do this is with three (or six) single-pole relays that each make or break a connection from point A to B, C, and D respectively. You only power one at a time and signal goes only there. You can do the reverse to recombine. The issue is high parts count which is why many production amps use LDRs.

You'd also need some sort of logic circuit to decide which one to power. To select between these channels I imagine you want an up/down button or single button to cycle through-- though I guess a 3-position switch on the amp's panel is an option. If you want footswitch buttons though you'd need logic, this can be done with discrete transistors or gate ICs though if you're a computer science type you might choose an Arduino or other small microprocessor.

[PRR]

>attached the 3 channel triple rec block diagram...
> according to the schem, I see that these are real relays, not opto's...

Let's see the actual schematic. The block diagram is conceptual, we want to steal/study the actual build.

[SolidStateCircuitReject]

>attached the 3 channel triple rec block diagram...
> according to the schem, I see that these are real relays, not opto's...

Let's see the actual schematic. The block diagram is conceptual, we want to steal/study the actual build.

I was searching for a link to one but I can't seem to find one any longer, so I attached it.

[PRR]

Sorry I asked!!

OK, this seems to be seriously well designed, and no opto-resistors.

There are missing details. On first look, the LM3914 does something, but what?? A couple pins go somewhere. But this is clearly not being used as intended (a bar-graph), so there is something clever.

There is a lot of logic in there, more than it looks because someone has gone through and Mickey-Mouse Logic-ed it thoroughly. (MML is a well known class of logic.) if A not B or C..... all diodes and series transistors, and maybe the '3914.

[sluckey]

Walk! Don't Run.

You still don't have a working power supply on paper to energize one relay yet. Get a transformer, bridge, capacitor, switch, relay, and a bit of wire. Temporarily solder it together and make the relay go click, click. Then get clever.

[SolidStateCircuitReject]

Walk! Don't Run.

You still don't have a working power supply on paper to energize one relay yet. Get a transformer, bridge, capacitor, switch, relay, and a bit of wire. Temporarily solder it together and make the relay go click, click. Then get clever.

Indeed. Also in the forefront of my mind.

So, as far as I can tell, optos were dropped in favor of relays because of there non-RoHS conformity... so I assume they'd have a hard time to sell them here in the EU. And the Dual Rec itself has sooooo many options that it requires somewhere around a dozen relays or more. Of course I don't want to get this crazy.

If having three channels forces me into logic gates etc, it is something that I will have to put on hold...  for now. More so because I want to understand the circuit I build, not just blindly follow schems.

I will return to a two channel for now to keep it simple (stupid!)...

Power supply for three relays.

I already have a small cache of Finder 6v relays, IN4007 diodes, 470uF caps (plus lots of others)... So returning to my original question then, how does one determine the capacitor value attached to the diode(s)? As I said I have dedicated 6.5V 3A CT'ed leads already.

Thanks for all the helpful information!!!!

[sluckey]

how does one determine the capacitor value attached to the diode(s)?

Plagiarize. No math or other brainstorming needed. The value is not critical other than the voltage rating must be larger than the voltage applied. 470µF@25VDC is plenty good for filtered dc voltage used to power a relay coil. I would put more importance on "will the physical size of this cap fit into the box?" than electrical characteristics.
 

[92Volts]

There are missing details. On first look, the LM3914 does something, but what?? A couple pins go somewhere. But this is clearly not being used as intended (a bar-graph), so there is something clever.

At first I thought an IC like that could be used as logic (if you were clever enough) but the only pins that seem to be connected/switched in any way are the "outputs" that drive LEDs-- I think it's just used as a current-limiting LED driver. Everything in that schematic seems to be controlled by the "loop assign rotary SW"-- the logic that is shown seems to make use of 3 independent positions of that switch. Achieving the same with foot-switchable buttons (for example) would require logic I don't think is shown in that document

[VMS]

This Aiken switch looks relatively easy:


https://sites.google.com/site/yourtubeamp/effects-and-accessories/mashall-jmp1-footswitch/aiken-switch-schematic

[VMS]

...and here is an interesting kit that could also be useful:


https://www.tube-town.net/ttstore/Kits/Misc-Kits/Kit-Midi-Channel-Switch::6193.html

[PRR]

> I think it's just used as a current-limiting LED driver.

It can't be even that if we don't connect more pins. There is Stuff Missing.

I've marked-up as much as I can figure around the LM3914 and don't get it.

[SolidStateCircuitReject]

...and here is an interesting kit that could also be useful:


https://www.tube-town.net/ttstore/Kits/Misc-Kits/Kit-Midi-Channel-Switch::6193.html

Living on Belgium, this is who I buy all of my supplies from and this kit was my original goto for the issue. While I finally got two channels (using two relays) going, this is probably the quick way forward for anything more intricate.

Thanks all for the help!!!

[dunner84]

http://www.londonpower.com/channel-switcher#intro

I have ordered a lot from London power. Kevin is a genius, and sells fantastic stuff. I'd highly consider it if you're looking for the best way to switch a lot of channels.

[VMS]

Kevin is great, i just wish he would write those hifi books already.