LDRs as pots?

[tbryanh]

Perhaps instead of relays, you could use a row of JFETs as switches, or even transistor arrays in IC form for even more compact.

They would work, but what about distortion? The are in the signal path. Alot of people do not want silicon in the signal path.

[Steve Conner]

There are hundreds of silicon-based digital potentiometer and analog switch and routing chips on the market. They can't be used in vacuum tube circuits because they are limited to +/- 15V signal swings at best. Tube preamps can see upwards of 50V p-p.

Mesa's LDR circuit is their attempt at making a cost-effective digital pot that can stand high voltages. It has its drawbacks, which you have found and discussed in this thread, but I can't think of anything else that would work better. The only other technology I can think of is the motorized pots that Fender used in the Cyber-Twin.

The mechanical relay and resistor ladder approach is feasible (it's basically a multiplying DAC) till you try implementing a log taper that way! There are also 400V MOSFET output optocouplers that you could investigate as an alternative to JFETs or relays. Regular optocouplers might even work as audio switches if you connected pairs of them back-to-back.

I guess you could make a circuit that attenuated the signals to digital-pot-IC-friendly levels and then boosted them back up again, but my gut feeling is that the clean channel would end up sounding weak and noisy, and the result would suck about as bad as the SR&D Rockman... It all adds up to enough of a design nightmare that the best solution I can think of would be to buy a used Triaxis!

Good luck getting an answer out of Perkin-Elmer!

[Gregg]

The resistor string-relays thing is the most "analog" one but multiply that by 6-10 controls and you'll end up with a 1 sq.ft "pot board".
Using FET switches sounds to me like using digital pots - I would prefer to attenuate the levels and use digital pots as in JMP-1.

I guess you could make a circuit that attenuated the signals to digital-pot-IC-friendly levels and then boosted them back up again, but my gut feeling is that the clean channel would end up sounding weak and noisy, and the result would suck about as bad as the SR&D Rockman...

I have a DIY version of JMP-1 /with digital pots/ and the clean channel is nothing like weak and noisy. Actually the drive channel sounded to me weak and thin that's why I rewired the circuit bypassing the SS drive section to make it all tube and now it looks like GT Trio Mean channel with a variable frequency /150Hz-1kHz/ shaping filter/boost in front of it.
The CF is loaded with a typicall Mrashall TS with flat response, then the signal is attenuated to ~1V and goes to the active tone stack.

I was thinking about loading the CF with a flat TS, attenuate the signal and connect /via buffer/ a second tone stack then boost the signal to it's previous levels which makes +1 tube. Now we can use LDRs and forget about most of the problems discussed here.

[Don Moose]

I'm not quite following this. It seems that if there are 8 taps in the resistor string, then 8 SPST relays are needed.

True enough - 8 positions just doesn't sound like enough to me.

I am thinking things are far simplier than you indicate and much more possible than you indicate.

Very likely. As we have already discussed, it is in my nature to fret and nitpick.

From what I get your saying this approach is not possible.

Not at all - I said it was a 'conceptually good, technically solid approach.' Just trying to think about all the issues implied.

I WANT your idea to work - that's why I'm pointing to the snags.
Anyway, hope this helps!

[tbryanh]

It all adds up to enough of a design nightmare that the best solution I can think of would be to buy a used Triaxis!

I’m not interested in buying a Triaxis.

The mechanical relay and resistor ladder approach is feasible (it's basically a multiplying DAC) till you try implementing a log taper that way!

The LDR Pot uses a resistor ladder too. So, if log taper is a problem, it is a problem with LDR Pots too. I don’t see log taper as a problem anyway. All you need is a calculator.

The resistor string-relays thing is the most "analog" one but multiply that by 6-10 controls and you'll end up with a 1 sq.ft "pot board".

I am not worried about the size. 1 sq.ft" is probably an overexageration.

I only see 2 issues at this point, distortion and switching time.

Does the LDR Pot distort? (Probably yes)

Does the “rotary-switch” Pot distort? (No)

Does the LDR Pot switch fast enough? (Probably yes)

Does the “rotary-switch” Pot, with its clamp muting, switch fast enough? (Probably not)

Can the “rotary-switch” Pot be made to switch fast without popping? (?)

[Don Moose]

Can the “rotary-switch” Pot be made to switch fast without popping? (?)

Yes - probably.

Critical thing #1 - make sure that in every position, every switching contact on every relay has a DC path to GND. It can be 3M3 or even 10M, but every contact must always have a DC path to GND.

Critical thing #2 - add a diode (cathode-to-(+)) and/or a cap across every relay's coil to contain the flyback currents (induced by collapsing coil field when you 'relax' and energized relay). Make these wires as absolutely short as possible.

Critical thing #3 - use 'low-signal' relays. Regular relays count on the signal current to keep the contacts closed. You won't have that luxury, so you need the relays with the contacts plated/coated with some suitable material. The Omron G5V-2 series, for instance.

Helpful thing to keep in mind - At least some of the pops relay circuits make are due to coil current di/dt getting coupled into the signal runs - (see Critical thing #2) - treat these runs like the noisy things they are - run them at right angles to the signal wires, keep them as far away as practical, and try to slow down the switching edges.

Helpful thing #2 - consider make-before-break switching, and 'walk' the control from current setting to new setting.

Hope this helps!

[tbryanh]

Helpful thing #2 - consider make-before-break switching, and 'walk' the control from current setting to new setting.

Critical thing #1 - Critical thing #3 are excellent as well as the helpful things. Thanks

I understand the make-before-break switching, but I am not sure what 'walking' means. Please explain if you don't mind?

Thanks again.

[Steve Conner]

About the log taper thing, I assumed that you'd use the relays in a binary weighted combination with a R-2R ladder, to save on relays. Binary weighting lets you have a 16 tap "pot" with only 4 relays, but this doesn't work for non-linear tapers. If you're not using this trick then I expect board sizes in the square feet too.

Running relay wires at right angles to signal wires is all very well, but eventually they have to meet at the relay where the coil sits right next to the contacts. I've got reasonable results using RC networks to slow the edge.

I think considering we make tube guitar amps because solid-state ones don't have enough distortion, worrying about LDR distortion is probably pointless. It'll be the same square-law distortion that was once claimed to give carbon comp resistors extra mojo, so if anything I'd expect it to improve the tone.

Did the JMP-1 actually use digital pots or multiplying DACs? I guess Marshall couldn't use the LDR method after Mesa patented it. We don't need to worry about patents.

[Tiago]

IMHO:

Switched resistor arrays seems a good way to go. Problem is the technology used to switch between them. Personaly, I would try Jfets and measure the results on scope, or even using LDRs in an On-Off state and then measure on scope.
The thing is that working in an On-Off state, when its "OFF" its impedance is so high compared to the impedances in the same place of the circuit that the distortion might be negligible, when its "ON" (shunt to ground) its even more negligible.
Its a good place to start experimenting.

1% Distortion we cannot afford even in a tube amp with a 30K impedance at a 1.4VRMS, not just because thats a lot of distortion for a "clean" sound, but because what we would need for a gain pot would be something like 500K and 2 to 7 VRMS, which would make the distortion so high that only in a "rectifryer" channel would be acceptable.

Despite of being a tube preamp design, if I was an engineer at a big company, and even maybe if I just had the time, I would try VCA's or a buffered Digital Potentiometer.
If some company came up with that it would blow the market.
I know, solid state, thats a BIG NO NO!! But listen:
At the first gain control we only need to control 2 to 7 VRMS. A +/-15Volt supplied buffered Digital Potentiometer is enough for that, I think that would affect much less the signal than a LDR.
The biggest problem is that one can't use a standard tonestack because there are no DigiPots with higher impedance than 200K that I'm aware off, so a 4 band active Equalizer would be the way to go.
Well there is a lot of solid-state on a design like this, but who can say it will sound bad no matter what?!!
And look at something, for one of the Big Four, an estimate 10,000 units like this, for sure they could make it cost less that 100$ on parts and even they sell it for four times that it would be affordable.
For an DIY, this could be a two month fun...

[Steve Conner]

Well, circuit theory tells us that if we make all of the resistors and pots in a tone stack 10 times smaller, and all the capacitors 10 times bigger, and scale the input and output impedances down by 10 times too, it will still have the same frequency response.

This is what I'd do if I were asked to make a digitally controllable tone stack, since I already have experience with hybrid amps and have got some good tones without breaking the +/-15V limit. I'd scale the tone circuitry from a classic amp to work with digital pots and multiplying DACs, and then make a completely fresh amp design to accommodate the lower signal voltages. I suspect the JMP-1 does something similar.

As an example, I don't think any tone pot in my old Ninja Toaster amp is above 100k or sees more than 30Vp-p. This was the result of using op-amps in the EQ. However, this introduction of silicon is the thin end of a wedge whose logical conclusion is a Pod.

Of course if you want an exact clone of a vintage amp that's digitally controllable, I guess you have to go down the road that this thread discusses. But to me that's like putting alloy wheels on a Model T Ford.

[Gregg]

Did the JMP-1 actually use digital pots or multiplying DACs?

JMP-1 used digital pots, now obsolete but you still can find them in bulk around the net. For Gain/Volume pot they used Toshiba's TC9176. For the tone stack they used TC9170 which was designed to be used in EQ circuits and has the S-shape response curve. Below are the data sheets for those digital pots:

http://www.datasheetarchive.com/preview/3342389.html

http://www.datasheetarchive.com/preview/3342399.html

JMP-1 schematic:

http://www.drtube.com/schematics/mar...jmp1-61-04.pdf

At the first gain control we only need to control 2 to 7 VRMS. A +/-15Volt supplied buffered Digital Potentiometer is enough for that, I think that would affect much less the signal than a LDR.
The biggest problem is that one can't use a standard tonestack because there are no DigiPots with higher impedance than 200K that I'm aware off, so a 4 band active Equalizer would be the way to go.

This is more or less done already - just check out the JMP-1 schematic. The pot is in front of the first stage so you control only - 0.2-0.4V and then you have the four band Marshall tone stack emulation.
And by the way there are 1M digital pots /frequency response is 6 kHz/ but they are low voltage.

Well there is a lot of solid-state on a design like this, but who can say it will sound bad no matter what?!!

It doesn't sound bad at all. Maybe I should go ahead and record few samples with my current DIY version of it.

Hughes and kettner have the Switchblade series fully programmable tube amps. I wonder how they did it? This is how it looks inside /sorry, the picture is small/.

[Don Moose]

I am not sure what 'walking' means. Please explain if you don't mind?

If you want to move from position 3 to position 7, _walk_ from 3 to 4 to 5 to ... rather than jumping straight to 7. This is probably another unnecessary complication, but my thinking is that it would keep any DC buildup minimized. I'd probably only mess with it as a last resort.

Hope this helps!

[loudthud]

Pots with relays

Clearly, on the bass and mid controls of the typical tone stack, a binary sequence of resistors can be used. (Those R/2R ladders always confuse me!) With 6 relays you could get 64 steps. The resistors might be 560K, 270K, 130K, 62k, 30K, and 15K (total 1.067meg). All resistors are in series and each one has a relay across it to short it out selectively. You get 63 possible resistances and a short. Lets call this a binary ladder. To emulate a log taper, 8 resistor/relays might be needed.

The same idea can be used to make a pot except you need twice as many relays and resistors. A binary ladder for the top portion of the pot and a binary ladder for the bottom portion of the pot. Although you could independently set the top and bottom resistance, it would be nice to keep the end to end resistance constant. To do that, you have to close (for example) the 62K relay on the top when you open the 62K relay on the bottom. So, you could use DPDT relays and arrange the contacts to do this automatically saving half the number of relays.

[tbryanh]

Pots with relays

Clearly, on the bass and mid controls of the typical tone stack, a binary sequence of resistors can be used. (Those R/2R ladders always confuse me!) With 6 relays you could get 64 steps. The resistors might be 560K, 270K, 130K, 62k, 30K, and 15K (total 1.067meg). All resistors are in series and each one has a relay across it to short it out selectively. You get 63 possible resistances and a short. Lets call this a binary ladder. To emulate a log taper, 8 resistor/relays might be needed.

The same idea can be used to make a pot except you need twice as many relays and resistors. A binary ladder for the top portion of the pot and a binary ladder for the bottom portion of the pot. Although you could independently set the top and bottom resistance, it would be nice to keep the end to end resistance constant. To do that, you have to close (for example) the 62K relay on the top when you open the 62K relay on the bottom. So, you could use DPDT relays and arrange the contacts to do this automatically saving half the number of relays.

Sounds good. do you have a diagram handy?

[tbryanh]

Of course if you want an exact clone of a vintage amp that's digitally controllable, I guess you have to go down the road that this thread discusses. But to me that's like putting alloy wheels on a Model T Ford.

Its not a matter of cloning vintage amps. Of course it will be used for that, but it will be used for modern and post modern designs too.

The issue I have is I want the best tone. Its true that if you have the best tone, you play your best. Its hard to be serious when you squeeze into a note that gives you that transistor fuzz. Sure it sounds pretty good overall, but that little fuzz takes alot of the excitement away.

It doesn't take an obvious large farty amount of transistor distortion to spoil things. Just a tiny amount of that transistor fizz still sucks.