To design/build an amp for a new electric instrument

[pjd3]

Hello there,

Following the successful build of a single channel Fender amp for gigging, I'm now venturing into designing a stereo tube amp for a guitar that I made for a very specific purpose.

I built a fretless guitar for the sole purpose of using an Ebow, only. It was set up and built as a single note melody instrument for playing cello like romantic melodies and also for building up faux string ensemble multi-track arrangements. I have zero interest in either picking or strumming on it.
The instrument completely surpassed my expectations as a viable expressive and unique melody instrument. It needs an amplification system that will enhance and highlight its unique qualities.

My loose plan at this point is to design a 2 channel amp that lives somewhere between a guitar amp and a Hi-Fi amp. I've been using this fretless Ebow guitar through a high end DI into a Neve preamp into a Kurzweil Stereo FX processor with decent conversion. This gets the sound pretty close to where I'd be happy with, but that has been just for recording. I'm up for creating a system for live performance that brings those recording qualities to a live stage.

The other aspect to this is that I'd like to tap off the output transformer secondaries to a "speaker-to-line" converter for sending into a nice stereo FX processor and back into a line level input that will mix in with one of the preamp gain stages. I insist on wide rich and opulent reverb and time base ambience live from the source. Jensen makes some nice transformers with this specific application in mind.

So, this is just the loose initial conception of this idea but, suffice it to say, I picture basing this off of traditional Fender gain stages, but perhaps experimenting with cathode and plate resistor values for say, lower distortion, or, for the "proper amount of distortion"!

I've pretty much decided this would include an Ultra-linear output tranny with Triode/Pentode switching. Where this is uncharted territory, I'm sure having these kind of options could be well worth the cost and work - unless during design/experimenting I find it to be undesirable. In addition, I may even choose to leave out the typical type EQ and implement a simple low pass filter to roll of high frequenies (this guitar generates very strong highs in the 8K to 9K area).

I picture this stereo amp living somewhere between a typical guitar tube amp and a Hi-fi tube amp. It can be a little of both, but also neither, and I'm totally cool with that. These things can be allowed to develop during the "design/experimentation phase". As a matter of fact, I've constructed a tube amp design station with 5 preamp sockets and a couple of Octal sockets all mounted on pcb's with parallel heater traces all there ready to go. I disassembled my Bogner Alchemist and used the pcb's as a tube bed. And added a couple of proto boards to hang in the slew of different caps and resistors that will be tried. I'm really looking forward to this.

If you've had the patience to read through this, I applaud you! thank you for your time and I very much look forward to your thoughts about what I'm doing, and perhaps some ideas or things to consider in the creation of this "Fretless Ebow guitar Stereo amplifier with build in FX send and input receive".

Thanks very much. Your thoughts are surely welcome and appreciated.
Phil Donovan

 

[leadbelly]

You need to pick a power level, that's a highly important design criteria.
I wouldn't consider triode/UL/pentode switching uncharted territory, it's nothing special.
As for keeping the sound clean without distortion, that's easily achieved by not having so many gain stages that tubes clip, and avoiding any tone shaping stages such as starved-plate.

 

[indianajo]

As fragile as tubes are to shock, and as heavy as they are, I can't imagine gigging with them in this day & time. I can understand the desire of guitar players to flat the top of the sine waves in a way similar to red plating of 6L6's , but there are other ways to do it. I think it was PB2 that gives away his plan to do it with an op amp? Much more sophisticated than simple clipping with a zener diode. I didn't save the thread link since I don't play guitar.
Depending on the size of the audience, a powered mixer offers a convenient source for a simple 2 driver speaker. Many have the DSP built in. If you're playing crowds over 100, then a reliable powered speaker with separate box for pickup interface (jfet op amp) tone shaping & reverb built in makes for quicker setup. I have sat in audiences at free shows where the time between bands goes over 15 minutes due to setup, and that is tedious.
I've got my 30 person audience rig down to a powered mixer 4" x 15" x 15", one 6.5" 60 W coaxial speaker in a 6"x6"x12" box, a camera tripod for the area coverage mike (which is lighter than a real mike stand) which all fits in 2 bags on the back of my bicycle. My autoharp & voice don't have much bass so I don't need a woofer. If I'm playing keyboard I need a woofer too. 100 person audiences, or instrument with content under 200 hz, you need a truck or station wagon or a 3 wheel trike with cargo box. A local harpist is doing 100 person gigs with a 1 box setup, plus a cable bag, plus her harp, which all fits in a Volkswagen station wagon. The harp has a pickup and she wears a headset mike. Her box is single channel I believe. She weighs about 90 lb and her roady friend is smaller, I believe. I weigh 160 lb and a Peavey KB3 would do what I need for keyboard, but it is too big to fit in a pannier bag.

 

[jjasniew]

"The other aspect to this is that I'd like to tap off the output transformer secondaries to a "speaker-to-line" converter for sending into a nice stereo FX processor and back into a line level input that will mix in with one of the preamp gain stages. I insist on wide rich and opulent reverb and time base ambience live from the source."

Sounds like a recipe for uncontrolled feedback. Now if you said "Output from channel A" through the FX processor, then "mix in with channel B's preamp gain stages" it wouldnt have that issue; i.e. feedback of a signal into itself.

Adjusted just so, such an arrangement might provide a wild ride in terms of reverb. However nothing is more embarrassing than uncontrolled feedback when on stage -

 

[chris661]

I think I'd find a decent active PA speaker and call it good.
The likes of a Yamaha DXR12 are hard to beat for the money.

Chris

 

[indianajo]

I think I'd find a decent active PA speaker and call it good.
The likes of a Yamaha DXR12 are hard to beat for the money.
Chris

DXR12mkII has 1 xlr input and 4 line level. Many musicians use 2 condensor mikes, one for instrument, one for voice. Requires phantom pwr. No fet input or vacuum tube grid for guitar pickup. So another box is required to convert instrument pickup to line level. + dsp requirement for reverb OP wants, + graphic equalizer controls to kill howls quickly. Roadies are $100 a day plus beer minimum around here for all that load.

 

[Gnobuddy]

H
...fretless guitar...Ebow...single note melody...cello like romantic melodies...

Do you have any sound clips we could listen to?

It sounds like an interesting project, but I have the feeling nobody on the thread (including myself) actually knows what you're hoping to achieve. Listening to one of your recordings might give us a better idea.


-Gnobuddy

 

[JMFahey]

successful build of a single channel Fender amp for gigging, I'm now venturing into designing a stereo tube amp for a guitar that I made for a very specific purpose.

Then build a stereo Fender amp.

I built a fretless guitar for the sole purpose of using an Ebow, only. It was set up and built as a single note melody instrument for playing cello like romantic melodies and also for building up faux string ensemble multi-track arrangements. I have zero interest in either picking or strumming on it.
The instrument completely surpassed my expectations as a viable expressive and unique melody instrument. It needs an amplification system that will enhance and highlight its unique qualities.

Cool.

My loose plan at this point is to design a 2 channel amp that lives somewhere between a guitar amp and a Hi-Fi amp.

Fender amps neet both requirements, since they are very clean guitar amps.

I've been using this fretless Ebow guitar through a high end DI into a Neve preamp into a Kurzweil Stereo FX processor with decent conversion. This gets the sound pretty close to where I'd be happy with, but that has been just for recording. I'm up for creating a system for live performance that brings those recording qualities to a live stage.

Then feed that signal into a flat power amp and speaker and you´re done.

The other aspect to this is that I'd like to tap off the output transformer secondaries to a "speaker-to-line" converter for sending into a nice stereo FX processor and back into a line level input that will mix in with one of the preamp gain stages. I insist on wide rich and opulent reverb and time base ambience live from the source. Jensen makes some nice transformers with this specific application in mind.

Do not go "backwards" on a signal chain with gain stages in between.

So, this is just the loose initial conception of this idea but, suffice it to say, I picture basing this off of traditional Fender gain stages, but perhaps experimenting with cathode and plate resistor values for say, lower distortion, or, for the "proper amount of distortion"!

You already have 2 solutions to your needs: stereo Fender amp and the processor chain you like ... why venture into uncharted teritory for no good reason?

I've pretty much decided this would include an Ultra-linear output tranny with Triode/Pentode switching. Where this is uncharted territory, I'm sure having these kind of options could be well worth the cost and work - unless during design/experimenting I find it to be undesirable. In addition, I may even choose to leave out the typical type EQ and implement a simple low pass filter to roll of high frequenies (this guitar generates very strong highs in the 8K to 9K area).

I picture this stereo amp living somewhere between a typical guitar tube amp and a Hi-fi tube amp. It can be a little of both, but also neither, and I'm totally cool with that. These things can be allowed to develop during the "design/experimentation phase". As a matter of fact, I've constructed a tube amp design station with 5 preamp sockets and a couple of Octal sockets all mounted on pcb's with parallel heater traces all there ready to go. I disassembled my Bogner Alchemist and used the pcb's as a tube bed. And added a couple of proto boards to hang in the slew of different caps and resistors that will be tried. I'm really looking forward to this.

I think you are already well beyond that point.

 

[pjd3]

Thanks for the responses everyone, lots of interesting views on this one.

I should probably mention that designing up and building a system for this is part of the whole thing that makes this fun and interesting. I'll be spending my energy on "how" to do it, not "if" I should do it. Sure, it would be easy to buy up some decent equipment such as a PA with everything already to go and make things easy on myself but that's just not what floats my boat with this fretless Ebow guitar. I want to bring a unique system to it that's tailored to its own strengths and weaknesses, and that can emerge over the design/experimentation period. With all due respect, this is "DIY Audio" website so, that happens to be why I came by, there is Sweetwater for anything else. I'm feeling pretty confident in the semantics of what I'm going for.

The FX aspect of this was one that had perplexed me. Normally, I would expect feedback if an amp was attempting to re-amplify its own signal in real time but, this particular approach was to use only an "effected only" send from a good reverb/delay processor stereo processor with non of the original and direct signal in the return. Does that sound better? In that case I figured the FX return would look no different than another instrument to the amp. Unless there is something I'm not seeing, Bringing an "effected signal only" mixed back into one of the preamp stages, in the same way that Marshall for instance did with the Plexi seems like it should work if proper gain staging and mixing resistors are implimented properly. I know there are also things like "zero loss" FX send/return kits that can be purchased and implemented for that. Perhaps thats a better way than tapping off the speaker and bringing just the "non-direct reverbed" signal back into the amp.

I will see what I can do about getting a video or sound clip on here to show the guitar. I measured the output with my meter last night to see what the maximum signal is. With volume and tone knob all the way up and the humbucker coils in series, I was reading upward of 2.5 volts ac and higher. I should really be looking at this in a scope to see whats really happening with the output but, just whacking the open strings measured around 30ma to 100mA so, relatively, the Ebowed output is startlingely higher that any picked or strummed sound you could make. So right away, I have to wonder how the input section would respond to that kind of voltage range, and if there is anything different that should be done in the input design.

How about turning my volume knob down?? And yup, that may be the right answer!

Hey everyone, have a great weekend. I'll try to dig up a clip of the guitar.

Thanks!
Phil D.

 

[Gnobuddy]

...Normally, I would expect feedback if an amp was attempting to re-amplify its own signal in real time but...approach was to use only an "effected only" send from a good reverb/delay... with none of the original and direct signal in the return.

Negative feedback - as used in the majority of Leo Fender's tube guitar amps, and virtually every solid-state power amp - IS making the amplifier re-amplify its own output signal in real time!

But to make this work, we take great care to keep the feedback negative over the entire frequency range of interest, and then we reduce amplifier gain at frequencies outside that range, to keep oscillations from starting. In other words, we have to watch the phase and amplitude of the feedback signal, so that we don't get smacked in the face by the Barkhausen stability criterion (formulated 99 years ago!): Barkhausen stability criterion - Wikipedia

Okay. Let's say we've built ourselves an amplifier, with negative feedback (output fed back to input), and we've paid our respects to Barkhausen, so the amplifier is nice and stable. Now let's do a thought-experiment to see what would happen if we inserted a delay into that nice stable negative feedback we already have.

For convenience, let's say you use a short "slapback" delay of 100 mS, or one-tenth of a second.

The period of a 10 Hz sine wave is also one-tenth of a second. If you take such a wave and delay it by exactly one-tenth of a second, the delayed wave looks identical to the original wave - it would overlay it exactly if you plotted both on a graph. In other words, the delayed signal is exactly in phase with the original signal.

So if you fed a 10 Hz sine wave into this amplifier (with the 100 mS delay in its feedback), the delayed feedback signal will be in phase with the input, rather than out of phase with it. What used to be negative feedback, has now become positive feedback. If there is enough gain in the amp, you will now have sustained oscillations at 10 Hz.

No sane designer gives a guitar amp lots of gain at 10 Hz. If we roll off the bass steeply below, say, 80 Hz, we can probably dodge this particular bullet. Whew! A lucky escape?

But wait, there's more...

What happens if you consider a 20 Hz sine wave? Its period is 1/20 seconds, or 0.05 seconds, or 50 mS. TWO of its periods add up to exactly 100 mS. If you delay a 20 Hz sine wave by 100 mS as before - that's exactly two periods - it will once again exactly overlay the original signal. Perfectly in phase.

So your amplifier will also try to oscillate at 20 Hz. Ugh.

A little thought will show that exactly the same reasoning applies at 30 Hz (100 mS will be exactly three periods of delay, once again turning negative feedback into positive feedback.) And at 40 Hz. And at 50 Hz. And so on.

So the formerly stable amplifier, with a 100 mS delay inserted in its negative feedback loop, will now be unstable and want to oscillate simultaneously at 40 Hz, 50 Hz, 60 Hz, 70 Hz, and every other positive integer multiple of 10 Hz.

Things are not looking so good for rockabilly music (slapback delay.) How about Pink Floyd? Let's say you use a longer delay in the fed-back signal. Suppose you use, say, 500 mS, in the vicinity of delays used by David Gilmour on some of his songs.

Well, a 2-Hz sine wave has a period of 500 mS, so your amplifier will want to oscillate at 2 Hz. Using exactly the same reasoning as above, the amp will also want to oscillate at 4 Hz, 6 Hz, 8 Hz, and so on, i.e. at every integer multiple of 2 Hz.

In other words, if I'm right about all this, adding delay in the feedback loop will not prevent the amplifier from oscillating. Rather, it results in positive feedback at a plethora of equally-spaced frequencies, and if there is sufficient loop gain at one or more of those frequencies, I would expect oscillation to occur.

As always with positive feedback, if you only use a smidge of it, feedback theory says you won't get outright oscillation. Whether this will be practical or not, I don't know.

While I hand-waved my way through the mathematical bits of this post, I can almost smell a more mathematically rigorous proof in the air, most probably involving Laplace transforms. It's been a long time since I was on friendly terms with those little beauties, though.

I expect your test-bench rig will tell you soon enough if your delayed-feedback idea is workable or not. Unless your bench-tests prove me wrong, though, my vote is with the others on this thread - I think it will cause instability problems.


-Gnobuddy

 

[pjd3]

Hey Gnobuddy, I see what you are saying and it's a good point. I may have erroneiously assumed that the returned signal would be to diffuse and disparate of the original realtime signal to enable any positive feedback but, I'm not going to bet my paycheck on it, yet. I've certainly had my fair share of inadvertantly setting up situations that caused outputs to regenerate into the input and it often was not too fun. Perhaps I'm expecting something different to happen that won't. But hey, one day it would be fun a learned to find out. Thomas Edison will be proud of me that, "I learned at least one more way this doesn't work".

Speaking of negative feedback, I'm thinking of modding in a NF adjustment or switching on my Fender build, which is actually Rob Robinettes single channel Vibroverb. It sound great but its so clean it could use an option to rat up the sound just a little! Of course I use many pedals for that but, It has the panel space and wouldn't take too much work or money to do.

Thanks Knobuddy, good insight there.
Phil D.

 

[pjd3]

Gnobuddy, would like to hear your take on this scenario.

This is a stereo amp. Would you expect any difference if "left speaker tap" was mixed into "right preamp gain stage"? Does that de-correlate things or, are we still dealing with similar overlapping frequency elements? Ha, have I simply added another feedback circle? Its late. Thinking is getting hard now.

Thank you,
Phil D.

 

[PRR]

...for playing cello like ...

Then build a stereo Fender amp....

+2.

I was VERY impressed by how a Twin Reverb handled cello. Dave made a side-career of it, and Ruth played a few gigs that way. I have also worked near a very high-end quartet with some amplified pieces, and a dense tech rider, and felt that a few healthy soulful guitar amps might serve them better.

Of course two Twins is a heavy load on the roadies.

 

[indianajo]

This is a stereo amp. Would you expect any difference if "left speaker tap" was mixed into "right preamp gain stage"? Does that de-correlate things or, are we still dealing with similar overlapping frequency elements? Ha, have I simply added another feedback circle? Its late. Thinking is getting hard now.
Thank you,
Phil D.

It is really best to do your waveform mods at the line level stage, before current amplification. That avoids the whole feedback problem.
We builders don't have to use pedals, we can put the waveform modification into the main chassis. The foot control can be a simple switch, not needing a power supply and being feet away from the liquid refreshment a performer may have. Analog switches can be used in the box to switch various mod circuits in and out of the power amp feed. Besides pedals are expensive, and the op amp copies cost $5-10 in parts. Digital circuits like reverb being a bit more expensive, of course.

 

[Gnobuddy]

This is a stereo amp. Would you expect any difference if "left speaker tap" was mixed into "right preamp gain stage"?

What is the complete signal chain? Guitar to left channel, and the feedback from left speaker to right channel input?

If so, there doesn't seem to be any actual feedback loop, but rather, a parallel effects channel. I don't think that will cause you any problems.


-Gnobuddy