Help learning reverb circuits

[hesamadman]

This is one of those times where I don't really know how to word the questions I need to have answered. Im back at reverb and want to install one in a build, but I dont want to just copy and paste a reverb and say "ok its in and works...done!" Im really trying to understand how this works. Whether or not its tube or solid state, a reverb pan has an input impedance and output impedance. Ok so impedance. Now thats a word that I still to this day am not sure I fully understand. So this is where I dont even know how to word questions. Lets use the reverb pan in a Blues Jr for example. Since I have one and know its input impedance is 800ohm and output is 2575ohm. I dont EXACTLY know what impedance is (other than resistance). Not sure how correct this is but I guess im thinking it is the measurement of the LOAD? So does the circuitry at the input of the pan have the the impedance of 800ohms and the output circuitry have to measure the 2757? If thats the case.....how do you even calculate impedance of such things. I mean resistors are easy math but capacitors are frequency dependent.....


How off am I here and can anyone help get me on the right track? Pretend youre explaining to a dummy (you are actually).


Included schematic for easier explanations 

[PRR]

Resistance is ratio of voltage and current.

Impedance is a V/I ratio which is not dead-constant like a Resistor.

Grab a pry bar. The long end moves easy but a long way. The short end moves a short way but with great force.

> I dont want to just copy and paste a reverb
> Lets use the reverb pan in a Blues Jr for example.

My advice: get over it. Steal the design. A full analysis of what Fender did there would keep you off the guitar for months.

Question you did not ask: is that 800 Ohm (at 1KHz?) impedance being driven with a low impedance source (as we often do in general audio), or a high-Z source (spring pans sometimes are)?

There's also tough questions of how much gain to put before, and after, the spring.

And how to tap-out and mix-back the reverb to the main path.

Bass-cut before? Treb-cut after?

Plagiarize, plagiarize, plagiarize.

[2deaf]

IC Op Amps such as the TL072 can make a rather impressive reverb driver.  Unfortunately, the Blues Jr. reverb driver isn't an example of one of those.  I don't hear people praising any of the Fender Op Amp reverbs very often, if ever.

IC Op Amps require a double power supply such as the +15V / -15V supply in the Blues Jr.  You will see examples of single supplies, but the "ground" that the Op Amp sees is elevated so the Op Amp is really operating with a double supply. 

People write whole books on Op Amps, but we only need to know a few things in order to use them.  Things like voltage gain, max. input and output voltages, max. output current, max. dissipation, input impedance and output impedance.

In order to understand this driver, you need to know what I mean by "impedance".  To me, impedance is a general term that refers to anything or any combination of things that impedes the flow of current.  This includes resistors, capacitors, and inductors.  The impedance of a resistor is typically called "resistance".  The impedance of a capacitor is typically called "capacitive reactance".  The impedance of an inductor is typically called "inductive reactance".  The input and output on the reverb tank are inductors, sometimes called "coils". 

[ratgon]

I'm really happy this thread is happening. I too have never had the concept of impedance click for me. I understand it's relationship to resistance and the fact that it's frequency dependent but when I hear how we use it I get flummoxed.

What alludes me is the concept of one stage having a low impedance "output" or high impedance "input".

Are we simply referring to the concept of a voltage divider where the bulk of a signal will be distributed to the high impedance input and less to the low impedance previous output stage? Or is that crazy talk?

And, how do we even know what an overall input and output impedance is for any given stage?

I'll be honest, a BSEE and BS Physics and I still never understood this stuff. Although to be fair, I bailed on engineering at graduation to try and be a rockstar. That didn't pan out either.

So, thanks for this thread.

[2deaf]

The voltage gain for an Op Amp is proportional to the impedance between pins 1 and 2 divided by the impedance from pin 2 to ground.  The impedance from pin 1 to pin 2 is the impedance of R46 (4.7K) in parallel with the impedance of the tank input coil.  This does not calculate out the same as two resistors in parallel, but it is similar.  The impedance from pin 2 to ground is essentially 47 ohms because C21 (22uF) is large enough to be ignored at the frequencies we use.

The impedance of the input coil is mathematically related to the frequency and the inductance.  The inductance doesn't change, so the impedance is directly proportional to the frequency.  If you double the frequency, the impedance doubles.  If you cut the frequency in half, the impedance drops in half.  You are given the impedance of 800 at 1KHz, so you can easily calculate the impedance at any other frequency.

As the frequency increases, the impedance between pins 1 and 2 increases and the voltage gain increases.  There is more voltage across the input coil, but its impedance is also higher so that the current through the coil doesn't change very much with frequency.  Current is what creates the flux that shakes the springs, so you would prefer to have the same current at all of the frequencies that you use for reverb drive.     

[2deaf]

Lower frequencies are generally restricted for reverb because they give the reverb an undesirable muddy sound.  C20 (330pF) serves to attenuate those lower frequencies before they get to the driver.  The input impedance of the TL072 comes into play here because it will change the voltage divider and the frequency response if it isn't significantly larger than R44 (120K).  As it turns out, it is 10¹² ohms, so there is no problem.

Higher frequencies create a couple of problems.  First, they give the reverb an undesirable shrill sound.  Second, the increase in voltage gain at higher frequencies will cause the output voltage to exceed the maximum.  R46 (4.7K) prevents the voltage gain from increasing linearly with frequency because it is in parallel with the input coil.  Just like resistors in parallel, the total impedance of a coil and a resistor in parallel cannot exceed the impedance of the component with the lowest impedance.  Not only that, the total impedance is influenced by the resistor more and more as the impedance of the coil increases.

Even though R46 will control voltage gain, the circuit has the potential to exceed the maximum voltage output of the TL072 and cause clipping.  Hard-clipped signals driving reverb yields what I regard as an unacceptable reverb.     

[EKDENTON]

Impedance is a measure of opposition to AC current flow in a circuit. Things that can attribute to impedance is resistance, inductance, and capacitance.  For a maximum load transfer the source and load impedance should match.

[ratgon]

Maximum load transfer is with matching impedance? So my voltage divider concept is clearly wrong. Hmmm.

[2deaf]

I don't know what the output impedance is of the Op Amps that I've used, but from experience I can tell you that it is very low.  They would have to be something on the order of 100 ohms in order to drive the capacitive loads that I have gotten them to drive.

The simplest model of the output of an Op Amp is a signal generator in series with an output resistor.  The output resistor is the output resistance for the device.  Your load impedance is then connected to the output resistor so that you have a voltage divider connected to a signal generator.  If your load impedance is much smaller than the output resistance, most of the signal will be dissipated across the output resistor.  If your load impedance is much larger than the output resistance, most of the signal will be dissipated across your load.

The tank input presents a low impedance load for the Op Amp, especially at lower frequencies where the tank may go as low as 66 ohms.  The Op Amp will still drive the tank at 82Hz, but it doesn't need much because the input signal has been seriously attenuated at that point.     

[2deaf]

One more thing about this driver is that the input coil is part of the feedback loop for the Op Amp, so both conductors going to the input coil must be isolated from ground.  One of the conductors going to the input coil is the shield of the Reverb Input cable, so it must be isolated from the tank housing and the amp chassis or any other ground.  The Reverb Output cable must have the shield grounded at the amp and connected to the tank housing.

[PRR]

> circuit has the potential to exceed the maximum voltage output of the TL072 and cause clipping.

I think not.

TP20, speaker shows 2.6V. We expect the amp clips near 12V (18W in 8r). So the test-levels are about 1/5th of maximum.

There's no user-knob in the reverb drive section.

TP22 is noted about 0.4V. When the amp makes all 18W, the level at the tank may be 5X more, 2V. Chip with 30V supply can do close to 10V. There's headroom in the reverb drive when the speaker stage is maxed-out. Reverb clipping will happen when the final is overdriven about 5X, and all signal subtlety is lost.

[Sonny ReVerb]

I've been looking at some of the old Supro and Ampeg circuits that used a transformer-less, cap driven design. What drove the decision to use one versus the other? (Interesting read.) How do they operate, perform, and sound compared to the classic Fender blackface reverb? Which should I choose for my 'new' design?

PRR made an excellent point in an earlier thread:

Fender found a glut of 15K:4 OTs made for old tube radios about the time that transistors started eating that business, so he used a modest tube plus a cheap transformer into a 4-10 Ohm tank.

Design decisions are sometimes (mostly?) driven by economics rather than optimum performance. Maybe I can use a better design for my amp, even though it might cost a bit more. So, I try to understand the circuit differences...

Then my head starts to hurt and I realize, if people way smarter than I am haven't come up with a better design in the last 55 years, go with what works. From one such person:

My advice: get over it. Steal the design.

Hope I didn't 'impede' the discussion  ;)

[sluckey]

I had the Ampeg Gemini II and the cap driven reverb sounds good. Sunn transformer driven reverb sounds good and Hammond reverb units sound good. Magnatone cap driven reverb sounds good. All the circuits that use the long Accutronics reverb tank sound good. But the only reverb I've ever heard that will do that Dick Dale sound was made by Fender. If you just want a good sounding reverb build any of them. But if you want the reverb to also do good surf sounds build the Fender circuit.

[2deaf]

> circuit has the potential to exceed the maximum voltage output of the TL072 and cause clipping.

I think not.

Reverb clipping will happen when the final is overdriven about 5X . . .

Doesn't that confirm what I said?

That 2.64Vac at TP20 could be peak because rms voltage is designated "V_rms" elsewhere.  If so, it would take 6.4x to get 18W.

I think the final only needs to be overdriven 3x to clip the reverb driver, but either way the output is way overdriven by the time the reverb clips.

Imagine if you had that amp cranked up and you slammed a chord then immediately dampened the strings.  You would be left with some real ugly reverb trailing off.

[2deaf]

In addition to limiting higher frequency gain, R46 (4.7K) also limits the driver voltage gain should you disconnect the tank.  Without it, the voltage gain with the tank disconnected is typically 200V / mV.

[2deaf]

C16 (.0022uF), R40 (430K), C35 (.0015uF), R41 (50K), C20 (330pF), R43 (910K), and R44 (120K) comprise a bandpass filter for the reverb driver input.  It is fairly flat from 700Hz to 2KHz then drops off 3dB at around 400Hz and 3KHz.  This is a very typical frequency response for reverb drivers for the reasons cited above. 

[2deaf]

The output impedance of that tank is 2575 ohms at 1KHz, which is pretty much all you can find these days.  The only good that number does for me is that it allows me to calculate the inductance of the output coil.  That inductance is then used in mathematical formulae that assist me in choosing components.

R47 (470K) gives the input of U1B a reference to ground when the tank has been disconnected so that the Op Amp doesn't freak out.  The output coil in conjunction with R47 will roll-off high frequencies.  The corner frequency of this roll-off depends on the resistance of R47 and the inductance of the output coil.  The inductance never changes, so we would select a resistor that puts the corner frequency well above the audio range.  High frequency roll-off could be a good thing to suppress high-frequency oscillation as long as the corner frequency is not too high, so we wouldn't want to select a resistor that moves the corner frequency way up.  220K is a value that works well for both considerations.     

[2deaf]

C22 (.0022uF) and the tank output coil have a resonant frequency that depends on the capacitance of C22 and the inductance of the output coil.  At this frequency, there is a spike in the output voltage from the tank.  The output voltage at frequencies on both sides of this spike are drawn up just like they are with an equalizer with a high value of Q.  In this case, we can calculate the resonant frequency using the inductance (410mH) that is derived from the output impedance of 2575 ohms at 1KHz and the capacitance of 2200pF and we get 5.3KHz. The net result is that the reverb sounds brighter.

The value of R47 will affect the magnitude of the resonant frequency voltage spike if the corner frequency of the coil/resistor high frequency roll-off is too low.  This is another reason to keep the corner frequency sufficiently high.

Some people put a resistor in series with C22 to reduce the magnitude of the spike and to increase the range of frequencies drawn up on both sides of the spike.   

[2deaf]

U1B is a standard Op Amp gain stage with a voltage gain of 166 that is set by R50 (330K) and R49 (2K).  I'm not sure what R48 (10K) does, but I have seen resistors used in this position before with the explanation that they restrict inrush current at start up.  I do know that these circuits will work fine without R48 because I have done it many times.  C24 (10pf) reduces gain at frequencies well above the audio range in order to eliminate high frequency oscillation.  I always use these.  You will note that U1A doesn't have one of these capacitors and it has no high frequency oscillation problems.  I always kind of thought the capacitance of the input coil takes the place of these capacitors.

R51 (50K-L) is the reverb level control with a bright capacitor C26 (.0033uf). 

R42 (220K) and R56 (470K) are the mixing resistors for the dry signal and the wet signal.  R51 is way smaller than R56 so that the load seen by the dry signal is about the same no matter where R51 is set at.

[sluckey]

Are you carrying on a conversation with someone?

[2deaf]

Are you carrying on a conversation with someone?

No.  Just answering the OP throughout the day while I am stuck at home recovering from surgery.  I'll knock it off.

[sluckey]

No need to stop. It just sounded like you were talking with someone but I was only seeing one side of the conversation. I've been following along but thought I might be missing something.

[PRR]

2deaf is answering hesamadman's question, a massive undertaking best served in small bites.

Heal well!

[sluckey]

I'm surprised that hesamadman has had nothing to say since his initial post. I know he's been on board.

[ratgon]

So, um, my output and following input impedance as voltage divider concept is wrong?

[PRR]

> BS Physics....
> voltage divider concept is wrong?

Distinguish between Voltage or Current and Power.

You want to grind grain quickly. You need Power.

Before power on wires, this was water-wheels.

(I don't know if water-wheel analogies are still legal in BS courses.)

You have a water-fall 50 feet high but a thin stream.

You have a wide-wide river but hardly drops at all.

Which will do more WORK? (More data needed.)

The "Matched Impedance" concept works when you are STUCK with a certain source impedance and need to extract maximum POWER. For my 120V and (say) 12 Ohm power line, max-power will be a 12 Ohm load and will get 60V. However the lights will be very dim at 60V. And there is as much heat in the wire as work in the grain grinder.

Utility electric power systems are NEVER run "matched". I actually have a 0.4 Ohm line and >4 Ohm load, 10:1 mismatch; 50:1 mismatch is more usual closer to civilization.

Audio systems are ALMOST never run "matched". But it is too late to go into the many reasons matching is usually folly.

Reverb drive is one of the few cases where we mis-match "the other way" (hi-Z feeds lo-Z). Again I plead bed.

[ratgon]

Alright, I'm gonna dive in. Physics degree be damned!

In trying to, now, after 30 years, understand the relationship between input and output impedance I keep seeing it like this. And it may be wildly inane but I'm still stuck with it as I try and learn.

In an amplifier which seems to be concerned with amplifying voltages, the relationship between any stage and the stage driving it seems to be, conceptually, akin to a simple voltage divider. if we want to maximize Vin as a perecentage of the existing Vout, then we'd use Vin = Vout * Zin/Zin + Zout. Which is a basic voltage divider. So we'd want input impedance much larger than the output impedance driving it.

I don't know. That's just as far as I've got with this whole thing. Clearly the science degrees didn't pan out which is why I'm back in school on the complete other side.

Anyway, thanks, truly, for trying to help me get this.

[PRR]

> In an amplifier which seems to be concerned with amplifying voltages

Yes. But no amplifier input is infinite impedance or zero Power. A naked vacuum tube grid may be >100Meg. But to use it well we need a grid-leak, typically <=1Meg. Capacitance everywhere and at the top of the audio band a high-gain tube may show C equivalent to about 200K. Guitar amp is not all gain, we have volume and tone pots. Tonal curve can be inductors or capacitors; since caps are cheaper all our tone-networks have declining impedance with frequency. 

So all our loads need real current and Power. We strive to minimize it. But it has to be accounted.

[ratgon]

Ok, got it. Well, starting to get it. But this helps a lot. If nothing else even in a terminology way and, for me, a way to visualize the whole as a series of "discrete" circuits.

[2deaf]

Heal well!

Thanks, man.

[2deaf]

So, um, my output and following input impedance as voltage divider concept is wrong?

As PRR said, you have it right for voltage.

Op Amps have an output impedance, but you don't see the voltage drop across it because of the negative feedback.  When there is a voltage drop across the output impedance, the negative feedback, which is taken from the output, becomes less so that the internal voltage gain becomes more and it compensates for the voltage drop.  The external gain remains the same no matter how much the voltage drop across the internal output impedance becomes as the result of a small external resistor until the device reaches one of its limiting factors.  The larger the output impedance, the sooner a limit will be reached with the same external resistor.

Other devices without negative feedback loops work exactly like you said. 

[hesamadman]

Man. I have some reading to do. Thanks all.

[2deaf]

Man. I have some reading to do. Thanks all.

As long as you are reading, here's a little more.

The impedance from pin 6 to ground on U1B at 1KHz is essentially the resistance of R49 (2K) because C23 (0.47uF) is large enough that the impedance of R49 in series with C23 is still just about 2K.  But at 200Hz, R49 in series with C23 has an impedance that is significantly higher than 2K.  This changes the ratio of the impedance from pins 7 to 6 to the impedance from pin 6 to ground so that gain decreases at 200Hz as compared to 1KHz.  R49 and C23 cause a low frequency roll-off just like C19 (.0033uf) and the 100K Reverb pot. on Doug's AB763 reverb recovery.

Actually, C21 (22uF) is not large enough to ignore at lower frequencies, but other factors come into play that just about cancel out that effect while creating other effects that reduce the current through the input coil.  We won't go into that here.

[jjasilli]

hesamadman,  the focus of your original question is not clear to me:  are you asking about impedance "matching" generally (with reverb tanks serving as the entree to that topic); or are you specifically asking only about reverb tanks? 


I put "matching" in quotes:  generally the design goal is to "match" either current, or voltage, from the source device to the input device.  Generally, current matching is not the goal of audio/guitar amps.  For current matching the source impedance should equal the input impedance.


Voltage "matching" is typically the concern for audio/guitar amps, their signal inputs and internal stages.  This is more appropriately called "bridging".  Generally, the input impedance should be  at least 10X the source impedance, or the source voltage and signal bandwidth will be degraded.   

There's lots of info if you search this Forum, and the web generally.  A multi-volume encyclopedia can be written on this topic covering such things as mic's; guitar amps; mixing boards; line out & DI; speaker driver or transducer (reverb tanks, for example), over the audio range; etc.  It's impossible to really cover the topic in a thread on a forum, unless there is a specific question.