A "Transparent" Buffer?

[Transmogrifox]

Looking at the 6G15 schematic
I posted about the cut off as a first simple thing to try
...
Also the output of this circuit looks like it was made to connect to a high input resistance like 1 meg(C1 and C10 being 220pf for 440pF that and C17 being .1uf the mixer being 250K? hard to read)

If you are looking to bypass the effect disconnect the top of the mixer control then short the output of C17 to the output jack wiper

The summary of this thread is not about the sound when bypassed.  It's about the sound of the dry "thru" component when verb is mixed in. 

The tone control and hpf output color the tone as you adjust the mix.  Also the mix attenuates the dry component while mixing in the verb.  OP considers this undesirable.

We have established that a mixer is wanted in which the dry component remains at unity while verb is mixed in.  He is already using TBP external so his bypass is as clean as a wire.

[Gus]

Transmogrifox

I thought part of this thread was about when the mixer control was set to dry it changed the sound.  There is no way around that in the stock circuit without changing the circuit but it should be less when the drive to the wet side is turned off.

You posted a few good circuits for minimizing the output interactions.

I guess the Fender Reverb was designed to be as inexpensive to to build as possible. I am guessing some people like the sound with the stock output interactions

[Transmogrifox]

Gus - I was under the same assumption at the beginning. 

I would agree that the mix control is part of the original effect and probably most who would build an emulator want to "get that sound" including the passive analog interaction around the mix knob.  In this case the OP wants "that sound" of the reverb without losing fidelity of the guitar signal going through.

[Loose]

To better describe the added high end to the signal: It's not harsh, it sounds like the midrange frequencies have been shaved and the high end was extended.

I tried adding another cable in front of the effect for added capacitance but it didn't achieve the desired effect.

I also added two 22n caps in parallel with the 22n caps on board, the result was additional midrange, but the sound lost some output and became thin.

Maybe I should remove the first amp stage all together and use an alternate way to add back the high end and midrange that were lost when I had only the mixer part?

Thanks again for the assistance guys

[Transmogrifox]

You're talking about the kind of thing where you need an oscilloscope and a frequency generator to map it out.

The poor man's version = Audacity + PC sound card.  Audacity can be set up to generate a frequency sweep.  You send that out to your buffer and record what's coming back.

If you could capture a frequency sweep of your sound card looping back on itself, then capture frequency sweep through the buffer then you would be able to see more scientifically whether the buffer + mixer circuit itself is a problem.

Set the sweep from about 60 Hz up to 15 kHz, logarithmic sweep, and duration about 2 minutes or more.

It is counter-intuitive that adding more 22n caps in parallel would make it lose some output and become thin.  That should make it more fat if it has any kind of audible effect at all.

[Loose]

So I found out I've been using 220n caps instead of 22n
I replaced them with 22n, but the loss of midrange and extended high end is still there. Then I put 2 more 22n in parellel, a bit better.
Then I went back to just the second amp stage, I don't know if there was much of a difference at this point.

It doesn't sound bad or harsh, but I would like it to be closer if possible.
I suspect it might be the capacitance as you suggested since that's what a buffer does. Also, how can I know if the impedance is different from my amp's input?

I think before figuring out the frequency sweep stuff I could try a 100n input cap and another cap to ground to take off capacitance. Thoughts?

Thanks!

[Transmogrifox]

In theory the 220n should have been ok once the circuit stabilized so I would be suspect of less ideal characteristics of the capacitor.

Anyway what amp are you using?  If a schematic is available then you can determine the input impedance by looking at all resistors to ground on the front-end signal path.

I think before figuring out the frequency sweep stuff I could try a 100n input cap and another cap to ground to take off capacitance. Thoughts?

Changing to anything larger than 22n on the first stage won't make an audible difference because it is already a gain reduction in the micro dB (yes, less than 1/1000th of a dB) at 82 Hz (Guitar low E).

On the second stage the roll-off is -0.15 dB at 82 Hz (low E).  I cannot detect 0.15 dB with my ears.

Anything larger than 22n should not be making an audible difference in the mid-range nor anywhere unless your resistor values are wrong (are you certain 470k isn't 47k and are you certain 4.7M isn't 470K?).

This definitely would make an audible difference if your resistors were 1/10th of the expected value.  Check the R's with a meter and make sure they are what you think they are.

[Loose]

Here's the schematic: http://www.blueguitar.org/new/schem/vox/ac15htvh1_handwired.pdf

My other amp is a 67 Blackface Bandmaster, AB763.

How do I calculate the input impedance by looking at the schematic? And if I would make sure the impedance on the mixer matches the amp, it still means it might not play as nice with other amps I plug it into?

I double checked and the resistor values are correct.

Thanks

[Transmogrifox]

LOW input is about 130k
HIGH input is about 1M

The high frequency roll-off on the worst case is >100 kHz so you aren't losing anything on the high end from the amp.

If you're still working from this:

Then change one of the 4.7M resistors to a 1.3M (or something close to that). A 1.3M parallel with 4.7M is 1.02M, but anything between 1.1M and 1.5M will get you close enough.

Then, if you still notice it being different from going straight to the amp then it's likely the cable interaction with your guitar.  Add about 30pF per foot.

A 600pF cap to ground on the input to the Surfy Bear would be a good starting point.

[Loose]

I did a bit of googling and looks like my Fender also has 1M input impedance on the high input, and Marshalls have it as well.
BTW: it seems like it would make more sense to match the pedal's output impedance with the amp's input impedance?

I am working off the first schematic that you've posted (the second opamp only), I put a 1.5M resistor in parallel with the 4.7M resistor to ground,  the measured resistance is 1.1M. I also put a 560pf cap in parallel, which I later changed to 220pf.

The sound is almost there, i'm just missing the extra punch I get when the pedal is bypassed. Should I add the first amp stage back in to assist with that?

Thanks

[Transmogrifox]

1M is pretty typical.  You do need that first buffer amp stage to get there.  If you eliminated the buffer then all you did was lower the input impedance from about 430k to about 334k and that would account for the loss of "extra punch".

My suggestion about adding the 1.5M in parallel was assuming you were still using the input buffer.  This is what allows you to get the impedance up to 1M.

No it does not make sense to match the pedal's output impedance to the amp's input impedance in the normal sense of impedance matching.  For these applications you want output impedance ideally zero (as low as possible) and input impedance as high as possible.

This impedance matching stuff is related to transmission line applications where you have are driving a long (100's of feet to miles) cable and you want to match impedances to prevent reflections and to maximize power to the load.

Impedance matching also has some meaning in the context of tube amps where the output impedance off the plate is pretty much a set value and you want to match that to a speaker impedance to maximize power to the load (and not minimize power dissipated in the tubes or transformer).

[Loose]

I finally had time to mess with the circuit again.
I added the buffer back in and left the 1.5M resistor in parallel with the 4.7M input resistor to ground. So that gives me 1.1M to ground from the input, 4.7M to VB, and 1M to ground on the surfy bear (R21). Is that the way it should be?

Thanks!

[Transmogrifox]

It isn't the way it should be to get a 1M input impedance. 

The input impedance is closer to 500k because you now have 1M from the Surfy Bear in parallel with 1M from the combination of the 2x4.7M and now-added 1.5M.  You can't test this in parts because it changes depending whether the surfy bear is there or not.

Suggesting that you add the 1.5M was assuming the Surfy Bear was being fed from the output of the buffer. 

If you leave the Surfy Bear connected to the input jack you have the following:

4.7M|1.5M|4.7M|1M = 477k (which for me & my rig would be fine -- don't think I can detect the difference between 477k input Z and 1M).

The 4.7M going to Vb still looks like a resistor to ground over the range of audible frequencies.  Vb is an AC ground, or known as a "virtual ground" because it is AC coupled to ground through the filter capacitor on the resistor divider.

If you remove the 1.5M in your current configuration here is your input impedance:
4.7M|4.7M|1M = 701k

If you relocate the surfy bear input to the output of the buffer and leave the 1.5M in place:
4.7M|4.7M|1.5M = 915k

What I was suggesting was to REPLACE one of the 4.7M resistors with a 1.5M, not simply adding it in parallel:
4.7M|1.5M = 1.14M


Just experiment with those and see if you can hear any difference.  Then if you find the 3 options don't really sound different then go with whatever is easiest and fits your layout the best.

If you can understand how I came up with all of the scenarios I presented above then I think you will be well on the way of understanding how to estimate input impedance of a circuit.  The main strategy is to treat all capacitors in series with the signal as short circuits (wire) and capacitors going to ground as open circuits. 

Resistors going off to places like "VB" are still shunt resistors that show up in parallel with input impedance because it is still a low impedance path to ground.

Now maybe you can answer your own question about whether it is the way it should be.

[Loose]

Thanks for the explanation.

I'm a bit confused about your suggestion to add the 1.5M assuming the Surfy Bear was being fed from the output of the buffer. 

In a previous post you said the Surfy Bear should come out of the same input as the buffer:



So are you now suggesting to change that?

[Transmogrifox]

We're mixing up ideas.  I'm suggesting you take the time to understand the options I presented above and go with one of them.  There are several ways to do this right and I think I have confused you with too many options.

To match the input impedance of an amp you need to get to something near 1 Meg.

That sketched-up schematic you re-posted was before attempting to get 1 Meg.  Presumably you tried that and you didn't like it.

This schematic that I posted that allows you to get a really high input impedance failed to clearly state the surfy bear input is fed from the buffer and thus your confusion:

For example, above schematic has input impedance of 4.7M|4.7M = 2.35M if you don't have any other connections than what is shown.  If you add other resistors to ground around the input (including surfy bear input) then those all show up in parallel with the 2.35M input impedance and lower it from there.

To get the benefit of the buffer now the mixer and the surfy bear inputs get fed off the buffer and then you can more exactly control the input impedance by changing the resistor to ground at the input of the buffer.

If you take the time to figure this out for yourself then you will have this licked within a few minutes.

The best option for getting 1M input impedance has 1.5M to ground -> capacitor -> 4.7M to VB -> buffer ->split buffer output to surfy bear and mixer, done.  The buffer just goes in front of all of it so you can get the advantage of being able to control input impedance.

I suggested several other ways that will get you close to 1M in different configurations, and some that won't. 

In the end the best advice is what my calculus teacher gave me:  "If you get to a spot where you don't understand something, just do stuff you know how to do".  In other words just try some stuff and you will start to understand just by getting your mind wrapped around the whole problem at hand.

In the end your ears will tell you if it's good.

[Loose]

I've changed the Surfy Bear input to come out of the buffer output and replaced the 4.7M resistor with 1.5M.
I've tried adding the cap between pins 6 and 7 of the second op amp, it sounds like this is a step in the right direction! I started with 100pf which took too much highs and i'm now experimenting with different values.
Is there any benefit in adding a cap between pins 1 and 2 of the first op amp?

Thanks!

[Transmogrifox]

If the first op amp is a buffer as drawn then the cap will not do anything.  It would be equivalent to twisting the 2 leads together and soldering it onto the board 

You can put a cap to ground at the input of the buffer to simulate additional cable capacitance.  Somewhere between 330pF to 680pF is probably the right range to experiment with.

[Loose]

I settled for a 50pf cap between pins 5 and 6. Sounds pretty close!

Thank you for all of the assistance Transmorgifox, I couldn't have done it without you!

One last question: I want to build an additional version that would just split the signal, so I can send the dry signal to one channel of the amp, and the wet signal to the second channel. I assume all I need to do is remove R4 and have the Surfy Bear output go to the second channel, right? And in that case, do I even need the second opamp? Maybe just to shape the dry sound?

[Transmogrifox]

I would think the best way to do this version would be without any kind of extra buffer circuitry.  Your "bypass" switch would just be the kill switch for the reverb output.

To get away from effecting your dry tone just do the following to the stock Surfy Bear board:
1) Remove C12
2) Change R21 to something really big like 10 Meg.
3)  Insert "kill switch" on the output jack.  All you need to do is make it short the output to ground when the switch is activated.  With a DPDT one switch pole can activate an LED and the other shorts/unshorts the output.
4)  Take the "through" jack directly off the input jack (won't matter which is "to instrument" or "to amp")

That will avoid a lot of fooling around and you won't have to build any floating circuit boards. 

The input capacitance of the JFET is equivalent to an extra foot of cable or switching to a different brand of cable so this is of no real effect.  Your guitar only sees a parallel 10 Meg with your amp AND your guitar is always connected directly to your amp so you don't hear any change in the dry sound when you hit the kill switch. 

You will probably end up liking the splitter version better than the buffered version.

[Loose]

Isn't a buffer essential when splitting signals? I've tried passive splitting my signal in the past and there was a drop in volume.

Thanks!