The best buffer... or at least my favorite!

[shredgd]

Hi,

I've been experimenting (again) with buffers lately.
Although you can find lots of information about them, I found that most proposed circuits overdo their main purpose, which IMO should be giving you the same tone and feel of your guitar through a good quality 5 meters cable to your amp: not less, not more. Why? Because pedals and amps are designed this way! So giving you extra-brightness or extra-anything "thanks to its billion-ohms high impedence" or something similar, might impress you at first, but after struggling with amps and pedals settings, you will realize that those kind of buffers alter your tone in a bad way!

After dozens of A/B tests, I've come to the same conclusion as Analog Man states in his Buffer Pedal page: nothing beats a bipolar transistor buffer in terms of tone and feel! But I searched even further. I started from the basic buffer schematics you can find online and compared them to the buffer schematics in manufactured non-true bypass pedals, analizing the role of each component and choices of components values. This way I realized 4 important things:

- leaving off or including the 1k input resistor of the typical Tubescreamer buffer (my starting platform) DOES give an audible difference in high frequencies content, but my final choice was to leave it in place: protecting your transistor is a good thing, and the buffer sounds and responds fantastic keeping it in place
- electrolytic DC decoupling caps DO eat some of your high frequencies (presence): a film cap parallel to the 10uF output cap is not just an option, but really necessary to "take the blanket off" (without adding any frequency you could already get with the guitar-cable-amp setup, which is your reference)
- higher gain transistors (2N5088, MPSA18), which should give you the closest to unity gain, give you a sort of "congested" tone: do yourself a favour and try a 2N2222 (my favorite choice) or a 2N3904, both of which have a typical Hfe of about 200. You will never go back to higher gain trannies for this application! No big surprise that 2SC1815s (the most used in buffers for Japanese pedals) have that same average Hfe...
- the resistor to ground after the output DC decoupling cap is a NEEDED component. You often don't see it in the "buffer schematics" online, however you always see it in manufactured non-true bypass pedals, even though they wouldn't need a pull-down resistor. I've had a buffer without that resistor for years in my rack setup pedalboard, because I thought that resistor wasn't really important. The rack includes a Mesa Triaxis preamp, not the worst out there... believe me, I checked the preamp various times, because something wasn't just quite right with my tone (a lack of bass and excessive presence), to discover that such a single, apparently redundant resistor makes the difference! Also, after A/Bing a 100k vs 10k value (do you remember the TS9 vs TS808 eternal struggle?), I found out that the 10k is the way to go: more bass and more definition to it, more "breathe". This is apparently in contrast to some technical discussion where such a resistor is suggested to be much bigger than the emitter resistor to avoid clipping of the negative half of your sine wave, but I consistently prefered 10k to 100k in my tests, even after repeating the comparisons days apart.

Sorry for the long post, but I wanted to share the results of my struggled experiments and tests to someone who, as I did, isn't still satisfied with his buffered tone.

Giulio

[Groovenut]

Do you suppose the differences you are hearing could be caused by the differing input impedance affect on the HPF that is formed between the input cap and the bias resistor? The lowering of the gain will change the input impedance and therefore the knee of the HPF at the input. I am wondering if the clarity you say you hear is a result of filtering out a good deal of the sub-harmonic frequencies.

Just a thought 

[shredgd]

You mean the difference between the lower and higher gain transistors? No, AFAIK there shouldn't be any significant difference in impedance between the two!

[antonis]

Obviously, we'll have to "fight" between theory & practice, so let's do it... !!

<there shouldn't be any significant difference in impedance between the two!>

Input impedance of an Emitter follower is: (h_fe + 1)(R_e + r_e) (ignoring bias resistors & R_load..)
Output impendace             //               : R_source / (h_fe + r_e)

From the above we have to conclude in: the higher the transistor gain the more "effective" the follower..!! 


<the resistor to ground after the output DC decoupling cap is a NEEDED component.>

I'll have to agree with you ONLY in cases we need an almost predictable HPF..
(i.e. with an unknown input impedance of next stage but with the certainity that its value will be much greater than our shunt resistor..)

[Groovenut]

You mean the difference between the lower and higher gain transistors? No, AFAIK there shouldn't be any significant difference in impedance between the two!

Yes that's what I was meaning. Since the input impedance is affected by the gain of the transistor, the differing gains will affect the HP filter knee at the input of the buffer. Granted the knee variations are in the sub-harmonic range but, there is a lot of energy in those frequencies (20Hz-35Hz).

[shredgd]

Yours are wonderful and very useful replies!
I didn't know that the hFE changed the impedance, maybe because it is still a high impedance, even with lower gain transistors, so you don't generally find it specified.

Too bad I don't have a setup to record high quality clips showing the A/B tests I did! But I tried many times to be sure I wasn't biased (actually I believed I needed the highest gain transistors and highest load resistor after the output cap, basing on what I had read) and I even kept the same position of my head relative to the amp speaker to be sure I always listened the same way! 😂

[antonis]

I even kept the same position of my head relative to the amp speaker to be sure I always listened the same way! 😂

That is called "Good Laboratory Testing Practice"...!!! 

[Groovenut]

I realize this isn't real world but here's a freq response graph for 3 but buffers using different transistors. You can see the differing input impedances affect on the bias voltage and on the HP knee. Subtle but measurable.

[shredgd]

Fantastic! I want that software!
Actually I use 0.1uF for the input cap 
Could you do the simulations again with that change? (If you saved it, don't bother drawing everything again for scratch if you didn't!)

[Groovenut]

Sure thing. It just shifts the knee lower but maintains the same differences between transistor gains.

[shredgd]

Thanks!
I knew, but as you can see, now the changes are really at the very limit of the audible range, which by the way should already be cut off by guitar loudspeakers...

[Groovenut]

Thanks!
I knew, but as you can see, now the changes are really at the very limit of the audible range, which by the way should already be cut off by guitar loudspeakers...

True but the loudspeakers are the last thing in the chain. To my knowledge there is no part of the guitar pickup circuitry that is specifically there to limit low frequency. Only the ability of the instrument to generate them. The lowest fundamental is 82.4Hz on a standard tuned 6 string, but there are harmonics created below as well as above the fundamental. There is a large amount of energy in those very low frequencies. Enough to dominate how a circuit behaves and you cant even really hear them. It's like having super high frequency oscillation causing a circuit to misbehave. You cant hear it and without a o-scope would never know it was there and causing the circuit to freak out, which affects the frequencies that you can hear. The same should go for low harmonics extending below the fundamental.

My 2 cents...

[shredgd]

I totally agree with you, and it also is the only explanation I can think of!

[antonis]

Lawrence, now you're a cheater..!!! 

Using 510k for Thevenin equivalence resistance of voltage divider isn't "flattering" for all 10k Emitter resistors... 
(those of higher gain BJTs feel more comfortable and cozy..)

You should be more unprejudiced if you've biased all BJTs with the same voltage on their bases..!!

[Groovenut]

Lawrence, now you're a cheater..!!! 

Using 510k for Thevenin equivalence resistance of voltage divider isn't "flattering" for all 10k Emitter resistors... 
(those of higher gain BJTs feel more comfortable and cozy..)

You should be more unprejudiced if you've biased all BJTs with the same voltage on their bases..!!

True I should have isolated each circuit from the other for the test (and I did) but it meant 3 separate images and was all that different a result. I also should have built the bias network instead of just using a 4.5 volt source (but I was lazy). It doesn't skew the results much. However in the interest of science, I'll post the results with everything added in for comparison.

[DrAlx]

[The lowest fundamental is 82.4Hz on a standard tuned 6 string, but there are harmonics created below as well as above the fundamental.

Depends on the instrument.  A plucked guitar string can't vibrate below it fundamental frequency.  To get a lower frequency than the fundamental, the fundamental needs to couple energy to some other lower frequency resonance on the guitar neck/body.
I've amplified my acoustic guitar with a piezo and got VERY low frequency feedback between body and speaker that is nothing to do with the strings vibrating.  The whole guitar top/back/cavity was resonating.  So an acoustic guitar can definitely resonate at a frequency below the low E.  Plucking notes that are multiples of that body resonance frequency will couple some energy to it.  I'm not sure solid body electrics are resonant to that extent though.  Maybe when I get home tonight I'll thud my electric guitar neck with the strings muted and see what sort of pitch the thud gives.

[anotherjim]

It's something even the hardened might forget. They did sit me at a bench long ago with various bit of stuff and then I had to chart all the hybrid parameters. I've forgotten most of all that stuff.

With an emitter follower, the emitter resistor voltage change must tend to change the base-emitter voltage - it opposes change in the base current = higher input impedance. So Hfe is an important factor. Input resistance is simply = Hfe x Re.
If you have x100 Hfe a 10k Re gives 1M into the base. That's low enough to change a guitar input bias resistance value. Not many small signal silicon transistors will be as low as 100, but some of the general purpose switch types can be and lower.
Hfe of 100 would drop the 510k bias to 340k.
That ought to be inaudible difference in bass with 1uF coupling caps, although the top end loading with humbuckers could very well be audible as a darker tone,

There is LF out of guitar during attack - especially palm mute heavy chugging. The more "slam" you can get, the better.

[antonis]

The more "slam" you can get, the better.

You have to admit it sounds a bit kinky.. 

[shredgd]

I measured the bias voltage in one of my buffers: with a power supply of 9.38v, I get 3.70v with a 2N2222 (hFE 206), 4.25v with a 2N5089 (hFE 745).

[Gus]

To add to Groovenut sims
Reply #7 first screen shot might be of help you can adjust the voltage divider to raise the emitter voltage(keeping in mind the current in R2) or set it to what you want with the transistor beta and input resistance you want.
http://www.diystompboxes.com/smfforum/index.php?topic=99884.msg876946#msg876946