Cornish Buffer Help

[yeeshkul]

Actually, what oscillation frequencies are we talking about?

[antonis]

Those at which load capacitive reactance practically(*) reaches zero ohms..
e.g. 100pF exhibit an impedance of 16R at 100MHz..

(*)always dependent on previous stage output impedance..

[yeeshkul]

And can oscillation on frequencies outside of the audible spectrum somehow disturb the function of the audio circuit?

[antonis]

And can oscillation on frequencies outside of the audible spectrum somehow disturb the function of the audio circuit?

Yes..!!

search for operational amplifiers Gain-Bandwidth product ..

[yeeshkul]

Thank you i am reading about it right now. Do these problems apply also for one stage transistor buffers like that one above here?

[antonis]

Thank you i am reading about it right now. Do these problems apply also for one stage transistor buffers like that one above here?

Of course they do..
Buffers DO exhibit gain..
They are considered current amplifies..

You know, Slew Rate can be refered on both voltage and current..

[Rob Strand]

Actually, what oscillation frequencies are we talking about?

The frequency depends on the capacitance in the circuit and the wiring inductance.   You can loosely think of the capacitance in the circuit is more or less fixed so the frequency depends inductance but it's a bit of an over simplification.

The inductance depends on the wiring.  External wiring, PCB traces (not only to the transistor base but also the power rails to the buffer) and even the length of the leads on the transistors.    You would probably end-up with frequencies in the 10's of MHz but it can go outside of that.  Adding the 1nF cap will push the frequencies down.

The maximum capacitances I gave previously are based on the circuit being stable for *any* inductance.     You don't know what the external wiring inductance is so you need to consider worst case.    When you have capacitive loads close to those maximum values the range of frequencies where the circuit can oscillation is very narrow.    If the inductance doesn't match that narrow range of frequencies the circuit won't oscillate even with the load capacitance at or greater than the maximums.       That's why you don't see potentially problematic circuits bursting into oscillation so often.    However one day you might plug something in which just happens to line up with the bad range of inductance and it mysteriously starts oscillating after working for years without a problem.

The technical side of this is fairly complicated and dependent on external factors.   Most designs put on a few precautionary parts like the 1k input resistor and the 150 ohm output resistor and hope for the best.    If it doesn't oscillate on the bench then everyone is happy.   People that design power amps are much more cautious.    Replacing transistors with different brands can often cause happy amplifiers to turn into troublesome oscillators.  That's why you see extra resistors and caps on some power amp designs.  DC laboratory power supplies are another example where a particular load causes the power supply to oscillate - even power supplies from reputable companies!

[yeeshkul]

Thanks Rob, it makes for very interesting reading.

[antonis]

DC laboratory power supplies are another example where a particular load causes the power supply to oscillate

And burn-out output transistors soon enough..

[bonehead1972]

hi guys,
i would like to share some thoughts..
thanks to you i got this done. it sounds good, bias voltages are good...but after more extensive playing and comparing i noticed some hint of extra presence {not highs} that i dont like. tweaked all the caps one at a time with no success.. any ideas..maybe some resistor values are not quite right..if so, which one should i start with?
thanks

[antonis]

On the basis of what particular schematic we're talking about..??
Could you plz write down your circuit items values refering on schematic below.??

[yeeshkul]

I was comparing this one and a common opamp buffer, and Cornish buffer came out too bright sounding. It can be indeed mitigated. I would be surprised if the schematic above was a standalone pedal. More like a buffer for a particular pedal.

[bonehead1972]

Thanks for your input Antonis! May be some resistor values are slightly off.. I used whatever parts  I had on hand and resistors/caps in parallel here and there in order to get it  close enough..bias voltages are good according to your erlier posts and it sounds good indeed.. but I think it could get better:)  i can check values when i get home.. it's based on this schematic- http://effectslayouts.blogspot.com/2016/03/cornish-buffer.html

[antonis]

May be some resistor values are slightly off.. I used whatever parts  I had on hand and resistors/caps in parallel here and there in order to get it  close enough..bias voltages are good according to your erlier posts and it sounds good indeed..

Almost all items values (despite its high part count, for a humble Emitter follower) are, more or less, critical either by their own or in relation with each other..

e.g.
R5/R6 ratio sets Base bias voltage while R5/R6 values set Emitter effective load 'cause they're set in parallel (via C4) with R7 for AC..
(intentionally neglecting R8 & R10 for the sake of simplicity..)
They could easily be 10 times lower, for a more "stiff" voltage divider, but then half of 7k5 (12k//20k//7.5 -> 3.75k) should be considered as Emitter resistor resulting into higher non-linearity and less gain hence less effective R4 bootstrapping hence lower total imput impedance..

[Rob Strand]

The act of adding a buffer changes the loading on the pickup and that's going to increase the high-end.   That's often why you use a buffer in the first place.   If you want the buffer to sound like a cable + amp then you need to add a resistor in parallel with a capacitor across the input terminals.   The buffer has about 1.5M input impedance.  A good first try would be to change the 10M resistor to 220k to 2.2M.     Beyond that, and more preferable, would be to use say 2.2M in parallel with say 470pF, tune the cap to taste.

[yeeshkul]

The act of adding a buffer changes the loading on the pickup and that's going to increase the high-end.   That's often why you use a buffer in the first place.   If you want the buffer to sound like a cable + amp then you need to add a resistor in parallel with a capacitor across the input terminals.   The buffer has about 1.5M input impedance.  A good first try would be to change the 10M resistor to 220k to 2.2M.     Beyond that, and more preferable, would be to use say 2.2M in parallel with say 470pF, tune the cap to taste.

Rob, when i want a buffer with a bit lower input impedance, i guess i can leave out the boot strapping, and use a standard simple buffer topology, right? Also why would you suggest to add a resistor across the input instead of lowering one of the base bias resistors? Just wondering

[Rob Strand]

Rob, when i want a buffer with a bit lower input impedance, i guess i can leave out the boot strapping, and use a standard simple buffer topology, right? Also why would you suggest to add a resistor across the input instead of lowering one of the base bias resistors? Just wondering

The input resistor directly changes the impedance.   The value you use is the load.   Changing that part doesn't affect the biasing.

Normally with bootstrapping in place the effect of all three bias resistors is  largely removed, so the 10M input resistor alone would set most of the impedance.   However, in this case the low value of the 120k to the base means that's not quite the case and the 120k contributes to the input impedance.  We could play with the 120k to the base to change the input impedance.  That will need an even lower base resistor.   For bootstrapping, the input impedance not only depends on the 120k but also depends on other factors.

Having such a low base resistor with bootstrapping isn't normal.  I can only guess the reason the Cornish buffer uses low base resistor values and bootstrapping is to ensure the bias point of buffer is maintained with any transistor gain.    There may be some advantage for noise when the source is high impedance - have not checked.

If we stripped out the bootstrapping and went with conventional biasing and if we then wanted a 1M input impedance we would end-up with the bias point depending on the transistor gain.    It's workable but you might notice the Boss pedal buffer using a BJT used a 220k to 470k base resistor and when they went to 1M input resistors the BJT was changed to a JFET.

[antonis]

Having such a low base resistor with bootstrapping isn't normal.

Can't get your track, Rob.. 

120k Base resistor value is a good compromise between voltage drop (about 300mV) due to Base current (added to about 190mV voltage drop on R5/R6 Thevenin equivalent resistance) and about 20M(*) apparent (bootstrapped) value..
(*) for a gain of 0.994, bootstrap factor is about 166, so 120k X 166 = 20M

Or maybe you refer on a totally different kettle of fish..?? 

[Rob Strand]

Can't get your track, Rob.. 

120k Base resistor value is a good compromise between voltage drop (about 300mV) due to Base current (added to about 190mV voltage drop on R5/R6 Thevenin equivalent resistance) and about 20M(*) apparent (bootstrapped) value..
(*) for a gain of 0.994, bootstrap factor is about 166, so 120k X 166 = 20M

Or maybe you refer on a totally different kettle of fish..?? 

The main point I wanted to make was if we set the input impedance to1MEG by reducing the120k base resistor the base resistor would end-up being quite small.   If you placed a significant load on the buffer the buffer gain reduces and that can have quite an effect on the impedance.   If you use the boost strapping to make the input impedance very high and set the impedance using a fixed resistor and changes in the output load have very little impact on input impedance.

As for the calculations, I didn't do any precise calculations.     The DC drops are fine like you said.    I've got some stuff from earlier in the thread.  I ended up with 12.5M for the impedance of the bootstrapped 120k and 0.9904 for the unloaded buffer gain.     Not as high as your value but in the scheme of things the main point is it's pretty high.     I don't think we disagree at all.

If we now load down the buffer and the buffer gain drops to 0.98 then we see the impedance for the bootstrapped 120k dropped to 6M.   So as is, the design is fine even with a load.

However, if we reduced the 120k  to set the input impedance things aren't so great.   If we ignore all the other impedance as just say we are setting the impedance of the bootstrapped 120k to 1M  impedance, then when the buffer is loaded down the impedance would drop to roughly 500k.   Which is probably too much change  if we are tweaking the input impedance.    Ideally we want to set the input impedance to some tweaked value and it doesn't care about what is on the output of the buffer.

(For a BJT with finite gain a load on the output is going to effect the input impedance even with the single resistor setting the impedance.  The change will be less than if we used the bootstrapped base resistor.)

[antonis]

The main point I wanted to make was if we set the input impedance to1MEG by reducing the120k base resistor the base resistor would end-up being quite small.

Ridiculously small, I'd outbid..

I don't think we disagree at all.

We never did..
Simply din't get the main point of your previous post..