Input and Output stages of Chorus/Flange/Delay

[YouAre]

Ok, now that I have a good idea of what's going on on the inner sides of Chorus/Flange/Delay (after you guys helped me on that last thread...), now I'm turning my attention to the input and output buffers.

The EHX Small Clone, Echo Base, and BYOC Flanger all have a non inverting input buffer, and inverting output buffer. It's my understanding that non inverting has a high input impedance that's great for a first stage, and that inverting stages are great for mixing. Are inverting buffers good for mixing because of their lower input impedance?

The Boss CE-2 seems to use inverting buffers for both it's input and output stages. Does anyone have any insight as to why?

I understand signal inversion does play a role in feedback for amplifiers, but does it have a role in here when feedback is usually done "in between" the buffers? (feedback is tapped before output before and fed after input).



Also on another note, what would be the implications of using transistor stages instead? I understand a Common collector/drain setup will have less than unity gain often times, which may be problematic for non inverting stages (whereas non inverting opamp buffers have at least a gain of 1). But surely a simple common emitter/source could make a suitable stage? Any insight/experience on this?

Thanks guys!

[amptramp]

You are correct in saying that the non-inverting amplifier has a high input impedance.

An inverting op amp stage usually has the non-inverting input sitting at a fixed voltage (usually Vcc/2 or half the power supply voltage).  It will attempt to drive the inverting input to the same voltage, so the inverting input acts as a virtual ground - it appears to have zero impedance connected to the same voltage as the non-inverting input.  (The virtual ground is actually at ground if the op amp is operated from ± supplies with the non-inverting input at ground.)  Thus, if it is used as a summing junction for a mixer, the current in any input resistor will be completely independent of the current in any other input resistor because all resistors appear to feed a node at a fixed voltage.  So yes, inverting buffers are good for mixing because ideally, the inputs feed a resistor to a point of zero impedance.

The input buffer on the Boss CE-2 is the single-transistor emitter follower.  The op amp stage following it is a filter that shapes the frequency response of what gets fed into the bucket brigade delay IC.  This stage provides a higher gain at high frequencies.  It is followed by a Sallen & Key lowpass filter and the output of the bucket brigade IC is also followed by an identical lowpass filter to take out the clock noise.

In order to get the chorus effect, you have to mix the original ("dry") signal with the processed signal, otherwise, you just have a vibrato.  Signal is fed forward, not back and the output stage is used as a mixer as long as the effect is on.  If the stompswitch commands it to be off, you just get the dry signal and it acts as a buffer.

You can use an emitter follower, but the input impedance may still be quite low.  It will be the bias resistor (470K on the CE-2 in parallel with a resistor that is hfe*emitter resistor).  Using a conservative value of 100 for hfe and 10K as in the CE-2, this is about 1 megohm.  This is in parallel with the 470K to give 319.7K - not bad, but if the hfe goes down, so will the input impedance.  To be accurate, the emitter resistor is in series with a silicon resistance constant that is 26 ohms at 1 mA and decreases as curent increases such that the product of stage current and effective ohms remains constant.  Of course, 26 ohms in series with 10000 ohms is negligible.

FET's have the advantage of very high input impedance, such that the input resistor is virtually the input impedance, but whereas the bias point on a bipolar transistor is set to one Vbe drop, the bias point on a FET can be... anything.  It is difficult to pinpoint the voltage and there is considerable variation, so the operating point could be sitting over a wide voltage range.  But bipolar and FET input and output buffers are quite common with bipolar being more commen due to the predictable operating point.  Bipolar emitter followers are used on many Boss pedals.

[merlinb]

Are inverting buffers good for mixing because of their lower input impedance?

No, its because they act as virtual earth amplifiers. That means you can mix multiples signals together without them affecting one another. For example, if you're mixing wet and dry together, you can switch one of them off or turn it up and down, but the other keeps passing through, completely unchanged.

Also on another note, what would be the implications of using transistor stages instead?

BJT buffers are easy to bias compared with FETs, but the input impedance is limited, noise is high, and most of all the headroom is usually pathetic. The typical Boss input buffer barely musters 1Vp-p headroom. A FET opamp as an input buffer makes a single BJT look like a sack of crap.

[YouAre]

So in the case of pedals with a an inverting input, what would be the changes if we were to replace it with this http://www.muzique.com/lab/splitter.htm ?

I know the parts count would increase, but I've heard of certain boutiquers using jfet input stages for mn3207 based chorus pedals. It was an idea I wanted to experiment with also because I noticed my tonepad small clone doesn't sound too great with a hot signal. I kinda like how fets take some drive better than opamps, so I figured the input stage could take a hot input signal, and be set to be relatively low gain so as not to overdrive a clean signal.

[Jazznoise]

There is other issues - if you want a buffer/gain stage an inverting amp be far more stable at higher gains.

You can have high-ish input impedence inverting amps, but it's very interactive and as far as I'm aware can lead to noise? Would like someone to verify that last point!

Then there's the issue of output impedence, which is very important in live scenarios as you need to drive and keep your impedence down before your cable sucks up all your signal. Now I'd be happy with just a non inverting buff here...UNLESS you have an inverting input!

Why? Well what if the person patches into a splitter and sends one signal to your chorus and an amp and one straight to another amp? Phasing! If both were equidistant (And the effect created no additional phase change, which would be daft - especialy in this scenario) they could cancel out completely! If not, the phasing will be pretty severe all the same ruin the musicians (However valid) master plan. And thus your pedal makes everything sound god awful. And suddenly no one's returning your calls. And you sit in the dark for fourteen years next to the phone wondering what you've done wrong.

[YouAre]

Jazznoise, I see your points, which is why I want to maintain the non-inverting input. It seems to be the tried and true method for many of these designs.

Can anyone give me some insight as to what some of the differences would be if i were to replace the unity gain non-inverting input stage of a pedal like the echo base with a splitter like the one I posted earlier?

Thanks again, and sorry for the super late bump!