Buffer - too good to be true?

[seanm]

Below is the input buffer stage of the Acoustic Control 360 bass amp.



It seems to be too good to be true since it provides a high impedance transistor input buffer. It seems to work great even at 9V even though it was designed for a 24V supply.

When I was first testing this I unplugged the power and it kept working! I tried another transistor and it still worked. It turned out that the 470uF cap I had tied from +Vcc to gnd was powering the transistor. This kept up for at least 10 minutes.

I have been looking for high impedance alternatives for a FET input buffer. It sounds good at 9V. Anybody see any negatives about running the circuit at 9V?

[Sir H C]

Less headroom, that is all.  You might want to make the two resistors on the input about 2.2meg each to keep the high input impedance and center the output a little better.  Another technique you will see used with such biasings is what is called a bootstrap to really kick up the input impedance.  That can get really high input impedance.

[seanm]

Less headroom, that is all.  You might want to make the two resistors on the input about 2.2meg each to keep the high input impedance and center the output a little better.  Another technique you will see used with such biasings is what is called a bootstrap to really kick up the input impedance.  That can get really high input impedance.

I have looked into bootstrap biasing, but it requires more parts. This seemed to be a simpler solution.

I will check out the headroom, thanks.

[slacker]

If you haven't seen it before check out The AMZ buffers page there's examples of that circuit with resistor values for 9volts and good explanations of the various types of buffer.

[seanm]

If you haven't seen it before check out The AMZ buffers page there's examples of that circuit with resistor values for 9volts and good explanations of the various types of buffer.

They are good, but the BJT versions are all fairly low impedance. IIRC they are about 100k input impedance.

[R.G.]

It's not perfect.

Part of its low current use is caused by the low current drain. The 100K emitter resistor only needs 45ua to bias it at mid supply. The base current is... what? At low emitter currents, bipolar current gain falls off a lot. True, it's only got to have a gain of twenty to make the base look like a 1M resistor, but the follower aspect is then pretty faulty, as it will only have an input to output voltage "gain" of about 0.9.

It's not as good for noise performance as it could be. The high value resistors on the base will add their thermal noise in this setup. A much better setup from a noise standpoint is to use lower-value resistors for the bias divider, then a high value resistor from the divider to the base. There is much less excess noise in that biasing setup.

And the "noiseless biasing" requires 1.000 more resistors and one more cap than the circuit as shown, while at the same time providing the opportunity for higher input impedance.

The 100pF from base to collector has exactly the same effect as the same value capacitor from base to ground, since there is no voltage gain to the collector to multiply it.

Another not-so-obvious problem is that the output impedance is no terribly low. In the downward direction for largish signals, it's 100K, as the 100K resistor is all that pulls the output down if the transistor is turned off. You get asymmetrical slew rates and signal loss under those conditions.

[R.G.]

They are good, but the BJT versions are all fairly low impedance. IIRC they are about 100k input impedance.

I haven't looked at Jack's buffers page, but it's quite easy to get input impedance over 100K with a BJT, bootstrapping being a primary candidate.

Just as a curiosity, it happens that if you use the same input cap and bias divider as well as output cap and pulldown resistor, you can put in a TL072 opamp  buffer and throw away the 100K emitter resistor and 100pF cap? It actually uses fewer parts and has higher input impedance.

[slacker]

yeah the ones on Jacks page are shown with 100k input impedance, I'd forgotten they were shown that low. He also shows the noiseless biasing method with 500k base to vref resistor and you can easily replace that with 1Meg or probably more.

[R.G.]

Jack finally picked that up, eh? Cool.

[brett]

Hi
my low-tech approach is to use 2 x 220k biasing and a 10k emitter resistors and a high hFE, low noise transistor for BJT buffers.
Why?
I'm no expert, but it seems that
220k input impedance is enough to prevent significant losses in pickups and 99.9% of circuits.
220k is low enough to minimise noise, both from the simple biasing setup and from high ouput impedance sources such as static
the 10k resistor keeps output impedance low
the high hFE low noise transistor (MPSA18/PN100) keeps input impedance high and minimises noise.

just my 2c worth
cheers

[Gus]

why do people want to always center the output?.   Use a scope your ears and some math.

   AOE has a good reason in a section not to center the emitter.

You need to think about input and output swing BE drop......

Look at the input buffer in the 3tran and rocket.

[seanm]

Just as a curiosity, it happens that if you use the same input cap and bias divider as well as output cap and pulldown resistor, you can put in a TL072 opamp  buffer and throw away the 100K emitter resistor and 100pF cap? It actually uses fewer parts and has higher input impedance.

I am trying to do it with a transistor for no good reason other than I want to. I am in a "transistor" mood right now.

As an aside, which would be lower power, the opamp or the transistor?

Use a scope your ears and some math.

I threw a scope on this circuit, and while I have not maxed out the input yet, the results have been stellar. I also listened to it briefly and it sounded good.

[Sir H C]

why do people want to always center the output?.   Use a scope your ears and some math.

   AOE has a good reason in a section not to center the emitter.

You need to think about input and output swing BE drop......

Look at the input buffer in the 3tran and rocket.

One reason I like to go for symmetrical swing is that you can then set the two input resistors to a lower value for each and still have a higher input impedance.  The 1meg and 4.7meg are in parallel giving an input impedance for just them of about 820k.  With two 2.2meg resistors you are at 1.1 meg and you don't have the noise of a 4.7 meg resistor in there.  With a 9 volt supply, the .7v slop I can deal with, but the 4.7 meg vs. 1 meg puts that about 2 volts from the top rail.

[seanm]

One reason I like to go for symmetrical swing is that you can then set the two input resistors to a lower value for each and still have a higher input impedance.  The 1meg and 4.7meg are in parallel giving an input impedance for just them of about 820k.  With two 2.2meg resistors you are at 1.1 meg and you don't have the noise of a 4.7 meg resistor in there.  With a 9 volt supply, the .7v slop I can deal with, but the 4.7 meg vs. 1 meg puts that about 2 volts from the top rail.

I measured 660k input impedance @ 100Hz with the 1M/4.7M combo. I hope to try the 2.2M/2.2M later today.

This brings up a question. What do people use as a resonable maximum input voltage?

I use passive basses and have hooked a scope up to the output. I don't have a storage scope so I just watched the output and tried to catch the peaks. I peaked at about 300mV p-p.

When I designed my preamp, I used 2V p-p but is 1V p-p a more resonable level?

[Sir H C]

The monster basses that are passive would be the Gibson EB-0 and the Ovations with those giant pickups.  I remember reading that the Gibsons easily put out a volt (in the Ampeg book as this is what they designed their bass amps to be able to drive clean).

So AC you have a 50k load on the follower.  Gain that up by the beta of the transistor (say 20-30 at such low currents?), you get a resistance of about 1.5 meg in parallel with the others.  So with the 1meg and the 4.7meg that would be about 600k, so beta is a bit better. 

[seanm]

Thanks Sir H.C.! So 1V p-p seems to be a reasonable value.

I have also been trying to calculate the output impedance of my bass. This is a stock passive P bass. The previous owner said the pickups where from a '70s era P bass.

I measured the output with the scope while hitting an open E (41Hz). It was about 200mV (V). I then placed a 100k resistor across the output and measured again. I got about 170mV (Vo). So using Zo = Rl(V-Vo) / Vo I got Zo = 17k.

I also tried it with the digital volt meter and got 13k. So in the same ballpark. I trust the scope value more since it was easier to see the peaks.

I also tried my other '62 reissue and 320mV/280mV = 14k. This bass has the tone control disabled.
My MIM with quarter pounders was 300mV/250mV = 20k. This bass has the tone control disabled.

So lets say 25k for a max impedance. With the 10x rule that means the input impedance should be at least 250k. I am starting to wonder if the standard 1M input impedance isn't overkill.

[seanm]

One more question... What is the purpose of the 100pF cap? It dosen't seem to affect noise or impedance by removing it. This is with the two base resistors at 1M.

Ok, two more questions... You usually see a large cap from plus to ground (and sometimes a smaller one). If you are running off battery only is this cap still a good idea? The problem I have is that the cap keeps the circuit going for a long time (minutes), so you can get a pop when you unplug. The stompbox will off battery only.

[brett]

Hi
I agree that for pickups with 13k outpuut impedance, anything over 130k is plenty.  As I said above, 2 x 220k bias resistors are a good compromise between impedance and noise for BJT buffers.

One more question... What is the purpose of the 100pF cap? It dosen't seem to affect noise or impedance by removing it. This is with the two base resistors at 1M.

This helps with RF interference.  Because the gain is unity in this buffer, the cap has minimal effect (the same as connecting it from the base to ground).  If your pickup has 13k ouput impedance, a  100 pF cap gives a rolloff frequency of about 120 kHz.  You'll see similar caps in discrete circuits like the Big Muff and in op-amp circuits like the tubescreamer.

cheers

[seanm]

This helps with RF interference.  Because the gain is unity in this buffer, the cap has minimal effect (the same as connecting it from the base to ground).  If your pickup has 13k ouput impedance, a  100 pF cap gives a rolloff frequency of about 120 kHz.  You'll see similar caps in discrete circuits like the Big Muff and in op-amp circuits like the tubescreamer.

That's what I thought it was for. Not shown in the schematic at the top is that in front of the 1uF input cap I have a 1k resistor in series with the input and a .001uF capacitor to ground after it. This helps cut down on the noise and would make the 100pF cap moot.

[R.G.]

I have also been trying to calculate the output impedance of my bass. This is a stock passive P bass. The previous owner said the pickups where from a '70s era P bass.
I measured the output with the scope while hitting an open E (41Hz). It was about 200mV (V). I then placed a 100k resistor across the output and measured again. I got about 170mV (Vo). So using Zo = Rl(V-Vo) / Vo I got Zo = 17k.
I also tried it with the digital volt meter and got 13k. So in the same ballpark. I trust the scope value more since it was easier to see the peaks.

Aye, there's the rub - it is indeed 13K - 17K, but only at 41Hz. Pickups are inductive. The impedance goes up with frequency. This is why you get tone sucking. So an input impedance of 13K for low E becomes much higher for any overtones. It's not as noticeable for bass where what you are trying to do is get low frequencies, but it's still there.

Depending on the sound you want from your bass, loss of treble may be good or bad. If you're trying to get all the harmonics and slap effects, you'll lose some with only 130K input impedances.