Transient Behavior of Coupling capacitors

[George Jefferson]

When coupling capacitors are used to remove DC offsets between stages, does
there exist a dc component during startup? I notice that when I watch the dc
after a coupling cap that it starts out with a large dc component which goes
to zero. I assume this is because the cap is charging and eventually charges
to the dc component which blocks it resuling in a net 0 voltage after the
cap?


Suppose I have an amplifier stage and I couple it to that of a jfet. In this
case the gate is "floating". We could add a large resistance to ground
it(lets say it requires a very large resistance to prevent loading... 10M or
more). But will the gate initially have a large dc component which could
potentially ruin the jfet?


e.g., I have a very high ouput impedence(~1M-10M) transducer with a high
supply voltage(~1kV) and a relatively large DC offset(> 100V). I would like
to create an amplifier.

I've tried a bjt but it simply loads the signal too much and I can't get the
transistor out of cutoff without drawing too much current through the
resistors(trying to use 1/4W or 1/2W). When I do load the signal/heat the
bias resistors I can get some amplification(finally get out of cutoff) but
the signal is distorted because of loading and probably because it is too
close to cutoff.

Therefore I figured I could use a common source amp. Of course now I have to
worry about blowing the gate. The signal switch is within spec but the dc
offset would easily ruin it if it gets through the cap.


Anything I could do? It would be a relatively straightforward problem if the
output impedance and voltages weren't so high. I would like to use basic
techniques if possible rather than some exotic method.

If I could in some way buffer the signal then I can get away from the
loading issues which is really the only problem I'm having.

Obviously some floating gate method would work but the only thing I can
think of is a transformer and I'm not willing to go there.

 

[George Jefferson]

I should be a bit more clear(sorry, been up all night)...

The problem is that the previous stage has high output impedance and all the
amplifier configurations requires some type of biasing. If the biasing
resistors are to form a stiff voltage divider to prevent loading then they
must be much larger than the output impedance of the previous stage. This
requires them to be on the order of about 100M to 10G depending on the amp
configuration. While I suppose this might work for some mosfets(such as the
common drain) it doesn't work with any bjt's. I'm not sure I have any
resistors above 10M to even test it and I imagine it would cause other
problems?

 

[George Jefferson]

If the source can drive enough current, yes. You can flow some current
into a JFET gate without damage (as long as you have enough resistance on
the drain), so you could protect the JFET gate with a series resistor, or
some clever combination of series resistors and shunt diodes.

Ok, I guess I have to ask the question different... I'll do another post.

If it's really truly 1M, and not effectively 1M, then a 100V offset would
only cause 100uA to flow -- that's not going to harm the JFET. OTOH, if
it's some nonlinear thing that'll flow TONS of current at first, then you
may have a problem.

ok, this is becauset he JFET's gate is much much greater so we effectively
have a voltage divider effect with the "gate voltage" basically being 0?


Basically your making it sound like ultimately what really matters is not
the absoltue maximim voltage but the current. If the max VGS is 100V and the
internal gate resistanec is 1M, just for calc, then really we simply can't
"force" more than 100uA into the gate?

Hence any transient voltage that may be much larger than 25V but doesn't put
more than 100uA in the gate won't hurt it?

I'll ask in another post so you might respond to that.

Remember that you're not going to put any more charge into the JFET gate
than will build up on the cap, so a smaller cap will help to some extent.
A small cap in conjunction with a careful voltage ramp to the transducer
supply will help more.


There are ways to get high input impedances from BJT stages. The easiest
from a circuit designer's point of view is probably to find an
off-the-shelf op-amp that'll do what you want.

Yes, I was thinking of using an op amp but I have read they can be bad in my
situation because of distortion. Something having to do with all the extra
bjt's in it that are not really needed.

What you're talking about building _is_ a buffer.

Yes, but the problems of biasing still exist. Buffer or Amplifier(common
drain/collector or source/emitter) still need to be biased which would cause
problems. This is because the signal is AC

A transformer to work with a 1M-ohm source would be absurd. You're on the
right track with a JFET, although if you can make it work with an op-amp
you should seriously consider that approach.

Yeah, your right... wouldn't be enough drive to get anything.


Thanks.