Thanks, Petter.
It is also interesting that the output stage is a source follower, which has low distortion and output impedence. But htis also means that all the voltage gain of 30 db must come from the single driver stage-- that is a lot for one stage. Perhape the later x amps, which is said to have a JFET imput stage, will perform better assuming it still has a MOSFT driver stage. It would indeed be interesting to see how that is implemented.
CT
It is also interesting that the output stage is a source follower, which has low distortion and output impedence. But htis also means that all the voltage gain of 30 db must come from the single driver stage-- that is a lot for one stage. Perhape the later x amps, which is said to have a JFET imput stage, will perform better assuming it still has a MOSFT driver stage. It would indeed be interesting to see how that is implemented.
CT
I don't consider a cascode as a gain stage (it's gain contribution
is generally insignificant), just as I also consider a differential
pair to be a single stage. These are perhaps my personal
prejudices, but usually I use the gain contribution as the
criterion.
The UGS has cascoded JFETs for inputs, and like the original
circuits on the X amplifiers, we have to do some level translation
to get rail-to-rail output. On the X600 and X1000 we did this
though a folded cascode (see the A75 for an example), and so
we think of that front end as being a single stage.
This requires an additional higher voltage supply rail, and is not
convenient, so on the X350 we used a current mirror for a level
shifter. You can set that up through the resistor values to either
have gain or not. With gain it's clearly a gain stage, and without
gain I would say it's a level shifter.
We did a similar thing on the UGS, and while it's more complex
than just a diff pair, the gain is essentially that of the JFETs
loaded into the Drain resistances we provide. To that end, it's
a single stage. Change the resistor values, and you can make
it into a 2 stage piece. But this is also a function of the loading,
and the last time I looked, the mirror gain was very low on the
power amps but about 3-4 dB on the preamps.
BTW, the patent has a rather complete schematic of the X600 /
X1000
is generally insignificant), just as I also consider a differential
pair to be a single stage. These are perhaps my personal
prejudices, but usually I use the gain contribution as the
criterion.
The UGS has cascoded JFETs for inputs, and like the original
circuits on the X amplifiers, we have to do some level translation
to get rail-to-rail output. On the X600 and X1000 we did this
though a folded cascode (see the A75 for an example), and so
we think of that front end as being a single stage.
This requires an additional higher voltage supply rail, and is not
convenient, so on the X350 we used a current mirror for a level
shifter. You can set that up through the resistor values to either
have gain or not. With gain it's clearly a gain stage, and without
gain I would say it's a level shifter.
We did a similar thing on the UGS, and while it's more complex
than just a diff pair, the gain is essentially that of the JFETs
loaded into the Drain resistances we provide. To that end, it's
a single stage. Change the resistor values, and you can make
it into a 2 stage piece. But this is also a function of the loading,
and the last time I looked, the mirror gain was very low on the
power amps but about 3-4 dB on the preamps.
BTW, the patent has a rather complete schematic of the X600 /
X1000
Mr. Pass,
Thank you for answering my question. Is the front end of the X amps of responsible for all the 29 db voltage gain (assuming that the output stage is a source folower)?
May I also ask why the X1000 and X600's frond end is only suitable for balanced inputs while that of the smaller X amps can take balanced or single-ended inputs?
Thank you for answering my question. Is the front end of the X amps of responsible for all the 29 db voltage gain (assuming that the output stage is a source folower)?
May I also ask why the X1000 and X600's frond end is only suitable for balanced inputs while that of the smaller X amps can take balanced or single-ended inputs?
It's a question of how intimately the Sources of the input differential are bound together and how stiff the resistor or current source under the tail of the differential is.
A differential with a few hundred ohms under the tail will work, but will not perform very well as a phase splitter. A few thousand ohms will do a great job. By the time you get to 5-10k or so, the results are virtually indistinguishable from a current source. Most people would be surprised at how little resistance is actually necessary; they reflexively think of a CCS and never even think to try a resistor. A decent current source has a very high Zout and will cause the differential to do a very good job of signal splitting.
Grey
A differential with a few hundred ohms under the tail will work, but will not perform very well as a phase splitter. A few thousand ohms will do a great job. By the time you get to 5-10k or so, the results are virtually indistinguishable from a current source. Most people would be surprised at how little resistance is actually necessary; they reflexively think of a CCS and never even think to try a resistor. A decent current source has a very high Zout and will cause the differential to do a very good job of signal splitting.
Grey
Grey and Mr. Nelson, thanks for the replies.
Speaking of the coupling between the two halves of a differential pair,
reducing the degree of coupling can also reduce the absolute DC offset (as in the case of the Aleph-x). But then the amp will only accept balanced inputs. Mr. Pass, how do you deal with the offset issue in the XA series (and in the X series, if there is any offset)?
Speaking of the coupling between the two halves of a differential pair,
reducing the degree of coupling can also reduce the absolute DC offset (as in the case of the Aleph-x). But then the amp will only accept balanced inputs. Mr. Pass, how do you deal with the offset issue in the XA series (and in the X series, if there is any offset)?
ctong said:
5 Ways:
1) Match parts for absolute and differential DC offset
2) Trim values with potentiometers for absolute and differential
DC offset
3) Common mode DC offset feedback from the outputs back to
the sources of the input diff pairs (for absolute DC offset)
4) Differential mode DC offset feedback (regular X feedback) for
differential DC offset
5) Resistive loading at the outputs to ground (about 100 ohms
or so) which minimizes absolute offset drift
Works like glue....
no, no way!
the correct way is yet to be applied
5 ways of glue is as good as none
the correct way is one
match the unmatched!
works like gold,rock solid!
5 ways is correct, however this is a balanced topology
and matching is paramount
mr pass thanks for the classic article,any i havent read would
be highly appreciated by mastertech
thanks in advance
the correct way is yet to be applied
5 ways of glue is as good as none
the correct way is one
match the unmatched!
works like gold,rock solid!
5 ways is correct, however this is a balanced topology
and matching is paramount
mr pass thanks for the classic article,any i havent read would
be highly appreciated by mastertech
thanks in advance
mastertech said:
Everything I've written can be found at www.passlabs.com or
www.passdiy.com or www.firstwatt.com
If there's anything you can't find, I can put it up on
www.passlabs.com/np (theres some miscellaneous files there
now if you want to look at them).
I am shortly going to release my collection of Threshold pics and
scans - just a few more to go.