I think I've got the key bits in there... like the common mode feedback the
right way around.

It's a diff amp, but most scopes have a SE front end. So, like most scopes,
the position knob goes to -in and signal to +in. A preamp might be a good
idea for the more sensitive ranges.

The 2N5179 pair should be fast enough that I get the ~20 gain I want up to
50MHz or so. '3904's are kind of pokey for this, but as emitter followers,
who cares. That just leaves the hope that the C1569's are as fast as
claimed.

And if not, I could replace the *******s with a 6CL6. That'll show 'em.

Tim

--
Deep Fryer: A very philosophical monk.
Website @ http://webpages.charter.net/dawill/tmoranwms

Tim Williams wrote:
I think I've got the key bits in there... like the common mode feedback the
right way around.

It's a diff amp, but most scopes have a SE front end. So, like most scopes,
the position knob goes to -in and signal to +in. A preamp might be a good
idea for the more sensitive ranges.

The 2N5179 pair should be fast enough that I get the ~20 gain I want up to
50MHz or so. '3904's are kind of pokey for this, but as emitter followers,
who cares. That just leaves the hope that the C1569's are as fast as
claimed.

And if not, I could replace the *******s with a 6CL6. That'll show 'em.

This reminds me of the vertical amplifiers in my old
Heathkit 5MHz scope. I perked up the high voltage stage by
cascoding those high voltage transistors. I tied their
bases to the +8 volt supply rail (+9 in this case) and added
faster, low voltage transistors in series with their
emitters. This gives the output stage something like a
pentode effect.

Yeah, I'm thinking of modding my IO-103, actually.

I suspect this circuit draws wee too much current for its supply though.
Suppose I could always add another transformer...

What kind of bandwidth did you get with the cascode?

Tim

--
Deep Fryer: A very philosophical monk.
Website @ http://webpages.charter.net/dawill/tmoranwms

"John Popelish" wrote in message
. ..
Tim Williams wrote:
I think I've got the key bits in there... like the common mode feedback
the
right way around.

It's a diff amp, but most scopes have a SE front end. So, like most
scopes,
the position knob goes to -in and signal to +in. A preamp might be a
good
idea for the more sensitive ranges.

The 2N5179 pair should be fast enough that I get the ~20 gain I want up
to
50MHz or so. '3904's are kind of pokey for this, but as emitter
followers,
who cares. That just leaves the hope that the C1569's are as fast as
claimed.

And if not, I could replace the *******s with a 6CL6. That'll show 'em.

This reminds me of the vertical amplifiers in my old
Heathkit 5MHz scope. I perked up the high voltage stage by
cascoding those high voltage transistors. I tied their
bases to the +8 volt supply rail (+9 in this case) and added
faster, low voltage transistors in series with their
emitters. This gives the output stage something like a
pentode effect.

On Tue, 4 Sep 2007 20:36:46 -0500, "Tim Williams"
wrote:

I think I've got the key bits in there... like the common mode feedback the
right way around.

It's a diff amp, but most scopes have a SE front end. So, like most scopes,
the position knob goes to -in and signal to +in. A preamp might be a good
idea for the more sensitive ranges.

The 2N5179 pair should be fast enough that I get the ~20 gain I want up to
50MHz or so. '3904's are kind of pokey for this, but as emitter followers,
who cares. That just leaves the hope that the C1569's are as fast as
claimed.

And if not, I could replace the *******s with a 6CL6. That'll show 'em.

Tim

The other trick is to add peaking inductors. The simplest is to add an
inductor in series with the final 1K collector loads. The ballpark
value is

L = K * R^2 * C

where C is the net capacitance (transistor + deflection plate +
wiring), R is your 1K, and K=0.5 to 1 maybe. Figure a 40% bandwidth
increase.

Next step is another inductor, in series with the outputs. Another 20%
maybe. Finally, an optimun t-coil can increase bandwidth by 2.8x over
no peaking at all.

As JP mentions, it helps to cascode the output stage, which reduces
feedback (Miller) C and allows you to use something really fast on the
bottom, like a gaasfet or a phemt maybe.

Hey, how about a fast opamp driving a phemt-C1569 cascode, with
peaking inductors?

John

Tim Williams wrote:
"John Popelish" wrote:

This reminds me of the vertical amplifiers in my old
Heathkit 5MHz scope. I perked up the high voltage stage by
cascoding those high voltage transistors. I tied their
bases to the +8 volt supply rail (+9 in this case) and added
faster, low voltage transistors in series with their
emitters. This gives the output stage something like a
pentode effect.

Yeah, I'm thinking of modding my IO-103, actually.

I suspect this circuit draws wee too much current for its supply though.
Suppose I could always add another transformer...

What kind of bandwidth did you get with the cascode?

I didn't even own a signal generator when I did this (poor
college student with 2 kids), but I was able to greatly
reduce a couple capacitor frequency boost networks between
pairs of emitters and still have a nice square wave response
with less ringing and squarer corners. I think I built a 1
MHz multivibrator as a test signal. The scope had a 60 Hz
voltage reference waveform generator that I used to get the
low frequency response back to where it started.

"John Larkin" wrote in message
...
The other trick is to add peaking inductors.

Yeah, reminds me, I took apart a bunch of monitors -- that's where I got my
BFQ221, 241, 225's and PH2369's from -- which used 1.3kohm 5W collector
resistors, and a little bit of inductance. I forget what it was, 10uH or
something. Dang, just the other day I took down the last board that I had
left up here. Ah well, it would do something like that in any case.

The simplest is to add an
inductor in series with the final 1K collector loads. The ballpark
value is

L = K * R^2 * C

Alright.

where C is the net capacitance (transistor + deflection plate +
wiring)

I'm going to guess in the ballpark of 10-50pF. How much capacitance does
300 ohm twin lead have, and what's a typical capacitance to chassis? Hmm, I
was going to look for a datasheet on the CRT, too. That would give me
deflection plate C.

Next step is another inductor, in series with the outputs. Another 20%
maybe. Finally, an optimun t-coil can increase bandwidth by 2.8x over
no peaking at all.

So series resonant with the deflection capacitance?

Please explain "t-coil".

As JP mentions, it helps to cascode the output stage, which reduces
feedback (Miller) C

Yeah, that was something else I was going to do- didn't draw it. Speaking
of things to do, is there any advantage to using collector or emitter drive?
Above, I show the outputs driven by an emitter follower. The original HK
schematic shows PNP collectors. Given the same bias current, slew rate is
about the same, so would I be better off going for the collector voltage
gain?

and allows you to use something really fast on the
bottom, like a gaasfet or a phemt maybe.

Hey, how about a fast opamp driving a phemt-C1569 cascode, with
peaking inductors?

I want to see waveforms /on the CRT/, not by the electromagnetic energy the
driver circuit is radiating! ;-)

Speaking of fast-assed parts, I'm going to need to get myself some snap
diodes some day. Got a supplier, or some extras?

Tim

--
Deep Fryer: A very philosophical monk.
Website @ http://webpages.charter.net/dawill/tmoranwms

On Wed, 5 Sep 2007 01:02:25 -0500, "Tim Williams"
wrote:

"John Larkin" wrote in message
.. .
The other trick is to add peaking inductors.

Yeah, reminds me, I took apart a bunch of monitors -- that's where I got my
BFQ221, 241, 225's and PH2369's from -- which used 1.3kohm 5W collector
resistors, and a little bit of inductance. I forget what it was, 10uH or
something. Dang, just the other day I took down the last board that I had
left up here. Ah well, it would do something like that in any case.

The simplest is to add an
inductor in series with the final 1K collector loads. The ballpark
value is

L = K * R^2 * C

Alright.

where C is the net capacitance (transistor + deflection plate +
wiring)

I'm going to guess in the ballpark of 10-50pF. How much capacitance does
300 ohm twin lead have, and what's a typical capacitance to chassis? Hmm, I
was going to look for a datasheet on the CRT, too. That would give me
deflection plate C.

Next step is another inductor, in series with the outputs. Another 20%
maybe. Finally, an optimun t-coil can increase bandwidth by 2.8x over
no peaking at all.

So series resonant with the deflection capacitance?

Yup. That helps a little more. The math is tricky enough that you may
as well simulate and diddle.



Please explain "t-coil".

That's a tapped inductor, sometomes with an added cap. You should be
able to google for that.

Jim Williams first Analog Circuit Design book, the 1991 one, has a
chapter on scope vertical amps. The classic of course is the Tektronix
Concepts series book "Vertical Amplifier Circuits"




As JP mentions, it helps to cascode the output stage, which reduces
feedback (Miller) C

Yeah, that was something else I was going to do- didn't draw it. Speaking
of things to do, is there any advantage to using collector or emitter drive?
Above, I show the outputs driven by an emitter follower. The original HK
schematic shows PNP collectors. Given the same bias current, slew rate is
about the same, so would I be better off going for the collector voltage
gain?

and allows you to use something really fast on the
bottom, like a gaasfet or a phemt maybe.

Hey, how about a fast opamp driving a phemt-C1569 cascode, with
peaking inductors?

I want to see waveforms /on the CRT/, not by the electromagnetic energy the
driver circuit is radiating! ;-)

Speaking of fast-assed parts, I'm going to need to get myself some snap
diodes some day. Got a supplier, or some extras?

Email me and I'll send you a few.

John

jjlarkin
approximately
highlandtechnology
dotcomthing