Driver to drive?

On Mon, 20 Mar 2017 20:59:26 -0000, "Kevin Aylward"
<kevinRemovAT@kevinaylward.co.uk> wrote:

wrote in message news:2g6uccheh2pto7113hegrstud5ro8r8paj@4ax.com...

On Sun, 19 Mar 2017 09:16:29 -0000, "Kevin Aylward"
kevinRemovAT@kevinaylward.co.uk> wrote:

"Kevin Aylward" wrote in message
news:kfudnRgqWLyQIlDFnZ2dnUU7-XvNnZ2d@giganews.com...

"bitrex" wrote in message news:9DdzA.62942$mb5.42260@fx19.iad...


I'm a liberal and yet, in some circumstances I do support the death
penalty.

I don't see that there can be much more of a cold-bloodied, calculated
murder, than having 12 people calmly sit on seats debating the merits of
killing someone over several days, with a state sponsored judge exposing
all sorts of "rational" arguments as to how it is ethically justifiable to
execute said person being debated. Said person is then dragged to a room
with gawking onlookers watching the deliberate injection of chemicals to
terminate said life. This is no less barbaric than at a Roman gladiator
ring
where the emperor points his thump up or down.

What is even more grotesque, is that large numbers of those barbarians
supporting state sponsored murder are alleged Christians, despite their
role
model, Jesus, emphatically instructing them that "thou shall not kill".
More, stunningly the xtians claim that it is they that there the morally
righteous ones.

The perp gave up his right to life by taking that of another. End of
story.

Ok. After the Jury, judge and executioners have killed the aforementioned,
we can now kill said Jury, judge and executioners because they have now
killed someone, or taken deliberate action that resulted in the death of
someone. i.e. murdered someone.

It's obvious you're illiterate.

<no point in reading further>

 

[rickman]

On 3/20/2017 6:54 PM, amdx wrote:

On 3/20/2017 5:27 PM, rickman wrote:
On 3/20/2017 6:21 PM, amdx wrote:
On 3/20/2017 3:34 PM, rickman wrote:
On 3/20/2017 2:16 PM, amdx wrote:
On 3/20/2017 11:29 AM, dagmargoodboat@yahoo.com wrote:

If you will look at my picture,
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163
you will notice I put the 470 ohm (turquoise resistor) on the other
side
of the shield, (I don't know why).
What affect would that have?

On Monday, March 20, 2017 at 9:36:51 AM UTC-4, amdx wrote:
If you want to get fancier
and even lower input capacitance we can bootstrap the drain, too. A
bootstrapped shield for your pass-through (where you bring the input
through your metal box) would help, too.

OK, I'd like to get fancy, but I don't understand "A bootstrapped
shield" for my pass through.
I would like to understand physically what I need to do.

I have seen this done for electrometers where the concern is leakage
current. There they call it a "guard" ring. As James says, it will
also reduce input capacitance.

I don't understand how you can bootstrap the drain though. Since it is
connected to the positive rail, I'm not even sure what that means.


Here's another picture showing the input, there is 5/8" hole in the
case
and I glued polystyrene sheet over it with the input wire coming out
the
center.
http://i395.photobucket.com/albums/pp37/Qmavam/Input.jpg
Ah, here's a picture of a previous input cap, I don't use anymore, but
it shows the input better.
http://i395.photobucket.com/albums/pp37/Qmavam/1cmx1cmspaced5mm.jpg

This latter image doesn't look like what is used in this image.

http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html

Was this just a temporary cap in place until you got the detail work
done?


Yes, that was a first iteration, nothing wrong with it, but it wasn't
as stable (two wires with a weights on the end) as my tiny piece of
double sided pcb.

Oh, so the free hanging copper is the old cap and the thing I can't see
is the PCB cap which you like better?


When you say can't see, do you mean it small and your looking side on
or do you mean you didn't see a link?
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163

btw the 0.3pf cap is a approximately 1/8" disc of Teflon pcb 0.32"
thick. PCB material is not critical. Easy enough to set size, check the
gain of your amp without the input cap, then reduce the size of your
input cap until gain equals 1.

I just meant it is small and I couldn't see what it was. Initially I
thought it might be a small value ceramic cap you put in place to check
out the amp before you optimized the cap. Now I realize it is a very
small piece of PCB and *is* the optimized cap. I also didn't know which
image was earlier and which was later.

Why would you adjust the overall circuit gain by trimming the cap? The
amp gain can be trimmed, no?

What frequency range is this amp intended to be used for? Nearly all
the stuff I've seen where Q matters is LF or MF. They use loop antennas
for higher frequencies, but with too high Q the bandwidth gets so narrow
even voice won't get through, just CW.

Some of the antennas Kleijer uses I would think would limit the
bandwidth of his circuits with Q of over 1000.

If the frequencies are not so high, why not use an opamp for the final
stage rather than the much more complex push-pull stage? I assume that
is intended to be in essence a "power" output stage and does not
otherwise contribute significantly to the characteristics of the rest of
the circuit.

--

Rick C

 

[Jim Thompson]

On Tue, 14 Mar 2017 18:58:32 -0700, Jim Thompson
<To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:

"Procedure" tomorrow (Wednesday) at 11:00AM:

Down the throat with a scope, check out the stomach, then into the
small intestine, use side-looking ultrasound on the end of the probe
(didn't know such a thing existed) to examine the common pancreas/bile
duct, go up it with a wire, then thread a balloon up that wire,
inflate and decimate the stones, then go on up and examine the gall
bladder.

Possible later procedure, after the nauseous, tiredness, yellowness
abates, go in thru an incision and remove the gall bladder.

Such fun >:-}

If I don't show up in a day or too...

...Jim Thompson

Well the docs tried to kill me, but I survived a very unpleasant trip.

They put a stent in my bile duct, but in recovery they discovered my
kidneys couldn't cope, and went into failure

So a one-day outpatient "procedure" ended up being a nasty 6-day
battle with helping the kidneys to recover.

Thanks to all for your good wishes!

...Jim Thompson
--
| James E.Thompson | mens |
| Analog Innovations | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| STV, Queen Creek, AZ 85142 Skype: skypeanalog | |
| Voice480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

Thinking outside the box... producing elegant solutions.

 

[amdx]

On 3/20/2017 7:55 PM, rickman wrote:

On 3/20/2017 6:54 PM, amdx wrote:
On 3/20/2017 5:27 PM, rickman wrote:
On 3/20/2017 6:21 PM, amdx wrote:
On 3/20/2017 3:34 PM, rickman wrote:
On 3/20/2017 2:16 PM, amdx wrote:
On 3/20/2017 11:29 AM, dagmargoodboat@yahoo.com wrote:

If you will look at my picture,
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163
you will notice I put the 470 ohm (turquoise resistor) on the other
side
of the shield, (I don't know why).
What affect would that have?

On Monday, March 20, 2017 at 9:36:51 AM UTC-4, amdx wrote:
If you want to get fancier
and even lower input capacitance we can bootstrap the drain,
too. A
bootstrapped shield for your pass-through (where you bring the
input
through your metal box) would help, too.

OK, I'd like to get fancy, but I don't understand "A bootstrapped
shield" for my pass through.
I would like to understand physically what I need to do.

I have seen this done for electrometers where the concern is leakage
current. There they call it a "guard" ring. As James says, it will
also reduce input capacitance.

I don't understand how you can bootstrap the drain though. Since
it is
connected to the positive rail, I'm not even sure what that means.


Here's another picture showing the input, there is 5/8" hole in the
case
and I glued polystyrene sheet over it with the input wire coming out
the
center.
http://i395.photobucket.com/albums/pp37/Qmavam/Input.jpg
Ah, here's a picture of a previous input cap, I don't use anymore,
but
it shows the input better.
http://i395.photobucket.com/albums/pp37/Qmavam/1cmx1cmspaced5mm.jpg

This latter image doesn't look like what is used in this image.

http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html

Was this just a temporary cap in place until you got the detail work
done?


Yes, that was a first iteration, nothing wrong with it, but it wasn't
as stable (two wires with a weights on the end) as my tiny piece of
double sided pcb.

Oh, so the free hanging copper is the old cap and the thing I can't see
is the PCB cap which you like better?


When you say can't see, do you mean it small and your looking side on
or do you mean you didn't see a link?
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163

btw the 0.3pf cap is a approximately 1/8" disc of Teflon pcb 0.32"
thick. PCB material is not critical. Easy enough to set size, check the
gain of your amp without the input cap, then reduce the size of your
input cap until gain equals 1.

I just meant it is small and I couldn't see what it was. Initially I
thought it might be a small value ceramic cap you put in place to check
out the amp before you optimized the cap. Now I realize it is a very
small piece of PCB and *is* the optimized cap. I also didn't know which
image was earlier and which was later.

Why would you adjust the overall circuit gain by trimming the cap? The
amp gain can be trimmed, no?

I'm sure the amp can be changed, but the end game was to make the input
cap small, I couldn't go much smaller.

What frequency range is this amp intended to be used for? Nearly all
the stuff I've seen where Q matters is LF or MF.

Kleijer says it's flat from 10kHz to 10 MHz.

They use loop antennas
for higher frequencies, but with too high Q the bandwidth gets so narrow
even voice won't get through, just CW.

Most of the stuff I'm looking at is crystal radio, by the time you
start driving a headphone your Q gets lowered a lot. I never hear the
crystal radio guys complain about lack of bandwidth.

Some of the antennas Kleijer uses I would think would limit the
bandwidth of his circuits with Q of over 1000.

Again not a common problem, but easy to fix with a load resistor.

If the frequencies are not so high, why not use an opamp for the final
stage rather than the much more complex push-pull stage? I assume that
is intended to be in essence a "power" output stage and does not
otherwise contribute significantly to the characteristics of the rest of
the circuit.

I'm sure it could be built with a couple of opamps. It is a 50 ohm
output. Could probably do better, this starts to distort at I think
8 Vpp. But I don't have any antenna that need much that headroom.

Mikek


---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

 

[amdx]

On 3/20/2017 8:08 PM, Jim Thompson wrote:

On Tue, 14 Mar 2017 18:58:32 -0700, Jim Thompson
To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:

"Procedure" tomorrow (Wednesday) at 11:00AM:

Down the throat with a scope, check out the stomach, then into the
small intestine, use side-looking ultrasound on the end of the probe
(didn't know such a thing existed) to examine the common pancreas/bile
duct, go up it with a wire, then thread a balloon up that wire,
inflate and decimate the stones, then go on up and examine the gall
bladder.

Possible later procedure, after the nauseous, tiredness, yellowness
abates, go in thru an incision and remove the gall bladder.

Such fun >:-}

If I don't show up in a day or too...

...Jim Thompson

Well the docs tried to kill me, but I survived a very unpleasant trip.

They put a stent in my bile duct, but in recovery they discovered my
kidneys couldn't cope, and went into failure

So a one-day outpatient "procedure" ended up being a nasty 6-day
battle with helping the kidneys to recover.

Thanks to all for your good wishes!

...Jim Thompson

So glad to hear from you Jim!
Good luck on your recovery!

Mikek

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

 

[Martin Riddle]

On Mon, 20 Mar 2017 03:45:26 -0700 (PDT), tabbypurr@gmail.com wrote:

On Wednesday, 15 March 2017 01:58:40 UTC, Jim Thompson wrote:
"Procedure" tomorrow (Wednesday) at 11:00AM:

Down the throat with a scope, check out the stomach, then into the
small intestine, use side-looking ultrasound on the end of the probe
(didn't know such a thing existed) to examine the common pancreas/bile
duct, go up it with a wire, then thread a balloon up that wire,
inflate and decimate the stones, then go on up and examine the gall
bladder.

Possible later procedure, after the nauseous, tiredness, yellowness
abates, go in thru an incision and remove the gall bladder.

Such fun >:-}

If I don't show up in a day or too...

...Jim Thompson

It's 5 days now.


NT

I'd give him at least a week. He might not be fully recovered if he
had his Gall bladder removed.

Cheers

 

[rickman]

On 3/20/2017 7:41 PM, billbowden wrote:

"rickman" <gnuarm@gmail.com> wrote in message
newsan4s4$99i$1@dont-email.me...
On 3/19/2017 7:31 PM, billbowden wrote:
"amdx" <nojunk@knology.net> wrote in message
newsam0ee$ch5$1@dont-email.me...
On 3/14/2017 11:18 PM, billbowden wrote:
Which is a better design. Suppose you have a 6 inch length of PVC pipe
with
numerous turns of wire that has an inductance of say 200uH. Now suppose
you
use the same (6 inch) piece of PVC with a ferrite rod in the core with
considerably fewer turns of wire. Which one would capture the most
signal
at the AM Broadcast frequencies (500K to 2 Megs) and produce the
greatest
signal output? Would it be more ferrite, or more wire?


I'll run the experiment.

Do you want it tuned?

If not, I have no way to measure the signals of my local stations.
I need the resonance peaking to see the signal.

What diameter PVC?

I have 1/2" OD polystyrene that will allow a little closer coupling
between the ferrite and the wire. 400 turns #28 = 203uh air core.

I have 1/2" CPVC. actual OD. 0.615"
290 turns #28 = 200uh air core


I have 1/2 PVC, actual OD. 0.832. 175 turns #28 = 205uh air core.

Pick one.

I'll also wind one with less turns and use my best Q rod that is 8" long
x
0.375" diameter.

I will check three frequencies, 590Khz, 1290kHz and 1430Khz.

I made a post last night of the wrong experiment (6"dia not 6" long)
It has not shown up this morning, so I'll repeat my measurement method.

To measure the signal I have a very high input impedance amp with a
gain
of 1.
I use the amp to drive a scope (ch 2) set at 50mV/div. I took the
channel
2 output from the back of the scope to drive a Boonton 92BD RF millivolt
meter. I use the scope to compare the visual to audio from a portable
radio to know where I am tuned.
Modulation causes a bit of amplitude bounce, but I do a visual average.

Let me know what you want.
Mikek


Actually, I'm just interested in comparing the response of two identical
loopsticks, one using an air core and the other using a ferrite core. I
could do the experiment since there is a 50KW station about 7 miles away
and
I can see the signal from the antenna loop directly on a scope. I can get
about 1 volt peak using a loop antenna of about 15 inches square. I just
thought someone would know the answer without a lot of experimenting. I
have
a portable car radio with a air core loop antenna mounted on the chassis
that measures 6.5 inch by 3/4 diameter and about 300 turns of small wire.
Works fine and gets stations 130 miles away. But it's a power hog and
draws
100mA from a 12 volt battery. I suppose a good test would be to use a
shorter ferrite rod and fewer turns to compare the results. But I'm lazy
and
just want to know which idea is better.

Connecting a scope directly to a loop antenna may cause a loss of Q. I'm
hoping not, as I am building a very high Q antenna that will depend on a
high impedance not sapping it, but it depends on the antenna. The point
is if your measurement saps the Q, then any impact on the Q by the ferrite
will not be noticed.


I just looked at my 15 inch square (air core) loop antenna on a scope. The
amplitude was about 1 volt P-P. I connected a 150K resistor across the loop
and the amplitude dropped to about half that. So, the impedance looks to be
about 150K. The scope input is 1 Meg and 30pF and I used a 10X probe. The Q
of the antenna was measured at 300 at the low end of the band (600Khz) and
200 at the high end. That was using a Boonton Q meter. So, it doesn't look
like connecting a scope has much effect on Q in this case. But possibly
doing the same test on a short ferrite loop would give different results.
Actually, the Q of the loop should be about 100 or less if you want a 10Khz
bandwidth.

I'm not looking for bandwidth. I don't think I can get the Q high
enough to cause a bandwidth problem.

--

Rick C

 

[rickman]

On 3/20/2017 9:09 PM, amdx wrote:

On 3/20/2017 7:55 PM, rickman wrote:
On 3/20/2017 6:54 PM, amdx wrote:
On 3/20/2017 5:27 PM, rickman wrote:
On 3/20/2017 6:21 PM, amdx wrote:
On 3/20/2017 3:34 PM, rickman wrote:
On 3/20/2017 2:16 PM, amdx wrote:
On 3/20/2017 11:29 AM, dagmargoodboat@yahoo.com wrote:

If you will look at my picture,
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163
you will notice I put the 470 ohm (turquoise resistor) on the other
side
of the shield, (I don't know why).
What affect would that have?

On Monday, March 20, 2017 at 9:36:51 AM UTC-4, amdx wrote:
If you want to get fancier
and even lower input capacitance we can bootstrap the drain,
too. A
bootstrapped shield for your pass-through (where you bring the
input
through your metal box) would help, too.

OK, I'd like to get fancy, but I don't understand "A bootstrapped
shield" for my pass through.
I would like to understand physically what I need to do.

I have seen this done for electrometers where the concern is leakage
current. There they call it a "guard" ring. As James says, it will
also reduce input capacitance.

I don't understand how you can bootstrap the drain though. Since
it is
connected to the positive rail, I'm not even sure what that means.


Here's another picture showing the input, there is 5/8" hole in the
case
and I glued polystyrene sheet over it with the input wire coming out
the
center.
http://i395.photobucket.com/albums/pp37/Qmavam/Input.jpg
Ah, here's a picture of a previous input cap, I don't use anymore,
but
it shows the input better.
http://i395.photobucket.com/albums/pp37/Qmavam/1cmx1cmspaced5mm.jpg

This latter image doesn't look like what is used in this image.

http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html

Was this just a temporary cap in place until you got the detail work
done?


Yes, that was a first iteration, nothing wrong with it, but it wasn't
as stable (two wires with a weights on the end) as my tiny piece of
double sided pcb.

Oh, so the free hanging copper is the old cap and the thing I can't see
is the PCB cap which you like better?


When you say can't see, do you mean it small and your looking side on
or do you mean you didn't see a link?
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163

btw the 0.3pf cap is a approximately 1/8" disc of Teflon pcb 0.32"
thick. PCB material is not critical. Easy enough to set size, check the
gain of your amp without the input cap, then reduce the size of your
input cap until gain equals 1.

I just meant it is small and I couldn't see what it was. Initially I
thought it might be a small value ceramic cap you put in place to check
out the amp before you optimized the cap. Now I realize it is a very
small piece of PCB and *is* the optimized cap. I also didn't know which
image was earlier and which was later.

Why would you adjust the overall circuit gain by trimming the cap? The
amp gain can be trimmed, no?

I'm sure the amp can be changed, but the end game was to make the input
cap small, I couldn't go much smaller.

What frequency range is this amp intended to be used for? Nearly all
the stuff I've seen where Q matters is LF or MF.

Kleijer says it's flat from 10kHz to 10 MHz.

They use loop antennas
for higher frequencies, but with too high Q the bandwidth gets so narrow
even voice won't get through, just CW.

Most of the stuff I'm looking at is crystal radio, by the time you
start driving a headphone your Q gets lowered a lot. I never hear the
crystal radio guys complain about lack of bandwidth.

Some of the antennas Kleijer uses I would think would limit the
bandwidth of his circuits with Q of over 1000.

Again not a common problem, but easy to fix with a load resistor.

If the headphones or load resistor limit the Q, what is the bleeping
point of all that Litz wire and custom tuning capacitor stuff? I
thought the whole point was to get the Q as high as possible. If you
start with a Q over 1000 and end up with a few hundred, I wouldn't
expect that to be distinguishable from a coil with a Q of some hundreds
which is where Kleijer started in his coil investigations.

If the frequencies are not so high, why not use an opamp for the final
stage rather than the much more complex push-pull stage? I assume that
is intended to be in essence a "power" output stage and does not
otherwise contribute significantly to the characteristics of the rest of
the circuit.

I'm sure it could be built with a couple of opamps. It is a 50 ohm
output. Could probably do better, this starts to distort at I think
8 Vpp. But I don't have any antenna that need much that headroom.

I designed an opamp circuit to output an 8 volt signal with a 12 volt
supply into 50 ohm cable. I didn't quite get 8 volts before it starts
to distort, but it was close. I believe the gain-bandwidth product of
the amp was over 30 MHz, so it should do ok at 10 MHz and no gain. But
then you need a little gain I guess. Opps, no, the GB was 15 MHz.

--

Rick C

 

[Tim Williams]

"Clifford Heath" <no.spam@please.net> wrote in message
news:58cf69a5$0$10743$c3e8da3$5d8fb80f@news.astraweb.com...

Interesting, I don't know. I suspect someone would have thought of
that during the >10years of research that went into this. Would
radiation kill the bias magnet? (of course, for this purpose it
wouldn't need to be built-in).

Hmm, I don't know offhand!

Looks like magnets are moderately sensitive:
https://accelconf.web.cern.ch/accelconf/r04/papers/TUKP06.PDF
I'm not sure offhand how many n/cm^2 corresponds to what Sv/Gy/rem/etc.
rate, or of course how that compares to beta or gamma radiation. But strong
reductions in the 10^15 range seems pretty sensitive, like a
knock-the-electron-out-of-place-and-she's-done ~quantitative kind of rate.

Hmm, I wonder how much that depends on the boron content. Ferrite magnets
wouldn't need to worry about that.

Anyway, no, you wouldn't need to bring the magnet with, not unless you need
that as a semi-permanent lock signal (as it's used in commerce).

There's also Wiegand wires, which I think are embedded (or omitted) in an
array, to give a pattern of blips as they slide past the pickup coil and
become magnetized.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com

 

[amdx]

On 3/20/2017 8:28 PM, rickman wrote:

On 3/20/2017 9:09 PM, amdx wrote:
On 3/20/2017 7:55 PM, rickman wrote:
On 3/20/2017 6:54 PM, amdx wrote:
On 3/20/2017 5:27 PM, rickman wrote:
On 3/20/2017 6:21 PM, amdx wrote:
On 3/20/2017 3:34 PM, rickman wrote:
On 3/20/2017 2:16 PM, amdx wrote:
On 3/20/2017 11:29 AM, dagmargoodboat@yahoo.com wrote:

If you will look at my picture,
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163

you will notice I put the 470 ohm (turquoise resistor) on the other
side
of the shield, (I don't know why).
What affect would that have?

On Monday, March 20, 2017 at 9:36:51 AM UTC-4, amdx wrote:
If you want to get fancier
and even lower input capacitance we can bootstrap the drain,
too. A
bootstrapped shield for your pass-through (where you bring the
input
through your metal box) would help, too.

OK, I'd like to get fancy, but I don't understand "A bootstrapped
shield" for my pass through.
I would like to understand physically what I need to do.

I have seen this done for electrometers where the concern is leakage
current. There they call it a "guard" ring. As James says, it will
also reduce input capacitance.

I don't understand how you can bootstrap the drain though. Since
it is
connected to the positive rail, I'm not even sure what that means.


Here's another picture showing the input, there is 5/8" hole in the
case
and I glued polystyrene sheet over it with the input wire coming
out
the
center.
http://i395.photobucket.com/albums/pp37/Qmavam/Input.jpg
Ah, here's a picture of a previous input cap, I don't use anymore,
but
it shows the input better.
http://i395.photobucket.com/albums/pp37/Qmavam/1cmx1cmspaced5mm.jpg


This latter image doesn't look like what is used in this image.

http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html

Was this just a temporary cap in place until you got the detail work
done?


Yes, that was a first iteration, nothing wrong with it, but it
wasn't
as stable (two wires with a weights on the end) as my tiny piece of
double sided pcb.

Oh, so the free hanging copper is the old cap and the thing I can't
see
is the PCB cap which you like better?


When you say can't see, do you mean it small and your looking side on
or do you mean you didn't see a link?
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163

btw the 0.3pf cap is a approximately 1/8" disc of Teflon pcb 0.32"
thick. PCB material is not critical. Easy enough to set size, check the
gain of your amp without the input cap, then reduce the size of your
input cap until gain equals 1.

I just meant it is small and I couldn't see what it was. Initially I
thought it might be a small value ceramic cap you put in place to check
out the amp before you optimized the cap. Now I realize it is a very
small piece of PCB and *is* the optimized cap. I also didn't know which
image was earlier and which was later.

Why would you adjust the overall circuit gain by trimming the cap? The
amp gain can be trimmed, no?

I'm sure the amp can be changed, but the end game was to make the input
cap small, I couldn't go much smaller.

What frequency range is this amp intended to be used for? Nearly all
the stuff I've seen where Q matters is LF or MF.

Kleijer says it's flat from 10kHz to 10 MHz.

They use loop antennas
for higher frequencies, but with too high Q the bandwidth gets so narrow
even voice won't get through, just CW.

Most of the stuff I'm looking at is crystal radio, by the time you
start driving a headphone your Q gets lowered a lot. I never hear the
crystal radio guys complain about lack of bandwidth.

Some of the antennas Kleijer uses I would think would limit the
bandwidth of his circuits with Q of over 1000.

Again not a common problem, but easy to fix with a load resistor.

If the headphones or load resistor limit the Q, what is the bleeping
point of all that Litz wire and custom tuning capacitor stuff? I
thought the whole point was to get the Q as high as possible.

Yes, it is. But you still need to get the maximum power out to drive
your headphones.
In a perfect system, you would drop your Q by half to get maximum
audio. But the diode gets involved and that I can't decipher.

If you start with a Q over 1000 and end up with a few hundred, I wouldn't
expect that to be distinguishable from a coil with a Q of some hundreds
which is where Kleijer started in his coil investigations.

This is radio dxing, grab every db of signal!

If the frequencies are not so high, why not use an opamp for the final
stage rather than the much more complex push-pull stage? I assume that
is intended to be in essence a "power" output stage and does not
otherwise contribute significantly to the characteristics of the rest of
the circuit.

I'm sure it could be built with a couple of opamps. It is a 50 ohm
output. Could probably do better, this starts to distort at I think
8 Vpp. But I don't have any antenna that need much that headroom.

That's 8Vpp open circuit, 4Vpp loaded.

I designed an opamp circuit to output an 8 volt signal with a 12 volt
supply into 50 ohm cable. I didn't quite get 8 volts before it starts
to distort, but it was close. I believe the gain-bandwidth product of
the amp was over 30 MHz, so it should do ok at 10 MHz and no gain. But
then you need a little gain I guess. Opps, no, the GB was 15 MHz.

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus

 

[Martin Riddle]

On Mon, 20 Mar 2017 18:08:35 -0700, Jim Thompson
<To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:

On Tue, 14 Mar 2017 18:58:32 -0700, Jim Thompson
To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:

"Procedure" tomorrow (Wednesday) at 11:00AM:

Down the throat with a scope, check out the stomach, then into the
small intestine, use side-looking ultrasound on the end of the probe
(didn't know such a thing existed) to examine the common pancreas/bile
duct, go up it with a wire, then thread a balloon up that wire,
inflate and decimate the stones, then go on up and examine the gall
bladder.

Possible later procedure, after the nauseous, tiredness, yellowness
abates, go in thru an incision and remove the gall bladder.

Such fun >:-}

If I don't show up in a day or too...

...Jim Thompson

Well the docs tried to kill me, but I survived a very unpleasant trip.

They put a stent in my bile duct, but in recovery they discovered my
kidneys couldn't cope, and went into failure

So a one-day outpatient "procedure" ended up being a nasty 6-day
battle with helping the kidneys to recover.

Thanks to all for your good wishes!

...Jim Thompson

Good to hear from you, complications suck. Hope your kidneys are doing
better, get some rest Jim.

Cheers

 

[rickman]

On 3/20/2017 10:23 PM, amdx wrote:

On 3/20/2017 8:28 PM, rickman wrote:
On 3/20/2017 9:09 PM, amdx wrote:
On 3/20/2017 7:55 PM, rickman wrote:
On 3/20/2017 6:54 PM, amdx wrote:
On 3/20/2017 5:27 PM, rickman wrote:
On 3/20/2017 6:21 PM, amdx wrote:
On 3/20/2017 3:34 PM, rickman wrote:
On 3/20/2017 2:16 PM, amdx wrote:
On 3/20/2017 11:29 AM, dagmargoodboat@yahoo.com wrote:

If you will look at my picture,
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163


you will notice I put the 470 ohm (turquoise resistor) on the
other
side
of the shield, (I don't know why).
What affect would that have?

On Monday, March 20, 2017 at 9:36:51 AM UTC-4, amdx wrote:
If you want to get fancier
and even lower input capacitance we can bootstrap the drain,
too. A
bootstrapped shield for your pass-through (where you bring the
input
through your metal box) would help, too.

OK, I'd like to get fancy, but I don't understand "A bootstrapped
shield" for my pass through.
I would like to understand physically what I need to do.

I have seen this done for electrometers where the concern is
leakage
current. There they call it a "guard" ring. As James says, it
will
also reduce input capacitance.

I don't understand how you can bootstrap the drain though. Since
it is
connected to the positive rail, I'm not even sure what that means.


Here's another picture showing the input, there is 5/8" hole in
the
case
and I glued polystyrene sheet over it with the input wire coming
out
the
center.
http://i395.photobucket.com/albums/pp37/Qmavam/Input.jpg
Ah, here's a picture of a previous input cap, I don't use anymore,
but
it shows the input better.
http://i395.photobucket.com/albums/pp37/Qmavam/1cmx1cmspaced5mm.jpg



This latter image doesn't look like what is used in this image.

http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html

Was this just a temporary cap in place until you got the detail
work
done?


Yes, that was a first iteration, nothing wrong with it, but it
wasn't
as stable (two wires with a weights on the end) as my tiny piece of
double sided pcb.

Oh, so the free hanging copper is the old cap and the thing I can't
see
is the PCB cap which you like better?


When you say can't see, do you mean it small and your looking side on
or do you mean you didn't see a link?
http://s395.photobucket.com/user/Qmavam/media/inside.jpg.html?o=163

btw the 0.3pf cap is a approximately 1/8" disc of Teflon pcb 0.32"
thick. PCB material is not critical. Easy enough to set size, check
the
gain of your amp without the input cap, then reduce the size of your
input cap until gain equals 1.

I just meant it is small and I couldn't see what it was. Initially I
thought it might be a small value ceramic cap you put in place to check
out the amp before you optimized the cap. Now I realize it is a very
small piece of PCB and *is* the optimized cap. I also didn't know
which
image was earlier and which was later.

Why would you adjust the overall circuit gain by trimming the cap? The
amp gain can be trimmed, no?

I'm sure the amp can be changed, but the end game was to make the input
cap small, I couldn't go much smaller.

What frequency range is this amp intended to be used for? Nearly all
the stuff I've seen where Q matters is LF or MF.

Kleijer says it's flat from 10kHz to 10 MHz.

They use loop antennas
for higher frequencies, but with too high Q the bandwidth gets so
narrow
even voice won't get through, just CW.

Most of the stuff I'm looking at is crystal radio, by the time you
start driving a headphone your Q gets lowered a lot. I never hear the
crystal radio guys complain about lack of bandwidth.

Some of the antennas Kleijer uses I would think would limit the
bandwidth of his circuits with Q of over 1000.

Again not a common problem, but easy to fix with a load resistor.

If the headphones or load resistor limit the Q, what is the bleeping
point of all that Litz wire and custom tuning capacitor stuff? I
thought the whole point was to get the Q as high as possible.

Yes, it is. But you still need to get the maximum power out to drive
your headphones.
In a perfect system, you would drop your Q by half to get maximum
audio. But the diode gets involved and that I can't decipher.


If you start with a Q over 1000 and end up with a few hundred, I
wouldn't
expect that to be distinguishable from a coil with a Q of some hundreds
which is where Kleijer started in his coil investigations.

This is radio dxing, grab every db of signal!


If the frequencies are not so high, why not use an opamp for the final
stage rather than the much more complex push-pull stage? I assume that
is intended to be in essence a "power" output stage and does not
otherwise contribute significantly to the characteristics of the
rest of
the circuit.

I'm sure it could be built with a couple of opamps. It is a 50 ohm
output. Could probably do better, this starts to distort at I think
8 Vpp. But I don't have any antenna that need much that headroom.

That's 8Vpp open circuit, 4Vpp loaded.

Well then, eazy peazy. I used a circuit that used positive feedback to
make the output impedance look higher than the resistor in the circuit.
I think I used 12.5 or maybe it was just 12 and got multiplied by 4.
This makes the calculation of the gain a little more complex so you
would not be able to adjust the gain with a pot or anything... unless it
was on an earlier stage.

The middle stage with the single transistor, anything special about
that? It has a connection with caps I'm not familiar with from the
emitter to the input. That would be a low level of positive feedback I
believe. Any idea why?

I designed an opamp circuit to output an 8 volt signal with a 12 volt
supply into 50 ohm cable. I didn't quite get 8 volts before it starts
to distort, but it was close. I believe the gain-bandwidth product of
the amp was over 30 MHz, so it should do ok at 10 MHz and no gain. But
then you need a little gain I guess. Opps, no, the GB was 15 MHz.

Wait a minute!!! That bootstrap of the input resistors won't work. The
470 ohm source resistor is there to set a bias point with the source
higher than the gate. The resistors are there to set the DC voltage of
the gate to ground. Perhaps a cap could be used to feed the point
between the two resistors?

I just realized that is the purpose of the two caps on T2 emitter to the
47 ohm resistor. It allows the resistors to set the DC operating point
but bootstraps their capacitance. No?

--

Rick C

 

On Monday, March 20, 2017 at 9:49:55 PM UTC-4, dagmarg...@yahoo.com wrote:

On Monday, March 20, 2017 at 6:15:37 PM UTC-4, amdx wrote:
On 3/20/2017 2:31 PM, dagmargoodboat@yahoo.com wrote:

The idea of bootstrapping is to provide a unity-gain buffer, then make
all the circuit nodes swing with the input voltage, so that the input
signal doesn't have to charge any of the nodes' capacitances.

This illustrates the ideas--

+12V +12V
-+- -+-
| |
| [22k] R4
\| |
Q1 |---+----------.
.<| | |
| [47k] R5 |
(shield) T1 |--' | |
------ BF256C | === |
----------+----->|--+----------------|---//
---+-- | | Vdd |
| [10M] R1 | -+- |
| | | | |
| [10M] R2 | |/ Q2 |
| | .-----+---| |
| | | | |>. C1 |
| '---' R3[470] | 100n |
| | +---+--||--'
| === |R6 |
| [1K] '--||--.
| | 100n |
| === C2 |
'---------------------------------+
|
[22k] R7
|
===

Q1 forces the drain to move up and down *with* the input signal,
cancelling Cgd. T1's source already follows the input signal,
reducing the effective Cgs.

I've shown Q2 driving a shield. It's optional--you likely don't need that.

Ok, so if I was using short coax to connect to my LC, Q2 would drive
the shield to null it's capacitance.

Yep.

I now use two 6" wires to clip to the LC. the two wire run about 2"
apart. I have wondered how much capacitance the leads add.

I suspect a lot. But you could always measure & be sure.

For a permanent test fixture with a tuning cap and a signal strength
meter, ready to plunk any coil into for testing this would be great use
of a fixed coax input.

Yes.

If our buffer (Q2) and T1 together manage a gain of 0.8, we'll cancel
roughly 80% of the input capacitance. (A more complicated buffer
getting closer to unity gain would provide even better input
capacitance cancellation.)

Still talking a about coax capacitance? Right

No, this applies to the whole scheme. The more closely all the nodes
follow the input signal voltage, the better our result.

Your author says his 0.3pF input cap and FET buffer form a 1:17 divider,
implying T1's effective input capacitance is about 5pF. You should be
able to improve that by a factor of five without breaking a sweat, by a
factor of ten with a little more care.

Now is that from the 20meg moving to the T1 source, or some more from
Q1 or both?

Both.

Note I added the missing 0.3pf cap to the schematic.

Yes. I was just sketching the relevant part of the front-end.

Can I tell this is working if my 1X gain increases?
The 17 to 1 divider of the input cap and the gate capacitance and the 17
times gain of the amplifier equals 1X.

Say I get 80% T1 gate cancellation (by moving the 20Meg), now we have
effectively 1pf.
1pf/0.3pf = 3.33 and the amp gain 17 / 3.33 = 5.1
So I would think my total circuit gain would increase to 5.1.
Or do I not get it?

You've got it perfectly. I don't expect a very large improvement from
bootstrapping the 20M alone though--a resistor's capacitance is pretty
low already, and two in series, even lower.

Your author's figures are inconsistent. He starts saying the input
capacitance is 1.4pF and the input coupling cap is 0.3pF, but then he
says the 0.3pF and FET T1's capacitances form a 17:1 divider. That can't
all be true--0.3pF should form a 5.7:1 divider with a 1.4pF input, not
17:1.

When I guesstimate a 5x improvement, I'm banking on the 17:1 being true,
c.in(eff) being 5pF, and getting that down to 1pF, roughly, with the
circuit I sketched.

If you're already really at 1.4pF the improvement will only be 1.0pF/1.4pF,
and not 1.0pF/5pF.

As I said before, a better buffer could do better--you could tweak the
bootstrap to perfect null--but then chances are you'd have an oscillator.

What I posted seemed like a reasonable compromise for a first try.

I thought about this a bit and came up with an improved follower.

The main limitation of the previous circuit was the FET's poor performance
as a voltage-follower. Unaided, the T1 has a gain of about 0.6. That hits
our bootstrapping from all sides. First, c(gs) (the largest capacitance)
is only bootstrapped by 60%, leaving 40% of the BC547C's ~5pF c(gs). Next,
we use that voltage to drive our less-than-unity Q2, which drives less-than-
unity Q1. This all adds up.

Changing T1's load to a current sink makes T1 into a much better follower,
increasing voltage gain from 0.6 to about 0.95. The better 'follower'
action now bootstraps away nearly all of c(gs) (T1's largest capacitance),
and gives us a better signal to drive the drain bootstrap as well. Good,
good, and good. And not terribly much trouble to do, either.

Vdd Vdd
-+- -+-
| |
| [22k] R5
Q1 \| |
BC547B |---+-------.
.<| | |
| [47k] R6 |
(shield) T1 |--' | |
------ BF256C | === |
>----------+----->|--. |
---+-- | | Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | +---| BC547B |
| | | |>. |
| | R3 [470] | |
| | | | | C3
| | | | | 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

Cheers,
James Arthur

 

That's just a Gilbert cell made of MOSFETs, isn't it?

No, it's a bridge topology, just like a diode bridge except without
mixing the control and signal currents (except via gate capacitance).

I was perhaps unclear--it's a ring topology, i.e. symmetrical, unlike a bridge rectifier.

Both are bridges in the strict sense, of course.

Cheers

Phil Hobbs

 

[Tim Williams]

"Clifford Heath" <no.spam@please.net> wrote in message
news:58d0bb96$0$39552$b1db1813$7968482@news.astraweb.com...

Since the gates are high-Z, you can use a large LO swing, which is mainly
what helps the IMD.

The problem with needing large LO swing is slew rate, of course.
It has to be quite substantial to get good rejection, because
it's not effective while still slewing.

The Tayloe approach uses a mux, but is the same basic idea, and aiui it
came along much later. So I think it should be called the Oxner mixer.

Same issue. You're pushing devices to switch at substantially
faster slew rates than required for the signal (large signal &
rapid transition compared to the frequency of interest).

You're ultimately limited by how much power you want to waste driving the LO
port, and how much voltage the gates (and substrate) can withstand.

The average gate and substrate voltages are normally biased independently,
to avoid substrate diode conduction (Vss is negative so-and-so), and to get
balance right. (This is in the appnote section of the aforementioned mixer
IC.)

I wonder what a vacuum tube design would offer. Pentodes should give
fantastic dynamic range. Though THD (and by extension, IMD) probably won't
be very good, because of the overall V^(3/2) characteristic.

But you can do even greater things in vacuuo. The venerable 7360 is the
greatest exemplar of the current-steering mixer (a half of a Gilbert cell,
e.g. CA2028). It works by physically steering the electron beam itself, via
electric field deflection. Look ma, no hands! (But I would guess the noise
performance still isn't as good as a traditional BJT Gilbert cell, simply on
account of being tube state.)

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com

 

[Phil Hobbs]

On 03/20/2017 10:17 PM, Clifford Heath wrote:

On 20/03/17 22:26, pcdhobbs@gmail.com wrote:
Thank you. I didn't realize they used series Schottky strings.
Could the need be somewhat averted with back bias instead?
Not so easy in a ring topology--the diodes are in antiparallel, not as
in a bridge rectifier.

That makes it easier - you can break the ring with a capacitor
(or a pair) and bias the whole string with one bias voltage.
They'll still be a ring at AC.

The LO switches the diodes hard, so it's tough to AC-couple them without
loading down the ring and so breaking the double-balanced condition.
It's probably possible, but it isn't obvious. It isn't like a sampling
bridge.

The MOSFET approach was developed by the late great Ed Oxner at
Siliconix in the late '70s, and published as an app note for the
Si8901 monolithic quad FET. (It might have been 8601--iirc they
re-used the p/n for a gate driver chip so the old one is hard to fin
info on.)
The idea is to put the FETs in a ring, cross-connect the gates in two
pairs and put the LO there via the usual transformer. The RF and IF
ports are the corners of the ring (1 and 3 for one and 2 and 4 for the
other).

That's just a Gilbert cell made of MOSFETs, isn't it?

No, it's a bridge topology, just like a diode bridge except without
mixing the control and signal currents (except via gate capacitance).

Since the gates are high-Z, you can use a large LO swing, which is
mainly what helps the IMD.

The problem with needing large LO swing is slew rate, of course.
It has to be quite substantial to get good rejection, because
it's not effective while still slewing.

That's one problem. In a Gilbert cell, the RF has to go into the bases
of a diff pair. It's the nonlinearity of the diff pair that causes the
issue in voltage-mode Gilbert cells such as the MC1496 (The CA3102 isn't
a Gilbert cell because it has only one diff pair instead of three.) A
real Gilbert cell, i.e. one with current inputs instead of voltage, is
somewhat better, but even with matched transistors there's still AC
nonlinearity due to the Early effect.

You can patch a lot of this stuff up by pouring on extra parts, but that
usually leads to limitations at high frequency.

The Tayloe approach uses a mux, but is the same basic idea, and aiui
it came along much later. So I think it should be called the Oxner mixer.

Same issue. You're pushing devices to switch at substantially
faster slew rates than required for the signal (large signal &
rapid transition compared to the frequency of interest).

No. The MOSFET approaches also separate the control and signal paths
and reduce the ON-resistance modulation even in the fully-ON state.

The biggest win is not letting the RF signals intermodulate by putting
them into a nonlinear stage without feedback, such as the bases of a
diff pair or a weak bipolar RF amp. (JFETs are somewhat better as RF
amps because their IMD is predominantly even-order, and even-order spurs
tend not to land in the IF passband the way odd-order ones often do.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

 

[Tom Gardner]

On 21/03/17 01:08, Jim Thompson wrote:

On Tue, 14 Mar 2017 18:58:32 -0700, Jim Thompson
To-Email-Use-The-Envelope-Icon@On-My-Web-Site.com> wrote:

"Procedure" tomorrow (Wednesday) at 11:00AM:

Down the throat with a scope, check out the stomach, then into the
small intestine, use side-looking ultrasound on the end of the probe
(didn't know such a thing existed) to examine the common pancreas/bile
duct, go up it with a wire, then thread a balloon up that wire,
inflate and decimate the stones, then go on up and examine the gall
bladder.

Possible later procedure, after the nauseous, tiredness, yellowness
abates, go in thru an incision and remove the gall bladder.

Such fun >:-}

If I don't show up in a day or too...

...Jim Thompson

Well the docs tried to kill me, but I survived a very unpleasant trip.

They put a stent in my bile duct, but in recovery they discovered my
kidneys couldn't cope, and went into failure

So a one-day outpatient "procedure" ended up being a nasty 6-day
battle with helping the kidneys to recover.

Thanks to all for your good wishes!

Best wishes again.

Good luck, and choose your future battles wisely to
minimise your use of micromorts

 

On Tue, 21 Mar 2017 01:32:19 -0500, "Tim Williams"
<tiwill@seventransistorlabs.com> wrote:

"Clifford Heath" <no.spam@please.net> wrote in message
news:58d0bb96$0$39552$b1db1813$7968482@news.astraweb.com...
Since the gates are high-Z, you can use a large LO swing, which is mainly
what helps the IMD.

The problem with needing large LO swing is slew rate, of course.
It has to be quite substantial to get good rejection, because
it's not effective while still slewing.

The Tayloe approach uses a mux, but is the same basic idea, and aiui it
came along much later. So I think it should be called the Oxner mixer.

Same issue. You're pushing devices to switch at substantially
faster slew rates than required for the signal (large signal &
rapid transition compared to the frequency of interest).

You're ultimately limited by how much power you want to waste driving the LO
port, and how much voltage the gates (and substrate) can withstand.

The average gate and substrate voltages are normally biased independently,
to avoid substrate diode conduction (Vss is negative so-and-so), and to get
balance right. (This is in the appnote section of the aforementioned mixer
IC.)

I wonder what a vacuum tube design would offer. Pentodes should give
fantastic dynamic range. Though THD (and by extension, IMD) probably won't
be very good, because of the overall V^(3/2) characteristic.

Double tetrodes has been used as high level _transmitter_ mixers in
SSB transmitters, especially when trying to avoid as many expensive
VHF or UHF gain stages as possible.

The QQE06/40 would be a nice high level mixer

I have no idea about the noise performance of small power tetrodes and
hence the suitability as _receiver_ mixer.

But you can do even greater things in vacuuo. The venerable 7360 is the
greatest exemplar of the current-steering mixer (a half of a Gilbert cell,
e.g. CA2028). It works by physically steering the electron beam itself, via
electric field deflection. Look ma, no hands! (But I would guess the noise
performance still isn't as good as a traditional BJT Gilbert cell, simply on
account of being tube state.)

Nice tube.

 

[Tim Williams]

"billbowden" <bperryb@bowdenshobbycircuits.info> wrote in message
news:58d068a7$0$59750$c3e8da3$460562f1@news.astraweb.com...

I just looked at my 15 inch square (air core) loop antenna on a scope. The
amplitude was about 1 volt P-P. I connected a 150K resistor across the
loop and the amplitude dropped to about half that. So, the impedance looks
to be about 150K. The scope input is 1 Meg and 30pF and I used a 10X
probe.

Mind that the probe acts like (9M || 11pF) + (100pF || 1M), but the cable
and scope input aren't actually 100pF but are lossy, having probably around
50 ohms ESR, and all the distributed L and C implied by a transmission line.

If you look at (10pF + 50) ohms, the parallel equivalent is 5M (at 1MHz, to
ballpark it), so the overall resistance is about 3M.

Still pretty big, but only 20 times bigger than 150k, not a huge amount.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Contract Design
Website: http://seventransistorlabs.com

 

On Tuesday, March 21, 2017 at 8:27:51 AM UTC-4, amdx wrote:

On 3/21/2017 2:19 AM, dagmargoodboat@yahoo.com wrote:
On Monday, March 20, 2017 at 9:49:55 PM UTC-4, dagmarg...@yahoo.com wrote:
On Monday, March 20, 2017 at 6:15:37 PM UTC-4, amdx wrote:

Can I tell this is working if my 1X gain increases?
The 17 to 1 divider of the input cap and the gate capacitance and the 17
times gain of the amplifier equals 1X.

Say I get 80% T1 gate cancellation (by moving the 20Meg), now we have
effectively 1pf.
1pf/0.3pf = 3.33 and the amp gain 17 / 3.33 = 5.1
So I would think my total circuit gain would increase to 5.1.
Or do I not get it?

You've got it perfectly. I don't expect a very large improvement from
bootstrapping the 20M alone though--a resistor's capacitance is pretty
low already, and two in series, even lower.

Your author's figures are inconsistent. He starts saying the input
capacitance is 1.4pF and the input coupling cap is 0.3pF, but then he
says the 0.3pF and FET T1's capacitances form a 17:1 divider. That can't
all be true--0.3pF should form a 5.7:1 divider with a 1.4pF input, not
17:1.

When I guesstimate a 5x improvement, I'm banking on the 17:1 being true,
c.in(eff) being 5pF, and getting that down to 1pF, roughly, with the
circuit I sketched.

If you're already really at 1.4pF the improvement will only be 1.0pF/1.4pF,
and not 1.0pF/5pF.

As I said before, a better buffer could do better--you could tweak the
bootstrap to perfect null--but then chances are you'd have an oscillator.

What I posted seemed like a reasonable compromise for a first try.

I thought about this a bit and came up with an improved follower.

The main limitation of the previous circuit was the FET's poor performance
as a voltage-follower. Unaided, the T1 has a gain of about 0.6. That hits
our bootstrapping from all sides. First, c(gs) (the largest capacitance)
is only bootstrapped by 60%, leaving 40% of the BC547C's ~5pF c(gs). Next,
we use that voltage to drive our less-than-unity Q2, which drives less-than-
unity Q1. This all adds up.

Changing T1's load to a current sink makes T1 into a much better follower,
increasing voltage gain from 0.6 to about 0.95. The better 'follower'
action now bootstraps away nearly all of c(gs) (T1's largest capacitance),
and gives us a better signal to drive the drain bootstrap as well. Good,
good, and good. And not terribly much trouble to do, either.

Vdd Vdd
-+- -+-
| |
| [22k] R5
Q1 \| |
BC547B |---+-------.
.<| | |
| [47k] R6 |
(shield) T1 |--' | |
------ BF256C | === |
----------+----->|--. |
---+-- | | Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | +---| BC547B |
| | | |>. |
| | R3 [470] | |
| | | | | C3
| | | | | 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

Cheers,
James Arthur

Thanks for the time.
As it is now constructed the enclosure is the shield.
Is that good or bad? ie. Should the enclosure be isolated from the shield?

The enclosure should be grounded! Let's not confuse that with
bootstrapping the input coax's shield (which you'll only do _if_
you use coax).

So yes, the enclosure should be isolated from the driven shield that is
shown in the schematic.

(My shield-driver is pretty wimpy, only suitable for a very short, low-
capacitance run. Might need beefing up.)

If you build it, it'll be fun to hear what output voltage you get from
this stage when you drive your 0.3pF input cap with, say, 50mV AC. If
you put 50mV into the 0.3pF and get 25mV out the back end, that means
our net input capacitance after bootstrapping is about the same as your
series 0.3pF.

Cheers,
James Arthur