Driver to drive?

On Saturday, March 25, 2017 at 3:20:31 PM UTC-4, rickman wrote:

On 3/25/2017 2:33 PM, dagmargoodboat@yahoo.com wrote:

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain closer
to 1, rather than 0.25.

The gain of 0.25 is not because of the voltage follower, it's from the
series cap.

The net gain's a result of the *divider* formed by the series coupling cap
and (the FET input capacitance + strays), multiplied by the follower gain.
I asked Mike to measure the follower gain so we could then calculate the
input capacitance.

Are you suggesting you can get the amp input capacitance
down to 10x less than the series cap?

Bootstrapping should reduce the FET input capacitance by a factor of
1/(1-G), where G = the gain of the bootstrap.

Another approach would be to put the series cap outside the box in the
probe tip as is done in o'scopes. Then the gain of the amp would be
sufficient to tie the coax shield to the output of the amp.

That's true. Mike could use a shielded connector and cable if he didn't
mind putting his series cap at the probe.

> What level do you think the amp input capacitance can reach?

Down to the strays, if we want, or even less. See above. This circuit
should be under 1/2 pF, by my estimate.

Cheers,
James Arthur

Based on your report I think the culprit limiting our improvement is likely
the drain driver, Q1 and related components. If you could 'scope (A) and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.

Cheers,
James Arthur

+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
>----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

 

[rickman]

On 3/23/2017 12:45 PM, krw@notreal.com wrote:

On Thu, 23 Mar 2017 06:42:31 -0000, "Kevin Aylward"
kevinRemovAT@kevinaylward.co.uk> wrote:

wrote in message news:0v46dc9bgb5pjf6lpc3jej19tl97qpu4db@4ax.com...

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.

Its obvious that you can't understand the issues involved.



Ho humm.

OK, you're not interested in the facts. Therefore, the only
conclusion possible is that you're trolling. Meet Slowman.

I gave you facts.

No, you don't even know the difference (I didn't read after your first
lie).

Well, there you go. It's pretty hard to have a conversation with
someone who isn't even going to listen to what you say. "Don't bother
me with the facts!"

--

Rick C

 

[rickman]

On 3/24/2017 7:46 PM, Cursitor Doom wrote:

On Fri, 24 Mar 2017 04:58:23 -0700, tabbypurr wrote:

You're wasting your time with him.

You don't know our Kev. He's not wasting his time one bit.


I'm all for sensible discussion but
once k & CD resort to stupid bs I don't see the point.

I haven't kept up with this thread and I don't know exactly what krw has
said, but I personally do maintain there is a perfectly solid case for
the death penalty when:

a) The crime is especially reprehensible

Do you realize that "reprehensible" is an emotional response?

I remember the Boston marathon bombing. One deli had the TV on for days
following every detail of the case. I asked the lady behind the counter
why this event was so riveting to her when a dozen people had died in
the Texas explosion just two days later. She couldn't explain other
than to relate how terrible the Boston bombing was. Clearly it was the
emotional aspects of a deliberate bombing by a terrorist.

Is that act really any different from other acts killing people? After
three years it was determined that the Texas explosion was also
deliberate by eliminating all possible accidental causes. No one seems
to be much concerned even though the nation remembers the Boston bombing
every year.

Which was "reprehensible"?

b) The identity of the perpetrator is beyond all question (not merely
"beyond reasonable doubt")

How do you establish "beyond all question"?

c) The mental capacity of the perpetrator was not impaired at the time

If that's what you call "stupid bs" then there's really nothing more I
can be bothered to say on the subject.

--

Rick C

 

[rickman]

On 3/23/2017 3:49 PM, Michael A. Terrell wrote:

Kevin Aylward wrote:
"Cursitor Doom" wrote in message newsaruc8$b5u$4@dont-email.me...

On Mon, 20 Mar 2017 20:23:37 -0400, krw wrote:

It's obvious you're illiterate.

no point in reading further

I feel I should at this point apologise for the remarks made by my
fellow
countryman, Kev. He and the poster "tabbypurr" are both singularly ill-
informed on this issue. Their ignorance is only matched by their
indefatigable ability to repeat the same dogma over and over and over
again. You will never win an argument against them; they simply won't
listen to reason. Do yourself a big favour and mark the thread "ignore"
in your newsreader. You'll save yourself from a huge amount of wasted
time.

I appreciate the support.

I did find the word "illiterate" somewhat amusing in as much as that it
is typically the "intellectuals" that present the reasoned arguments on
ethical issues rather than the redneck southerners.


I would rather have a 'redneck' on my side in a fight, than an
'ineffectual' intellectual.

Which is better would depend on whether the fight was on the Charlie
Rose show or Jerry Springer!

--

Rick C

 

[rickman]

On 3/24/2017 6:02 AM, Jeroen Belleman wrote:

On 2017-03-23 20:49, Michael A. Terrell wrote:


I would rather have a 'redneck' on my side in a fight, than an
'ineffectual' intellectual.


A redneck can help you win a fight, but you need the intellectuals
to have a chance to win the war.

How about redneck intellectual?

--

Rick C

 

[rickman]

On 3/24/2017 9:07 AM, Michael A. Terrell wrote:

Jeroen Belleman wrote:
On 2017-03-23 20:49, Michael A. Terrell wrote:


I would rather have a 'redneck' on my side in a fight, than an
'ineffectual' intellectual.


A redneck can help you win a fight, but you need the intellectuals
to have a chance to win the war.


How many intellectuals put their lives on the line, to win that war?
They sit in offices and decide what are acceptable levels of casualties
among the actual fighters.

Or they crack codes that save thousands of lives and ultimately win the
war!

--

Rick C

 

On Saturday, March 25, 2017 at 3:44:39 PM UTC-4, amdx wrote:

On 3/25/2017 1:33 PM, dagmargoodboat@yahoo.com wrote:

Based on your report I think the culprit limiting our improvement is likely
the drain driver, Q1 and related components. If you could 'scope (A) and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.


I adjusted both probes to be flat, then I connected both probes to the
same point (T1d) and adjusted *Var. for exact overlay of the two
waveforms. Then I move one probe to (T1s), I can not see any difference
in the signals.They still overlay, maybe a little thicker trace. I have
the brightness turned way down to make a thin trace, focus adjusted.
Looking for any small difference.

I'm a little confused by my scope. There is not a clear subtract (A-B).

The manual says ADD then INVERT to get a difference**.
See page 3-5 in the manual.
https://www.scribd.com/document/25186404/Tektronix-2465B-Options-and-Operator-Manual
Following the method in the manual, I get about 4mV of difference signal.
If this is not satisfactory, I have another Tektronix scope (475), I
can put on the bench. I think that has subtract.

* this was a very minor tweak
** this seems a poor method if your wave form is slightly asymmetrical.
But I don't see that.
Mikek

That's good technique, but the result is a little puzzling--with
bootstrapping gain this close to unity we'd expect better cancellation
of the FET capacitances.

The FET-follower into a current-sink load has a gain of 0.98 (measured).
That should cancel 98% of Cgs, the largest FET capacitance (5pF).

The drain-follower circuit also has a gain of roughly
(250mV-4mV)/250mV = .98, which, following the .98 gain of the FET stage,
produces a drain bootstrap gain of 0.96. That should cancel 96% of the
BF256C's Cdg (1.2pF), the remaining input-related capacitance.

Either there's close to 1pF of stray capacitance to ground at the input
node, or ... I've missed something. Like, what's your test frequency?
I've been assuming 1MHz (crystal-radio band).

Cheers,
James Arthur

 

[rickman]

On 3/25/2017 2:33 PM, dagmargoodboat@yahoo.com wrote:

On Saturday, March 25, 2017 at 10:29:57 AM UTC-4, amdx wrote:
On 3/25/2017 12:13 AM, dagmargoodboat@yahoo.com wrote:
On Friday, March 24, 2017 at 11:46:48 PM UTC-4, amdx wrote:
On 3/24/2017 11:40 AM, dagmargoodboat@yahoo.com wrote:
On Friday, March 24, 2017 at 10:13:18 AM UTC-4, amdx wrote:
On 3/24/2017 8:44 AM, amdx wrote:
On 3/23/2017 11:58 PM, amdx wrote:
On 3/23/2017 6:48 PM, dagmargoodboat@yahoo.com wrote:
On Thursday, March 23, 2017 at 2:52:11 PM UTC-4, amdx wrote:

OK, The circuit is built, I haven't tested it. I'm testing just the
circuit just the circuit posted.
I want to know how to drive the input.
Say I connect a 9 inch piece of RG58, that's 19pf, is that to much?
I want to drive it with a 50 ohm sig/gen.
Can I terminate the RG58 with 50 ohms or do I need to feed the RG58
with
a high impedance? Maybe a series 1Mohm.
I don't see how to separate the sig/gen cable from the input cable.
May I just need to connect it to the LC and see what I get.
Thank, Mikek


How about one step at a time? First, calibrate your input cap:

C1
.---. 0.3pF
| ~ |--------||----+----> A
'---' |
10V --- C2
--- 100pF
|
===

Measure V(A), calculate (don't forget to add your measuring instrument's
c.in to C2).


Easier said than done.
I have a pretty good capacitance tester. I made a .3pf cap :-/
and used an 82pf cap plus the probe, together they measured 106.7pf.
I applied 16Vpp and measured 0.06Vpp at A.
16 / 0.06 = 266.6 The voltage ratio is 266.6.
106.7 / 266.6 = 0.4
That makes my cap look like 0.4pf

Yes.

I measured the capacitance of a 6" x 8" piece (48sq in) of the PCB I
make my caps from, it measured 990 pf. 990 / 48 = 20.63pf per sq in.
My cap is 1/8" diameter or 0.01227 sq in.
So, 1sq in / 0.01227 sq in = 81.53
20.62pf/sq in / 81.53 = 0.253pf per 1/8" dia.
So fringing and lead to lead capacitance, I'm in the ballpark, just not
sure what field.
If you have any suggestions, I'm willing.
Way past my bedtime, good night.
Mikek

You're close enough for our purposes. The first measurement may be being
affected by stray capacitances, dunno. In the end all we'll care about
(when you use this thing) is getting 100mV out for 100mV in.

PS. PCB is Rogers Duroid 5880, I could not find a capacitance per sq in
number online.


I got out my better probe (Tek 6122 11pf) and made curly cues for the
probe tip.
Here's a picture of the tightened up circuit.
https://www.dropbox.com/s/sixe91dvxs40f1z/P1010193.JPG?dl=0


The new total capacitance, probe plus 82pf cap is 94pf. (measured)
16Vpp input, V(A) is 0.068Vpp. Then 16/0.068 = 235.3
235.3 / 94 = 2.5pf.

No, that should be 94pF * (.068V / 16V) = 0.4 pF.

This is a tighter measurement and also agrees with my calculation of
0.253 that I calculated for the pcb capacitance per sq in.
Rather than fight with a new piece of pcb material just slightly over
1/8" in diameter.
I'm going to go with a new value input cap of 0.25pf.
If you see this as a bad idea, let me know and I'll fumble around
and make a larger cap. I used a 1/8' paper punch to make this cap,
I could use a 1/4" and grind it down, but it's a chore.

It looks to me like your input cap is good, and 0.4pF. No need to change
it.

Now for the next measurement.
Is this with a T1 in the completed circuit?
Or, do I add the probe + 82pf back in to get a ratio?
Ya, I'm now confused.

Yes, it's time to use your calibrated in the completed, working circuit.
We'll want to measure the a.c. signal at the FET source first, then the
FET's drain.

I think we might want the voltage at the gate, but I
think that would be an impossible measurement. ie putting 11pf
probe across a 0.3pf cap ?
Thanks, Mikek



Ok, I tried this, out of circuit, just the FET.



T1 |--
.---. 0.3pF |
| ~ |--------||------>|--.
'---' | T1.s = 0.06Vpp
16Vpp |
|---probe tip
82pf –--
--- 12pf
|---probe ground
|
---
///

So, 16Vpp / 0.6Vpp = 266.6
266.6 / 94 = 2.84pf
2.84pf - 0.3pf = 2.54pf for the Cgs.
I think that's to low, I thought it was 5pf,
but I'm looking for your thoughts.

Thanks Mikek

Sorry, I wasn't clear enough. We're measuring in-circuit.

To check the effective input capacitance we want to know the attenuation
of the input a.c. voltage, as measured at point (A).

Rationale: The input voltage will be divided across your 0.40pF input
capacitor and the circuit's input capacitance:

Ccoupling
0.40pF .. .. ..
Vin >---||----:--+--o V(A)
: |
: --- Ceff
: ---
: |
: ===
'.. .. ..

We can't measure at the gate--that's too high impedance--so we're measuring
at FET T1's source terminal, point (A).

If 63% of the voltage is lost across the 0.40pF series cap, then the
effective input capacitance is dropping 37%, which means the series
cap is 63% of the total reactance and Ceff is 37% of the total reactance.

More formally, Ceff = 0.40pF * (Vin - V(A)) / V(A).

(We're trying to measure Ceff.)

You can measure Vac at the FET source on the original circuit for comparison,
too. That's a good way to assess the relative performance of the new circuit
vs. the original (after correcting for any differences in the input coupling
caps, of course.)

(Kleijer said his input division ratio was 17:1, indicating his circuit's
input capacitance (Ceff) was 16 units, and his coupling cap was 1 unit of
capacitance, or Ceff = 16 x Ccoupling.)

We want to know the a.c. voltage @ (B) just to gauge how well our drain
portion of the bootstrap is working. The closer Vac(B) is to Vac(A), the
better we're doing. 1:1 would be ideal. 0.9 would be okay, less than 0.9
indicates a problem.


AC Voltage at B just slightly less than AC Voltage at A. I had to look
several times just to make sure there was a difference.

Okay, that means the drain bootstrap should be working.

Here's the circuit, for reference, with probe points (A) and (B) marked:

(Remember, I'm showing a boot-strapped coax on the input, but we're not
using that yet. We want to keep the comparisons apples-to-apples, and
change only one thing at a time.)


Preliminary data.
I have just a wire input like the original.
I have added my measured DC voltages to the schematic below.
With 1Vpp input, T1s has 0.25Vpp and the output is 0.24Vpp.
So, I think that means the bootstrap is working, I don't know how much
change to expect, but that seems good.

The voltage gain, input-to-output, is about 1/4, or roughly 4x better than
Kleijer's 1/17. But I was expecting to do a little better still and would
like to figure out what's up, if you're game for poking and prodding this
thing a bit more.

Your measurements mean we're dropping .75V across the 0.4pF coupling cap,
and 0.25V across our input capacitance at the FET gate. Our input
capacitance is thus about three times the coupling cap, or about 1.2pF.


In the morning I will compare this to the original.
Thanks, Mikek

You can compare to the original easily--temporarily connect R3 to ground
instead of to R4. That disables all of the bootstrapping, which makes the
new circuit operate as the original. Measure, and compare output voltage
without bootstrapping to the value obtained with bootstrapping. The
ratio tells us how much better we're doing than the original.

Everything checks as expect.
With bootstraping T1s is 0.225 Ratio about 4
When R3 is grounded T1s is 0.60 Ratio about 17
(Just to be complete, I also lift the base of Q2)

This also is identical to the first circuit I built.
ie, Ratio 17. 1 / 0.6 = 16.6666
I measured it this morning.

Also interesting: for the full circuit, 1V p-p input,
o Output voltage w/C2 connected vs. disconnected? (Tells us how effectively
the drain bootstrap is working.)

w/C2 coonected Vpp = 0.245
w/C2 disconnected Vpp = 0.13


o Output voltage with 100pF temporarily shorting the 0.4pF? (So we can measure
the FET-follower voltage gain.)

I think you want me to put 100pf in parallel with my 0.4pf cap.
When I do that the T1s voltage .98 Vpp
When I remove parallel 100pf, I get 0.25 Vpp.

(T1s seems to be ever so slightly increasing, I'm not sure if that is
the sig/gen or the scope warming up, or some other change.
I'll ignore it for now.)

Thanks, Mikek

Okay, nice work on the measurements.

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain closer
to 1, rather than 0.25.

The gain of 0.25 is not because of the voltage follower, it's from the
series cap. Are you suggesting you can get the amp input capacitance
down to 10x less than the series cap?

Another approach would be to put the series cap outside the box in the
probe tip as is done in o'scopes. Then the gain of the amp would be
sufficient to tie the coax shield to the output of the amp.

What level do you think the amp input capacitance can reach?

Based on your report I think the culprit limiting our improvement is likely
the drain driver, Q1 and related components. If you could 'scope (A) and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.

Cheers,
James Arthur


+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

--

Rick C

 

[Joerg]

On 2017-03-25 10:49, rickman wrote:

On 3/24/2017 4:36 PM, Joerg wrote:
On 2017-03-24 10:59, rickman wrote:
On 3/24/2017 1:14 PM, Joerg wrote:
On 2017-03-20 02:06, Tauno Voipio wrote:

[...]

Please do not use PVC as RF insulation or support pieces.
It is lossy to extremely lossy at RF.


Plasticized PVC ain't stellar but not horrid.

http://g3ynh.info/zdocs/comps/part_6.html

Or do they add a lot of vispipuuro into the mix in Finland?

Are you reading the same page? It says PVC Tanδ is 0.04 - 0.14 at 1
MHz. That is in no way acceptable for the sort of high Q circuits that
are being discussed. That's comparable to wood, 0.059.

This page even lists PVC in the "Lossy" group as defined by Tanδ ≥ 0.01.


The losses in ferrite rods are nothing to sneeze at in comparison. I
have used both in ham radio a lot when I was young. I built kilowatt
level RF power amps, impedance matching boxes and similar gear. Ferrite
rods in transmitters sometimes became so hot that you could barely touch
them while inductors wound on some random piece of PVC pipe remained
cool. I don't remember the PVC type other than that it was remnants they
sold for pennies in the plumbing department so that was very likely well
plasticized.

Maybe I don't recall the context of the ferrite, but the receiving
antennas in much of this thread and described in detail at Kleijer's web
site showed measurable losses with better plastics than PVC. Even
cardboard was a poor support relatively speaking. Seems he would use a
foam type plastic to get some rigidity with a minimum of loss. Foam PVC
was not bad, but his Q measurements increased significantly with foam
polypropylene. Obviously foam is better because there is less actual
material. Solid PVC would show more pronounced losses.

He does some amazingly detailed work. You might find it interesting.
The LC experiments were reported in a series of web pages. Page 10
doesn't have a link to page 11, but otherwise they are all chained.

Looks like there isn't a page 11.

http://www.crystal-radio.eu/enlctest.htm

He gets good Q with PVC. Worse with cardboard but who knows what kind of
stuff he used. It looks like carton material. That is often
post-consumer recycled and you never know what's in it. Water content is
another factor, got to bake it out and then lacquer it immediately
afterwards.

The huge jump in Q in part 3 shows that he used the wrong measurement
setup until then. You can't do this sort of stuff with a 100:1 scope
probe, it has to be at least a high-end FET probe. However, hat off, he
achieved amazing Q values in his experiments.

Look at the Q values he gets in part 7. Those are all regular plumbing
PVC pipes. L16 has lower Q but that PVC tube isn't even particularly
clean. It is important to keep or make them pristine when building
high-Q stuff.

I found it very illuminating his tests with the variable capacitors.
Even parts like bushings were found to make a difference.

I remember that from my ham radio days. Dirty or oxidized rotor contacts
.... phssssst ... PHUT ... *BANG* and worst case that could take the
tubes into the abyss along with it. I also spent part of a Wenol
polishing paste tube keeping the final Pi-filter inductor shiny so it
wouldn't heat up too badly and unsolder a connection to it (which could
also result in a loud bang).

BTW, while such high Q isn't all that useful in practice if you ever
needed that there are simpler means to achieve it. An old method is the
"Q-multiplier". In essence the resonance circuit sits inside an
amplifier stage that is deliberately pulled very close to oscillation.
That shrinks the bandwidth big time.

--
Regards, Joerg

http://www.analogconsultants.com/

 

[amdx]

On 3/25/2017 1:33 PM, dagmargoodboat@yahoo.com wrote:

On Saturday, March 25, 2017 at 10:29:57 AM UTC-4, amdx wrote:
On 3/25/2017 12:13 AM, dagmargoodboat@yahoo.com wrote:
On Friday, March 24, 2017 at 11:46:48 PM UTC-4, amdx wrote:
On 3/24/2017 11:40 AM, dagmargoodboat@yahoo.com wrote:
On Friday, March 24, 2017 at 10:13:18 AM UTC-4, amdx wrote:
On 3/24/2017 8:44 AM, amdx wrote:
On 3/23/2017 11:58 PM, amdx wrote:
On 3/23/2017 6:48 PM, dagmargoodboat@yahoo.com wrote:
On Thursday, March 23, 2017 at 2:52:11 PM UTC-4, amdx wrote:

OK, The circuit is built, I haven't tested it. I'm testing just the
circuit just the circuit posted.
I want to know how to drive the input.
Say I connect a 9 inch piece of RG58, that's 19pf, is that to much?
I want to drive it with a 50 ohm sig/gen.
Can I terminate the RG58 with 50 ohms or do I need to feed the RG58
with
a high impedance? Maybe a series 1Mohm.
I don't see how to separate the sig/gen cable from the input cable.
May I just need to connect it to the LC and see what I get.
Thank, Mikek


How about one step at a time? First, calibrate your input cap:

C1
.---. 0.3pF
| ~ |--------||----+----> A
'---' |
10V --- C2
--- 100pF
|
===

Measure V(A), calculate (don't forget to add your measuring instrument's
c.in to C2).


Easier said than done.
I have a pretty good capacitance tester. I made a .3pf cap :-/
and used an 82pf cap plus the probe, together they measured 106.7pf.
I applied 16Vpp and measured 0.06Vpp at A.
16 / 0.06 = 266.6 The voltage ratio is 266.6.
106.7 / 266.6 = 0.4
That makes my cap look like 0.4pf

Yes.

I measured the capacitance of a 6" x 8" piece (48sq in) of the PCB I
make my caps from, it measured 990 pf. 990 / 48 = 20.63pf per sq in.
My cap is 1/8" diameter or 0.01227 sq in.
So, 1sq in / 0.01227 sq in = 81.53
20.62pf/sq in / 81.53 = 0.253pf per 1/8" dia.
So fringing and lead to lead capacitance, I'm in the ballpark, just not
sure what field.
If you have any suggestions, I'm willing.
Way past my bedtime, good night.
Mikek

You're close enough for our purposes. The first measurement may be being
affected by stray capacitances, dunno. In the end all we'll care about
(when you use this thing) is getting 100mV out for 100mV in.

PS. PCB is Rogers Duroid 5880, I could not find a capacitance per sq in
number online.


I got out my better probe (Tek 6122 11pf) and made curly cues for the
probe tip.
Here's a picture of the tightened up circuit.
https://www.dropbox.com/s/sixe91dvxs40f1z/P1010193.JPG?dl=0


The new total capacitance, probe plus 82pf cap is 94pf. (measured)
16Vpp input, V(A) is 0.068Vpp. Then 16/0.068 = 235.3
235.3 / 94 = 2.5pf.

No, that should be 94pF * (.068V / 16V) = 0.4 pF.

This is a tighter measurement and also agrees with my calculation of
0.253 that I calculated for the pcb capacitance per sq in.
Rather than fight with a new piece of pcb material just slightly over
1/8" in diameter.
I'm going to go with a new value input cap of 0.25pf.
If you see this as a bad idea, let me know and I'll fumble around
and make a larger cap. I used a 1/8' paper punch to make this cap,
I could use a 1/4" and grind it down, but it's a chore.

It looks to me like your input cap is good, and 0.4pF. No need to change
it.

Now for the next measurement.
Is this with a T1 in the completed circuit?
Or, do I add the probe + 82pf back in to get a ratio?
Ya, I'm now confused.

Yes, it's time to use your calibrated in the completed, working circuit.
We'll want to measure the a.c. signal at the FET source first, then the
FET's drain.

I think we might want the voltage at the gate, but I
think that would be an impossible measurement. ie putting 11pf
probe across a 0.3pf cap ?
Thanks, Mikek



Ok, I tried this, out of circuit, just the FET.



T1 |--
.---. 0.3pF |
| ~ |--------||------>|--.
'---' | T1.s = 0.06Vpp
16Vpp |
|---probe tip
82pf –--
--- 12pf
|---probe ground
|
---
///

So, 16Vpp / 0.6Vpp = 266.6
266.6 / 94 = 2.84pf
2.84pf - 0.3pf = 2.54pf for the Cgs.
I think that's to low, I thought it was 5pf,
but I'm looking for your thoughts.

Thanks Mikek

Sorry, I wasn't clear enough. We're measuring in-circuit.

To check the effective input capacitance we want to know the attenuation
of the input a.c. voltage, as measured at point (A).

Rationale: The input voltage will be divided across your 0.40pF input
capacitor and the circuit's input capacitance:

Ccoupling
0.40pF .. .. ..
Vin >---||----:--+--o V(A)
: |
: --- Ceff
: ---
: |
: ===
'.. .. ..

We can't measure at the gate--that's too high impedance--so we're measuring
at FET T1's source terminal, point (A).

If 63% of the voltage is lost across the 0.40pF series cap, then the
effective input capacitance is dropping 37%, which means the series
cap is 63% of the total reactance and Ceff is 37% of the total reactance.

More formally, Ceff = 0.40pF * (Vin - V(A)) / V(A).

(We're trying to measure Ceff.)

You can measure Vac at the FET source on the original circuit for comparison,
too. That's a good way to assess the relative performance of the new circuit
vs. the original (after correcting for any differences in the input coupling
caps, of course.)

(Kleijer said his input division ratio was 17:1, indicating his circuit's
input capacitance (Ceff) was 16 units, and his coupling cap was 1 unit of
capacitance, or Ceff = 16 x Ccoupling.)

We want to know the a.c. voltage @ (B) just to gauge how well our drain
portion of the bootstrap is working. The closer Vac(B) is to Vac(A), the
better we're doing. 1:1 would be ideal. 0.9 would be okay, less than 0.9
indicates a problem.


AC Voltage at B just slightly less than AC Voltage at A. I had to look
several times just to make sure there was a difference.

Okay, that means the drain bootstrap should be working.

Here's the circuit, for reference, with probe points (A) and (B) marked:

(Remember, I'm showing a boot-strapped coax on the input, but we're not
using that yet. We want to keep the comparisons apples-to-apples, and
change only one thing at a time.)


Preliminary data.
I have just a wire input like the original.
I have added my measured DC voltages to the schematic below.
With 1Vpp input, T1s has 0.25Vpp and the output is 0.24Vpp.
So, I think that means the bootstrap is working, I don't know how much
change to expect, but that seems good.

The voltage gain, input-to-output, is about 1/4, or roughly 4x better than
Kleijer's 1/17. But I was expecting to do a little better still and would
like to figure out what's up, if you're game for poking and prodding this
thing a bit more.

Your measurements mean we're dropping .75V across the 0.4pF coupling cap,
and 0.25V across our input capacitance at the FET gate. Our input
capacitance is thus about three times the coupling cap, or about 1.2pF.


In the morning I will compare this to the original.
Thanks, Mikek

You can compare to the original easily--temporarily connect R3 to ground
instead of to R4. That disables all of the bootstrapping, which makes the
new circuit operate as the original. Measure, and compare output voltage
without bootstrapping to the value obtained with bootstrapping. The
ratio tells us how much better we're doing than the original.

Everything checks as expect.
With bootstraping T1s is 0.225 Ratio about 4
When R3 is grounded T1s is 0.60 Ratio about 17
(Just to be complete, I also lift the base of Q2)

This also is identical to the first circuit I built.
ie, Ratio 17. 1 / 0.6 = 16.6666
I measured it this morning.

Also interesting: for the full circuit, 1V p-p input,
o Output voltage w/C2 connected vs. disconnected? (Tells us how effectively
the drain bootstrap is working.)

w/C2 coonected Vpp = 0.245
w/C2 disconnected Vpp = 0.13


o Output voltage with 100pF temporarily shorting the 0.4pF? (So we can measure
the FET-follower voltage gain.)

I think you want me to put 100pf in parallel with my 0.4pf cap.
When I do that the T1s voltage .98 Vpp
When I remove parallel 100pf, I get 0.25 Vpp.

(T1s seems to be ever so slightly increasing, I'm not sure if that is
the sig/gen or the scope warming up, or some other change.
I'll ignore it for now.)

Thanks, Mikek

Okay, nice work on the measurements.

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain closer
to 1, rather than 0.25.

Based on your report I think the culprit limiting our improvement is likely
the drain driver, Q1 and related components. If you could 'scope (A) and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.

I adjusted both probes to be flat, then I connected both probes to the
same point (T1d) and adjusted *Var. for exact overlay of the two
waveforms. Then I move one probe to (T1s), I can not see any difference
in the signals.They still overlay, maybe a little thicker trace. I have
the brightness turned way down to make a thin trace, focus adjusted.
Looking for any small difference.

I'm a little confused by my scope. There is not a clear subtract (A-B).

The manual says ADD then INVERT to get a difference**.
See page 3-5 in the manual.
> https://www.scribd.com/document/25186404/Tektronix-2465B-Options-and-Operator-Manual
Following the method in the manual, I get about 4mV of difference signal.
If this is not satisfactory, I have another Tektronix scope (475), I
can put on the bench. I think that has subtract.

* this was a very minor tweak
** this seems a poor method if your wave form is slightly asymmetrical.
But I don't see that.
Mikek

Cheers,
James Arthur


+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

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

 

On Saturday, March 25, 2017 at 7:18:00 PM UTC-4, rickman wrote:

On 3/25/2017 6:25 PM, dagmargoodboat@yahoo.com wrote:
On Saturday, March 25, 2017 at 3:20:31 PM UTC-4, rickman wrote:
On 3/25/2017 2:33 PM, dagmargoodboat@yahoo.com wrote:

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain closer
to 1, rather than 0.25.

The gain of 0.25 is not because of the voltage follower, it's from the
series cap.

The net gain's a result of the *divider* formed by the series coupling cap
and (the FET input capacitance + strays), multiplied by the follower gain.
I asked Mike to measure the follower gain so we could then calculate the
input capacitance.

Yes, of course, but the lost voltage is across the input cap. The
voltage follower gain will be nearly 1 in all variations. The
capacitance of the amp input is a function of it's intrinsic capacitance
and the gain. Closer to 1 the gain is the smaller the bootstrap makes
the effective input capacitance.


Are you suggesting you can get the amp input capacitance
down to 10x less than the series cap?

Bootstrapping should reduce the FET input capacitance by a factor of
1/(1-G), where G = the gain of the bootstrap.

That's not what I asked. What is important in determining the circuit
gain is the amp's effective capacitance value compared to the input cap.
To get the gain of the input amp with input cap close to 1, the
effective input capacitance of the amp would need to be ball park 10
times lower than the input cap.

If the gain of the input amp were made more than 1, (to compensate for
the lost gain in the cap divider) the circuit could easily become
unstable. A follower won't have gain of 1 or more, only approach 1.

I don't know how close you can get to a gain of 1 for the amp and input
cap, but I think it will be hard to get the amp effective input
capacitance anywhere near 0.04 pf. If you can do that, you really don't
need the input cap. I believe it's purpose was to mitigate the input
capacitance of the amp so it didn't impact the circuit being measured.
If you can get the amp effective input capacitance below 0.4 pF, you can
do away with the input cap and it's voltage division.

Remember what the input cap is for. That's why I suggested it might be
useful to put it in the probe end. Using a coax with the shield as
guard, there should be little problem with noise. Heck, you are
measuring antennas/coils. Kleijer found a radio station in his
measurements because the coil was picking it up.


Another approach would be to put the series cap outside the box in the
probe tip as is done in o'scopes. Then the gain of the amp would be
sufficient to tie the coax shield to the output of the amp.

That's true. Mike could use a shielded connector and cable if he didn't
mind putting his series cap at the probe.

What level do you think the amp input capacitance can reach?

Down to the strays, if we want, or even less. See above. This circuit
should be under 1/2 pF, by my estimate.

Cheers,
James Arthur

Based on your report I think the culprit limiting our improvement is likely
the drain driver, Q1 and related components. If you could 'scope (A) and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.

Cheers,
James Arthur

+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===



--

Rick C

I think you might want to re-read the thread. With a low Cin(eff),
net gain should be close to unity, *including* the coupling cap.

Cheers,
James Arthur

 

On Friday, 24 March 2017 23:49:21 UTC, Cursitor Doom wrote:

On Fri, 24 Mar 2017 04:58:23 -0700, tabbypurr wrote:

I'm all for sensible discussion but
once k & CD resort to stupid bs I don't see the point.

I haven't kept up with this thread and I don't know exactly what krw has
said, but I personally do maintain there is a perfectly solid case for
the death penalty when:

a) The crime is especially reprehensible
b) The identity of the perpetrator is beyond all question (not merely
"beyond reasonable doubt")
c) The mental capacity of the perpetrator was not impaired at the time

If that's what you call "stupid bs" then there's really nothing more I
can be bothered to say on the subject.

no it's something entirely different I called stupid bs. The above is fair enough until you look much closer at the issues in the real life process. It's riddled with them.


NT

 

On Saturday, 25 March 2017 13:32:03 UTC, Cursitor Doom wrote:

> The main problem with Kev is that he just enjoys a good argument. Kev

no, he enjoys a stupid one.

will argue the hind leg off a donkey just for the sake of it. You will
never get the better of him; he will just wear down any interlocutor by
attrition until they give up from exhaustion. It's what he does. If I may
quote from an early 80s movie, "The only way to win [against Kev] is not
to play".
HTH.

 

On Saturday, March 25, 2017 at 7:27:57 PM UTC-4, amdx wrote:

On 3/25/2017 6:00 PM, dagmargoodboat@yahoo.com wrote:
On Saturday, March 25, 2017 at 3:44:39 PM UTC-4, amdx wrote:
On 3/25/2017 1:33 PM, dagmargoodboat@yahoo.com wrote:

Based on your report I think the culprit limiting our improvement is likely
the drain driver, Q1 and related components. If you could 'scope (A) and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.


I adjusted both probes to be flat, then I connected both probes to the
same point (T1d) and adjusted *Var. for exact overlay of the two
waveforms. Then I move one probe to (T1s), I can not see any difference
in the signals.They still overlay, maybe a little thicker trace. I have
the brightness turned way down to make a thin trace, focus adjusted.
Looking for any small difference.

I'm a little confused by my scope. There is not a clear subtract (A-B).

The manual says ADD then INVERT to get a difference**.
See page 3-5 in the manual.
https://www.scribd.com/document/25186404/Tektronix-2465B-Options-and-Operator-Manual
Following the method in the manual, I get about 4mV of difference signal.
If this is not satisfactory, I have another Tektronix scope (475), I
can put on the bench. I think that has subtract.

* this was a very minor tweak
** this seems a poor method if your wave form is slightly asymmetrical.
But I don't see that.
Mikek

That's good technique, but the result is a little puzzling--with
bootstrapping gain this close to unity we'd expect better cancellation
of the FET capacitances.

The FET-follower into a current-sink load has a gain of 0.98 (measured).
That should cancel 98% of Cgs, the largest FET capacitance (5pF).

The drain-follower circuit also has a gain of roughly
(250mV-4mV)/250mV = .98, which, following the .98 gain of the FET stage,
produces a drain bootstrap gain of 0.96. That should cancel 96% of the
BF256C's Cdg (1.2pF), the remaining input-related capacitance.

Either there's close to 1pF of stray capacitance to ground at the input
node, or ... I've missed something. Like, what's your test frequency?
I've been assuming 1MHz (crystal-radio band).

Cheers,
James Arthur


Yes, 1 MHZ.
I've checked my circuit layout many times, But still, any
thing that could be wrong to cause this?
I'll go check resistor values while waiting for a response,
maybe I got two of them reversed, I don't expect it though.
Thanks, Mikek

From the d.c. voltages you posted, the resistors seem fine. Maybe a photo
of the layout would clear things up? We can take this to email rather than
clutter s.e.d. if you'd like--my email address is good.

Rick's right about the shield-driving scheme--it'll never fly as drawn.

Solutions: move the 0.4pF to the probe tip, or, at some point, you could
use a much-larger series cap., e.g. 10pF.

Kleijer's 0.3pF input cap was to reduce his FET's 5pF loading to 0.3pF.

But if we ever get our FET down to 0.3pF, the series cap's absolute value
becomes less important. 10pF coupling in series with 0.3pF at the FET, for
example, is still 0.3pF loading at the input, but would make the shield
drive ~= Vg(T1) ~= Vin, allowing us to cancel the coax capacitance.


---- 10pF Vg(T1)
Vin >---------||---.
--+- | (input C, post-bootstrapping)
| --- 0.3pF Ceff
| ---
| |
| ===
|
'----< shield drive

Cheers,
James Arthur

 

[rickman]

On 3/25/2017 3:44 PM, amdx wrote:

On 3/25/2017 1:33 PM, dagmargoodboat@yahoo.com wrote:
On Saturday, March 25, 2017 at 10:29:57 AM UTC-4, amdx wrote:
On 3/25/2017 12:13 AM, dagmargoodboat@yahoo.com wrote:

+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

You can compare to the original easily--temporarily connect R3 to
ground
instead of to R4. That disables all of the bootstrapping, which
makes the
new circuit operate as the original. Measure, and compare output
voltage
without bootstrapping to the value obtained with bootstrapping. The
ratio tells us how much better we're doing than the original.

Everything checks as expect.
With bootstraping T1s is 0.225 Ratio about 4
When R3 is grounded T1s is 0.60 Ratio about 17
(Just to be complete, I also lift the base of Q2)

This also is identical to the first circuit I built.
ie, Ratio 17. 1 / 0.6 = 16.6666
I measured it this morning.

Also interesting: for the full circuit, 1V p-p input,
o Output voltage w/C2 connected vs. disconnected? (Tells us how
effectively
the drain bootstrap is working.)

w/C2 coonected Vpp = 0.245
w/C2 disconnected Vpp = 0.13


o Output voltage with 100pF temporarily shorting the 0.4pF? (So we
can measure
the FET-follower voltage gain.)

I think you want me to put 100pf in parallel with my 0.4pf cap.
When I do that the T1s voltage .98 Vpp
When I remove parallel 100pf, I get 0.25 Vpp.

(T1s seems to be ever so slightly increasing, I'm not sure if that is
the sig/gen or the scope warming up, or some other change.
I'll ignore it for now.)

Thanks, Mikek

Okay, nice work on the measurements.

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain
closer
to 1, rather than 0.25.

Based on your report I think the culprit limiting our improvement is
likely
the drain driver, Q1 and related components. If you could 'scope (A) and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.


I adjusted both probes to be flat, then I connected both probes to the
same point (T1d) and adjusted *Var. for exact overlay of the two
waveforms. Then I move one probe to (T1s), I can not see any difference
in the signals.They still overlay, maybe a little thicker trace. I have
the brightness turned way down to make a thin trace, focus adjusted.
Looking for any small difference.

I'm a little confused by my scope. There is not a clear subtract (A-B).

The manual says ADD then INVERT to get a difference**.
See page 3-5 in the manual.
https://www.scribd.com/document/25186404/Tektronix-2465B-Options-and-Operator-Manual

Following the method in the manual, I get about 4mV of difference signal.
If this is not satisfactory, I have another Tektronix scope (475), I
can put on the bench. I think that has subtract.

* this was a very minor tweak
** this seems a poor method if your wave form is slightly asymmetrical.
But I don't see that.

ADD with INVERT should be exactly the same as subtract. I'm not sure
what you mean about the asymmetry.

4 mV sounds good, but what it the context? What is the amplitude of the
signal at this point? Is this the 0.25 Vpp measurement?


--

Rick C

 

[John Robertson]

On 2017/03/25 12:32 AM, Michael A. Terrell wrote:

Kevin Aylward wrote:
wrote in message news:mnq8dcpnbekopn07j0cu3crq12hq0905l1@4ax.com...

On Thu, 23 Mar 2017 14:58:23 -0400, "Michael A. Terrell"
mike.terrell@earthlink.net> wrote:

krw@notreal.com wrote:
On Sun, 19 Mar 2017 14:19:57 -0000 (UTC), Cursitor Doom
curd@notformail.com> wrote:

On Sat, 18 Mar 2017 20:03:50 -0400, krw wrote:

That's impossible, which is the whole "reasonable" thing.

Nothing whatsoever impossible about it!

Not at all. There is always doubt. It may be an unreasonable doubt
but there is _always_ doubt (what if the Earth was really flat?).
Hence, "beyond reasonable doubt".


What do you think of this sorry piece of work?

http://www.wftv.com/news/local/ayala-to-explain-why-she-wont-seek-death-penalty-against-murder-suspect-markeith-loyd/503151996



She should *immediately* be removed from office for violating her oath
of office to follow the constitution and laws of the state of Florida.
That decision isn't hers to make.

Ho hummm..

Like, what specific law says that anybody has to demand the death of
someone else? Dah....


It's her job tpoo follow the laws, not decide which ones she likes.
In that case, he murdered a police officer in front of a lot of people,
and he has a long history of violence, If she can't or won't do her job,
she needs to be fired.

She only said she wouldn't execute the man, not that she was going to
release him in a school yard!

Yes, the man sounds like a total waste of space, but the state doesn't
have to compound the issue by sinking to his level and dragging the jury
and judge down with him.

Isn't it better to take the high road about murder rather than the
vigilante route? And we know how vigilantes end up - the KKK used to be
full of them.

John

 

On Sunday, March 26, 2017 at 6:40:17 AM UTC+11, Cursitor Doom wrote:

On Sat, 25 Mar 2017 18:05:27 +0000, Kevin Aylward wrote:

I may not always agree with Bill, but I have never seen any evidence
that Bill lies.

boggle

Cursitor Doom isn't a far gone as krw, but he does believe what he reads in the Express, so his grasp of what constitutes truth and what might be falsehood isn't great.

--
Bill Sloman,

 

[amdx]

On 3/25/2017 3:34 PM, rickman wrote:

On 3/25/2017 3:44 PM, amdx wrote:
On 3/25/2017 1:33 PM, dagmargoodboat@yahoo.com wrote:
On Saturday, March 25, 2017 at 10:29:57 AM UTC-4, amdx wrote:
On 3/25/2017 12:13 AM, dagmargoodboat@yahoo.com wrote:

+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

You can compare to the original easily--temporarily connect R3 to
ground
instead of to R4. That disables all of the bootstrapping, which
makes the
new circuit operate as the original. Measure, and compare output
voltage
without bootstrapping to the value obtained with bootstrapping. The
ratio tells us how much better we're doing than the original.

Everything checks as expect.
With bootstraping T1s is 0.225 Ratio about 4
When R3 is grounded T1s is 0.60 Ratio about 17
(Just to be complete, I also lift the base of Q2)

This also is identical to the first circuit I built.
ie, Ratio 17. 1 / 0.6 = 16.6666
I measured it this morning.

Also interesting: for the full circuit, 1V p-p input,
o Output voltage w/C2 connected vs. disconnected? (Tells us how
effectively
the drain bootstrap is working.)

w/C2 coonected Vpp = 0.245
w/C2 disconnected Vpp = 0.13


o Output voltage with 100pF temporarily shorting the 0.4pF? (So we
can measure
the FET-follower voltage gain.)

I think you want me to put 100pf in parallel with my 0.4pf cap.
When I do that the T1s voltage .98 Vpp
When I remove parallel 100pf, I get 0.25 Vpp.

(T1s seems to be ever so slightly increasing, I'm not sure if that is
the sig/gen or the scope warming up, or some other change.
I'll ignore it for now.)

Thanks, Mikek

Okay, nice work on the measurements.

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain
closer
to 1, rather than 0.25.

Based on your report I think the culprit limiting our improvement is
likely
the drain driver, Q1 and related components. If you could 'scope (A)
and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it myself...

Nice work Mike. You've made something that's 4x better, even as it is.


I adjusted both probes to be flat, then I connected both probes to the
same point (T1d) and adjusted *Var. for exact overlay of the two
waveforms. Then I move one probe to (T1s), I can not see any difference
in the signals.They still overlay, maybe a little thicker trace. I have
the brightness turned way down to make a thin trace, focus adjusted.
Looking for any small difference.

I'm a little confused by my scope. There is not a clear subtract (A-B).

The manual says ADD then INVERT to get a difference**.
See page 3-5 in the manual.
https://www.scribd.com/document/25186404/Tektronix-2465B-Options-and-Operator-Manual


Following the method in the manual, I get about 4mV of difference
signal.
If this is not satisfactory, I have another Tektronix scope (475), I
can put on the bench. I think that has subtract.

* this was a very minor tweak
** this seems a poor method if your wave form is slightly asymmetrical.
But I don't see that.

ADD with INVERT should be exactly the same as subtract.

I'm not sure what you mean about the asymmetry.
After further review, that is nonsense.

4 mV sounds good, but what it the context? What is the amplitude of the
signal at this point? Is this the 0.25 Vpp measurement?

Yes, still the 0.25Vpp measurement.
Drain and Source AC voltage are very close.
I'm using a magnifier to look at a signal that is about one small division.
I just made a second measurement, it is about 2.5mvpp.
Looking at the circuit, I think I might get by use a 1X scope probes.
That would get me a factor of ten increase in size of my difference
waveform. thoughts?

Mikek



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

 

[rickman]

On 3/25/2017 5:12 PM, amdx wrote:

On 3/25/2017 3:34 PM, rickman wrote:
On 3/25/2017 3:44 PM, amdx wrote:
On 3/25/2017 1:33 PM, dagmargoodboat@yahoo.com wrote:
On Saturday, March 25, 2017 at 10:29:57 AM UTC-4, amdx wrote:
On 3/25/2017 12:13 AM, dagmargoodboat@yahoo.com wrote:

+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

You can compare to the original easily--temporarily connect R3 to
ground
instead of to R4. That disables all of the bootstrapping, which
makes the
new circuit operate as the original. Measure, and compare output
voltage
without bootstrapping to the value obtained with bootstrapping. The
ratio tells us how much better we're doing than the original.

Everything checks as expect.
With bootstraping T1s is 0.225 Ratio about 4
When R3 is grounded T1s is 0.60 Ratio about 17
(Just to be complete, I also lift the base of Q2)

This also is identical to the first circuit I built.
ie, Ratio 17. 1 / 0.6 = 16.6666
I measured it this morning.

Also interesting: for the full circuit, 1V p-p input,
o Output voltage w/C2 connected vs. disconnected? (Tells us how
effectively
the drain bootstrap is working.)

w/C2 coonected Vpp = 0.245
w/C2 disconnected Vpp = 0.13


o Output voltage with 100pF temporarily shorting the 0.4pF? (So we
can measure
the FET-follower voltage gain.)

I think you want me to put 100pf in parallel with my 0.4pf cap.
When I do that the T1s voltage .98 Vpp
When I remove parallel 100pf, I get 0.25 Vpp.

(T1s seems to be ever so slightly increasing, I'm not sure if that is
the sig/gen or the scope warming up, or some other change.
I'll ignore it for now.)

Thanks, Mikek

Okay, nice work on the measurements.

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain
closer
to 1, rather than 0.25.

Based on your report I think the culprit limiting our improvement is
likely
the drain driver, Q1 and related components. If you could 'scope (A)
and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it
myself...

Nice work Mike. You've made something that's 4x better, even as it is.


I adjusted both probes to be flat, then I connected both probes to the
same point (T1d) and adjusted *Var. for exact overlay of the two
waveforms. Then I move one probe to (T1s), I can not see any difference
in the signals.They still overlay, maybe a little thicker trace. I have
the brightness turned way down to make a thin trace, focus adjusted.
Looking for any small difference.

I'm a little confused by my scope. There is not a clear subtract (A-B).

The manual says ADD then INVERT to get a difference**.
See page 3-5 in the manual.
https://www.scribd.com/document/25186404/Tektronix-2465B-Options-and-Operator-Manual



Following the method in the manual, I get about 4mV of difference
signal.
If this is not satisfactory, I have another Tektronix scope (475), I
can put on the bench. I think that has subtract.

* this was a very minor tweak
** this seems a poor method if your wave form is slightly asymmetrical.
But I don't see that.

ADD with INVERT should be exactly the same as subtract.

I'm not sure what you mean about the asymmetry.
After further review, that is nonsense.


4 mV sounds good, but what it the context? What is the amplitude of the
signal at this point? Is this the 0.25 Vpp measurement?

Yes, still the 0.25Vpp measurement.
Drain and Source AC voltage are very close.
I'm using a magnifier to look at a signal that is about one small division.
I just made a second measurement, it is about 2.5mvpp.
Looking at the circuit, I think I might get by use a 1X scope probes.
That would get me a factor of ten increase in size of my difference
waveform. thoughts?

Yeah, it will get you 10x, but add a few pF to the circuit. Certainly
it's worth giving a try, but I think knowing 2.5 mV is good enough.
That's 100:1 which is about what I would expect. Maybe drive a larger
signal to the input? I guess you don't want too large of a signal.

I don't really have a feel for how low this voltage can go. It would be
driven by the gain of the FET, but with the added drive of the bipolar
devices the voltage can go even lower. Right now I would say 100:1 is
pretty good as a goal. It means the effective capacitance of the input
is 100 times lower. But your capacitance measurements don't agree with
that.

--

Rick C

 

[amdx]

On 3/25/2017 4:44 PM, rickman wrote:

On 3/25/2017 5:12 PM, amdx wrote:
On 3/25/2017 3:34 PM, rickman wrote:
On 3/25/2017 3:44 PM, amdx wrote:
On 3/25/2017 1:33 PM, dagmargoodboat@yahoo.com wrote:
On Saturday, March 25, 2017 at 10:29:57 AM UTC-4, amdx wrote:
On 3/25/2017 12:13 AM, dagmargoodboat@yahoo.com wrote:

+12V +12V
-+- -+-
| |
| [22k] R5
Q1 \| | 8.0v
BC547B |---+-------.
7.4v .<| | | input 1Vpp
| | | T1s 0.25Vpp
(B)| [47k] R6 | output 0.24Vpp
(shield) T1 |--' | |
----- BF256C | === |
----------||-+----->|--. |
---+- .4pF| (A)| Vdd --- C2
| | | -+- ---100n
| | | | |
| R1 [10M] | |/ Q2 |
| | 3.4v +---| BC547B |
| | | |>. |
| | R3 [470] |2.7v |
| | | | | C3
| | | | (C)| 100n
| +----||---+-----+-------+-----||---> to ampl.
| | C1 | |
| R2 [10M] 100pF R4 [470] --- C4
| | | --- 100n
| === === |
| |
'------------------------------+
|
Cin ~200fF [2.2k] R7
|
===

You can compare to the original easily--temporarily connect R3 to
ground
instead of to R4. That disables all of the bootstrapping, which
makes the
new circuit operate as the original. Measure, and compare output
voltage
without bootstrapping to the value obtained with bootstrapping. The
ratio tells us how much better we're doing than the original.

Everything checks as expect.
With bootstraping T1s is 0.225 Ratio about 4
When R3 is grounded T1s is 0.60 Ratio about 17
(Just to be complete, I also lift the base of Q2)

This also is identical to the first circuit I built.
ie, Ratio 17. 1 / 0.6 = 16.6666
I measured it this morning.

Also interesting: for the full circuit, 1V p-p input,
o Output voltage w/C2 connected vs. disconnected? (Tells us how
effectively
the drain bootstrap is working.)

w/C2 coonected Vpp = 0.245
w/C2 disconnected Vpp = 0.13


o Output voltage with 100pF temporarily shorting the 0.4pF? (So we
can measure
the FET-follower voltage gain.)

I think you want me to put 100pf in parallel with my 0.4pf cap.
When I do that the T1s voltage .98 Vpp
When I remove parallel 100pf, I get 0.25 Vpp.

(T1s seems to be ever so slightly increasing, I'm not sure if that is
the sig/gen or the scope warming up, or some other change.
I'll ignore it for now.)

Thanks, Mikek

Okay, nice work on the measurements.

Summary:
o The FET-as-voltage follower has a gain of 0.98, which is quite
decent for our purposes.
o Input capacitance is 1.2pF.
o We've reduced the input capacitance by a factor of about 4 compared
to Kleijer, but it's still not as low as expected.

The shield driver + BNC isn't going to work, not until we get the gain
closer
to 1, rather than 0.25.

Based on your report I think the culprit limiting our improvement is
likely
the drain driver, Q1 and related components. If you could 'scope (A)
and
(B), a.c.-coupled, with the 'scope set to subtract the two channels,
(A)-(B) would tell us if our drain bootstrap is up to snuff.

I might have to cobble one of these together so I can probe it
myself...

Nice work Mike. You've made something that's 4x better, even as it
is.


I adjusted both probes to be flat, then I connected both probes to the
same point (T1d) and adjusted *Var. for exact overlay of the two
waveforms. Then I move one probe to (T1s), I can not see any difference
in the signals.They still overlay, maybe a little thicker trace. I have
the brightness turned way down to make a thin trace, focus adjusted.
Looking for any small difference.

I'm a little confused by my scope. There is not a clear subtract
(A-B).

The manual says ADD then INVERT to get a difference**.
See page 3-5 in the manual.
https://www.scribd.com/document/25186404/Tektronix-2465B-Options-and-Operator-Manual




Following the method in the manual, I get about 4mV of difference
signal.
If this is not satisfactory, I have another Tektronix scope (475), I
can put on the bench. I think that has subtract.

* this was a very minor tweak
** this seems a poor method if your wave form is slightly asymmetrical.
But I don't see that.

ADD with INVERT should be exactly the same as subtract.

I'm not sure what you mean about the asymmetry.
After further review, that is nonsense.


4 mV sounds good, but what it the context? What is the amplitude of the
signal at this point? Is this the 0.25 Vpp measurement?

Yes, still the 0.25Vpp measurement.
Drain and Source AC voltage are very close.
I'm using a magnifier to look at a signal that is about one small
division.
I just made a second measurement, it is about 2.5mvpp.
Looking at the circuit, I think I might get by use a 1X scope probes.
That would get me a factor of ten increase in size of my difference
waveform. thoughts?

Yeah, it will get you 10x, but add a few pF to the circuit. Certainly
it's worth giving a try, but I think knowing 2.5 mV is good enough.
That's 100:1 which is about what I would expect. Maybe drive a larger
signal to the input? I guess you don't want too large of a signal.

I don't really have a feel for how low this voltage can go. It would be
driven by the gain of the FET, but with the added drive of the bipolar
devices the voltage can go even lower. Right now I would say 100:1 is
pretty good as a goal. It means the effective capacitance of the input
is 100 times lower. But your capacitance measurements don't agree with
that.

I increased the input to 5Vpp this gave me about 1.236Vpp on the
Drain and 1.25Vpp on the Source. For a difference of 14mVpp.

Mikek

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