RE: differential probes

["Tesla list" <tesla-at-pupman-dot-com>]

Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>

Hi Jimmy,

I worked more on compensation tonight.  I am getting real close.  Here is 
the Tek probe (blue) and our probe (yellow):

http://hot-streamer-dot-com/temp/03072701.gif

There is a little bump in the beginning and a "birdie" at 33MHz...  But 
darn close for a pure resistive input probe ;-))  I have 2.2K resistors in 
the feedback so my gain is 10% high.  I'll try and find a trim cap to 
provide a proper compensation adjustment.  Don't want to get too carried 
away till I get the "real" resistors.

I grounded Vocm but the darn offset is still very sensitive to supply 
voltage.  Probably have to bite it there and use +-5V voltage 
regulators...  A 1Meg trim pot and 100K resistor going to Vocm provide a 
very nice offset adjustment.

Here is the exciting test setup:

http://hot-streamer-dot-com/temp/P7270012.jpg

Here is what the board looks like these days ;-))

http://hot-streamer-dot-com/temp/P7270014.jpg

More below...

At 07:05 PM 7/27/2003 -0700, you wrote:
>Hi Terry,
>
>I am using the neutral for the MOV ground right now,
>is there anything wrong with that?

I the MOVs ever were to fire, they may have to dump a lot of current into 
the ground with a lot of voltage behind it.  I was going to dump that back 
into the coil's ground system rather than the scope's ground.  Thus the 
separate grounding...


>more questions/comments below
>
>--- Tesla list <tesla-at-pupman-dot-com> wrote:
> > Original poster: "Terry Fritz" <teslalist-at-qwest-dot-net>
> >
> > Hi Jimmy,
> >
> > I got the boards today and was working on them
> > tonight.  I had to chop
> > things up a "little" :o)) but I think they will work
> > very well.  Here are
> > the changes the boards could use so far.  I had
> > better write them down
> > before I forget ;-)
> >
> > 1.  Lets take the MOVs off the board.  It is best if
> > the board just has
> > solder eyes for the HV wires.  We can just solder
> > the MOVs to the fuse
> > holders and ground terminals rather than clog the
> > board with them.
> >
> > 2.  The 1K feed back resistors need to be 2K.
>
>I noticed today that it is a 2000x probe. If I want to
>make my other channel a 10x probe, then I would change
>it to 200k right?

I think 20K would be 10X and 200K would be 100X.  Right now it is 2000K of 
combined input resistors with 2K of feed back resistors for 1000X.  Be 
careful of power dissipation in the input resistors and all that.

> >
> > 3.  There is some offset and I think we can use the
> > Vocm pin to fix that.
> >
>
>Grounding that pin worked perfectly for me.

I still found that it was super sensitive to supply voltage.  Regulated 
supplies and an adjustment seem to be needed.


> > 4.  We need the expensive 10nF caps after all.
> >
> > 5.  The 2K resistors need ~~50pF of compensation.
> >
> > 6.  To go to a BNC cable, we need to put 50 ohm in
> > series with the output
> > to match the cable.
> >
> > 7.  The feedback path was very noisy,  I had to chop
> > it all out.  Need to
> > get ride of the ground plane near the input and
> > reposition the 2K resistors
> > right on the input.
>
>Can you explain this one? Does it have something to do
>with capacitive coupling to the ground plane?

Ohhh!! Yeahhh!!!  We need to eliminate as much input "trace" as 
possible.  Move the components to almost a star junction.  Possibly mount 
them on their sides to really pack them in...  Capacitance to ground and 
anything else of the op-amp input node is "bad".


> > 8.  I replaced the 1 Meg 1/2 resistors with low
> > voltage surface mount
> > ones.  The big resistors were very noisy.  Hard to
> > wait for DigiKey to get
> > the right ones...
>
>What makes the resistors noisy?

The discrete resistors have a spiral element (inductor) and they are "big" 
so stray capacitance goofs with them.  They have all kinds of 
resonances.  Looks like a mini secondary coil at high frequency.  Surface 
mount types eliminate all that.

Cheers,

         Terry