signal pickoff

The waveform at B is good if C4 is 50 pF.

C is fine, but it's only 120 millivolts.

I just have no idea if C1 is really 0.1 pF.

e:

te:

nsen

n Larkin:

d like to

rs/Pickoff.JPG

, roughly

have to

2 can be

pick the

dd

e bottom

cut out

ld be

o move the

er, lest the

C

TI

eds

at

00nS,

ant is

ived,

e

).

e

B).

r,

he 1G

oss

d

er

Right, obviously. That's why I put in V(C)--to illustrate the perfectly-siz ed compensating cap.

My point in posting this was to show that my scheme settles 20x faster. But your original values and a 50pF trimcap might be the ideal solution.

An approximate compensation cap AND lower equivalent resistance divider mig ht be an alternative. It'll over- or undershoot if there's a capacitance mismatch, but the transient settles out much faster.

I'd hope it's 1/5th that or less, but it's not going to be nil; compensatio n will be needed one way or the other.

Cheers, James Arthur

Effect of Gigohm and 1/10th environmental stray C

RL

Version 4 SHEET 1 1380 680 WIRE 816 -80 272 -80 WIRE 16 0 -48 0 WIRE 96 0 16 0 WIRE 192 0 96 0 WIRE 272 0 272 -80 WIRE 272 0 192 0 WIRE 384 0 272 0 WIRE 416 0 384 0 WIRE 512 0 496 0 WIRE 528 0 512 0 WIRE 656 0 608 0 WIRE 720 0 656 0 WIRE 816 0 816 -80 WIRE 880 0 816 0 WIRE 912 0 880 0 WIRE 1008 0 992 0 WIRE 1024 0 1008 0 WIRE 1152 0 1104 0 WIRE 1216 0 1152 0 WIRE 96 16 96 0 WIRE 272 16 272 0 WIRE 720 16 720 0 WIRE 1216 16 1216 0 WIRE 16 32 16 0 WIRE 192 32 192 0 WIRE 384 32 384 0 WIRE 880 32 880 0 WIRE -48 96 -48 0 WIRE 16 128 16 96 WIRE 96 128 96 96 WIRE 96 128 16 128 WIRE 192 128 192 96 WIRE 272 128 272 96 WIRE 272 128 192 128 WIRE 384 128 384 96 WIRE 720 128 720 96 WIRE 720 128 384 128 WIRE 880 128 880 96 WIRE 1216 128 1216 96 WIRE 1216 128 880 128 WIRE 96 160 96 128 WIRE 96 160 16 160 WIRE 144 160 96 160 WIRE 272 160 272 128 WIRE 272 160 192 160 WIRE 320 160 272 160 WIRE 720 160 720 128 WIRE 720 160 640 160 WIRE 832 160 720 160 WIRE 1216 160 1216 128 WIRE 1216 160 1136 160 WIRE 1328 160 1216 160 WIRE 96 176 96 160 WIRE 272 176 272 160 WIRE 720 176 720 160 WIRE 1216 176 1216 160 WIRE 16 192 16 160 WIRE 192 192 192 160 WIRE 640 192 640 160 WIRE 1136 192 1136 160 WIRE -48 208 -48 176 WIRE 16 272 16 256 WIRE 192 272 192 256 WIRE 640 272 640 256 WIRE 1136 272 1136 256 WIRE 96 288 96 256 WIRE 272 288 272 256 WIRE 720 288 720 256 WIRE 1216 288 1216 256 FLAG -48 208 0 FLAG 272 288 0 FLAG 192 272 0 FLAG 320 160 A FLAG 720 288 0 FLAG 640 272 0 FLAG 832 160 A2 FLAG -48 0 V1 FLAG 512 64 0 FLAG 656 64 0 FLAG 1216 288 0 FLAG 1136 272 0 FLAG 1328 160 B2 FLAG 1008 64 0 FLAG 1152 64 0 FLAG 96 288 0 FLAG 16 272 0 FLAG 144 160 B SYMBOL voltage -48 80 R0 WINDOW 3 -162 235 Left 2 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR Value PULSE(0 1200 0 2uS 2uS 50uS) SYMATTR InstName V1 SYMBOL res 256 0 R0 SYMATTR InstName R1 SYMATTR Value 1G SYMBOL res 256 160 R0 SYMATTR InstName R2 SYMATTR Value 100k SYMBOL cap 176 32 R0 SYMATTR InstName C1 SYMATTR Value .1pF SYMBOL cap 176 192 R0 SYMATTR InstName C2 SYMATTR Value 1nF SYMBOL res 704 0 R0 SYMATTR InstName R6 SYMATTR Value 0.33G SYMBOL res 704 160 R0 SYMATTR InstName R7 SYMATTR Value 100k SYMBOL cap 368 32 R0 SYMATTR InstName C5 SYMATTR Value .1pF SYMBOL cap 624 192 R0 SYMATTR InstName C6 SYMATTR Value 1nF SYMBOL res 512 -16 R90 WINDOW 0 -6 92 VBottom 2 WINDOW 3 -30 34 VTop 2 SYMATTR InstName R8 SYMATTR Value 0.33G SYMBOL res 624 -16 R90 WINDOW 0 -3 77 VBottom 2 WINDOW 3 -27 20 VTop 2 SYMATTR InstName R9 SYMATTR Value 0.33G SYMBOL cap 496 0 R0 WINDOW 0 -12 51 Left 2 WINDOW 3 29 51 Left 2 SYMATTR InstName C7 SYMATTR Value .01pF SYMBOL cap 640 0 R0 WINDOW 0 -22 18 Left 2 WINDOW 3 -42 48 Left 2 SYMATTR InstName C8 SYMATTR Value .01pF SYMBOL res 1200 0 R0 SYMATTR InstName R3 SYMATTR Value 0.033G SYMBOL res 1200 160 R0 SYMATTR InstName R4 SYMATTR Value 10k SYMBOL cap 864 32 R0 SYMATTR InstName C3 SYMATTR Value .1pF SYMBOL cap 1120 192 R0 SYMATTR InstName C4 SYMATTR Value 1nF SYMBOL res 1008 -16 R90 WINDOW 0 -6 92 VBottom 2 WINDOW 3 -30 34 VTop 2 SYMATTR InstName R5 SYMATTR Value 0.033G SYMBOL res 1120 -16 R90 WINDOW 0 -3 77 VBottom 2 WINDOW 3 -27 20 VTop 2 SYMATTR InstName R10 SYMATTR Value 0.033G SYMBOL cap 992 0 R0 WINDOW 0 -12 51 Left 2 WINDOW 3 29 51 Left 2 SYMATTR InstName C9 SYMATTR Value .01pF SYMBOL cap 1136 0 R0 WINDOW 0 -22 18 Left 2 WINDOW 3 -42 48 Left 2 SYMATTR InstName C10 SYMATTR Value .01pF SYMBOL res 80 0 R0 SYMATTR InstName R11 SYMATTR Value 0.1G SYMBOL res 80 160 R0 SYMATTR InstName R12 SYMATTR Value 10k SYMBOL cap 0 32 R0 SYMATTR InstName C11 SYMATTR Value .1pF SYMBOL cap 0 192 R0 SYMATTR InstName C12 SYMATTR Value 1nF TEXT 32 360 Left 2 !.tran 150uS

.............plot file

[Transient Analysis] { Npanes: 1 { traces: 5 {524290,0,"V(a)*1.05"} {589831,0,"V(v1)/10000"} {524291,0,"V(a2)*1.1"} {589830,0,"V(b)*.95"} {524292,0,"V(b2)*.9"}

Y[0]: ('m',0,-0.01,0.01,0.14) Y[1]: ('_',0,1e+308,0,-1e+308) Volts: ('m',0,0,0,-0.01,0.01,0.14) Log: 0 0 0 GridStyle: 1 } }

OK, we call that a "trimpot." My records show that we've used over

90,000 trimpots so far, mostly surface mount. I don't think they have caused any problems. We buy sealed, cermet trimpots. Old unsealed carbon pots tended to get flakey.

A decent single-turn trimpot can be set, with care, to 0.1% of its range, and pretty much stays there if you whack the board. 0.5% is a very conservative design assumption, easy to set.

I agree that trimpots are over-used, especially in online amateur schematics. But sometimes you need to set a gain or something, where dacs and eeproms and software aren't convenient. The nonvolatile digital pots still need a programmer and software, and most are bandwidth limited to not much past audio.

We use one little pot that works fine up to 1 GHz or so. That's a lot of performance for seven cents.

Less than 10c each in quantity eg: digikey 490-7795-2-ND

--
  \_(?)_

[snip!

I got a fair estimate of the parasitic capacitance of a 1206 SMD resistor by just assuming it's a flat plate capacitor with all the field confined to the SiO2 substrate. I calculated 30fF and subsequent measurement indicated 50fF.

For what it's worth..

Jeroen Belleman

I thought they were usually Al2O3.

Cheers

Phil Hobbs

That's basically the three-terminal capacitance measurement trick. Like the Hi/Lo Z trick, it helps the frequency response but not the noise, and the C loading to ground may be a problem in itself. I much prefer to make the c apacitance stable and predictable using air.

Cheers

Phil Hobbs

Erm, yes. That's my reward for trying to make sense on a Monday morning before coffee.

Jeroen Belleman

That's a useful notion, despite Phil's catch about the substrate material.

k(Al2O3)=9.7, vs. k(SiO2)=3.9, FWIW.

John's 2010 and most of the other form factors all have the same

2:1 aspect ratio, so the same capacitance per that approximation.

Too bad.

Cheers, James Arthur

That's worth a lot!

The PCB material will add some more, and a tiny bit more from the air above. So 0.1p might be in the ballpark.

I guess a 2512 above a ground plane shouldn't have much capacitance through the FR4.

I'll try to measure that.

I guess I'll put a fixed cap and a trimmer cap across the lower resistor, and have testing tweak each unit to get a nice looking pulse. We have a cute little Johanson 5-30 pF trimcap in stock, which the fixed cap can pad up if necessary.

Given a 1200 volt signal, noise is not a concern here!

That was fun. The strays have horrible time constants if they're isolated by high resistances.

In the Dept. of Silly Walks department, dividers made entirely of a single resistor type are inherently perfectly compensated.

Vin --- | [R1] | [R2] | +-----+--> Vin/5 | | [R3] [R4] | | === ===

Vin --- | [R1] | [R2] | [R3] | +-----+-----+--> Vin/10 | | | [R4] [R5] [R6] | | | === === ===

Vin --- | [R1] | +------. | | [R2] [R3] | | === +-----. | | [R4] [R5] | | === +-----. | | [R6] [R7] | | === +----> | [R8] | ===

Cheers, James Arthur

Right, that's this one, or almost:

I guess it was designed to be cheap, and accidentally works up into the GHz range. If it was called a microwave trimpot, I'd gladly pay

20x the asking price.

If you'd used carbon track ones it would be a very different story.

NT

Right. Most surface mount resistors are made on 20 mil alumina, with about the same aspect ratio, so their capacitance (and their thermal resistance, and power dissipation capability [1]) is independent of size.

[1] assuming you heat sink the end caps

I have files of all the resistor values and all the dividers that you can make from a quad resistor pack.

Here's one of them

Not around here. Testing people are not expected to solder.

If I wanted to adjust capacitance in steps, I'd use a dip switch, or one of those IC multi-step capacitors.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Note that this is a 1Gohm, 1 watt resistor.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com