Switching small capacitive loads

[Andrew Holme]

A colleauge of mine is building an automated test jig for measuring rise
times under different capacitive loads ranging from 5 to 70pF. Switching
small capacitances with relays is difficult/impossible and I proposed he use
electronic switching. Here are the schemes we've considered:

.-- 1V step from 100-ohm source impedance
| 10ns rise time (unloaded)
| -------o-----------------o----------------o----> To 'scope
| | | |
--' --- --- ---
C1 --- C2 --- C3 --- R3 10k
| | | ___
.-----o | o----|___|-> +10V
| | | |
L1 C| | PIN |/ |/
1uH C| V Diode .----| Q1 .----| Q2
C| - D1 | |> | |>
| | | | | |
.-. | .-. | .-. |
R1 | | === R2 | | === R3 | | ===
120 | | GND 1k | | GND 1k | | GND
'-' '-' '-'
| | |
| | |
3.3V Level CMOS Logic Control Lines

C1/C2/C3 in 15 - 50pF range
Q1/Q2 = 2SC4774 or BFS20W

Our first thought was to use PIN diode switching; but we see problems with
the above circuit in simulation when the diode is supposed to be off. Large
negative DC bias is required to stop the fast rise time of the signal
turning on the diode; and, unless an impractically large inductance is
specified, choke L1 passes a sizeable AC current, causing severe waveform
distortion.

The bipolar transistor circuit seems much better. There is no waveform
distortion except for the expected RC low-pass filtering effect of the
selected load capacitance. The output capacitances of the 2SC4774 and
BFS20W are both specified as about 1pF at Vcb = 10V; however, it seems to
work equally well in simulation with or without DC collector bias.

Any comments or advice?

TIA

 

[David L. Jones]

A colleauge of mine is building an automated test jig for measuring
rise times under different capacitive loads ranging from 5 to 70pF.
Switching small capacitances with relays is difficult/impossible and

Pickering do a low capacitance SIL relay (<0.1pF):
http://www.pickeringrelay.com/dropdown/103series.htm

Dave.

 

[Jamie]

A colleauge of mine is building an automated test jig for measuring rise
times under different capacitive loads ranging from 5 to 70pF. Switching
small capacitances with relays is difficult/impossible and I proposed he use
electronic switching. Here are the schemes we've considered:

.-- 1V step from 100-ohm source impedance
| 10ns rise time (unloaded)
| -------o-----------------o----------------o----> To 'scope
| | | |
--' --- --- ---
C1 --- C2 --- C3 --- R3 10k
| | | ___
.-----o | o----|___|-> +10V
| | | |
L1 C| | PIN |/ |/
1uH C| V Diode .----| Q1 .----| Q2
C| - D1 | |> | |>
| | | | | |
.-. | .-. | .-. |
R1 | | === R2 | | === R3 | | ===
120 | | GND 1k | | GND 1k | | GND
'-' '-' '-'
| | |
| | |
3.3V Level CMOS Logic Control Lines

C1/C2/C3 in 15 - 50pF range
Q1/Q2 = 2SC4774 or BFS20W

Our first thought was to use PIN diode switching; but we see problems with
the above circuit in simulation when the diode is supposed to be off. Large
negative DC bias is required to stop the fast rise time of the signal
turning on the diode; and, unless an impractically large inductance is
specified, choke L1 passes a sizeable AC current, causing severe waveform
distortion.

The bipolar transistor circuit seems much better. There is no waveform
distortion except for the expected RC low-pass filtering effect of the
selected load capacitance. The output capacitances of the 2SC4774 and
BFS20W are both specified as about 1pF at Vcb = 10V; however, it seems to
work equally well in simulation with or without DC collector bias.

Any comments or advice?

TIA

Q1 seems to be more practical if you add a discharge circuit at the
source when it's goes low. You may already have that in mind? I can't
tell if your source signal is actually pulling to common when off. This
is keeping in mind, that Q1 remains selected so that discharge can
take place.

Q2, will introduce voltage on the source line in the deselected
transition. You may want to think about that.

Those are my observations at first glance.

 

[Jeroen Belleman]

Pickering do a low capacitance SIL relay (<0.1pF):
http://www.pickeringrelay.com/dropdown/103series.htm

It may have <0.1pF between open contacts, but each terminal
still has >1pF to GND!

I have the measurements somewhere, but I'd have to dig them out.

Jeroen Belleman

 

[David L. Jones]

It may have <0.1pF between open contacts, but each terminal
still has >1pF to GND!

The "Guard screen only" (103G) version is speced at 0.1pF contact to coil
capacitance and <0.1pF open contact capacitance.
The magnetic screen version (103M) is the worst at 0.45pF per contact to
coil, so close to your claimed 1pF figure.

So it's possible to get the correct version with only 0.2pF total
capacitance to the coil. Unless you are saying they don't meet their specs?

Dave.

 

[Jeroen Belleman]

The "Guard screen only" (103G) version is speced at 0.1pF contact to coil
capacitance and <0.1pF open contact capacitance.
The magnetic screen version (103M) is the worst at 0.45pF per contact to
coil, so close to your claimed 1pF figure.

So it's possible to get the correct version with only 0.2pF total
capacitance to the coil. Unless you are saying they don't meet their specs?

They *do* meet their specs. With the (grounded) guard screen
in between, the coil-to-contact capacitance is indeed very small.
It's the contact-to-guard capacitances I was thinking of. I dug up
my measurements: It's about 2.8pF for each contact.

Anyway, in view of the intended application, the measurement of
rise time with varying capacitive loads, it may be useful to
keep in mind that the open contacts of the relay with guard
screen look like a short piece of open-ended transmission line.
It's up to the OP to judge if that's relevant.

Jeroen Belleman