Creating capacitance on PCB board for feedback of OPA657?

[[email protected]]

I am designing a detector using transimpedance amplifier - OPA657. We
are using simple 2 layer PCB board. On TI's datasheet they suggest
using 0.1pF capacitor in parallel in 200K feedback resistor.
http://focus.ti.com/lit/ds/symlink/opa657.pdf
On the previous design we had - 1pF(?!) was used in parallel with ~130K
feedback resistor using OPA655.
http://focus.ti.com/lit/ds/symlink/opa655.pdf
1pF(?I - not sure if it was 1pF as it is impossible to measure it)
capacitor was obtained simply using parasitic capacitance of the empty
pad. If the empty pad was not there, the gain would be very low (10x
lower), so the feedback resistance had to be increased significantly
(5x). How do I get ~0.1pF parasitic capacitance for OPA657? And if you
know a good guide for calclulating parasitic capacitance of PCB traces
- it would be great as well.


Thanks.

 

[Fred Bartoli]

I am designing a detector using transimpedance amplifier - OPA657. We
are using simple 2 layer PCB board. On TI's datasheet they suggest
using 0.1pF capacitor in parallel in 200K feedback resistor.
http://focus.ti.com/lit/ds/symlink/opa657.pdf
On the previous design we had - 1pF(?!) was used in parallel with ~130K
feedback resistor using OPA655.
http://focus.ti.com/lit/ds/symlink/opa655.pdf
1pF(?I - not sure if it was 1pF as it is impossible to measure it)
capacitor was obtained simply using parasitic capacitance of the empty
pad. If the empty pad was not there, the gain would be very low (10x
lower), so the feedback resistance had to be increased significantly
(5x). How do I get ~0.1pF parasitic capacitance for OPA657? And if you
know a good guide for calclulating parasitic capacitance of PCB traces
- it would be great as well.

Use 3 caps in a T connection.
This is equivalent to 3 caps, Delta connected.
Only one cap matters to your feedback network, the 2 others are absorbed
into the output and input parasitics.
You can make the central cap adjustable, with the added advantage that it
have it's frame grounded.

0.1pF is pretty easy to achieve, provided you've correctly done the layout
(not too much parasitics from output to input).

 

[Winfield Hill]

[email protected] wrote...

I am designing a detector using transimpedance amplifier - OPA657.
We are using simple 2 layer PCB board. On TI's datasheet they
suggest using 0.1pF capacitor in parallel in 200K feedback resistor.
http://focus.ti.com/lit/ds/symlink/opa657.pdf

Don't forget the feedback resistor's self capacitance. The
through-hole resistors I've measured have been between 0.07
and 0.20pF of self capacitance, with most between 0.10 and
0.15pF. The SMDs I've measured have been slightly higher.

As far as the best desired value, this depends as much on
the sensor and node-wiring capacitance as it does on the
feedback resistor. Using the appropriate formula works
better than an ad-hoc suggestion.

 

[[email protected]]

[email protected] wrote...

Don't forget the feedback resistor's self capacitance. The
through-hole resistors I've measured have been between 0.07
and 0.20pF of self capacitance, with most between 0.10 and
0.15pF. The SMDs I've measured have been slightly higher.

That is a bit odd. My impression was that the parallel resistance of
L-trimmed surface mount resistors was appreciably lower than that of
spiral=trimmed axial lead through-hole resistors - an impression formed
by a Philips video pre-amp on a plumbicon camera tube I ran into on the
Cambrdige Instruments Chip-Check mask inspection tool which was based
on a Metals Research Quantimet image analysis system.

The original Quantimet system had used a Cambridge Instruments
pre-amplifier built with through resistors, and the feedback resistor
had been limited to 130k by its paralllel capacitance and the
requirement to maintain a video bandwidth. The Philips pre-amplifier
that came with the new, improved camera tube, had something closer to
500k of SMD feedback resistance and still gave the same bandwidth.

Nobody had noticed, that this resistor was larger, which messed up all
the biassing arrangements, and gave us rotten noise levels - I got
slung in as a trouble-shooter, and found and fixed the problem a day or
two, which did my reputation no harm at all.

As far as the best desired value, this depends as much on
the sensor and node-wiring capacitance as it does on the
feedback resistor. Using the appropriate formula works
better than an ad-hoc suggestion.

The OPA657 data sheet has a good wirte-up, apparently taken over fron
the old OPA655 data sheet, which I've recommedned here before.

http://focus.ti.com/lit/ds/symlink/opa657.pdf

 

[[email protected]]

[email protected] wrote...

Don't forget the feedback resistor's self capacitance. The
through-hole resistors I've measured have been between 0.07
and 0.20pF of self capacitance, with most between 0.10 and
0.15pF. The SMDs I've measured have been slightly higher.

That is a bit odd. My impression was that the parallel resistance of
L-trimmed surface mount resistors was appreciably lower than that of
spiral=trimmed axial lead through-hole resistors - an impression formed
by a Philips video pre-amp on a plumbicon camera tube I ran into on the
Cambrdige Instruments Chip-Check mask inspection tool which was based
on a Metals Research Quantimet image analysis system.

The original Quantimet system had used a Cambridge Instruments
pre-amplifier built with through resistors, and the feedback resistor
had been limited to 130k by its paralllel capacitance and the
requirement to maintain a video bandwidth. The Philips pre-amplifier
that came with the new, improved camera tube, had something closer to
500k of SMD feedback resistance and still gave the same bandwidth.

Nobody had noticed, that this resistor was larger, which messed up all
the biassing arrangements, and gave us rotten noise levels - I got
slung in as a trouble-shooter, and found and fixed the problem a day or
two, which did my reputation no harm at all.

As far as the best desired value, this depends as much on
the sensor and node-wiring capacitance as it does on the
feedback resistor. Using the appropriate formula works
better than an ad-hoc suggestion.

The OPA657 data sheet has a good wirte-up, apparently taken over fron
the old OPA655 data sheet, which I've recommedned here before.

http://focus.ti.com/lit/ds/symlink/opa657.pdf

 

[Winfield Hill]

[email protected] wrote...

That is a bit odd. My impression ...

I wasn't too surprised after considering the considerably-closer
spacing of a surface-mount part's endcaps, compared to thin-lead
1/4-watt axial-lead resistors, etc. Ohmite's larger SlimMOX HV
resistors get down to under 0.04pF, far better than the miniature
SMD blokes.

 

[Vitaliy]

Hi Fred,

I am not quite sure how I would do T connection in this case. Could you
please explain?
I will be using all SMT componentes, and I haven't even seen the
adjustable SMT resistors (do they exist)?

Thanks,

Vitaliy

 

[Vitaliy]

I have SMDs and was aware of their parasitic effects, never knew their
exact values, so thanks for sharing the information.

I will be doing something similar to Figure 3 on OPA657 datasheet, so
I will definitely go over that.

Thanks for your help guys.
Vitaliy