Hi,
I have a few questions concerning proper (or maybe optimum
) physical layout of an opamp circuit with not-so-easy to tame chips and tiny input signals.
1) What to do with internally-not-connected-pins? Should they be grounded or left floating?
2) If I use a groundplane, is it wise to have the area under the chip grounded, or better leave it out? (I've seen both)
3) The datasheet of opa627 says to use a guard ring around the inverting input. Is that guard referenced to ground?
thanks,
Rüdiger
I have a few questions concerning proper (or maybe optimum
) physical layout of an opamp circuit with not-so-easy to tame chips and tiny input signals.1) What to do with internally-not-connected-pins? Should they be grounded or left floating?
2) If I use a groundplane, is it wise to have the area under the chip grounded, or better leave it out? (I've seen both)
3) The datasheet of opa627 says to use a guard ring around the inverting input. Is that guard referenced to ground?
thanks,
Rüdiger
1 Leave them be, safest also in the future if you want to use some other type. Notice that in some case like TPA6120 you should connect them in order to get better cooling but as always, read the datasheet carefully.
2 It's normally not a problem. One square centimeter = 3 pF only. This is valid for to areas of the same shape. In real life you are dealing with a fraction of a pF. In some cases it is important but not in normal audio circuits.
3 This is only important for high impedance signal sources, >> 1MOhms forget it in audio circuits, but no harm to use it either.
2 It's normally not a problem. One square centimeter = 3 pF only. This is valid for to areas of the same shape. In real life you are dealing with a fraction of a pF. In some cases it is important but not in normal audio circuits.
3 This is only important for high impedance signal sources, >> 1MOhms forget it in audio circuits, but no harm to use it either.
The main issue with these fast chips is to get power rail decoupling caps as as close to the pins as possible. Surface mount components are good if you can get them.
Indeed, groundplane and tight placed decoupling is good.
Check C6 and C13. Notice that the pcb has now the groundplane connected on both sides.
http://home5.swipnet.se/~w-50674/hifi_pics/hifi_100pr/qrv07r0_TPA6120.jpg
Check C6 and C13. Notice that the pcb has now the groundplane connected on both sides.
http://home5.swipnet.se/~w-50674/hifi_pics/hifi_100pr/qrv07r0_TPA6120.jpg
peranders said:
P-A, interesting detail. Do C6, C13 connect directly to the groundplane on the underside or did you establish a star-wired route?
Rüdiger
Hi,
the guard ring should be connected to a source of equal voltage.
This is to minimise the leakage current from your input.
You can often find examples of guard rings in the data sheets.
the guard ring should be connected to a source of equal voltage.
This is to minimise the leakage current from your input.
You can often find examples of guard rings in the data sheets.
Now I have "crosses" on both sides. The first batch had this error.Onvinyl said:
http://home5.swipnet.se/~w-50674/hifi_pics/hifi_100pr/qrv07pcbcomp.jpg
http://home5.swipnet.se/~w-50674/hifi_pics/hifi_100pr/qrv07pcbsold.jpg
Decoupling question:
In an active crossover, would i have to use two 0.1uf MKP decoupling caps for each opamp?
In an active crossover, would i have to use two 0.1uf MKP decoupling caps for each opamp?
Yes, it's a good rule, no need to save parts. If you happen to use fast opamps, then it's even more important.
AndrewT said:
Hi Andrew,
equal to what? Vcc? Vin? (It's DC, I guess)?
If I use an opamp in inverted mode (and thus the signal feeds in the inverted input), is it wise to reference the shield of the signal cable with that voltage (rather then ground)?
Rüdiger
P-A, I don't get it. From what I see, C6/C13 go nowhere. How are they connected now?peranders said:
Rüdiger
Rüdiger, this you'll have to study more. Check Walt Jung's Opamp book but seriously, this applies to signal sources with very high internal impedances', like pH-probes, photo detectors. The this is to eliminate voltage between two traces and by that leakage currents. 478 GOhms a couple of volts plus some dirt => current.
The picture shows a groundplane on the "solderside" plus an another one on the "component side" which didn't get connected 100% right but now it's fixed.Onvinyl said:
peranders said:
Yes, it's clear you have two groundplanes. What I don't see is, where do C6/C13 go, as it on the photo it *looks* like the pads are not connected to either groundplane.
Rüdiger
- Status
- Not open for further replies.