De-coupling supply to 747 op amps

[Chris Wilson]

Old "in a can" stlye CT747 op amps, the supply is + and - 15V. They have no supply de-coupling and RF may be getting into them. I know how to de-couple a single rail op amp with a 0.1uF ceramic across the supply pins, but how do I decouple a two rail device please? Schematic is here:

http://www.gatesgarth.com/H60002.jpg


Thanks.

[MagicSmoker]

You decouple a bipolar power supply the same way as a unipolar, except you need 2 capacitors: one from V+ to common, the other from V- to common. If you are feeling extra paranoid you can also wire a capacitor directly across V+ and V-.

[Zero999]

I agree with the above: connect a capacitor from either rail to the common and if that's not enough, another between the +&- for good measure.

Note that the circuit already has some decoupling from C1 & C2. Presumably the circuit doesn't work at a high frequency or draw any sudden current surges, so the designer decided it wasn't necessary to cover the board in decoupling capacitors.

[TimFox]

With normal dual-rail op amps, the minimum required is a cap from V+ to V-.  Thereafter, a cap from V- to ground is a good idea, since the NPN transistor that develops the output voltage (supplied to the output complementary follower) is referenced to V-.

[danadak]

Not all ceramics are equally good. Pay particular attention to datasheet
f vs z curve.  Lead length critical at RF as well.


http://www.atceramics.com/Userfiles/capacitors_in_bypass.pdf


https://www.microwaves101.com/encyclopedias/capacitors


Regards, Dana.

[Chris Wilson]

Thanks for the info and the links, very useful, and much appreciated!

[SeanB]

If you are getting RF into a 741 that will be hard, but if decoupling does not help and you are sure it is RF then a 100p capacitor between pins 2,3 and pin 4 will get rid of it, at the expense of slowing down the slew rate, but a 741 is not exactly a high frequency performer.  Before doing anthing else replace C1, C2 and C4, that will do a lot more to get Rf out if they are bad, 2 of these opamps are already used as integrators.

[Chris Wilson]

Thanks for the reply, I have already changed all electrolytics on the board, and I now believe I was in error considering RF might be getting in, I have subsequently found the coarse control pot for output voltage has an intermittent open in its wire winding. The thing is poorly designed (says he who could barely design a crystal set....) in that an open control gives max volts out. I would have thought it would have been safer had an open given minimum or no volts out? Thanks SeanB you are a credit on here with your tireless sharing of hard won experience!

[SeanB]

I must have changed hundreds of those TO100 can 741's in equipment, as the application they were in abused the offset null as a way to trim the system offsets, not the opamp offset only. This would invariably lead to parameter shifts in the opamp offset, and the cure ( aside from redesign to get the offset away properly) in most cases was to simply replace the stressed opamp with a new one with less stress in it. I tended to simply take the board and set the offset to zero with input shorted ( thank you for the manufacturer making extender cards with links on them, and with each link having the ability to either be through or open, using little U shaped jumpers that would accept test leads, and also for putting spare sockets both sides so you could probe there or use as grounding jumpers for the inputs. All aC signals with synchronous rectifiers, using a lot of older DG series analogue switches, discrete JFET switches, Unijunction transistors and such to do sample and holds that acted as rectifiers as well, then into an 18 bit ADC for digitisation into a serial data stream into some MK4007 serial memories, 1026 bits of CCD storage for everything, all recirculated once per cycle for operation.

[Chris Wilson]

I must have changed hundreds of those TO100 can 741's in equipment, as the application they were in abused the offset null as a way to trim the system offsets, not the opamp offset only. This would invariably lead to parameter shifts in the opamp offset, and the cure ( aside from redesign to get the offset away properly) in most cases was to simply replace the stressed opamp with a new one with less stress in it. I tended to simply take the board and set the offset to zero with input shorted ( thank you for the manufacturer making extender cards with links on them, and with each link having the ability to either be through or open, using little U shaped jumpers that would accept test leads, and also for putting spare sockets both sides so you could probe there or use as grounding jumpers for the inputs. All aC signals with synchronous rectifiers, using a lot of older DG series analogue switches, discrete JFET switches, Unijunction transistors and such to do sample and holds that acted as rectifiers as well, then into an 18 bit ADC for digitisation into a serial data stream into some MK4007 serial memories, 1026 bits of CCD storage for everything, all recirculated once per cycle for operation.

I'm Googling hard, but first reaction is wow, that sounds clever