Well the power supply decoupling caps do need GND. Which can be done with a wire if you want the decoupling caps on the adapter PCB closest as possible to the IC pins as required with fast opamps.
Wire - Wikipedia
It will be compatible as anything on many boards created by whatever person provided there is a GND point nearby but this is more rule than exception on PCB's with DIL opamps, with just the addition of a thin wire. Since SMD-to-DIP adapters don't win beauty contests anyway the wire won't be an obstacle. It is just practical design, you could of course try it out and see if you can use it in your designs.
Wire - Wikipedia
It will be compatible as anything on many boards created by whatever person provided there is a GND point nearby but this is more rule than exception on PCB's with DIL opamps, with just the addition of a thin wire. Since SMD-to-DIP adapters don't win beauty contests anyway the wire won't be an obstacle. It is just practical design, you could of course try it out and see if you can use it in your designs.
Last edited:
I should have patented it.The LM4562 has a bipolar input stage, and a bipolar differential pair goes into clipping somewhere around 52 mV. Hence, a 1 V step is bound to drive the input stage into clipping. Clipping of the input stage causes slew-rate limiting of the complete op-amp, as your simulations show.
Does it help to put a little RC low-pass filter (220 ohm- 220 pF or so) at the input? The left op-amp then sees a low source impedance at high frequencies. I didn't have any stability problems the (only few) times I used an OPA2134 as a voltage follower, but I usually had an RC low-pass at the input.
HOLD THE PHONE!! You seem to be saying that every LM4562 CLIPS with an input signal >52mV. That cannot be true.
I have a couple of suggestions.
I do not know, but I believe the OPA2134 has a common-emitter output stage. If that is true, then its gain is proportional to the load, and if the output is unloaded, then the gain becomes too high for the compensation. Each op amp should have some reasonable load to ground on its output. 10K ohm for instance. If you look at the buffer circuit Mark Johnson showed in post #4, it has 5.6K ohm to ground on the output.
It is also possible that the 2nd filter in Rod Elliott's circuit is too capacitive a load for the 1st op amp. 50 ohms in series with C4 should take care of that without screwing up the filter alignment. I doubt that this is the problem.
Only try the second fix if the first one doesn't work. I think some loading to GND on each output is the cure.
Last edited:
Not quite, only the input stage clips in response to a very fast step greater than 52 mV. At the output you see that as slew-rate limiting.
Marcel is talking about the size of the input error voltage, not the size of the signal. If the step is fast enough, the amp cannot keep up, and the error voltage can exceed 52 mV. This is the correct number for bipolar inputs with no emitter degeneration. The amplifier will go to its full slew rate to attempt to minimize the error voltage.
He is not really talking about anything destructive here. This is how op amp slew rates are tested. Square wave inputs to an op amp buffer will always cause this to happen.
He is not really talking about anything destructive here. This is how op amp slew rates are tested. Square wave inputs to an op amp buffer will always cause this to happen.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.