You can try any NPN safely. Some will be better, some worse.
Even a Chinese S8050 was already much better than the 337
Here's the updated and corrected last version of the Dienoiser designed for assembling on a prototype pcb.
It's designed for through hole parts only, and the parts terminals are to be used as the circuit traces, twisting and soldering the wires to interconnect the parts.
Please disregard the other design I had uploaded some days ago, as it had one trace missing. Also this time, and considering some feedback from others that used the Dienoiser and seemed to have had some oscillations, I added a compensation resistor space in series with the 22n capacitor.
As soon as possible I will assemble that proto pcb here, with the parts I have, and add it to some of the regulators I also have.
As you can see, there's a vertical line in the middle, which you can use to separate the positive and negative Dienoisers, accordingly to what you need. Or if you have only positive regulators, as might be the case, repeat another positive Dienoise on the right side.
This is as easy as this project can get.
It's designed for through hole parts only, and the parts terminals are to be used as the circuit traces, twisting and soldering the wires to interconnect the parts.
Please disregard the other design I had uploaded some days ago, as it had one trace missing. Also this time, and considering some feedback from others that used the Dienoiser and seemed to have had some oscillations, I added a compensation resistor space in series with the 22n capacitor.
As soon as possible I will assemble that proto pcb here, with the parts I have, and add it to some of the regulators I also have.
As you can see, there's a vertical line in the middle, which you can use to separate the positive and negative Dienoisers, accordingly to what you need. Or if you have only positive regulators, as might be the case, repeat another positive Dienoise on the right side.
This is as easy as this project can get.
Attachments
Last edited:
I tried the dienoiser with only 22n compensation and without compensation cap to no avail.
Yes, originally I had 33R in series with 22n cap. Actually it had the worst oscillations. In Spice simulations the dienoiser has a resonance at ~40MHz without compensation. Adding compensation cap just moves the resonance to a lower frequency. E.g. with 22n compensation the resonance is at ~5Mhz.
I'm not sure what's your point. Here is another view of the Spice transient analysis which more clearly shows the bursting into oscillation. This is taken from the "ground" leg of Cadj (C12).
Attachments
Only that if the control system is unstable, and if Nyquist oscillation is observed on node "AAA", we also expect to observe it on BBB, CCC, DDD, and EEE too.
Among them, I expect the greatest amplitude at BBB and EEE. Probing them ought to give the easiest-to-see oscillatory waveforms. Plotting what's easy to see, feels like a useful decision.
_
Among them, I expect the greatest amplitude at BBB and EEE. Probing them ought to give the easiest-to-see oscillatory waveforms. Plotting what's easy to see, feels like a useful decision.
_
Attachments
The tried both 10n and 22n as compensation. Both oscillated in Spice. The Cadj (100u or 220u) should have not impact on the oscillations.
