Ahaa, didn't realize Clare sold individual parts, a logical move though, based on the optos stuff. That's reassuring that we have at least two manufacturers of depletion mode DMOS.
EDIT: Also the geometric mean of the threshold voltages listed is 2.5V, a bit higher than the other parts.
Last edited:
I did figure this out. But I was hoping for models for the Supertex FETs you recommend in
Scott, when you do this for #1397, could you also explain the reasons for various choices? I would much like to know the reasons for
- the fancy output stage compared to simple Class B as in 990
- why FET i/p instead of simple BJT in your rehash of 990
High voltage devices to begin with will allow the SW-OPA to be adadted to power amp use is a very good idea. Truely universal design that way... My plane is about to leave.... check in when I'm on the other side of the world. -RNM
Just tried something else in my version of SW-OPA, looking at PSRR. At 20kHz, only a bit over 15dB on the -ve supply, 30dB on the +ve when working at a gain of 20. So an extremely well regulated supply will be necessary for the circuit as is ...
Frank
Frank
Last edited:
Not to knee-jerk defend, but are you measuring this based on your closed-loop gain of 101?
????? ...
Feedback resistors of 2k, 40K. Tried also 2k, 10k, so gain of 6, that improved performance by about 10dB ...
Frank
And of further interest, a member PM'ed me about a breadboard he's already knocked up, and his PSRR figures, measured at lower frequencies, correlate beautifully with what the sim predicts ...
Frank
Frank
Frank,
dadod's version sims at about ~-25db at 20KHz on the negative side with your 20K/2K feedback values. Switching to K170/J74 models for the front end @ 7mA, everything else otherwise the same, brought it up around ~-31db at 20KHz. Low frequency is about -89db for the LSK389/J109 with the knee at about 65Hz. The K170/J74 was in the -96db range with the knee around 70Hz. High side PSRR is better.
All based on the LTSpice sim of course… Anybody else get similar numbers from it? It would be nice to know it works the same in multiple locations, even if just in the sim at this point.
Dave
dadod's version sims at about ~-25db at 20KHz on the negative side with your 20K/2K feedback values. Switching to K170/J74 models for the front end @ 7mA, everything else otherwise the same, brought it up around ~-31db at 20KHz. Low frequency is about -89db for the LSK389/J109 with the knee at about 65Hz. The K170/J74 was in the -96db range with the knee around 70Hz. High side PSRR is better.
All based on the LTSpice sim of course… Anybody else get similar numbers from it? It would be nice to know it works the same in multiple locations, even if just in the sim at this point.
Dave
This circuit is shown here, and discussed close to the end of the Audio Xpress interview with Walt Jung
WaltsBlog
Probably off topic here, but worth a new thread in the right place...
PSRR is something to look at and there could be some tweaks. I would expect non-cascoded inputs to be fairly sensitive to source impedance mis-match. Welcome to the design process.
EDIT- OK we need to track down the discrepancy, the very first circuit (all N flavor FET's but input cascoded) shows me flat -100dB RTI on both supplies out to 20K.
Last edited:
General appeal: could people please reference a post # when referring to a design? It would burden things unduly to attach the schematic each time, but at least we can use the number and go back and confirm what we're talking about.