Toshiba MOSFETs with a twist

[nige2000]

Think it's just Derek and Pearse

[jkeny]

Think it's just Derek and Pearse

Ah, OK, I'll post it

[nige2000]

Richard can we use the k339 and j327 as substitutes of j74/k170 in all those buffer and gain circuits out there?

[abraxalito]

I see the absolute max voltage for 1387 from d/s is 6V - i'm thinking of 2 batteries to give 6.6V - have your tried this high a V?

No, but a guy from Korea who PM'd me on DIYA said he'd gone up to 7V and it sounded better. I'll stick with 6V.

So a pot substituted for 56 or 62ohm could adjust for V offset or what's the best approach?

I'm capacitively coupling my output so I'm only concerned about offset to the degree it compromises my output swing. What I did was measure the offset then calculate a parallel resistor across one or other of the CCSs to minimize that. Soldered it on top of the existing R. The offset takes maybe 30s to stabilize after powering up, partly because of my big cap on pin7 (1500uF for two TDA1387).

[abraxalito]

Richard can we use the k339 and j327 as substitutes of j74/k170 in all those buffer and gain circuits out there?

The ones with gates of N and P FETs tied together you can't because these MOSFETs are enhancement types. So neither will turn on in that configuration.

I'm no fan of those purely complementary configurations, I much prefer SE topologies. The second schematic you're showing should work fine with a substitution - looks like a folded cascode. I will try it out in LTspice and report back later. Gate biassing arrangements will need to change - resistor to 0V will have to become a resistor to a higher voltage, necessitating coupling caps.

[fas42]

Very informative thread, Richard - good to see the detail laid out like this!

Frank

[abraxalito]

I'm no fan of those purely complementary configurations, I much prefer SE topologies. The second schematic you're showing should work fine with a substitution - looks like a folded cascode. I will try it out in LTspice and report back later.

It tried it (and no, its not a folded cascode) and the PSRR was really poor. So I don't think that circuit can be rescued. I am building up a two MOSFET gain stage with both MOSFETs acting as buffers (source followers) and the voltage gain coming from a current mirror in a similar way to how the AD844 gets its voltage gain (bipolar transistors though as they match better). In fact the circuit looks rather like an AD844 cut in half left to right and the bottom half, instead of being a mirror image of the top, is replaced with current sources. The CCSs give excellent PSRR which is the reason for using them.

AD844.png (24.03 KiB) Viewed 982 times

Skipping through the AD844's datasheet the one thing very conspicuous by its absence is the customary PSRR plot vs frequency. I wonder if the omission is by design due to it looking like a dog's dinner? On DIYA sometimes heated debates arise between devotees of CFB amps and the more traditional VFB (LTP input, popularized by Doug Self) and even the proponents of CFB admit that their amps suck in PSRR.

[nige2000]

Is the psrr the upmost or only priority or is there other factors?
could decent ps carry a "poor psrr amp" past a "better psrr amp" subjectively ?

is variable gain a possible solution for volume control?
or how can we have volume control without raising impedance?

[abraxalito]

I'd not say that PSRR is the only thing that matters but for now (at my current level of understanding) its what I'm focussing on and its delivering pleasing sonic results. As to whether a decent PSU coupled to average PSRR amp can beat off a high PSRR amp paired with a lousy PSU I don't know, its still early days with this philosophy. I reckon its simpler, cheaper and more compact to make DACs, line stages and lowish power amps with high PSRR than to build kick-ass supplies. I can always add on an enhanced PSU later to see if it brings further improvements to my high PSRR designs.

Variable gain I believe has been done by Doug Self but he'll have been using opamps. It normally calls for an inverting configuration so as to get below 0dB. Jury's still out on the best way to do volume, I'm veering towards tweaking pin7 on my TDA1387s...

[nige2000]

Yea...
Ok

Last time we were at the 1387 I put a variable voltage divider between 6.6v and gnd into pin 7 to change volume or force the reference voltage
There was little to no negative effects and worked fine