DIY Audio Engineering

Does anyone have a spice model for 6Н3П that works in MicroCap 9 ?

That looks interesting.  A while back I downloaded the Quartus II software, which looks pretty powerful for VHDL-oriented stuff, but I didn't stop to think that the earlier tools might have more analog-oriented features.  I'll definitely have a look.

So far, I've been pretty happy with KiCad, but I do miss the schematic SPICE verification aspect.  One can export SPICE netlists from it, but simulation with just a netlist is a real PITA.

Well I got a reasonable amount done this weekend. The complete amp module is now up and running (with the MOSFET output devices) and it works fine. The PCB will need a revision though to fix the ground issue mentioned above and some other small issues. I have changed the base stopper wiring to the second EF pair in the diamond buffer driver to a better arrangement. I got a really nice full amplitude 200kHz squarewave (and very clean 200kHz sine clipping / overdrive recovery) out of it before the zobel resistor started smoking  . The next step will be to bolt on a heat sink and make some proper measurements.

[ Guests cannot view attachments ]
[ Guests cannot view attachments ]
[ Guests cannot view attachments ]

I would imagine that many of you have come across this site,
if not, here is the link:
http://www.alistapart.com/about/

Well I finally got this PSU up and running this evening. All worked fine, except that the overload shutdown/latch circuit worked too well. On power up the charging current of the 1000uF bypass capacitors on the amplifier module triggered the latch. I had to increase value of the "trigger delay" capacitors (C17-C20 in the PSU schematic) to stop this from happening. However that was an easy fix and the protection circuity is working really well.

As for the amplifier module, I've got that to the stage of running minus the power output MOSFETs (for preliminary testing a temporary jumper lead connects the output of the diamond buffer driver to the nfb pick-off point). When testing out a new design I always like to get the "front end" up and running at first, minus the power output devices. If there are any issues with that part of the design (oscillation, etc) it's always a lot easier to troubleshoot and rectify without the complication of the high power part.

I spent hours chasing an 30MHz (unstable, repetitive bursts) oscillation problem. For some reason I was convinced that the problem was originating in the diamond buffer driver and spent ages stuffing around with base/collector stoppers to no avail. The problem turned out to be, following Self’s recommendation, my omission of the usual 10R resistor linking the signal ground to the locally bypassed power ground on the PCB.

In the past I’ve always relied on this resistor (fitted to each amplifier module) to act as the necessary power ground termination for the signal current. This resistor is always “star earthed” at the amplifier module PCB along with the speaker ground lead. This has always worked very well for me, but Self’s recommendation of seperately returning the signal ground and the power ground (as well as the speaker ground return) off the PCB to star points off the main reservoir capacitors looked even better in theory.
In his book he shows a diagram of the signal ground and the bypass capacitor/power ground being made to the PCB with two separate ground wires. There is no connection at all between the signal ground and the power ground on the PCB. This is what I am now doing, but I have found that omitting the now seemingly redundant 10R resistor between these two grounds on the PCB is definitely not a good idea!

The oscillation issue was entirely due to this earthing arrangement and was eliminated entirely simply by fudging in a 100R resistor to connect the two grounds on the PCB. The amplifier that was oscillating all evening is now rock stable as it should be.
The only other fix that worked at all was to reduce the length of both the signal ground and the power ground return wires to an impracticably short length (to minimize the length and inductance of the total signal ground return loop). Ideally, it would be nice not to have that 100R resistor there to inject some degree of current into the signal ground from the dirty power ground, but I think on a practical level its it impact would be next to nil as this resistor is effectively shunted in parallel by off-board signal ground return wire. What is really important is to have a low inductance bypass to this large wiring loop, and the 100R resistor on board does that very adequately.