Amp stability unclear

[Khron]

Greetings everyone...

I'm in the process of putting together 4 of Dr Self's "Load-invariant" amps, to be used in some bi-amped speakers. They're going to handle under 2KHz signals only.

I made some component value alterations due to my somewhat restricted part inventory.

The thing i'm slightly nervous about is the fact that in an LTspice4 AC analysis, the amp still has about 6dB gain when it reaches -180deg, at the output (but before the output R//L). Is this just some simulation (in)accuracy that's fueling this "paranoia", or should i be safe in real-life?

[ostripper]

The thing i'm slightly nervous about is the fact that in an LTspice4 AC analysis, the amp still has about 6dB gain when it reaches -180deg, at the output (but before the output R//L). Is this just some simulation (in)accuracy that's fueling this "paranoia", or should i be safe in real-life?

You really won't know until you build it , the ones that
didn't sim correctly for me DID oscillate when I built them..

try a very fast rise time (10e-9+) square wave at 10k
check for ringing... Lpt, OPs

BTW , 10pf cdom is a little low. Post .asc file

Attached is pix of 'right' gain plot and a "ringing" shot
with amp undercompensated..
OS

[Khron]

I had to "flip" the polarities of the whole front end, to be able to use some MPSA18's i have on hand, so the Cdom is the 330pF on top. That 10pF cap is something else; can't remember off the top of my head, but it's some "tweak" from Dr Self's book.

[ostripper]

It rings.. everywhere.(attached)

How did you check open loop gain ?? I see no sweep source.
I looked at the original in the self book and you changed
a lot of things..

It might work, but it also triagulated a 10KHZ SW ,so
It might not sound good, also it showed a HUGE gain
"spike" at 1MHZ so I'm 99% sure it would oscillate in real
life.

Is this you first amp ?? Have you thought of doing a simpler
one like the standard self amp.. it is much more tolerant
about device changes.

OS

[Khron]

I didn't check the open loop gain, but as far as i know, the "high end" of that should look the same, open loop or closed loop.

I did a sim with a 10KHz square wave (2Vpp), and i'm attaching the result.
V(n007) = VAS input
V(n015) = before output inductor
V(n016) = after output inductor

Indeed, there is a bit of ringing (around 2MHz), but it gets damped fairly quickly.

I know i changed a lot of things. I *DID* mention i turned the whole front end "upside down".

[Khron]

Here's a zoom in on the VAS input trace.

[fjr]

C13 >> 10...100pF

[ostripper]

Indeed, there is a bit of ringing (around 2MHz), but it gets damped fairly quickly.

But when you build it and drive a real reactive load , that
is where things get interesting.

The "change" that made the most difference is subbing
the BD's for the MJE340/350's. For a current source or
a vbias generator this is not as critical , but for the
VAS itself , a world of differance (Cob/Hfe).

"Flipping it around" makes no difference at all as far as
stability and/ or compensation.
OS

[Khron]

Ok then, duly noted. It's 1:30am here now, but i'll try to find some decent MJE340/350 tomorrow and take another shot at some sims.

The assistance and guidance is appreciated, OS

[Rafael L]

Hello

You must remove input subsonic filter to make measurements, the subsonic filter change the curve of the graph but does not cause instability (not in the feedback link)

I have simulated the Low Tim seems strangely unstable >180 in 0dB (without filters)

The circuit.asc LTC can be downloaded site :

http://users.ece.gatech.edu/~mleach/lowtim/