Re: Win11 ??

DANG!! NOT!!

Glitch just happened at minute 84.

> On Sep 3, 2025, at 17:42, David McClain (as dbm at refined-audiometrics dot com) <[email protected]> wrote:
> 
> Full Circle…
> 
> I looked carefully at the 48 kHz recordings. The glitch shows really prominently in the Left channel, and there is no corresponding glitch in the Right channel IRIG-B audio. Nada!
> 
> The Left channel shows what appears to be an insertion of 6 ms of garbage, followed by a phase discontinuous resumption of the expected signals. How??
> 
> All devices in the system are being driven by a common reference clock - the GPSD-OCXO, using its 10 MHz reference signal, split out into a bank of instrument amplifiers for deep isolation among the clients.
> 
> The clients are:
>   - The radio receiver, IC-7700
>   - The reference carrier function generator (Agilent or Siglent)
>   - The 48 kHz TTL Word Clock being fed to the ADC (a MOTU 828 mkIII)
> 
> The MOTU ADC is being driven by its WORD Clock (indirectly from the GPSD-OCXO).
> 
> The glitching happens whether or not the GPSD-OCXO is locked to GPS. 
> 
> So I tried reversing the L/R ADC channels - feed the radio signal to the right channel, and the IRIG-B to the left channel. So far, I have not seen any glitching. But it is still early in the trial.
> 
> But what I’m beginning to think is that there may be a problem in the Window/11 MOTU driver. How else can one insert an extra 6 ms of junk into a recording in the Left channel, followed by a phase discontinuous resumption of the expected signal - all while leaving the right channel perfectly intact??
> 
> You would have to believe that the two signal generators - designed and built more than a decade apart, and in two vastly different engineering shops (one in USA, the other in China), would both behave nearly identically on some small shift in the reference clock being fed to them. Them both seeing a glitch in the reference oscillator, then restarting from phase zero after a 6 ms recovery period. Both of these generators behaving like that???
> 
> I find that very hard to believe. But I can believe that a glitch in buffer filling might have left-over garbage for some reason (think wild C pointers in Windows/Driver) and some non-zero starting offset into a buffer. This might happen periodically for some, as yet, unknown reason. And it would only affect one, or a few other, channels, but *not* channel Analog 8 where the IRIG-B had previously been fed.
> 
> Any glitching of the IRIG-B signal will now be most likely ignored by the very forgiving demodulator that I built in Lisp.
> 
> We have a few minutes left in the 76 min period. Maybe the glitching will continue with these input channels reversed. But I’m beginning to believe not…
> 
> [ I am seeing that the injected reference carriers, from the DSB modulated function generators, shows a sensitivity to whether or not I’m in the room nearby. When I’m over there, the demodulated reference carriers shows a higher level of variation from one measurement period to the next. And when I’m out of the room, I see the monitor trace settle down to about half the variance. Just an interesting aside…]
> 
> So it really is looking like a Win11 problem.
> 
> [ I do see that M$ fixed the problem of Explorer (the Win11 display manager) having unbounded memory consumption. It is now well contained from the outset. ]
> 
> And we have just passed 80 min of glitch-free data acquisition !! Sure looks like a Win11/Driver issue to me...
> 
> - DM
> 
> 
>> On Sep 3, 2025, at 15:05, Christopher Stacy <[email protected]> wrote:
>> 
>> On 9/3/25 5:08 PM, David McClain (as dbm at refined-audiometrics dot com) wrote:
>>> I think we can well rule out any Lisp involvement in the glitching. So I can take this offline now.
>> It was a fun ride.
>> 
>> 
>> 
>