开云体育

ctrl + shift + ? for shortcuts
© 2025 Groups.io
  1. Duncanampspsud
  2. Topics

Enhabcement

 

开云体育

Hi Duncan,

My PC has 32GB and I get OOM messages when I try to simulate over 80s delay.

Additionally, is updating the screen with calculation status slowing the final data?






Seems slightly lethargic given the hardware?
-- 
- Ian
 

On 12/12/2021 12:14 PM, Ian Eales wrote:
I get OOM messages when I try to simulate over 80s delay.
Seems slightly lethargic given the hardware?
For what circuit??

Using default circuits, even with extreme capacitor, on an older 8GB 3MHz Win7 PC, 99 Second simulation: no error and 7 or 8 seconds.

Also: what do you expect to change at 80 seconds?

Memory utilization rises from 3.38GB to 3.73GB. PSUD set for High accuracy.

I think more details of your circuit are needed; maybe the PC also.
 

On 12/12/2021 12:14 PM, Ian Eales wrote:
I get OOM messages when I try to simulate over 80s delay.
Sorry; attachment fell off. Showing 7 sec for 99 sec.

PSUD-OOM-Dec2021---42.gif

 

开云体育

On 12/12/2021 09:40, Paul Reid wrote:
On 12/12/2021 12:14 PM, Ian Eales wrote:
I get OOM messages when I try to simulate over 80s delay.
Seems slightly lethargic given the hardware?

For what circuit??



Also: what do you expect to change at 80 seconds?
Tubes to get hot, voltages to drop, currents to stabilize.


Memory utilization rises from 3.38GB to 3.73GB. PSUD set for High accuracy.
Back in the day when bytes mattered, we didn't make a humongous table for simulations, but used a looping algorithm to preserve memory. Since PSUD results are not time scrollable, after xx seconds should essentially be unlimited.

16bit legacy issue ???


I think more details of your circuit are needed; maybe the PC also.

...
 

Hi Ian,

This circuit is problematic in a significant number of places.

R1 - 10 milliohm resistor goes into 40uF capacitor with series resistance of 4.24 ohms and an inductor with resistance of 25 ohms. I'm really not getting what that 10 milliohm resistor is doing for you? All it will do is force the simulator to review the tiny RC constant of 10milliohm/40uF into smaller and smaller timesteps until it breaks or you run out of memory.

Similar with C2/C3/C4 and R2/R3 balancing them. Again, this will force the timesteps to be very small and consume memory. These are loops with potentially high currents that can only be balanced with microscopic time steps (long sim times and lots of memory). C2/C3/C4 can be replaced with one 600uF capacitor with ESR of the three resistances in parallel. R2/R3 can be removed - they are not contributing materially to the PSU output.

A power supply shouldn't take 80 seconds to stabilise, this will exacerbate the problem. As will running high accuracy with the aforementioned values that the simulator will struggle to balance.

Regards,
Duncan
 

开云体育

Hi Duncan,

The small resistances are the wire.

The circuit is as built in the amp. I know I can parallel up the parts and had done so in the initial run.
effective saves 10s in execution, gains 10s in delay.


I ran into OOM issues with much simpler circuits and similar delays.

The power supply does take a while to stabilize on a plain old VOM from a cold start.

AND PLEASE fix Save As... to the same folder as the orig and preserve the name to allow me to make a Rev change if I load from recent on the File menu ?

C:\Users\%user%\AppData\Roaming\psud\examples??? Never!!!!
- Ian
On 12/12/2021 12:01, Duncan Munro wrote:

Hi Ian,

This circuit is problematic in a significant number of places.

R1 - 10 milliohm resistor goes into 40uF capacitor with series resistance of 4.24 ohms and an inductor with resistance of 25 ohms. I'm really not getting what that 10 milliohm resistor is doing for you? All it will do is force the simulator to review the tiny RC constant of 10milliohm/40uF into smaller and smaller timesteps until it breaks or you run out of memory.

Similar with C2/C3/C4 and R2/R3 balancing them. Again, this will force the timesteps to be very small and consume memory. These are loops with potentially high currents that can only be balanced with microscopic time steps (long sim times and lots of memory). C2/C3/C4 can be replaced with one 600uF capacitor with ESR of the three resistances in parallel. R2/R3 can be removed - they are not contributing materially to the PSU output.

A power supply shouldn't take 80 seconds to stabilise, this will exacerbate the problem. As will running high accuracy with the aforementioned values that the simulator will struggle to balance.

Regards,
Duncan

1 - 6 of 6