Deutsch   English   Français   Italiano  
<vrfeqi$1o9mh$1@dont-email.me>

View for Bookmarking (what is this?)
Look up another Usenet article

Path: ...!weretis.net!feeder9.news.weretis.net!news.quux.org!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!eternal-september.org!.POSTED!not-for-mail
From: Martin Brown <'''newspam'''@nonad.co.uk>
Newsgroups: sci.electronics.design
Subject: Re: Too much time on their hands!
Date: Wed, 19 Mar 2025 22:00:50 +0000
Organization: A noiseless patient Spider
Lines: 59
Message-ID: <vrfeqi$1o9mh$1@dont-email.me>
References: <vr94br$42no$1@dont-email.me>
 <6vdgtjhdeni42cbj5jipqc4anb44si5dms@4ax.com> <vrbd6g$25gno$1@dont-email.me>
 <bb2jtjp8h5bdslkv2jr3fbdrdejjumj23n@4ax.com> <vre3id$is1i$1@dont-email.me>
 <lhkltj9ncpfdl6lemab5677qo3a21tpbhi@4ax.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Wed, 19 Mar 2025 23:00:51 +0100 (CET)
Injection-Info: dont-email.me; posting-host="ba1cf95e20e6e96fb8e11a71d62d6e91";
	logging-data="1844945"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+OyTwdbZFv18p80h69XkRke6tCwB/a2IUC9J4zmLV13Q=="
User-Agent: Mozilla Thunderbird
Cancel-Lock: sha1:cb+4z6Aqtpq+JJwfoQvq2qPeSn8=
In-Reply-To: <lhkltj9ncpfdl6lemab5677qo3a21tpbhi@4ax.com>
Content-Language: en-GB
Bytes: 3743

On 19/03/2025 14:32, john larkin wrote:
> On Wed, 19 Mar 2025 09:42:37 +0000, Martin Brown
> <'''newspam'''@nonad.co.uk> wrote:
> 
>> On 18/03/2025 15:03, john larkin wrote:
>>> On Tue, 18 Mar 2025 09:08:32 +0000, Martin Brown
>>> <'''newspam'''@nonad.co.uk> wrote:
>>>
>>>> Unless you do a lot of video editing or 3D rendering the GPU built into
>>>> the modern Intel chips is entirely adequate for 2D business graphics.
>>>
>>> I wish they would help with Spice. Yesterday we were running a pretty
>>> simple power supply sim at around 100 us/s. It takes many minutes to
>>> settle out, and it's hard to learn with such delayed feedback.
>>
>> Have you tried running two instances of Spice at the same time?
>>
>> On a suitably beefy machine with plenty of ram it might be possible to
>> run two different sets of parameters at the same time on the performance
>> cores without saturating memory or disk IO bandwidth.
>>
>> It will depend critically on how big the matrix problem gets but for
>> some smaller problems it might possibly be an option.
> 
> Two instances would be confusing. One use of Spice is to train one's
> instincts and iterate a design.

You could try two alternative values for one parameter at the same time.

> At times yesterday, the power supply sim was running at picoseconds
> per second. LT Spice allows one to set the max time step, but not the
> minimum time step.

That is usually an indication that there is something stiff about the 
differential equations being solved and that the algorithm has halved 
the time step many times in a desperate attempt to control the error 
budget. I once had a plot job from solving a very stiff set of equations 
cancelled by the operator "because the red pen began to work loose". The 
line segments approaching zero got very very short indeed!

> The Gear solver and some relaxed tolerances seem to be better for this
> case.
> 
> In one recent case the sim kept stalling. I added a 1K resistor off to
> the side, one end grounded and the other end connected to nothing.
> That fixed things.

That is odd. I can imagine adding a 1M resistor between some pair of 
nodes might take the edge off it.

> Inductors, especially coupled inductors, are quirky. A little ESR
> often helps, but that may be random, like the 1K resistor.

Resonant tank circuits with high Q can sometimes cause trouble.


-- 
Martin Brown