[time-nuts] Parallel voltage regulators

[Don Collie]

----- Original Message ----- 
From: "Bruce Griffiths" <bruce.griffiths at xtra.co.nz>
To: "Discussion of precise time and frequency measurement" 
<time-nuts at febo.com>
Sent: Friday, October 26, 2007 12:36 PM
Subject: Re: [time-nuts] Parallel voltage regulators


> ); SAEximRunCond expanded to false
> Errors-To: time-nuts-bounces+donmer=woosh.co.nz at febo.com RETRY
>
> Magnus Danielson wrote:
>> From: Bruce Griffiths <bruce.griffiths at xtra.co.nz>
>> Subject: Re: [time-nuts] Parallel voltage regulators
>> Date: Thu, 25 Oct 2007 17:42:51 +1300
>> Message-ID: <47201ECB.6050702 at xtra.co.nz>
>>
>>
>>> Don Collie wrote:
>>>
>>>> Hi Tom,
>>>>     If you really want to regulate the oven`s supply voltage, my 
>>>> National
>>>> Voltage Regulator handbook shows that the LM317T will supply over 2 
>>>> Amps,
>>>> with an input/output differential of between 5, and 12.5 Volts. A 
>>>> single one
>>>> of these should do the job OK.
>>>> Cheers!,.................................Don Collie jnr.
>>>>
>>>>
>>>>
>>> Never rely on typical specs always use the minimum spec which is 1.5A
>>> not quite enough.
>>>
>>
>> You want design-margin. Some of that toll will be in less than optimum 
>> heating,
>> some will be in less heating in the first place (compared to upper limit) 
>> and
>> for a power regulating aspect, headroom allows better regulations.
>>
>> In one design we had to parallel the regulators since the regulator the
>> designer put in just barely was able to regulate the CPU core voltage.
>> It worked, but at just rebooted at some vauge point an the memory tests.
>> What actually happend was that as soon as it started to actually do 
>> anything,
>> the regulator was running at its limit and output voltage dropped as the
>> current was rising and the voltage supervision pulled the RESET.
>>
>> That's what you get from reading the typical reading on the CPU current 
>> and
>> match that with the maximum rating of the power regulator. A no margin 
>> design.
>> That designer had a few more flaws which was creeping around in that 
>> design,
>> but let's not bring that can of worms open here. :)
>>
>> The 5A LM338 will be just fine. Infact, you can pull 8A out of it under
>> certain conditions.
>>
>> Cheers,
>> Magnus
>>
>>
>
> Hej Magnus
>
> The LM338 thermal design is also much easier (it has a much lower
> junction to case thermal resistance than an LM317) especially if the
> circuit is intended to operate over wide temperature (0 -40C or more)
> and mains voltage ranges (+-10% or more).
>
> The alleged problem with the high short circuit current is easily solved
> by using diodes with adequate current ratings in conjunction with a fuse
> to protect the transformer if it isnt rated to produce an 8A dc output.
>
> The startup current of the load (rubidium standard) may also vary with
> temperature and /or input voltage.
> Either  find the manufacturers specifications or allow adequate margins.
>
> Worst case design is desirable even for one off circuits especially if
> the circuit is published.
> When the design is  publicly available one is in effect transferring the
> production run problems associated with a marginal design to many
> individuals rather than a single factory or production line.
>
> Bruce
>I don`t think the higher current created when a fuse is used instead of 
>near-instantaneous current limiting is "alleged", but rather a real problem 
>that can cause damage further down the line. Fast acting current limiting 
>is preferable to all but the fastest fuses that are designed to protect 
>semiconductors. Current limiting plus thermal shutdown in the regulator
will protect both load, and regulator [and resovior capacitor, and diodes, 
and transformer] Commonly available fuses won`t give much protection to the 
load - especially the delay types often necessary with large filter 
capacitors.
    A precision, proven, high performance, low noise regulator like the 723 
using an external pass transistor [or preferably a darlington], to avoid 
chip heating, and a well bypassed reference would be a lovely solution.
Cheers!,............................................Don C.

>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to 
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.