> -----Original Message-----
> From: Brent Hilpert via cctalk <cctalk@classiccmp.org>
> Sent: Monday, May 1, 2023 11:23 PM
> To: General Discussion: On-Topic and Off-Topic Posts
> <cctalk@classiccmp.org>
> Cc: Brent Hilpert <bhilp...@shaw.ca>
> Subject: [cctalk] Re: Rainbow H7842 PSU Fault
>
> On 2023-May-01, at 2:25 AM, Peter Coghlan via cctalk wrote:
> > It seems a bit odd that a power supply from someone like DEC of that
> > era would be designed to depend so critically on the absolute value of
> > a rail used for startup purposes.
>
> Further to Peter's point above, the 1988 NatSemi databook - which is to
say,
> from the era of this power supply - specs the 7812 output min-max to be
> 11.4 to 12.6V (+/-5%). Your measured Vstart=12.4V is well within this.
> Looking at the schematic, nothing stands out where the distinction from 12
or
> 12.1 would matter.
>
> You still haven't reported the IC power pin connections. If the neg-supply
> pins are supplied by -12 rather than GND, it could explain the odd voltage
> seen on the E3d +input.
Sorry, I thought I had done this. The two LM339 comparator power pins are
connected to Vstart and GND.
The reason for the odd output of E3d seems to be that the positive input is
actually higher than the negative one, but only by a small amount, and that
small amount would seem to correspond to the slightly higher value of
Vstart, because the negative input is the output of a voltage divider from
Vstart to ISense -12V+, which is GND.
>
> There are 3 explicit components in the design which provide -12V at
startup.
> They didn't throw those components in there just to fill up board space
and
> look pretty. Why would you expect the control circuit to be testable for
valid
> startup state when you haven't provided the startup environment?
Not quite sure what you mean here. I had advice from a friend to bench test
the control module by providing 14V to the input of the 7812. On the good
PSU I can see the PWM operate, on the bad one the PWM is shutdown.
Regards
Rob
