[time-nuts] Re: HP Z3801A project update

[paul swed]

Ed thats quite a story and good to read.
But a bit hard to follow to actually answer your question that appears not
to need a answer now.
The diode switch makes sense. But I think you need another diode to stop
the converter from seeing the large filter cap. A thought at least.
Switching converters can become unstable when a large capacitor is added.
My 2 cents.
The other thing I have done is use an actual relay for isolation and a
holdup cap after the relay. Pretty simple and very safe. Relay dropouts
about 250 ms or less.
Regards
Paul

On Wed, Jan 19, 2022 at 5:02 PM ed breya <eb at telight.com> wrote:

> Since back in October, I've been working on getting the Z3801A in a
> 5065A carcass project closer to completion. I mostly finished the clock
> control board and other details, and moved on to the power system. I
> decided to completely redo a lot of the mechanical arrangement, wiring,
> and the line powered section with nicer layout and all the facilities
> for battery backup.
>
> A couple weeks ago, I had a major setback. I was really on a roll,
> having wrapped up the external battery control section, and ready to
> start on the internal battery interface and charging system. I had
> already built in the facilities for external DC input years ago, so it
> was mostly ready to go, once the controls were set up. I was doing some
> checkout on external powering, when the DC-DC converter in the booster
> crapped out - specifically, during hand-off tests between AC powered and
> external DC powered modes. I had incorporated all sorts of protections
> against faults, but apparently not enough, or I missed something that
> could happen in certain conditions. After inspection and rethinking, I
> still can't determine for sure how the converter could have blown out,
> but I have some theories.
>
> First, I should mention that this Z3801A is of the "BTS" type, which
> runs on 20-30 VDC. This is handy for battery backing operation. The
> internal battery system is two 12 V, 12 A-hr SLAs in series. When
> backing up, it is used directly, without any extra conversion loss, just
> the forward drop of the blocking diode.
>
> The AC power system uses the original 5065A transformer and choke, plus
> a bunch of other stuff to make everything work. The main output is
> called "Vs," and is the central point for all the power, running from
> about 21-30 VDC, depending on line voltage and battery modes and
> conditions. Everything goes into Vs. Its blocking diode is the rectifier
> bridge back at the transformer output.
>
> The external source was intended to be 12 V nominal (11-15 range), like
> a car battery. For this mode, I had stacked a DC-DC converter output on
> top of the supply voltage, with all sorts of diodes to prevent (so I
> thought) damage under all conditions. The regulated output of the
> converter was isolated 12 V, and the input could run on anything from 9
> to 36 V. It was really nice, simply adding 12 V to the external supply
> voltage, giving around 23-27 V, less the blocking diode drop.
>
> Anyway, it ruined my day when it crapped out. I had a pretty good deal
> going, with exactly three of these identical DC-DC converters - two
> deployed, and one spare. The other converter runs from Vs, to make the
> regulated 12 V wrt chassis ground, that runs all the peripheral
> circuitry. The common line for all the power side circuitry is separate
> from the chassis (but clamped by diodes), to better control any ground
> loops.
>
> So, pondering the situation, I could have replaced the converter, and
> scrutinized everything thoroughly and added better protection. Or, I
> could have just dropped the feature for now, to be figured out some time
> in the future (likely never). I chose a third option, to build a fresh
> boost converter into it, where I know and control exactly what's going
> on inside, rather than guessing the limits of the not all that well
> specified last DC-DC converter module.
>
> I'm happy to report that I finally managed to squeeze everything into
> the very limited space, and it's up and running fine so far. The new
> deal also opens up other options for improving and simplifying operation.
>
> The converter is based on the LT1070, made civilized and protected by
> various circuitry. I'll have more to say next time about what's in
> there, and details of how all the stuff plays together.
>
> Ed
>
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com -- To unsubscribe send
> an email to time-nuts-leave at lists.febo.com
> To unsubscribe, go to and follow the instructions there.
>