Tom Van Baak writes:

+1 on the power-polery.

I really wish somebody would do a version of this with power-pole:

https://www.mini-box.com/Y-PWR-Hot-Swap-Load-Sharing-Controller

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

Bill, most modern double-oven OCXO's (including I think the BVA under
discussion) are under 5W average consumption after warmed up.

Most UPS's are rated for near-peak-power output (most of a kW or more) for
5 or 10 minutes to give time to shut computers down gracefully and are not
sized well for keeping a single 5W device powered up for hours (days?
depends on your requirement) although you can use them that way with a
substantial loss of efficiency.

Figure out your needed hang-time, and size 24V gel-cell or AGM battery to
keep the OCXO's ovens warm and oscillator humming. Diode-or'ing works well
and if you need to account for a 24V gel cell really being 28V under float
charge, use multiple series diodes in that leg to get voltage drop to it
being under 24V. The spec about frequency shifting under Voltage change is
not really relevant if you aren't using the output during the power outage.
You just want to keep the oscillator ticking and ovens warm to prevent
having to re-age after an outage.

Tim N3QE

On Fri, Sep 25, 2020 at 3:38 AM Bill Notfaded notfaded1@gmail.com wrote:

Hi

How often do you loose power? If it’s once a year for < 1 hour …. how big an
issue is this to you? Do you need to do precision measurements every day?
Is restarting a run once a year a major issue? Option one is to simply restart
after an outage. Option two is to protect against it.

To protect a full bench setup, some sort of UPS is going to be needed. With counters
and all the other “stuff” there really is no other practical approach. That immediately
gets to how much power for how long. A bench that pulls 800W and a typical
outage of 4 hours is going to mean a pretty big battery pack on the UPS.

If you go shopping and expect a typical outage of 4 hours, shop for something rated
at 6 hours (at your power level). If the bench appears to pull 800W go for something

in both numbers.

When you get into run time dimensioned in hours, things get nutty expensive. A generator
(natural gas or propane powered) with an auto start is often the lower cost approach
compared to giant batteries. You still need the UPS, but it only has to run for a few
minutes while the generator fires up.

======

In the house before this one, the generator was the way to go. That neighborhood
was deep in the woods. If a big storm blew in off the ocean and a lot of trees went down,
our little cluster of houses was not very high on the list for repair. We could be out
for a week or more every couple years. An outage of > 2 hours was pretty normal
several times a year.

=====

When we first moved in here, the main feed line into town had “issues”. Just about
any time of year, a modest amount of wind could shut down the feed line. It was
reasonable to expect multiple outages a month. In the windy months multiple
outages a day did happen. As a result, we had a lot of UPS’s scattered around
the house. The outages were mostly in the 5 seconds to two minutes range. Very
normal UPS units did the trick.

A convenient tornado came through a few years ago. It took out about 1/3 of the
feed line and the sub-station feeding the line. As a result they replaced the entire
feed line. Since then, there has been very little need for the UPS fleet.

One aspect of keeping a heavily used setup running is wear and tear on the batteries.
They don’t last as long as you might hope they would, at least not to anywhere near
the full capacity rating. Going to LIthium’s instead of sealed lead acid’s might help
this, but at a pretty steep price up front. Are they cheaper long term? We’ll know in
20 years …. :)

Bob

Hi,

On 2020-09-24 23:47, Tom Van Baak wrote:

That's how we do it in Telecom, but on the 48V level. I managed to drive
my company into do it with 48V all the way to the various boards,
because that way the protection switching out there handled multiple
faults. Also, for some reason there is this line of DCDC converters from
48V to about anything. We kept doing that since, even if the diodes now
been replaced with MOSFETs to lower losses.

If you look into say the 5065A that's how it's done there too.

As for power-pole, those are great connectors, but I need to keep 12V,
24V and 48V in the lab, so I need to get the different color codes not
to interchange them. You usually react when you see a yellow-black
trying to mate with a red-black. So, I recommend folks to do the same.
Once one got started with the Anderson PowerPole, it becomes more and
more a solution.

But yeah, thanks for reminding me that I need to progress on the
power-pole and power supply projects. I'll do that after the PiDP-11
project.

Cheers,
Magnus

Pondering the backup power issues for my BVA a bit more,  I am thinking a dedicated DC battery bank (maybe 5 or 6 nominal 6 volt batteries in series) powering a suitable linear regulator circuit is probably the direction I will go in.  I suspect there are more elegant and or simpler approaches but I think from my perspective as a hobbyist this is probably the best direction for me.

I like the idea of using a diode arrangement to facilitate changing the power source for the BVA.  I expect I will also add some form of over voltage protection as well.

I need to ponder the likely voltage drops in the voltage regulator and diodes along with the voltages the batteries will provide as they discharge under load.

It seems I have another winter project.

Thanks all for the suggestions.

Mark Spencer
mark@alignedsolutions.com
604 762 4099

Mark Spencer writes:

If you are after the low noise, be weary of "integrated" lithium batteries of 12V and higher, many of them have a built in buck-converter to take the real battery voltage down to 12V and provide over-current protection.

If you go lead-acid, don't forget rule #1:

Your fuses can never be too numerous or too close to the battery.

Lead-Acid batteries float-charge at 13.8V[1] so two 12V in series you gets you 27.6V.

Subtract the drop over a series diode and add a linear voltage regulator and you have nice and quiet 24V.

(A lot of professional fire and burgler-alarms run on such configs, and the hardware is actually pretty nice, being a nieche and high-margin business, look out for it in scrap-heaps.)

When the charger looses power, you will drop below 24V after some time (1-60 minutes depending on battery capacity), but if the rest of your lab is dark anyway, that can still keep the BVA warm until power comes again.

If on the other hand you want to disconnect the charger to make low-noise measurements running from battery, you will need more voltage headroom.

I would not go with five 6V batteries.  6V batteries are almost universally lower quality than 12V, probably because nobody "serious" uses them, so manufacturers do not get yelled about on quality.

Three 12V blocks will put you at 41.4V when float-charging, which is a lot to swallow for a linear regulator, both in terms of voltage and power.

Poul-Henning

[1] Tons of footnotes on this but find and trust the manufacturer is a very good starting point, unless you want yet another hobby.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

Thanks Paul.

The bit I am struggling with re using a 24 volt battery system in this application is what happens when AC power is removed and the terminal voltage of the battery  starts to fall.

Between voltage drop in the regulator, the voltage drop in any extra diodes etc, a nominal 24 volt lead acid battery doesn't seem enough to me if the goal is to provide the BVA with a stable 24 volt power source for a lengthy period of time.  I acknowledge that one could perhaps decide to run the BVA from a some what lower voltage and still be within the voltage spec, but I don't want to do that and in my view that probably won't really solve the issue if one wants to fully use the capacity of typical lead acid batteries.  (I expect a typical 24 volt lead acid battery system could provide useable power down to 22 or perhaps even 21 volts ?).  The voltage regulator and diodes will add voltage drop.

Maybe I am missing something ? (Maybe fancy regulator schemes that automatically bypass themselves when the input voltage goes below a preset level ? But from my hobbyist perspective that seems more complicated than simply using a higher voltage battery system.)

Thanks

Mark Spencer
mark@alignedsolutions.com
604 762 4099

Hi

Properly charging batteries is a bit complicated. Why does “properly” matter? You want them to be fully
charged, but not damage them in the process. That generally involves some sort of variable voltage
for the charging source. If you decide to go with Lithium based parts, you can get BMS (battery management)
IC’s that can help with the charging and the balancing ( = you want them all at equal voltages) of the cells.

If you look at older gear, this sort of thing didn’t get a lot of attention. Most of the backup battery setups
worked poorly as a result.

======

One “cute” alternative is to do a battery + boost converter ( = switcher) instead. If the device only operates
during a power outage, noise is not a big deal. Everything that is attached to the OCXO output is dead anyway ….

The advantage of a switcher is obviously efficiency. A fully charged “12V” LiFePo4 stack could easily be at 14.8V.
Fully discharged it might get down to 10.4V. Allowing for that 1.4:1 output ratio with a linear regulator means a
lot of (expensive) battery energy goes up in heat ….

Bob

Hi

If you run a straight linear regulator setup, then you do indeed have a battery voltage
that is (nominally) much higher than the “target” voltage. If you run a diode voting input,
then the “normal” voltage will be even higher. You can easily get into the 35 to 40V range
doing this.

Bob