My battery backup setup

[sdb7802]

1.) I ran input voltage tests on 3 different A10 AutoSet units. The maximum voltage I was willing to subject the machines to was 30 volts. With the blower running, all of the units function normally as the input voltage is varied between 20.5 and 30 volts. Below 20.5 volts, the machines shutdown completely and restart as the voltage is increased to 22.3 volts. The machines were tested while configured with a minumum load (4cm pressure, humidifier off, tubing heater off) to a maximum load (20 cm pressure, max humidification, max tubing temp). Fixed pressure and APAP modes were tested. Flow restriction was also varied.

From an unpowered state, the machines require at least 22.3 volts to power up to the standby (blower off) mode.

Input Voltage Range (to power up): 22.3V to 30V
Input Voltage Range (blower off/on): 20.5V to 30V

Machine variation: -3%

I'll look at the battery manual to see if the batteries can meet these specs.

2.) I think a 10A fast-blow fuse will work fine.

3.) I would change the 3.3 volt regulator to a LM2936Z-3.3/NOPB. This part has a 40V breakdown voltage which provides a greater safety margin than the 28V part. It is available in a TO-92 case which is easier to work with. Need to change the caps for this new part. I'll look into that.

4.) I agree with Mongo about removing the comparator circuit from the diagram. I'd remove all the stuff about the resistor values, the patent, the 90W converter, the ferrite bead, and anything but the part descriptions and labeling. Just, what it is, what it does, and how it goes together. I'd also re-arrange your diagram to flow left-to-right (batteries on the left); it's a little easier to follow that way.

5.) I'd prefer not to be credited in the diagram. I've done this stuff all of my adult life and now I just enjoy helping others.

6.) I like Mongo's idea about the schottky diode bridge for protection. I'll round up some part numbers.

[justMongo]

I think a lead-acid, 6 cell battery essentially at full discharge at 11 Volts under no-load. So, 22 Volts would be a minimum seen by 12 cells. (Depends a bit on temperature.)

[verbatim1]

Couple of points. Inside the Blower unit electronics -- you have a comparator. It's output should not connect to the power (central pin) like that. The power (central pin) connects to Vin. The comparator output supplies a logic signal that shuts down the blower if the wrong power supply is indicated. Best just remove the comparator circuit from the diagram.

OK. I'll remove the comparitor circuit from the diagram.
I see Steve has input that he suggests the diagram change a lot, so, it will be part of the major rewrite.

If you really want to get fancy, you idiot proof where the batteries connect by using a 4 schottky diode bridge such that any way the batteries are connected; the positive side is always steered to the positive leg of the circuit (same for negative leg.)

That idea to optionally idiot proof the output seems perfectly reasonable.

1. I looked up Schottkey Diodes, and they seem to have, according to Wikipedia (which may not be a viable reference) "a low forward voltage drop (e.g., in the range of 0.15 V to 0.46 V) and a very fast switching action". https://en.wikipedia.org/wiki/Schottky_diode

We don't "need" either of those two features, although they're both good things. Is there another reason you suggested a "Schottkey" diode instead of "normal" diode?

2. Since the nominal current capacity of the ResMed power supply is 3.75 amps, do you think we might want to specify diodes that handle double that, at 6 or 7 amps?

Power diodes might be hard to come by and perhaps expensive. I don't know. My only experience with diodes are with the tiny see-through orange-tinted glass ones and the small black ones with a gray stripe. Those diodes would blow up with this kind of current.


3. While I understand the concept of idiot-proofing the design in case someone hooks the battery to the cable backward, I don't understand how the blocking diodes would be hooked up in a "bridge" configuration (which I have always seen only for AC situations).

Since we're dealing 100% with DC current, I'm wondering why we can't just put a single (high power) diode in the positive lead pointing in one direction, and another similar diode in the negative lead pointing in the other direction, such that both diodes would stop the current if the batteries are hooked up backward to the cable?

Googling for "blocking diodes", the best I can find is this diagram, but it uses only a single bridge:


http://www.edn.com/design/analog/4368527...ltage-drop

[verbatim1]

1.) I ran input voltage tests on 3 different A10 AutoSet units.

Thanks Steve, for risking your expensive equipment. I was worried that you were going to blow them, since excess current capacity at the power supply is nothing, but excess voltage is, as you know, the danger.

The maximum voltage I was willing to subject the machines to was 30 volts.

I would think 30VDC should be plenty, bearing in mind the Microchip voltage regulator is also rated at 30 volts maximum (28VCD nominal) and keeping in mind that a 12VDC battery is something like 13.8VDC nominally (times two would be around 28 volts).

With the blower running, all of the units function normally as the input voltage is varied between 20.5 and 30 volts.

Whew! Thanks for risking your machines!

Half of 20.5 is 10.25, which is within the norm of a heavily loaded 12VDC battery, but, at the current draws we're subjecting it to (less than 3.75 amps), I would consider a 10.25VCD battery pretty run down.

So, the range of 20.5 to 30 seems reasonable to me, given that corresponds to a 10.25VDC to 15VDC set of batteries.

Would it be too hard to design in a 24VDC voltage regulator? I' would guess the power requirement is the problem (with 3.75Amps of current being nominal).

Below 20.5 volts, the machines shutdown completely and restart as the voltage is increased to 22.3 volts.

This is interesting that the machines exhibited a sort of hysteresis, shutting down at a lower voltage than starting back up.

I guess that would mean that we probably wouldn't want the batteries to drop below 11 volts each for a 22 volt minimum goal (assuming the batteries are evenly matched in brand, model, age, charge, and condition to start with).

The machines were tested while configured with a minumum load (4cm pressure, humidifier off, tubing heater off) to a maximum load (20 cm pressure, max humidification, max tubing temp). Fixed pressure and APAP modes were tested. Flow restriction was also varied.

Wow. That was thorough.

From an unpowered state, the machines require at least 22.3 volts to power up to the standby (blower off) mode.

Input Voltage Range (to power up): 22.3V to 30V
Input Voltage Range (blower off/on): 20.5V to 30V

My take is that the sweet spot is something like 22 to 28 VDC.

While I can easily see the danger to the machine if the voltage is too high, is there a danger when the voltage is too low?

[verbatim1]

2.) I think a 10A fast-blow fuse will work fine.

Thanks for confirming.
We don't want metal fatigue to blow too many fuses, and, we want the fuse to blow at about double the amperage, but we also want to get a standard sized fuse, so, 10A seems like it's reasonable then.

3.) I would change the 3.3 volt regulator to a LM2936Z-3.3/NOPB. This part has a 40V breakdown voltage which provides a greater safety margin than the 28V part. It is available in a TO-92 case which is easier to work with. Need to change the caps for this new part. I'll look into that.

Thanks for this suggestion of a voltage regulator with a higher breakdown voltage. I remember the Microchip diagram said 30VDC was the maximum, but, there's no harm in adding a safety factor.

When the circuit is rewritten, I'll include that regulator instead.

I think we pay for power, so, luckily, the voltage regulator for the 3.3VDC sense line doesn't have to handle more than a few milliamps, so, we have a lot of leeway on that voltage regulator.

4.) I agree with Mongo about removing the comparator circuit from the diagram.

I agree with both of you.

I'd remove all the stuff about the resistor values, the patent, the 90W converter, the ferrite bead, and anything but the part descriptions and labeling.

That makes sense also. I was trying to be "all inclusive", but, it's not meaningful, especially since I doubt even ResMed has the 30W and 60W and mains power supplies. They only seem to have the 90W power supply, so, the only resistor value that matters is the 2.7K ohm resistor.

We can probably draw it more like a schematic anyway, which would be easier than the way I've been doing it. I don't have circuit design software, but, I can draw a schematic with a rudimentary "mspaint-like" program.

Just, what it is, what it does, and how it goes together. I'd also re-arrange your diagram to flow left-to-right (batteries on the left); it's a little easier to follow that way.

Yeah. I know. Power on the left, load on the right. The diagram evolved in this thread, so, I knew all along it was backward from conventions. I agree with you.

5.) I'd prefer not to be credited in the diagram. I've done this stuff all of my adult life and now I just enjoy helping others.

We can remove all the credits then, as I don't need the credit either.

6.) I like Mongo's idea about the schottky diode bridge for protection. I'll round up some part numbers.

Thanks for volunteering for that. I'm familiar with a wheatstone bridge and a full-wave rectifier bridge, but, I'm not sure how this schottky circuit would work (see my prior post).

Also I'm used to diodes that are at best 1 amp, so, these diodes will need to be rather large.

[justMongo]

We need the lower Schottky forward Voltage drop. The blower unit will shutdown sooner in the discharge cycle if Silicon diodes are used.

The advantage of a bridge is to permit operation no matter which way the battery is connected. To simply protect from reversed battery polarity, one could use a single diode. You wouldn't need one in both legs -- just the positive leg.

[sdb7802]

10 Amp Schottky diodes cost about $0.75 each.

I'm going to use the two 12V 8AH batteries in my fairly new UPS to test the circuit. I've ordered some Schottky diodes, so hopefully I can run the test by the end of the week.

If the 24 volt battery circuit doesn't work; I would recommend using a single 12 volt battery and the ResMed converter instead of building a 24 volt buck/boost regulator.

Verbatim, I think your drawing will work well for visualizing how to build the circuit. I can draw the regular electronic schematic to augment that.

[verbatim1]

We need the lower Schottky forward Voltage drop. The blower unit will shutdown sooner in the discharge cycle if Silicon diodes are used.

Thanks Mongo for explaining the reason for the Schotkey diodes in a bridge configuration.

Here's a basic circuit I just hacked out with unlabeled components without the bridge. (I apologize that I'm using Paint.NET on Windows, which doesn't seem to have a snap-to-grid feature.)

The advantage of a bridge is to permit operation no matter which way the battery is connected. To simply protect from reversed battery polarity, one could use a single diode. You wouldn't need one in both legs -- just the positive leg.

I will try to figure out the configuration of the 8 diodes for the two bridges, but, I don't know what direction they go at the moment, so, I haven't inserted it into the diagram.

[surferdude2]

Only one bridge needed. Regardless of input polarity, Vout will always be positive.

[verbatim1]

10 Amp Schottky diodes cost about $0.75 each.

Wow. OK. That's super cheap. I didn't know high-power diodes were that readily available. That means that 8 of them (for the two bridges) should be no problem (especially since they're optional).

I just don't know the configuration. Is it as simple as this?
I'm not sure where it goes though, so I didn't put it in the schematic.

I'm going to use the two 12V 8AH batteries in my fairly new UPS to test the circuit. I've ordered some Schottky diodes, so hopefully I can run the test by the end of the week.

That would be great.
On average, the ResMed A10 should use about an amp at 24VDC, right?
So, that would last you all night.

If the 24 volt battery circuit doesn't work; I would recommend using a single 12 volt battery and the ResMed converter instead of building a 24 volt buck/boost regulator.

At some point, using the $100 (in toto) walled-garden approach dictated by ResMed marketing would be the easiest thing, I agree.

Still, this helps to UNDERSTAND what ResMed did, which in and of itself, is a worthwhile endeavor.

Verbatim, I think your drawing will work well for visualizing how to build the circuit. I can draw the regular electronic schematic to augment that.

I think the most important concept was when you drew what the power connector looked like, and what the connections were to the voltage regulator.

I apologize that I'm using Paint.NET on Windows to do the schematic drawing. I will google for a freeware schematic program (if they exist), which would make things far simpler to draw (and better looking).