I'll use a 1.25F supercapacitor just to be on the safe side. I will probably go with a common 27uF capacitor, to eliminate any issues.

In all seriousness, unless you have specific reasons for more capacitance, a 1nF cap is all that is required and has never given me any problems. In products I have designed there must be millions of these in the world now so my recommendation of 1nF is well qualified

 

I know voltage is temperature and load dependent. But you could set it to something conservative like 12V and it would probably discharge to less than 100% DoD. It may not be perfect but would probably be good enough.

Look at the battery performance curves in the data sheets and you might be disappointed, or maybe not. It is your call of course, I only wanted to make you aware of the issues so your decision is informed and you are not surprised and at the end of the day, you will know more about SLAB than the vast bulk of people using them who just think they know about these things

 

Look at the battery performance curves in the data sheets and you might be disappointed, or maybe not. It is your call of course, I only wanted to make you aware of the issues so your decision is informed and you are not surprised and at the end of the day, you will know more about SLAB than the vast bulk of people using them who just think they know about these things

I know that there are many downsides, but I think it still makes the most sense here, from a cost and practicality standpoint. I will make sure to adjust based on load and temperature, and hopefully it will last a long time.

In all seriousness, unless you have specific reasons for more capacitance, a 1nF cap is all that is required and has never given me any problems. In products I have designed there must be millions of these in the world now so my recommendation of 1nF is well qualified

The voltage may change suddenly due to load changes, etc, and I want the reference to be stable. Wouldn't a larger capacitance ensure that it has time to adjust? I know at a certain point it doesn't really matter, but isn't it better to be on the safe side, with a few extra uF? And also there may be parasitic capacities, which could lead to issues. A larger capacitance would mostly negate these effects, right?

 

I also want to get a battery that will have a decent lifetime when discharged to 30% SoC. I have looked at many 12V 35AH batteries, and they all have essentially the same DoD capacity graph. It generally looks something like this:

Given that it is unlikely that I'll find anything better, I will just go with the original UB12350s I was looking at. In less demanding applications, I can try 50% DoD, and hopefully get at least a few hundred cycles with a decent capacity. I know most of the amazon reviewers are not by any means experts, but 800, mostly positive reviews seems convincing enough to get these. I do not really want to pay more for the same specs.

Also, I will get this CCA wire to make the connections. I will probably use some nuts and bolts to connect them, too.

 

I know that there are many downsides, but I think it still makes the most sense here, from a cost and practicality standpoint. I will make sure to adjust based on load and temperature, and hopefully it will last a long time.
The voltage may change suddenly due to load changes, etc, and I want the reference to be stable. Wouldn't a larger capacitance ensure that it has time to adjust? I know at a certain point it doesn't really matter, but isn't it better to be on the safe side, with a few extra uF? And also there may be parasitic capacities, which could lead to issues. A larger capacitance would mostly negate these effects, right?

You understand the importance of limiting the DoD and the current and temperature dependencies involved so the rest is at your discretion and my opinion is of no relevance. If I have helped you understand the issues then I have achieved and that is gratifying and good.

One idea though, you might improve the DoD situation by going to a bigger battery, budgetary constraints not withstanding .

On the assumption that there is always enough cathode current available for the 431 to regulate (>1mA) and the load on the reference voltage the 431 is generating is either constant or varying at a rate far slower than the 431 can track then the 1nF cap serves only to quiet electrical noise. And to do that a small ceramic or plastic cap is better than a small value electrolytic. Although that statement should be qualified with conditions of the bandwidth of the noise and the troublesome frequency bandwidth (if any) for the load on the reference.

As you mentioned varying supply voltage I will say that a large shift in supply voltage may require a large modulation of the cathode current and the 431 will have a limit to how fast it can do that. I have never had to consider that situation to the extreme degree you rightly indicate is a possibility. Variations such as those I would normally have filtered out in a preceding stage because the reference is not the only circuit element sensitive to such fluctuations. Do you think you should do the same and for the same reason? Could fast and sharp dips and spikes effect other circuit functions as well?

As for the stability, the 1nF number is comfortably below the threshold that makes the 431 unstable so parasitic capacitance that adds to the 1nF is profoundly unlikely to add sufficiently to make the 431 unstable. The parasitic capacitance even on strip board would only be in 10's of pF at most. Longer traces may represent series inductance that will thwart the 1nF cap but to do that at any frequency the 431 is going to be fast enough to react to at all is also profoundly unlikely I think. The only sensitive node is the reference input of the 431 so if it is being used as a reference then stray coupling to other signals on that node could be an issue unless it is hard wired (by a 2mm trace between the pins of the part) to the cathode in which case the sensitivity is a mute point. Since the 431 is stable without any shunt capacitance at all then if the shunt capacitance happens to be not as close to the 431 as is desirable it still won't make it unstable. So that is my reasoning but it is your application so only you can know if any of this is actually relevant. If you go with an electrolytic for sound technical reasons rather than assumptions and personal preferences then I have succeeded. Sorry if that sounds pompous.

My purpose here is to put as many relevant facts as I know at your disposal so you can make a better informed decision. You don't always agree and that is absolutely fine. As far as I can see you are making good and informed decisions either way and that is gratifying whether you have agreed with me or not.

 

I also want to get a battery that will have a decent lifetime when discharged to 30% SoC. I have looked at many 12V 35AH batteries, and they all have essentially the same DoD capacity graph. It generally looks something like this:

Given that it is unlikely that I'll find anything better, I will just go with the original UB12350s I was looking at. In less demanding applications, I can try 50% DoD, and hopefully get at least a few hundred cycles with a decent capacity. I know most of the amazon reviewers are not by any means experts, but 800, mostly positive reviews seems convincing enough to get these. I do not really want to pay more for the same specs.

Also, I will get this CCA wire to make the connections. I will probably use some nuts and bolts to connect them, too.

I agree that you should get the cheapest of the set that have the same life cycle graphs, just beware that the life cycle graph is actually applicable to the battery you buy. Online selling does not appear to always comply with consumer protection laws You are probably already well aware of all that and I am 'preaching to the choir' so please ignore if that is the case.

The CCA cable looks very nice! Soft and flexible, expensive and requires crimping tools. If you don't have crimping tools or have not dealt with cable of this gauge before, don't be tempted to solder it. Buy, borrow or steal a hex crimper if you don't already have one. Hand operated at this gauge is still an option. The one I got years ago was not expensive either and handles 6mm2 to 50mm2 and they are probably even cheaper now.

The number of failed joints I have seen with cables of this sort of size because solder was involved is crazy. That professionals who theoretically handle wire of this size routinely and who have seen fit to incorporate solder in the joint and defend that decision even when confronted with the same joint failure multiple times just beggars belief. Sheesh!

 

I also want to get a battery that will have a decent lifetime when discharged to 30% SoC. I have looked at many 12V 35AH batteries, and they all have essentially the same DoD capacity graph. It generally looks something like this:

Given that it is unlikely that I'll find anything better, I will just go with the original UB12350s I was looking at. In less demanding applications, I can try 50% DoD, and hopefully get at least a few hundred cycles with a decent capacity. I know most of the amazon reviewers are not by any means experts, but 800, mostly positive reviews seems convincing enough to get these. I do not really want to pay more for the same specs.

Also, I will get this CCA wire to make the connections. I will probably use some nuts and bolts to connect them, too.

There is one other thing I have not mentioned (and it may be redundant anyway) which is that the cycle life graph is for fixed DoD but that DoD is not specified as being the DoD from the initial capacity (the capacity printed on the side of the battery) or the residual capacity (the capacity that remains as the battery gets weaker.

I'm guessing that you'd require the battery to be able to deliver a given amount of energy (number of Ah) irrespective of the battery capacity. Say for arguments sake, you want a discharge of 10Ah from a 30Ah battery which is a 33% DoD. As the battery ages and its capacity drops the 10Ah represents a higher and higher DoD. When the battery capacity has reduced to 20Ah your load is now a 50% DoD so the aging of the battery actually accelerates because of the aging. Nasty. But also reality.