AVR Freaks

As shown in the schematic, I have a 10000 uF cap to supply surge power (charged to 24V). The 754410 H-driver provides the directionality. If I don't use the boost, it works fine. But with the boost, there is not enough juice in the 10,000 uF to kick the solenoid, yet the commercial device uses only 2200 uF,  so I'm trying to figure out where I'm going wrong.

I have a scope, though even a 1 ohm resistor in series with the solenoid keeps it from working with 24V, so I'll need to rummage up a smaller resistor if I want to measure current through the selenoid. 

I mean, I kinda know where I am going wrong ... not enough current. But I'm trying to figure out why 10,000 uF isn't giving the required kick; it can kick pretty hard. Maybe I'll have to visit the local thieves and get a smaller resistor and scope the current.

You mean decoupling caps? Yes. It went badly without them :)

I have tested it. When hooked up to lots of input power, it works. When used with the boost, it does not. Nothing else changes between these two. Both use the same voltage (at idle....) Now energy is both voltage, current *and* time, and I might have missed the time dropping due other factors, like power sag on the low voltage rail. I hope it is that, because that is easy to fix :) 

Well, I came here specifically for the speculation :) Alas, everyone speculated about what kind of solenoid it was (I already knew), what it took to drive it (I already knew).  That's fair, the person with the problem (me) has often made some bad assumptions, but in this case I tested that part pretty throughly. The discussion still gave me some ideas. And yah, I could have scoped everything, but I'm not fond of random troubleshooting, and it didn't occur to me to scope the low voltage rail.... clearly I'm rusty. 

That driver looks pretty interesting, but this is a personal project and it is make-it-with-what-you-have as much as is sensible (and the 754410 does get the job done, if not as efficiently as is possible). When I run out of of the current H-bridge, I'm definitely going to look at trying those instead, in part because it has a feature I really would love to have: 3V compatible inputs, so at least on the low voltage side I could do with only a 3.3V rail in some instances, which would be very nice. Thanks for the pointer.

No, I'm not assuming that at all. I know there is not enough energy is being delivered to the solenoid (that is not assumption, that is fact), and I know that lots of power in does work, and when I switch to a boost+large cap, it doesn't. That is what I know. The actual question originally was, and is, why. I'm not assuming the booster is at fault, but I know introducing it causes the problem; I know full well there are artefacts, this is why I was asking in the first place. The question never was why doesn't a booster supply enough power. It was why doesn't this circuit work. Note the circuit includes a large cap to provide surge energy, because clearly the peak current by the booster is insufficient, by its own spec sheet, and its input power (the USB power supply) doesn't have enough even if the booster could handle it.

However, you assumed I assumed you assumed is way more debate then electronics, so I'll just try some new stuff this aft (separate rails for H-bridge and booster) and get back on its success! One unfortunate is that I switched boosters (as that one fails like crazy) and the new one might work out of the box, which would be good for success but bad for learning.

I did actual measurements, including waveforms. I built up everything and got it all fully working (including firmware for the tiny, and a test routine that checks operation, etc) with a bench PS and that involved a fair bit of mucking around to get the right params. Then I introduced the boost circuit. I didn't realize this was a test where I had to prove I tested. I asked for some ideas on what might be wrong, that is all. I don't want to assemble a 20 page white paper proving my methodology. I do this for fun!

I didn't post them. I felt asserting that the circuit worked with a power supply of the same voltage would be sufficient.

No, it is the unit linked to up above. It has the requisite inductor, etc. The booster works (well, worked...), the circuit it is in does not. Note it does move the solenoid some (in circuit), just not enough.

4.3 ohm Tried two different peak-reading multi-meters (Fluke 87V and Mooshimeter (sampling at 8kHz)), both sometimes caught about ~2.7A surge on actuation, but clearly the surge was very very short as they only catch that once in a while. When the solenoid is wet it takes more to move it, that could affect some of the readings as well.

Added: The surge is so short the bench top PS current limit (set at around 500mA) doesn't catch it. Mind you, it is a nice-ish one, but not top of the line or anything.

As a matter of interest. Supply set at 24V or so, current limit set to 1A. Pulse width 20ms. Solenoid activation. CH1 is the 24V rail. This is good activation of the solenoid. The PS recovery is interesting.

Well, good news / bad news.

The good news is that the new booster will actually drive it (with a cap). The new one is XL6009 base and is specifically this one:

https://www.amazon.ca/gp/product...

It has much nicer components then the previous one, and is cheaper to boot (at least if you buy 4 of them...). 

The bad news is now I'll never be sure what the problem was with the previous one. But, WTH, lets see some scope output since I said I would.

For your amusement: Booster set to 24V, driven off 5V shared rail (which is driven by USB PS). No extra caps. I don't expect this to work (not enough juice in that 5V supply) and it didn't. 20 ms pulse Here are the 5V rails and the 24V rail:

As you can see, the 24V rail crashes and burns. The 5V rail is hurting.

Here, I add a 4700 uF cap to the 24V rail. For the record, it is a Panasonic, sourced from Digikey, so probably not fake. This setup actually works, seems to solid turn on the valve, even with water pressure.

Note the time scale is different then the above one.

You can see that the cap discharges for ~22 ms (cursor probably isn't accurate), then starts charging again. The 5V rail suffers for this period. The other devices on the rail (2 micro controllers) are ok with a volt here or there, but in theory the low of the 5V rail is outside 754410 spec, but it seems to be OK. Once it charges up again, we are OK.

It would be interesting to completely disconnect the booster just before firing to stop the 5V sag. I think there is enough juice in the cap, or if I switch the 10,000 uF one there will be. Would require another FET or something. That would also be good if I wanted to run it off battery so I could charge the cap only when needed. Might be interesting to limit the charge rate of the cap but let it discharge full tilt; not sure the best way to go about that. Another 'feature' that would be good would be to have enough stored juice to run the 5V rail and the 24V rail long enough to turn off the valve on power failure; one of the big disadvantages of these types of valves is that they latch open and if you lose power, they don't close.... my micro closes it on power up, but it would be nice to detect power loss and kill the valve on the way down. Tricky though, the direct controller is outta pins :) And of course that 5V sag looks like power loss....

Anyway, many thanks.

Ah, good idea. There will probably be a Pi ZeroW on the bus eventually (well, still up in the air vs XBee's.. I like my XBee, but one of them is more expensive then an entire Pi ZeroW, and both get the job done for me)., and it is pretty picky about power. Could drive it with something a bit "raw", schottky + cap + reg for the sensitive bits, raw directly to boost for the loose bits. Hmm. Or I could keep it simple with the XBee, (sure wish that 75441 was 3.3V friendly.... yah, I do need a better bridge) Definitely rusty :)