5v PSU - Booster Noise?
-
Having finally arrived at a compact solution for two Nodes with ultrasonics (JSN-SR04-2.0) sensors on tanks powered by a 5v booster (as featured in the Store/Buyer's Guide), found recently when a tank was emptied (sewage) that consistent levels are detected at 650mm instead of 1860.
This did not occur previously with a linear supply which took up considerable space, logically noise on the 5v line is somehow interfering with the ultrasonic board. The booster is set with the ground plane toward the the rest of the arrangement to part shield RF, removing the LED from the board should not have caused any detrimental effect.Having read mixed results and opinion on filtering, the prevailing solution was to use a linear VR such as the 7805, but this would require ca 6.5v booster to cater for the voltage drop.
No problem ordering up replacement boosters if it works and retains the compact format, any opinions or recommendations?
-
I almost hate to ask this but did you follow the manufacturers datasheet when implementing the voltage regulator?
I would usually add an electrolytic across the output and a 10n or 100n (or both) across the input to the regulator. You can put a low value series resistor in between as well as it might help, but will use a little more power.
I have had a lot of issues with the 662k type switching regulators, but never had an issue with the AMS1117 and an output cap in the 22uF-47uF range.
-
@skywatch Thanks for hating to ask
I haven't see a data sheet on it so simply used it raw aside removing the LED, only when this recent issue turned up, it caused me to question things...
This is the version I used link text
Quite happy to pop on some electrolytics or ceramics to see if that helps...
-
This post is deleted!
-
As it is a module it seems reasonable to expect that the manufacturer added the required passive components to make it work to specification.
Boosters are noisy as they are 'pumping' the power to get to the higher voltage.
Is it not possible to do 'something' so that the booster is no longer required?
That said, it wouldn't hurt to try a 100-1000uF ceramic/electrolytic across the output an see it that helps. It depends on what is causing the problem, switching 'ripple' or switching spikes. So while you are at it add 10n+100n as well to reduce the spikes. It just might work. An oscilliscope would be useful to see what is happening but I realise that not everyone has access to one.
-
@skywatch The problem is the old chestnut of a 3.3v Node and a 5v ultrasonic, the type needed was a single transducer (a la parking sensor) as it is a high humidity environment, only available as 5v.
The original solution (V1) was a 5v promini on a 4xAA supply running the ultrasonic in a separate enclosure and sending results over I2C to the main Node, but the Node's 2xAA batteries lasted 4 months and would occasioally hang or cause problems with the RTC.The Booster version (V2) for the water tank was in a single enclosure, and ran in parallel for 6 months with exceptionally low decay on the batteries. When V1 had a brainfart and flattened the 2xAA suddenly after 2 months, confidently changed it over to the same setup and was delighted until a few weeks later when the sewage tank got emptied it threw this anomaly.
Hence my query over a linear VR such as 7805 on a higher booster as read this would worked well. Additional power losses are no serious issue as the load is max 10 secs every hour, otherwise isolated on a relay.
One of these days I'll get a scope, but it's way down my list.
The only other alternative for secondary supply is to revert to a second enclosure with 4xAA and a low drop VR...
-
Linear regulators are not best suited for battery nodes as they dissipate excess power as heat.
-
@skywatch Agreed for a continuous situation, but the question is would it be effective? I've never dealt with SMPS circuits, 4cx250Bs supplies yes, hence I need to tap into younger advice and experience...
The only device running and going to sleep is the node on it's own PSU, down into the microamp domain...
The secondary circuit is controlled via a latching relay, zero consumption in the off state, so in perspective, On for 1 or less to max 10 secs, Off for 3590 secs, every hour.
Even were the secondary circuit 50% efficient, the drain should not dramatically reduce mAh for such short bursts, but would it cure the problem?
Or am I missing something?
-
@zboblamont You'll need to experiment and see.
Try with ceramic SMD or electrolytic caps for smoothing/reserve current and 10n +100n in parallel for decoupling/spike reduction and see how it goes.
-
@zboblamont
for filtering output of switching converters, it's easier when you know the parts used (reference of ic, inductor). then you can get some "filtering" calc vs load, freq etc.
It can help using low esr ceramic capacitors (ideally x7r dielectric). 10n+100n like skywatch said won't hurt sure, I would add maybe 22n and 22-47uf. It might improve. Otherwise, try adding 1-2 more low value capas.
Then if not enough, I would a ferrite bead in serie and add another 10n/10u combo. But you need the right reference else it can be worse (3 datas are important for this choice: impedance and freq, low resistance, and max current). But it's easier when you have an oscillo for checking.
Else a good high psrr LDO+a few caps for very sensitive circuit (like those for rf..) but sometimes there is no other choice than switching converters (they are handy too).
And depending on the circuit, there are additional things to do (using tvs etc), but I don't think that's your case here.I hope this helps a bit!
-
@scalz @skywatch Thanks both for thoughts, but managed to solve the original problem when I stumbled across why the JSN-SR04-2.0 would not work at 3.3v. The standard pulse of 10 microseconds changed to 15 had it working perfectly rather than intermittently. All that pain over two years for the sake of 5 microseconds.
Will try playing with a high psrr LDO post booster for potential future needs, as the interaction completely threw me. The error only showed up when the ultrasonic range extended (tank emptied), blissfully unaware until then.
I don't have the skills to design a booster, and components on the commercial ones are difficult if not impossible to identify, so post treatment it will have to be.
