On the death of an Imp

It is with a heavy heart that I inform you that one of my Imps is no longer with us. 0c2a690020d5 stopped blinking and responding last night while powering one of my new prototypes. May it rest in peace.

I am autopsying the circuit to determine the cause of death. The Imp was in an April board, powering a simple transistor relay circuit, and switching the relay at the time of its passing. Here is all of the information I can think of that may have contributed to it.

I was powering the april board from this ACDC:

The DC output was connected to one of the GND pins, and to VIN, and the jumper was set to USB.
An LED (that I have no specs for) and a 1K resistor were connected to VIN and GND.

I was powering a 2N3904 transistor from Pin 9 with a 1K resistor, and using it to drive this:
I had a diode across pins 3,4 on the SSR… and thinking about that now… I don’t have any idea if an SSR needs a diode.

The circuit was working just fine, the Imp would switch the relay correctly, however I did notice this: I wasn’t able to clear the wifi configuration or change it to a new one, it was completely ignoring any attempt at blink-up.

I’ve gone over the DC parts of the circuit with a meter, and everything looked to be fine, but I made these changes:
I moved the 3.3V DC+ to the P+ pad, and the jumper to BATT, and the unknown LED is no longer in the circuit. The imp in the circuit now clears and blinks up properly.

@Hugo I am very sorry for killing an Imp, if it was indeed my fault. It was a good Imp, and it will be missed. Any ideas about what I might have done?

The wiring seems fine, you don’t need a diode with an SSR because there’s no inductance there (which is what the catch diode is designed to deal with in relay circuits).

What ACDC variant did you use? (the link shows several). If you are feeding 3.3v into the april, you don’t want to do it on the P+ pad really, as that’s for unregulated input.

The last bit of the post is a bit confusing, are you using another imp now?

I was using the 3R3U… the 3.3V version. I’ll put DC V+ back to VIN.

I do have another imp in the circuit now, and it is working just fine, I can clear and reconnect wifi, and it operates the SSR correctly.

In my first post I said that the imp was switching the relay when it died, but to be clear, I meant that it had been able to switch the relay the last time I sent a command, but it wasn’t actually doing any switching when the agent logged that it went offline, it was just powered on.

Still baffled by what happened, and what might have caused it to start ignoring blinkup before it went completely out.

I’m a bit worried that the 3.3v ACDC might have a noisy output. That could affect blinkup as it’s analog.

Using a 5v ACDC and relying on the circuitry on the april to clean up the supply is likely safest. I don’t know what the quality of the ACDC output is…

Thanks, that makes sense. The datasheet has a typical application circuit that I assume is to clean up the input, output and provide some surge protection. I don’t yet have any of those components in the circuit.

I’m looking at a few different ACDC’s, and I am coming to the same conclusion, that using a higher voltage ACDC and then using the DCDC on the April it a good cheap option. I’m not really simplifying my circuit with the 3.3V ACDC’s since they don’t contain all the components for clean power within the package. Crydom does have an ACDC that supplies up to 1000mA, that would be really handy for circuits with high power needs.

Would there be any reason not to run the ACDC 3.3V output into the DCDC on the April to clean it up?

The DCDC on the april can’t do much if you don’t give it any voltage headroom; it’ll pretty much just turn the switch on 100% of the time. To get any noise isolation, it needs to actually be switching (ie, dropping voltage) - I’d say 3.7v or above.

@Hugo What is your opinion on directly powering an Imp/impee with this:

(The 3.3V version)

From what I can see, the only thing I would really need to add is a fuse. It isn’t cheap but I looks like it would nicely take care of all my power needs.

Actually, that one has a very long lead time… perhaps this one instead…

If you’re going to do that, then you need to ensure you have enough local capacitance on the imp side. Wires are inductance, this is not good when you have a load like the imp, which can produce some very fast current transients (eg when the wifi transmitter turns on and off). These do not play well with inductance.

It’s still safer to use an april and feed it 5v, if you can.

I’ve used the attached RAC03-based circuit quite successfully for a few projects … it’s just based on the Recom design quidelines, as well as some circuits copied from Hannah. I use a 5v device (for some LED needs), then just linearly regulate down to 3.3v … (I haven’t been overly concerned with efficiency given the AC source). I’ve got the pwb it resides on also laid out for an alternate CUI device, but have never used/tested one … the Recom part has never given me any problems.

  • Larry

As Hugo indicates, I do utilize fairly large output caps to handle transients.

Thanks, both of you. I see that the 5V version of the Recom ACDC is stocked by Digikey, and slightly cheaper.

Does anyone know of any Eagle libraries that might have the components for the LDO layout on the Amber board? I’m working on learning to build parts in Eagle, but that has been slow going so far. It seems like Eagle may have changed quite a bit in the recent updates, and some of the tutorials I’m looking at are quite a bit different. Smartmaker has the imp002 built for Eagle, which is really handy.

Edit… I found a similar LDO in the burr-brown library.