Shop OBEX P1 Docs P2 Docs Learn Events
Propeller Held in Reset — Parallax Forums

Propeller Held in Reset

I've got the propeller in a device that sits on the end of a 3 meter cable. I want the controlling unit at the other end of the cable to have access to the propeller's reset line incase things go bad and the controller needs to restart the device. I have the reset line pulled high with a 10k resistor that's part of a resistor network pulling things up that go out on the cable to the world. Oddly enough, the prop boots IF I have the prop clip connected, but as soon as I remove the prop clip, it is being held in reset. If I remove the prop clip and disconnect the reset line from the cable, it boots again. Is the solution here a smaller value pull up for the reset line? Seems like 10k should have done the trick. There is a tranzorb and series 10k series resistor in-line as well for protection.
«1

Comments

  • 10k+10k sounds like you have a divider effectively but a circuit or photo would help. I just wouldn't run the reset that far but it would not have hurt to just use a transistor reset there anyway as it would still work cascaded but it stops the reset being so exposed. One glitch and you're gone from a reliability point of view.
  • Hey Peter - Here's a sketch of what I've got. I may decide exposing the reset is too risky, but it would be nice for the controller to be able to pull the plug if things are going horribly wrong.

    2761 x 1190 - 376K
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2016-10-10 02:52
    Just as I thought, you're circuit might just reset the Prop but it can only pull down the reset line to half-supply due to your 10k+10k divider. Either drop the series resistor value or simply put the pullup on the other side of the series resistor. The transzorb is probably not really necessary due to current limiting anyway but I still don't see where your reset problem lies. If you must run it that far than consider the transistor I mentioned instead of this circuit.

    Btw, it could be that capacitive crosstalk from the tx line is coupling across to the reset due to the length. Very first transition on this line and the Prop gets reset before it can go any further.
  • If the Prop brownout circuit is enabled, then the Prop itself provides a 5k pullup resistor to Vdd. So the voltage divider is really 3.3k:10k. At fiirst glance the stronger pullup and series resistor should make it harder to effect a reset. The Prop Plug of course does have the transistor circuit that goes directly to the reset pin, so no trouble pulling down the 3.3k. But when the Prop Plug releases the transistor that pin should be back to the 3.3k pull-up so what else is there that could be holding it in reset? Is it cycling in and out of reset, or solidly reset?
  • It appears to be solidly reset - at lease the current it's pulling from the supply doesn't change on the timescale of the current meter update rate. I'll try jumping the series resistor tonight and see what happens.
  • How is the your Propeller powered? Does the power come through the same cable as the reset input? What else, besides the Propeller's Vss, is the ground of your power supply connected to?

    -Phil
  • Phil

    There is battery power (~12 VDC) coming down the 3m cable. That cable terminates on a small PCB that has the input protection diodes, series resistors, and a 3.3V switching regulator. That 3.3 V, reset, etc, goes to the main PCB on a 6" ribbon cable. The main PCB has the Prop, accelerometer/magnetometer, ADC, temperature sensor, and several analog amps.
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2016-10-10 19:55
    Is the battery's negative terminal connected to anything besides the circuit you describe, e.g. earth ground, computer case, etc.?

    What I'm trying to get at is whether connecting the Prop Plug (presumably with a computer attached) completes a ground circuit that's otherwise open. IOW, the Prop may not be held in reset; it may just not be getting power.

    -Phil
  • jmgjmg Posts: 15,173
    geo_leeman wrote: »
    I've got the propeller in a device that sits on the end of a 3 meter cable. I want the controlling unit at the other end of the cable to have access to the propeller's reset line ...

    I'd be wary about connecting any 3m cable too directly to the reset pin. That's a great aerial you have made.
    An irony here is the connection to reset, can cause what you hope to avoid....

    Try a digital transistor, or even an opto coupler, which isolates things, and ensures nothing too fast can make it to the Prop RST pin.

  • @Phil - The negative is connected to the case ground, but that isn't connected to mains ground and it's operating off a laptop unconnected to mains ground. If I remove the 3m cable connection to the reset terminal it boots up fine, so the prop is getting power.

    @jmg - I'm thinking that the transistor solution is probably the way to go, but will know more as soon as I test the divider idea in a few hours.
  • kwinnkwinn Posts: 8,697
    jmg wrote: »
    geo_leeman wrote: »
    I've got the propeller in a device that sits on the end of a 3 meter cable. I want the controlling unit at the other end of the cable to have access to the propeller's reset line ...

    I'd be wary about connecting any 3m cable too directly to the reset pin. That's a great aerial you have made.
    An irony here is the connection to reset, can cause what you hope to avoid....

    Try a digital transistor, or even an opto coupler, which isolates things, and ensures nothing too fast can make it to the Prop RST pin.

    +1

    I was going to suggest the optocoupler on the propeller end of the reset cable as well. It eliminates any RFI and EMI concerns as well as protecting the propeller from ESD. By far the most reliable and interference free method of remotely resetting a system.
  • Any suggestion of specific parts? I'd prefer the opto to only be on for a reset since I'm power sensitive in this application.
  • kwinnkwinn Posts: 8,697
    I used a TLP175A for a similar task recently, but almost any opto with a bipolar or fet transistor output would work. Since the led needs to be on for the output transistor to pull the reset down the led only needs to be on for the reset, and the trigger current is only 1mA.
  • Hmm ok - I tried last night and the problem didn't reproduce - maybe different RF environment, cable position, alignment of the planets, etc. But the divider effect does prevent a proper reset and it's not good that the behavior is so unpredictable. I think the opto might be the best shot. Luckily it's a small and cheap PCB to spin! I guess the moral is put the series resistor after the pull-up (doh)!

    With the opto, do people still recommend a 10k pull up to make the reset less susceptible to mis-triggers?
  • kwinnkwinn Posts: 8,697
    I would recommend the 10k.
  • As Tracy pointed out, if you've enabled the brownout reset, there's an internal 5K pullup resistor connected to /RST. So with a 10K series resistor, you will still not be able to pull /RST low enough to get a reliable reset.

    -Phil
  • kwinnkwinn Posts: 8,697
    As Tracy pointed out, if you've enabled the brownout reset, there's an internal 5K pullup resistor connected to /RST. So with a 10K series resistor, you will still not be able to pull /RST low enough to get a reliable reset.

    -Phil

    Sorry, I thought he meant a 10k pullup. Definitely don't want or need a 10k between the transistor and the /RST pin.
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2016-10-11 18:25
    Low impedance looking back down the cable is most important if the cause of reset happens to be noise pickup. A parallel pullup resistor can serve that purpose, but it might as well be 1k instead of 10k, which gives 833Ω in parallel with the internal 5k. Then the device on the other end (or the spurious noise source) will have to supply almost 4mA in order to pull RST\ to ground. Or, a replace the 10k pullup with a 0.1µF capacitor, and use a smaller series resistor, say 330Ω. That provides a lowpass filter for RF and coupled noise but still allows reset with a longer intentional pulse. The same thing applies if there is a transistor added for isolation, it needs to see a low impedance looking back down the cable. An opto-coupler by its nature provides that due to the current needed by its LED.

    Is there any possible source of crosstalk on the cable itself? It's good that the problem went away but, what can you do about the alignment of the planets!?

  • Tracy AllenTracy Allen Posts: 6,664
    edited 2016-10-11 18:28
    There is this:
    That cable terminates on a small PCB that has the input protection diodes, series resistors, and a 3.3V switching regulator. That 3.3 V, reset, etc, goes to the main PCB on a 6" ribbon cable.
    I'm wondering if the switching regulator could be a source of noise/crosstalk.
  • I'll have a look at the noise on the power rail on the scope. I'm out of town today (Thursday), but will be back for the weekend and will upload a scope shot.
  • If you don't have a low esr cap on the input terminal of the regulator then you may find the input line very noisy although I would not think that would have enough of a swing to affect the reset.

    I still think that all you need is an npn circuit rather than venturing into opto-isolation.

    btw, you should always be using your scope but just make sure you attach the ground correctly and directly otherwise you may see false noise.
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2016-10-14 16:12
    I agree that crosstalk noise is probably not an issue on that long input cable, especially given that the 10k input resistor didn't even allow a proper reset.

    Isn't troubleshooting fun? Phil mentioned a possible weak ground connection via the Prop Plug. I ran into a situation like that yesterday. The setup was one master Prop talking to 4 slave props via 3V RS232 out on 15 inch cables. I got the firmware running on a Mac/bst and then we went to test it also on a PC, but nothing. Back to the Mac, it ran fine, that is, until I disconnected the Mac from its line wall wart. Weird, no other connections to earth ground. Could it be some kind of grounding issue?. However, probing with the 'scope revealed a weak RS232 signal from the Prop Plug and the slave tx data was also showing up as big triangle waves on the master data line. All that when away when the Mac wall wart was reattached. The short of it is, changing to a new Prop Plug solved the problem on both the Mac and the PC. I guess the first Prop Plug was defective, although I don't see any breaks on the connectors, which is the usual stress condition for those. It is as if it were running on parasite power. One of those out of left field troubleshooting issues. I don't know why the Mac wall wart made a difference.
  • Just for fun I took a look at the power rail. Maximum of 20 mV noise, but probably less given my probing technique wasn't incredibly pure and there are some flying wires with a 250 Hz square wave on them in the enclosure.

    I found an older post from Tracy talking about using pre-biased transistors for the reset logic. It looks like the recommended part is being EOL'd, but the PDTC114ET,215 is a drop in replacement.
  • I have some other lines (control lines, etc) coming down the same cable. Currently they are pulled up at the prop side. Would the transistor solution be recommended there too?
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2016-10-17 00:56
    I use the DDTD pre-biased NPNs for reset and also for switching heavier loads up to 500ma but as for the "other control lines" I would think you would have to provide details of this even fro the start.

    I find that a lot of problems could be solved if the details weren't held so close to the chest, as if someone on the forum is going to "steal" ideas and make a fortune :)

    All the stuff that is really security sensitive is dealt with without the help or knowledge of the forum. My advice then is if you need the forum then just share all the details right up front, you will be sure to get the best help. Most of the time a market advantage is not due to technology, but due to being in the right place at the right time and knowing the right people.
  • Peter - not trying to be secretive, just trying not to provide needless detail. Okay - here's the story. This is an instrument to measure ground tilt at the micro radian level. I've attached a photo of the setup and enclosure I've built so far. There is a 10-pin water proof connector on the end of the cable that has power (6-28 VDC), ground, RS232 TX/RX and six "control lines". Of these six, three have defined functions: trigger reading, power down (hold system in RC slow state until we need it), and reset. The other three (control 1-3) have no defined purpose right now, but are there for future use. I'm planning on dedicating one to a GPS PPS signal so that the internal time-base can be locked to real time. I don't plan on using them for anything high speed, just to help future proof the design if I want to add features. Hopefully that helps clear things up. Once I get this issue straight, I'll put up the schematic of things (it's going to be open-source anyhow) for any feedback :)

    720 x 540 - 150K
    1080 x 810 - 333K
  • Oh yes - one more thing. I want someone to be able to connect this to power and a RS232-USB adapter and run just connected to a computer. Connecting it to my logging setup will utilize the power down/control lines. The board in the cap of the enclosure is the protection/power reg board for any spikes or outside trouble. The main board in the base has all the expensive bits like the ADC and other sensors.
  • Tracy AllenTracy Allen Posts: 6,664
    edited 2016-10-17 16:59
    Nice housing! A picture, 1000 words.

    I'm still not clear about the "controlling unit" that will be connected at the outside end of the cable. Is that also Propeller-based? And where does the option come in for the user to "connect this to power and a RS232-USB adapter and run just connected to a computer"?

    Currently for a prebiased NPN I use the MUN5211 family from ON Semi, with variations in SOT23 and SC70. It (and a host of other) are equivalent to the DTC... part.
  • Thanks! I'm neither a mechanical or electrical engineer, but have managed to learn enough to be dangerous :) The case has been tested to IP67 and did very well!

    Onto your question. This will likely be hooked up to some customer's in-place telemetry systems. Often that's USB to a computer or hooked into some funky satellite data link. I'm leaving that up to them, but "stock" you power it up and data comes shooting out the RS232 lines on the connector.

    I plan on making a compatible logger (prop based) that will go with this and log the data to an SD card as well as enabling PPS lock, etc. One of the best features is I can use the trigger line to put the unit to sleep for people that want very sparse sampling rates, but need low power. (Like tidal researchers that need a reading every 5-10 minutes, but on batteries over the Arctic/Antarctic winter in the most extreme example.) My logger will handle all of that, but undoubtedly some adventurous folks will try to rig their own setups together, hence why I want to make all of the extra lines (trigger, etc) be pretty robust to someone accidentally hooking it up to a 12 V battery or some other nonsense.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2016-10-18 00:44
    RS232 should have been shot a long long time ago. Down with RS232, long live serial!

    There is no good reason at all to inflict +9V to -9V transitions just to pass serial data which we do quite well at much higher speeds at much lower voltages unbalanced as in the case of I2C etc. If you must have RS232 then don't feed those "RS232" signals down a shared data cable, just have a small active RS232-TTL header at the remote end. You will find though almost without exception that you can feed inverted logic levels into "RS232" and simply use a current limit resistor of around 33k (+pulldown) or so to receive RS232 (also good protection) since none of the proper RS232 receivers are proper anymore and are designed accept to logic signals.
Sign In or Register to comment.