PDA

View Full Version : PropController Ethernet - Construction question



Matt_Edwards
06-11-2011, 04:08 AM
Ben,
I want to check Q1.
I think your PWB footprint is wrong.
http://doityourselfchristmas.com/forums/attachment.php?attachmentid=10489&d=1307779637
I think pins 2 and 3 are swapped. On the PWB pin2 and connected to ground.

DynamoBen
06-11-2011, 01:54 PM
Ahhh interesting, it looks like the text labels in the schematic are wrong. The collector, emitter, and base are all correct but the drawing isn't representing the pin assignments correctly.

chilloutdocdoc
06-11-2011, 03:16 PM
My PCB works as soldered, matching the footprint on the PCB (some transistors can be made backwards though.)

Matt_Edwards
06-11-2011, 09:52 PM
My PCB works as soldered, matching the footprint on the PCB (some transistors can be made backwards though.)

Well yes it will still work. Q1 configuration in the schematic is a "Common Emitter". If you fit Q1 as per the PWB overlay, Q1 configuration becomes a "Common Base". the sneaky thing is both will work, albeit with different characteristics.

DynamoBen
06-11-2011, 09:57 PM
This is just a text issue on the schematic, the PCB is correct.

DynamoBen
06-11-2011, 10:55 PM
This is just a text issue on the schematic, the PCB is correct.

Matt is right I am wrong, nice catch! The schematic is correct in that its wrong, that particular part has a different pinout than what appears in the schematic. I'm reving the board now.

In addition as Matt pointed out the circuit will still function, its just not correct.

Matt_Edwards
06-11-2011, 11:15 PM
For your new layout, another thing to consider is the the centre pin is usually on curved side not the flat side of the component. In fact a fair number of transistors in TO-92 case come with their legs pre-bent in this format.

DynamoBen
06-11-2011, 11:22 PM
For your new layout, another thing to consider is the the centre pin is usually on curved side not the flat side of the component. In fact a fair number of transistors in TO-92 case come with their legs pre-bent in this format.

I'll have to change the library for that one since this is the default pad config.

budude
06-12-2011, 01:35 AM
So - - are you saying that for those that have built this (and I assume the DMX version as well) that this part has probably been installed incorrectly? I'm using the PropPlug so not too concerned about it myself but it would be good to know for others.

DynamoBen
06-12-2011, 11:55 AM
So - - are you saying that for those that have built this (and I assume the DMX version as well) that this part has probably been installed incorrectly?

It's installed correctly but two pins are swapped (2 & 3) on the PCB. Long story short, the board will function correctly even with this mistake.

That transistor is used to toggle the reset on the prop when programming it. I've used this design with this mistake for 2 years and have never had an issue.

RPM
06-13-2011, 08:43 PM
It's installed correctly but two pins are swapped (2 & 3) on the PCB. Long story short, the board will function correctly even with this mistake.

That transistor is used to toggle the reset on the prop when programming it. I've used this design with this mistake for 2 years and have never had an issue.

Sound kind of like my issue with my Propeller based board, but mine had the V+ and V- to the Propeller IC backwards... this was due to an issue with the library file I was using.
I had done a pre-check of voltages before inserting the IC's, but still missed the swapped pins.
Testament to the durability of the Propeller, even after running the board with the voltages reversed to the IC, it still works! I can say I was amazed at this!

Ben, if you are going to change the layout of the PC board, might I suggest you feed the input of the 3.3 regulator from the output of the 5v regulator?
Right now the 3.3V regulator is fed from the same source as the 5V regulator, and on my board the 3.3v reg seems to run a little hot.

DynamoBen
06-14-2011, 12:28 AM
Ben, if you are going to change the layout of the PC board, might I suggest you feed the input of the 3.3 regulator from the output of the 5v regulator?
Right now the 3.3V regulator is fed from the same source as the 5V regulator, and on my board the 3.3v reg seems to run a little hot.

I did try that on one of my boards and they both run hotish. Beyond that I'm indifferent about changing it.

budude
06-14-2011, 01:34 AM
I would be hesitant to pull the 3.3v off the 5v regulator as I prefer to supply 5v to the daughterboards and we have to ensure we keep enough reserve to power all of them. It may be more than enough but we should probably keep power management in mind. I'm guessing also that powering the 3.3v bus in addition to external boards may just shift the heat from the 3.3v regulator to the 5v regulator.

I'm wondering if there are any drop-in switching style regulator modules that could be used? I know I used one for my Ren48LSD v3b board that worked very well (at 5v) - perhaps they have one for 3.3v as well.

pmscientist
06-14-2011, 11:49 PM
I'm guessing also that powering the 3.3v bus in addition to external boards may just shift the heat from the 3.3v regulator to the 5v regulator.

Yah, this seems the crux of the issue. I've been seriously considering this question for another project. From a pure thermal perspective, it would be more efficient to shed the heat from the large voltage drop once as opposed to twice. On the other hand, it probably doesn't take all that much 3.3V current, to reach the point where the additional heat sink capacity capacity needed may outweigh any potential efficiency gains.

Just as an example, using a 12V source and assuming input current draw of 1A@5V and 0.5A@3.3V, you only shed approx. 0.95W less in a cascaded regulator configuration than in a parallel configuration. However, this savings comes at the cost of adding about 2.8W of heat to lose at the 5V regulator. In this case, I don't see a strong case for a redesign by rearranging linear regulators.


I'm wondering if there are any drop-in switching style regulator modules that could be used? I know I used one for my Ren48LSD v3b board that worked very well (at 5v) - perhaps they have one for 3.3v as well.

You hit the nail on the head there. Moving to a switching regulator would be the way to go, should a redesign of the power supply be considered. There are definitely 3.3V switching regulators, and regulators with 2+ outputs that can do both in the same package. The pitfall with switching controllers seems to be choosing one that may add more complexity to the supply than is really necessary for the application. That looks easy to do with the multiple output regulators. There is alot to be said for the simplicity of linear dropout ICs.

dirknerkle
06-15-2011, 12:05 AM
Wow, all this development stuff is going so quickly that by the time I get a chance to ding around with the PropController boards, the originals I bought will be so far out of date that they probably won't work. They sure are super-looking boards, though! :smile:

budude
06-15-2011, 12:14 AM
I'm wondering if there are any drop-in switching style regulator modules that could be used? I know I used one for my Ren48LSD v3b board that worked very well (at 5v) - perhaps they have one for 3.3v as well.

Well - indeed they do: https://dcomponents.com/?content=details&idpart=8385

It's a pricey upgrade though ($9.02 qty 1)! I do know that switching my 7805 to the 5v version of this made it run nice and cool afterwards, whereas the 7805 was at the maximum operating temperatures in the LSD... This is only rated for 1A but there is a 1.5A version as well. It does save space over the DIY version of the switcher regulators like I did with the v3c version of the LSD so the extra cost might be worth it - it's unlikely you're going to have a large number of PropControllers in your setup since they support higher channel counts.

Matt_Edwards
06-15-2011, 01:33 AM
looking at the tracks on the pwb, they were never intended for large currents, I would suggest you change both to NS Smartswitchers

chilloutdocdoc
06-15-2011, 08:19 AM
What about use a 5V switching regulator, and then a 3.3V LDO? The switching regulator would save the heat from the Input -> 5V and the 3.3V could maybe handle the 1.7V drop from there?

DynamoBen
06-16-2011, 12:25 AM
What about use a 5V switching regulator, and then a 3.3V LDO? The switching regulator would save the heat from the Input -> 5V and the 3.3V could maybe handle the 1.7V drop from there?

Wouldn't that increase the cost dramatically?

chilloutdocdoc
06-16-2011, 11:44 AM
it would, just tossing ideas out there.

DynamoBen
06-16-2011, 11:52 AM
it would, just tossing ideas out there.

That's cool, I had considering using a switching supply on a previous project and the barrier was cost. I though perhaps the cost had changed since then.

budude
06-16-2011, 12:06 PM
That's cool, I had considering using a switching supply on a previous project and the barrier was cost. I though perhaps the cost had changed since then.

True enough it would add some cost, but as I said before, most folks aren't going to have 10-12 PropControllers - more like 1 or 2 so the extra cost can be spread out over the total channel count. You would also be increasing the reliability of it as well since the parts would be running cooler and more efficiently.

Matt_Edwards
06-16-2011, 03:59 PM
Has anyone done any current measurements yet of the 5v and 3.3 regs?
How much current needs to be considered for J1, J2, J3, J4 and J5?

I support Brian's comment, making the regs run cooler has to be better for reliability.
How does ~$2 for a LM2576-005 and ~$4 for a 100uH 3A inductor, $1 for a 1000uF and 70c for a LM1117-3.3 compare to the overall cost of the finished board?
Adding $8 for a switching reg and a LDO reg or $14 for 2x switching regs is not unreasonable.

chilloutdocdoc
06-16-2011, 05:19 PM
Considering these changes would be for a Rev 2 board... has anybody considered using a regulated 5V supply on the input, and anything that requires an AC zero crossing signal can have the required H1 chip onboard?

(Something similar to http://www.sparkfun.com/products/8269) This wouldn't be all that expensive considering you already need to spend about that much on a 6-9V ac supply anyway.

Josh

pmscientist
06-16-2011, 08:21 PM
I support Brian's comment, making the regs run cooler has to be better for reliability.
How does ~$2 for a LM2576-005 and ~$4 for a 100uH 3A inductor, $1 for a 1000uF and 70c for a LM1117-3.3 compare to the overall cost of the finished board?
Adding $8 for a switching reg and a LDO reg or $14 for 2x switching regs is not unreasonable.

It may be even less of a difference than that. Since the two regulators spec'd now run about $3 total, the total increase would only be about $5 per unit for a switching+LDO combination. Unless I'm double subtracting the existing regulators, which has been known to happen :) This seems a good option for an on-board supply.


Considering these changes would be for a Rev 2 board... has anybody considered using a regulated 5V supply on the input, and anything that requires an AC zero crossing signal can have the required H1 chip onboard?

Personally, this is an option I really like. I suspect many people working with RGB pixels probably would as well, as they'll have regulated DC power already available. Only a 3.3V regulator would be necessary, and since there would be virtually no AC current requirement, one could use a really cheap wall wart as the AC source for the ZC signal. For me, this would save a few bucks, as I don't have a 9-12VAC supply w/1A capacity, though I have 4 or 5 in the 250mA to 800mA range.

I'm fairly certain the main board is the best location for the ZC signal. While it certainly could be implemented at the daughter board level, it's simpler to implement it once, and then use it as many times as necessary. If you don't need it at all, the line is already tied high via R8 and the part can be omitted.

It appears that using a DC supply is doable now by dropping out the diodes and wiring regulated 5-12V power in directly. If a 5V source is used, pins 1 and 3 of the 5V regulator could be wired together in place of the regulator. The A/C would remain wired as designed, it just wouldn't have the same current requirements.

Are those correct assumptions? If so, I'll wire up one of mine this way and see how it goes.

As a dev platform, it seems like an on board supply is a good thing to have. Although it would be nice to have a place to solder on some terminals for an external DC supply. A jumper to bypass the 5V reg if it's not needed might be nice, tho soldering a wire there is just as good really. It seems like there's plenty of board space to add this without increasing the footprint.

DynamoBen
06-17-2011, 12:06 PM
BTW I expected that some folks might run this board with a 9-12V DC supply, which works w/o any changes. As far as regulator changes and direct access to the 5V rails, I'm open to board changes if they make sense.

chilloutdocdoc
06-17-2011, 12:19 PM
WooHoo! One less Transformer I need in my show!

chilloutdocdoc
06-17-2011, 07:47 PM
Question, what way does the DC get plugged in? Where gets + and where gets -? Rather ask then be stupid and blow up a controller :)

Matt_Edwards
06-17-2011, 07:54 PM
Well pull out D1 to D4, & place a wire link in D1 and D4, the "+" for C1 then indicates "+" for J12 too!

pmscientist
06-18-2011, 02:20 PM
BTW I expected that some folks might run this board with a 9-12V DC supply, which works w/o any changes. As far as regulator changes and direct access to the 5V rails, I'm open to board changes if they make sense.

Maybe this would be a set of refinements to revisit after the holidays are over and some experience has been gained. Certainly you've made it plenty flexible enough already to implement various power options without board modification.

DynamoBen
06-18-2011, 02:31 PM
Maybe this would be a set of refinements to revisit after the holidays are over and some experience has been gained. Certainly you've made it plenty flexible enough already to implement various power options without board modification.

Absolutely, I expect that the design will change over time and will be become a springboard for new designs....after all that is the point. ;)