Author Topic: Introducing Afterburner - MOSFET drive technology for Falcon F8 / F16 Boards  (Read 4173 times)

Offline corey872

  • Supporting Member
  • ******
  • Join Date: Oct 2013
  • Location:
  • Posts: 485
  • Kudos: 16
    • Link to the sale...
Introducing Afterburner - a MOSFET driver kit for Falcon F8 / F16 boards.  A demonstration video is live at:

http://youtu.be/c035HnwbD-E

Afterburner is designed as a simple 'drop in' kit to upgrade Falcon F8 / F16 boards to MOSFET drive technology like that used in the uSC - Micro String Controllers.  This will boost the Falcon FXX to Node distance from 20-30 feet to 50-60 feet or more.

The kit is a simple hardware board requiring no programming, no setting changes, no PIC processors, etc.  One Afterburner kit would upgrade a block of 8 channels - ie one kit for Falcon F8, one or two kits for Falcon F16.  You can run the Falcon F16 output 'split' if desired (ie one side using the original op-amp output for local node strings, one side with the Afterburner for more remotely located strings) without diminishing the performance of either side.  If you wish to boost less than an 8 channel block, the uAmp signal conditioner/amplifier would be one option.
Corey

2017 uSC, Afterburner, uAmp co-op is CLOSED.
uAmp is SOLD OUT for 2017
Remaining boards are now FOR SALE

Online CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 7,941
  • Kudos: 142
Looks great.  I have been thinking about using a F16 v1 to drive my window frames and house outline and this should give me the distance I would need for the far windows and gutter strings.
-
Chris

Offline JerryPlak

  • Supporting Member
  • ******
  • Join Date: Jul 2013
  • Location: Sunbury, OH
  • Posts: 487
  • Kudos: 1
I am Putting this on my wish list.
Jerry Plak

Offline Gary

  • Supporting Member
  • ******
  • Join Date: Jan 2015
  • Location: Chilliwack, BC Canada
  • Posts: 342
  • Kudos: 3
    • Diamond Crescent Musical Christmas Lights
This looks awesome. And a simple install to boot!

But I have some questions (what I love to do)...  :)
  • You mention that 20-30 feet transmitting distance could turn into 50-60 feet with the Afterburner. On the Falcon16 v2 which uses the same "default" driver circuitry as the v1, David Pitts mentioned 25 feet for flex strips or 15 feet for nodes  (http://falconchristmas.com/forum/index.php/topic,2310.msg25010.html#msg25010). Would this mean that when using an Afterburner, the distance to the first pixel on a flexstrip would be 55-ish feet, but would only be maybe 35 feet on nodes? Why is there a distance difference between nodes and flex strips?

  • Will the Afterburner work in any other equipment, such as the Falcon16 v2 since it uses the same drivers as the v1? My guess the answer is "no" because you would have mentioned it already.

  • Are there plans to create Afterburners compatible with other equipment such as the Falcon16 v2 or Falcon16-B? My guess/hopes are that the answer is "yes".

  • While a small component like this would be a fraction of our budget, I'm curious to know how much a kit will cost?

  • Not related to the Afterburner, but I noticed in the video that there seems to be 3 RJ-45 Jacks on the board; looking closely, the board is labelled v1.03. I suppose it's for one input and 2 outputs versus the v1.01 assembly manual on the wiki (http://falconchristmas.com/wiki/index.php/Build:_F-16) that had only 1 output? Are both outputs for Pixelnet, or can one be DMX? I'd be giddy if one can be DMX.  ;D

P.S. The Afterburner name is awesome!

Offline corey872

  • Supporting Member
  • ******
  • Join Date: Oct 2013
  • Location:
  • Posts: 485
  • Kudos: 16
    • Link to the sale...
This looks awesome. And a simple install to boot!

But I have some questions (what I love to do)...  :)

You mention that 20-30 feet transmitting distance could turn into 50-60 feet with the Afterburner. On the Falcon16 v2 which uses the same "default" driver circuitry as the v1, David Pitts mentioned 25 feet for flex strips or 15 feet for nodes  (http://falconchristmas.com/forum/index.php/topic,2310.msg25010.html#msg25010). Would this mean that when using an Afterburner, the distance to the first pixel on a flexstrip would be 55-ish feet, but would only be maybe 35 feet on nodes? Why is there a distance difference between nodes and flex strips?

The data distance can vary quite a bit.  Probably worthy of a video in its own right, but to try and simplify: 

There are several things which go into consideration:

- Inductance / capacitance of the data line (generally increase with length)
- What type of driver is sending data (PIC chips are the worst, the F8/F18 op-amp is better, the uSC MOSFET is generally best)
- How much power is being drawn and what size/type of wire is used. (more power / thinner wire = less distance)


Quote
  • Will the Afterburner work in any other equipment, such as the Falcon16 v2 since it uses the same drivers as the v1? My guess the answer is "no" because you would have mentioned it already.
We were looking at putting the mosfet drive directly on the V2, but Dave mentioned some issues.  I've not heard what the ultimate resolution was.  If the same op-amp is used, then the Afterburner should technically work, though I don't know about board clearances, etc.  There is also the issue of having support for four wires to the nodes, meaning 2 drivers for each string - you start to need a LOT of drivers.  So the short answer is 'no' it is not designed to work with F16V2.

Quote
Are there plans to create Afterburners compatible with other equipment such as the Falcon16 v2 or Falcon16-B? My guess/hopes are that the answer is "yes".

Not really at this time.

Quote
While a small component like this would be a fraction of our budget, I'm curious to know how much a kit will cost?
I'm trying to keep the cost below $10 per kit...though as always 'economies of scale' come into effect pretty quick.

Quote
Not related to the Afterburner, but I noticed in the video that there seems to be 3 RJ-45 Jacks on the board; looking closely, the board is labelled v1.03. I suppose it's for one input and 2 outputs versus the v1.01 assembly manual on the wiki (http://falconchristmas.com/wiki/index.php/Build:_F-16) that had only 1 output? Are both outputs for Pixelnet, or can one be DMX? I'd be giddy if one can be DMX.  ;D

This would be best answered by Dave.  They're labeled as pixelnet outputs, I don't really have any DMX stuff, so not sure if they can be split or not.


Offline Gary

  • Supporting Member
  • ******
  • Join Date: Jan 2015
  • Location: Chilliwack, BC Canada
  • Posts: 342
  • Kudos: 3
    • Diamond Crescent Musical Christmas Lights
The data distance can vary quite a bit.  Probably worthy of a video in its own right, but to try and simplify: 

There are several things which go into consideration:

- Inductance / capacitance of the data line (generally increase with length)


Is inductance generally increased when you have a bunch of cables carrying Pixelnet (and 12v) running parallel to each other, so crosstalk and data degradation occur? I suppose the situation gets better using twisted-pair CAT5 cable? I think I remember seeing one of your distance test videos where you achieved greater distance using CAT5 vs. 3-conductor cable similar to what nodes use. Mind you, from what I've learned doing some Googling (http://en.wikipedia.org/wiki/Twisted_pair) is that twisted pair helps with maintaining signal quality only when an equal and opposite signal is sent over each pair. But if there is only 1 conductor used for the data signal would it make a difference, or is it that the wires used for +12V and -12V wouldn't cause as much interference on the data conductor if they are twisted?

And doing some Googling/Wiki-ing, it looks like capacitance increases by using larger cables.

But then... I remember your video where you just re-purposed a bunch of 120V extension cords to run Pixelnet and you got something like 80 feet. There is a relatively small number of twists in extension cords, and the wire gauge is much thicker than what we generally use for running pixels, so where does that fit into this inductance/capacitance explanation?   ??? 

Quote
- How much power is being drawn and what size/type of wire is used. (more power / thinner wire = less distance)

I wasn't thinking that issue with distance was power and voltage drop, but the data signal loss. From what I gathered from other peoples' comments, it was implied that the pixel type (not the cable type going to the pixel) is what determined the achievable distance  to the first pixel. Would it be that the circuitry in nodes tends to be less tolerant of a "dirty" signal than what strips expect? You'd think that there would be variation from manufacturer to manufacturer.

Offline corey872

  • Supporting Member
  • ******
  • Join Date: Oct 2013
  • Location:
  • Posts: 485
  • Kudos: 16
    • Link to the sale...

Is inductance generally increased when you have a bunch of cables carrying Pixelnet (and 12v) running parallel to each other, so crosstalk and data degradation occur? I suppose the situation gets better using twisted-pair CAT5 cable?

Books could (have) been written on this subject, but I will try to keep things concise here.  Technically inductance would be increased by a group of cables together - I've never really tested it as I don't have groups of cables running for long distances all bundled together.  If you do and can get away with it, then it should be no problem.  If you're having data Tx issues, then a large mass of bundled cables might be something to avoid if possible.  As mentioned, the PIC would be most susceptible, op-amp less so, and mosfet drive least likely to have an issue.

Just using cat-5 probably offers minimal benefit in this situation, the data is not sent as a differential balanced signal (ie equal/opposite signals on a twisted pair).  Though if you use "STP" or shielded twisted pair cat5, and take care to ground the shield, there may be some slight benefit.

Quote
I think I remember seeing one of your distance test videos where you achieved greater distance using CAT5 vs. 3-conductor cable similar to what nodes use. Mind you, from what I've learned doing some Googling (http://en.wikipedia.org/wiki/Twisted_pair) is that twisted pair helps with maintaining signal quality only when an equal and opposite signal is sent over each pair. But if there is only 1 conductor used for the data signal would it make a difference, or is it that the wires used for +12V and -12V wouldn't cause as much interference on the data conductor if they are twisted?

I have not noticed external interference to be a huge issue.  For general purposes, the 'twisted pair' is totally useless for any Tx after the uSC because as mentioned above, the data is not a differential/balanced signal.  Though I have 100's of feet of cat5 so I often use it as there is certainly no detriment to using it either.

Quote


And doing some Googling/Wiki-ing, it looks like capacitance increases by using larger cables.

But then... I remember your video where you just re-purposed a bunch of 120V extension cords to run Pixelnet and you got something like 80 feet. There is a relatively small number of twists in extension cords, and the wire gauge is much thicker than what we generally use for running pixels, so where does that fit into this inductance/capacitance explanation?   ??? 

It was basically just a demonstration.  Inductance / capacitance are not the largest portion of the issue for the mosfet drive.

Quote
I wasn't thinking that issue with distance was power and voltage drop, but the data signal loss. From what I gathered from other peoples' comments, it was implied that the pixel type (not the cable type going to the pixel) is what determined the achievable distance  to the first pixel. Would it be that the circuitry in nodes tends to be less tolerant of a "dirty" signal than what strips expect? You'd think that there would be variation from manufacturer to manufacturer.

Pixel type would figure into the large ball of factors, too.  But in general, back to the old mantra of PIC being susceptible to inductance / capacitance issues after just a few feet.  Op-amp is generally good for several 10's of feet,  Mosfet drive can generally handle the inductance/capacitance, but will run into issues with ground line loading. 

The heart of the "ground line loading" centers on the issue of voltage drop.  If you start with 12V at the power supply and measure 10V at the nodes, you say "2 volt drop" - but that is really 1 volt drop on the power line leading out to the nodes and 1 volt drop on the ground line leading back to the power supply.  The nodes see 11V on the power wire and float at 1V on the ground wire, which is 10V total. 

Nodes are usually happy with this, but the problem comes in referencing the data signal to their floating ground.  Data carries so little power, it can oscillate between almost the full 0-5V over these distances.  The node floating at 1V 'thinks' the data signal is actually swinging between -1V and 4V.  The issue gets worse with more power drawn or smaller wire gauge - either of which mean more voltage drop.  If a full 128 nodes are turned on all-white, the voltage drop may be 2.5-3V at the nodes...again split equally on both ground and power wires.  The ground may be floating at 1.5V and the node would 'think' the data is oscillating -1.5V to 3.5V.  This is the point where the nodes likely can't read the data anymore, so they would 'lock' on all white.  This is also why we used to get rainbows of color half way down the string...the ground line loads up as more and more nodes dump power on it, so if it's floating at 1V for the first node, it may be 1.2-1.3V half way down the string...so that node has trouble reading the data and passes incorrect data along.

So again...smaller wire, longer distances, higher node count, higher current draw, brighter nodes / all white nodes, etc. all contribute to the ground line loading.  Conversely, this is one of the issues the uAmp cures, because it references the outgoing data signal to the local ground, so acts as a data clarifier / regenerator / line leveler.

Offline Gary

  • Supporting Member
  • ******
  • Join Date: Jan 2015
  • Location: Chilliwack, BC Canada
  • Posts: 342
  • Kudos: 3
    • Diamond Crescent Musical Christmas Lights
Thanks for the wordy explanation--just how I like it!   ;D

Offline sirloinofbeef001

  • Sr. Member
  • ****
  • Join Date: Oct 2013
  • Location:
  • Posts: 453
  • Kudos: 0
  • believe in miracles
The after burner is for making the data go further, I'm still not sure with the 55ish distance Gary was speaking about would we need to inject power at these distances or will this depend on the amount of nodes? I'm hoping to use 150 nodes at 35' without having to inject power for all of these props. Or do you think if we use the 3wire 18 awg wire from our suppliers be ok with these distances?
Its a wonderful life.

Offline JonB256

  • Supporting Member
  • ******
  • Join Date: Mar 2013
  • Location: Granbury, Texas
  • Posts: 3,849
  • Kudos: 81
    • Granbury Christmas Lights
I'm hoping to use 150 nodes at 35' without having to inject power for all of these props. Or do you think if we use the 3wire 18 awg wire from our suppliers be ok with these distances?

That may be pushing the power limits a bit. 150 pixels, particularly full intensity and White, may reduce your voltage too much. Other than the "eyeball" test, you can check the voltage at the 150'th node. If it is still near or at 11vdc, you should be good. A few years back, I incrementally dialed down the voltage on a 120 count string of TM1804 bullet pixels and started to lose the last few pixels (flickering / reddish tint) when the INPUT voltage reached 10vdc. Voltage at the end was 9vdc.

A 16 awg wire would a pretty sure bet, but you would probably have to fab that up yourself.

"I squared R" losses in a wire are just something we have to deal with.

Offline sirloinofbeef001

  • Sr. Member
  • ****
  • Join Date: Oct 2013
  • Location:
  • Posts: 453
  • Kudos: 0
  • believe in miracles
These nodes I'm speaking of are 1804 and they have been work horses. I used them last year for my driveway arches and they worked great but only 17' to first node. I'm actually changing them to use for some 6' white wireframe trees and they have the same distance as last year so I will try them this evening to see if I can stretch the power out on them without power injection.

Offline Bshaver

  • Developer
  • ******
  • Join Date: Aug 2014
  • Location: Denver, CO
  • Posts: 1,267
  • Kudos: 21
HI, when would this be available: or did I miss this?

Offline sirloinofbeef001

  • Sr. Member
  • ****
  • Join Date: Oct 2013
  • Location:
  • Posts: 453
  • Kudos: 0
  • believe in miracles
HI, when would this be available: or did I miss this?
The afterburner? The coop is done already. Maybe Corey may have some left over.

Offline sirloinofbeef001

  • Sr. Member
  • ****
  • Join Date: Oct 2013
  • Location:
  • Posts: 453
  • Kudos: 0
  • believe in miracles
I'm hoping to use 150 nodes at 35' without having to inject power for all of these props. Or do you think if we use the 3wire 18 awg wire from our suppliers be ok with these distances?

That may be pushing the power limits a bit. 150 pixels, particularly full intensity and White, may reduce your voltage too much. Other than the "eyeball" test, you can check the voltage at the 150'th node. If it is still near or at 11vdc, you should be good. A few years back, I incrementally dialed down the voltage on a 120 count string of TM1804 bullet pixels and started to lose the last few pixels (flickering / reddish tint) when the INPUT voltage reached 10vdc. Voltage at the end was 9vdc.

A 16 awg wire would a pretty sure bet, but you would probably have to fab that up yourself.

"I squared R" losses in a wire are just something we have to deal with.
I have a question. When a f16 is in test mode is it simulating data?
I've just tested 2 sets of 1804 with 150 nodes each @ 11.75 v at the p/s. One string had 17' distance to first node the power dropped to about 8v average. The other was  30' to first node and it dropped to 7v. But neither dimmed at the last node or any discoloration. So I dropped the 30' string to 120 nodes and i gained 1v. I'm not sure of how to test for the amperage to see if im over the amperage per channel or port.

Offline tbone321

  • Hero Member
  • *****
  • Join Date: Oct 2014
  • Location:
  • Posts: 1,335
  • Kudos: 35
Just use either an amp meter or a multi-meter capable of 10A or better.  These meters are connected in series with the load so you disconnect the +12V lead from the nodes and connect the positive lead of the meter to the power source and the negative to the nodes +12V input then fire them up.

 

Back to top