Author Topic: Soap router - HW specs too good to be true?  (Read 97578 times)

0 Members and 2 Guests are viewing this topic.

Offline madires

  • Super Contributor
  • ***
  • Posts: 7695
  • Country: de
  • A qualified hobbyist ;)
Re: Soap router - HW specs too good to be true?
« Reply #125 on: July 12, 2014, 09:29:41 pm »
I had someone (who claimed to be an engineer) tell me that Ethernet was too unreliable, because:
  • If the cable is longer than 100 meters, you will blow the amplifiers.
  • If the cable is shorter than 1 meter, you will blow the amplifiers.
  • If the cable has a bend with a radius of less than an inch, the packets can't make it through.

Actually, there's some truth in the third point. Bending the ethernet cable too much degrades it's transmission properties resulting in more FCS errors (broken ethernet frames) for example.
 

Offline rob77

  • Super Contributor
  • ***
  • Posts: 2085
  • Country: sk
Re: Soap router - HW specs too good to be true?
« Reply #126 on: July 12, 2014, 09:47:35 pm »
I had someone (who claimed to be an engineer) tell me that Ethernet was too unreliable, because:
  • If the cable is longer than 100 meters, you will blow the amplifiers.
  • If the cable is shorter than 1 meter, you will blow the amplifiers.
  • If the cable has a bend with a radius of less than an inch, the packets can't make it through.

longer than 100m might not work because of attenuation of the signal and picking up noise - the receiver side might not be able to decode, other factor might be the delay caused by >100m . but i had fast ethernet links on high quality shielded cat5 with a length of 140m - without issues (there was no option to install a repeater and cost was a blocking factor for fiber).

shorter than 1 meter - that's valid only for long range optics - if you connect 2 SFPs designed for a 100km fiber link with a 1m cable - then you'll burn the receiver. but for copper.... that's bullshit.

bending the cable too much will cause errors as mentioned above, but will not burn the transceivers. bending the fiber optics will attenuate the signal , bending  it way too much will crack the fiber
 

Offline Rasz

  • Super Contributor
  • ***
  • Posts: 2616
  • Country: 00
    • My random blog.
Re: Soap router - HW specs too good to be true?
« Reply #127 on: July 12, 2014, 09:52:26 pm »
Link aggregation on layer 2 requires links with the same speed. Since WLAN connections change the link speed all the time they can't be used for link aggregation. The obvious solution would be to use IP load sharing, but that requires a sophisticated algorithm for links with different and changing speeds. Another issue with available IP load sharing solutions is, that they can't fully utilize the links. The utilization depends on the mix of IP connections. If someone develops an algorithm solving all those issues he would get a nice job offer from Cisco or Juniper ;) For me, it's just more marketing hogwash.

its coming http://en.wikipedia.org/wiki/Multipath_TCP
Who logs in to gdm? Not I, said the duck.
My fireplace is on fire, but in all the wrong places.
 

Offline duncan711

  • Newbie
  • Posts: 1
Re: Soap router - HW specs too good to be true?
« Reply #128 on: July 12, 2014, 09:54:30 pm »
I just wanted to mention that kickstarter rely purely on the backers to judge if a campaign is a scam. As long as they respect the ToS they are fine.

Any criticism on a kickstarter  project from people like you should be encourage because it is literally the only way to find out if a project is a scam or not.
 

Offline madires

  • Super Contributor
  • ***
  • Posts: 7695
  • Country: de
  • A qualified hobbyist ;)
Re: Soap router - HW specs too good to be true?
« Reply #129 on: July 12, 2014, 10:05:59 pm »
its coming http://en.wikipedia.org/wiki/Multipath_TCP

And what about ICMP, UDP, SCTP and all the other protocols? ;) A load sharing method has to support all protocols.
 

Offline BartManInNZ

  • Contributor
  • Posts: 46
  • A Geek In New Zealand
    • YouTube Channel
Re: Soap router - HW specs too good to be true?
« Reply #130 on: July 16, 2014, 07:14:11 am »
An update, English is not their first language I am guessing?
"The Bad News

We were hoping we would have our assembled board today to show off to everyone, but there has been a delay (we know its lame)"

The word should be: "it's"

The end comes soon?
« Last Edit: July 16, 2014, 07:15:56 am by BartManInNZ »
"If you want to build a ship, don't herd people together to collect wood
and don't assign them tasks and work,
but rather teach them to long for the endless immensity of the sea." - Antoine de Saint Exupery
 

Offline echen1024

  • Super Contributor
  • ***
  • Posts: 1660
  • Country: us
  • 15 yo Future EE
Re: Soap router - HW specs too good to be true?
« Reply #131 on: July 16, 2014, 12:40:51 pm »
An update, English is not their first language I am guessing?
"The Bad News

We were hoping we would have our assembled board today to show off to everyone, but there has been a delay (we know its lame)"

The word should be: "it's"

The end comes soon?
We're gonna get this line quite often.

Crowdfunding scheme has delay, no reason=no news
I'm not saying we should kill all stupid people. I'm just saying that we should remove all product safety labels and let natural selection do its work.

https://www.youtube.com/user/echen1024
 

Offline dext0rb

  • Contributor
  • Posts: 49
  • Country: us
  • grind it out
Re: Soap router - HW specs too good to be true?
« Reply #132 on: July 16, 2014, 03:27:03 pm »
Their grammar has been bad throughout the campaign; I wouldn't start harping on them now, though.

I actually do believe their latest delay story - they probably went to the assembly house with 3 (or less?) boards or something and the house was probably like "bollocks to this one-off shite!!!"
EE and Tinkerer.
 

Offline rob77

  • Super Contributor
  • ***
  • Posts: 2085
  • Country: sk
Re: Soap router - HW specs too good to be true?
« Reply #133 on: July 16, 2014, 07:28:53 pm »
I actually do believe their latest delay story - they probably went to the assembly house with 3 (or less?) boards or something and the house was probably like "bollocks to this one-off shite!!!"

does it even make sense to ask an assembly house to populate one or two boards ? isn't it better to do it by hand ? even BGAs can be soldered by hand with hot-air ;) or ask the assembly house just to solder the BGAs and do the rest by hand "in house".
when doing it by hand - one could do some basic verification during assembly - such a board will never work 100% correctly on the very first attempt.
 

Offline Precipice

  • Frequent Contributor
  • **
  • Posts: 403
  • Country: gb
Re: Soap router - HW specs too good to be true?
« Reply #134 on: July 16, 2014, 07:46:47 pm »
does it even make sense to ask an assembly house to populate one or two boards ?

They have money but a shortage of time, so I'd say it was definitely worth subbing out the assembly. Bringing up a prototype when you're not sure what's soldered down right is a nightmare. That double-pad-ring QFN they've got is more scary than the BGA.
While it's just about possible to hand-place BGAs, yield isn't 100% (far from, in many cases, especially if you're struggling to get the solderpaste perfect), this is a double sided, reasonably annoying component mix, board.

Of course, subbing out the assembly still takes time and effort, and makes it harder to just populate one section of the board at a time, if you're desperately short of silicon, or want to bring up a subsystem at a time.
Best of luck to them - it can be a hell of a long and expensive journey from first bare PCB to a product.
 

Offline dext0rb

  • Contributor
  • Posts: 49
  • Country: us
  • grind it out
Re: Soap router - HW specs too good to be true?
« Reply #135 on: July 16, 2014, 09:16:08 pm »
I wonder how the costs compare between what they are doing (board house hand assembly) vs. having a PNP machine run thru a lot of 10, lets say.
Does the NRE for all that equal labor cost for hand-assembly? They will have to come back again and have a PNP program set up eventually...

I would shoot for as close to final manufacturing process as possible in the prototype runs, so that problems can be identified and corrected sooner (Correct PNP program, solderpaste composition, stenciling...whatever else could go wrong....Murphy! :P )
Maybe I'm off though?
EE and Tinkerer.
 

Offline Precipice

  • Frequent Contributor
  • **
  • Posts: 403
  • Country: gb
Re: Soap router - HW specs too good to be true?
« Reply #136 on: July 16, 2014, 09:54:26 pm »
For what it's worth, I've got both PnP machines and manual gear.
For a run of 10 of those boards, I'd go for PnP - but I don't have access to minimum-wage labour to do the hand assembly - that might swing things back to manual. Perhaps all the easy components (passives, easy silicon) by machine, and the tricky chips and connectors by hand (well, silicon on the BGA / CSP placement rig, connectors by hand). That volume and complexity does put the job safely into 'there's no ideal way, so it's going to be a bit expensive' category.
Component sourcing for small batches is a pain, too - stuff tends to turn up in utterly machine hostile forms - short strips, single chips in bags, and just plain wrong, all of which pushes the balance towards manual assembly. Having the machines toss away components because they're wrong is more annoying than having hand assembly stall while people scratch their heads for a while.
Going to 'real' production methods too early (and this really does feel like there are a few spins to go) can just be a distraction. Still, it'll give the hardware team something to do while the softies work their way through that enormous and ever-growing 'spec'.

Edit: Obviously, this is just me. The Soap guys are obviously much closer to the project and know much more detail.
« Last Edit: July 16, 2014, 10:00:49 pm by Precipice »
 

Offline dext0rb

  • Contributor
  • Posts: 49
  • Country: us
  • grind it out
Re: Soap router - HW specs too good to be true?
« Reply #137 on: July 16, 2014, 10:02:48 pm »
For what it's worth, I've got both PnP machines and manual gear.
For a run of 10 of those boards, I'd go for PnP - but I don't have access to minimum-wage labour to do the hand assembly - that might swing things back to manual. Perhaps all the easy components (passives, easy silicon) by machine, and the tricky chips and connectors by hand (well, silicon on the BGA / CSP placement rig, connectors by hand). That volume and complexity does put the job safely into 'there's no ideal way, so it's going to be a bit expensive' category.
Component sourcing for small batches is a pain, too - stuff tends to turn up in utterly machine hostile forms - short strips, single chips in bags, and just plain wrong, all of which pushes the balance towards manual assembly. Having the machines toss away components because they're wrong is more annoying than having hand assembly stall while people scratch their heads for a while.
Going to 'real' production methods too early (and this really does feel like there are a few spins to go) can just be a distraction. Still, it'll give the hardware team something to do while the softies work their way through that enormous and ever-growing 'spec'.

Cool, thanks for the good insight.
EE and Tinkerer.
 

Offline BartManInNZ

  • Contributor
  • Posts: 46
  • A Geek In New Zealand
    • YouTube Channel
Re: Soap router - HW specs too good to be true?
« Reply #138 on: July 16, 2014, 10:07:38 pm »
The backers on kickstarter are starting to ask the hard questions - without answer from the soap crowd... how long before this overflows into IGG?

I would really like to believe that Soap can deliver on all they have promised but I have serious doubts whether they will actually come up with anything other than excuses.
"If you want to build a ship, don't herd people together to collect wood
and don't assign them tasks and work,
but rather teach them to long for the endless immensity of the sea." - Antoine de Saint Exupery
 

Offline Smokey

  • Super Contributor
  • ***
  • Posts: 2540
  • Country: us
  • Not An Expert
Re: Soap router - HW specs too good to be true?
« Reply #139 on: July 16, 2014, 11:09:50 pm »
I can take a guess as to how much assembly would be for a small run. 
We just went through a small prototype run of boards.
Our Board stats = around 80 line items, around 350 components total, 3in x 8in PCB, mostly 0402s TSSOPs and a 150+ pin TQFP.  No BGAs.

NRE = 2 stencils (top and bottom) + programming + documentation = around $1200USD.  The crappy part of doing a small prototype run is that you pretty much KNOW you will be changing something so you will have to pay this again next go.
Machine Assembly Cost = $350 USD per board.  I'm not sure how much extra x-ray inspection of the BGAs would be.  You would need to add that in, plus I guess it's technically optional so you could be cheap and just pray they all got soldered right.

So assuming a run of 5 boards, that's somewhere around $3000USD for the build, just for assembly cost.  You have to add in BOM cost to that too, so if you have $200USD in parts cost per board because you could only order small expensive quantities for your prototype build, that's another $1000 USD right there.

Call it a round $4000USD total to get 5 boards built.  That's my wild ass guess.  By doing this small run, you not only get a set of boards to confidently debug but you also have a first pass at working out any potential assembly issues that might come up.  You don't want to find out you put a connector too close to the edge so they can't cut the rails off or forgot the fiducials after you already ordered 10,000 PCBs and have the machines all loaded up.  You want to work that out in the first 5.

One last assumption:
Assuming they are actually legitimately trying to make a real product in the real world.... which is a BIG assumption...
For a group of guys that took in hundreds of thousands of dollars in funding money, I think that's an easy decision to do a small professional assembly prototype run to work the bugs out quickly.
 

Offline dext0rb

  • Contributor
  • Posts: 49
  • Country: us
  • grind it out
Re: Soap router - HW specs too good to be true?
« Reply #140 on: July 18, 2014, 09:01:12 pm »
http://venturebeat.com/2014/07/17/listening-to-the-crowd-proves-a-slippery-slope-for-soap/

Quote
Yet Jones appears deaf to reasonable criticism that Soap is growing out of control. Following his passion for devices that use mesh networks to increase their overall computing power and reliability, he touts the fact that multiple Soap devices can leverage this ability. When we asked why someone would want more than one Soap, he said, “When you buy two of these devices they’ll work together as one superior model,” but to our follow-up question of why a router would need to be so powerful in the first place, he replied, “I actually agree with you on that. It’s that these people love these numbers, they love these specs to the teeth. We were trying to appeal to that early adopter.”

Translation: "Let's come up with as much bullshit as possible."
 :--
« Last Edit: July 18, 2014, 09:03:30 pm by dext0rb »
EE and Tinkerer.
 

Offline madires

  • Super Contributor
  • ***
  • Posts: 7695
  • Country: de
  • A qualified hobbyist ;)
Re: Soap router - HW specs too good to be true?
« Reply #141 on: July 18, 2014, 10:06:09 pm »
And another goodie:

Quote
Your normal router is a thousand times more vulnerable than Soap.

First provide a working Soap with all software features and then ask a team of security professionals to perform a penetration test. And don't forget that most SOHO routers run Linux. On which OS is Android based? Ah, yes, Linux. And what about the Soap apps? Are you able to write 100% secure code? How do you make sure that the apps are secure? The costs for that are much higher than your KS funding. Please stop promising or claiming things you can't deliver!
 

Offline dext0rb

  • Contributor
  • Posts: 49
  • Country: us
  • grind it out
Re: Soap router - HW specs too good to be true?
« Reply #142 on: July 25, 2014, 03:58:53 pm »


Quote from: chris.indiegogo1
Question: isn’t the micrel ksz8895mq only capable of 10/100 Ethernet speeds?

Quote from: Micrel
KSZ8895MQ
Integrated 5-Port 10/100 Managed Ethernet Switch with MII/RMII interface
Huh. Where did Gigabit LAN go?

 :-//
EE and Tinkerer.
 

Offline Kean

  • Supporter
  • ****
  • Posts: 2054
  • Country: au
  • Embedded systems & IT consultant
    • Kean Electronics
Re: Soap router - HW specs too good to be true?
« Reply #143 on: July 25, 2014, 04:05:31 pm »
Quote from: chris.indiegogo1
Question: isn’t the micrel ksz8895mq only capable of 10/100 Ethernet speeds?
Yeah, I just re-posted that Q on the KS comments.  It does kind of affect one of their main features.
Mind you, that is probably the least of their problems at this stage...
 

Offline rob77

  • Super Contributor
  • ***
  • Posts: 2085
  • Country: sk
Re: Soap router - HW specs too good to be true?
« Reply #144 on: July 25, 2014, 04:15:10 pm »
2 weeks to populate one side of a single pcb ? they should have sourced the parts for the very first prototype as soon as possible (even from various sources in single piece quantities) to verify the PCB design... now they spent 2 weeks and still might realize the PCB needs a re-spin....  :palm:
and actually dropping the GigE ? no comment ;) at the end of the day they will most probably have to drop 70% of the promised features to deliver at least something without a 2 years of delay.
 

Offline dext0rb

  • Contributor
  • Posts: 49
  • Country: us
  • grind it out
Re: Soap router - HW specs too good to be true?
« Reply #145 on: July 25, 2014, 04:43:14 pm »
I'm sitting here looking at that big honkin' 128 TQFP Micrel part and I'm thinking to myself "Man, what a huge and old looking package..."

Checked it at Digi-key...
Quote
Part Status: End of Life

 :palm:

Are you guys serious? You put in a EOL part on a new design...and it's not even 10/100/1000 ??? What the shit?
Man, if you really had "partners", and you met with them, they should be telling you, "No, don't put that Micrel into the design, it's going EOL soon. Here's an alternative part."


 
EE and Tinkerer.
 

Offline Kean

  • Supporter
  • ****
  • Posts: 2054
  • Country: au
  • Embedded systems & IT consultant
    • Kean Electronics
Re: Soap router - HW specs too good to be true?
« Reply #146 on: July 25, 2014, 05:04:20 pm »
Are you guys serious? You put in a EOL part on a new design...and it's not even 10/100/1000 ??? What the shit?
Man, if you really had "partners", and you met with them, they should be telling you, "No, don't put that Micrel into the design, it's going EOL soon. Here's an alternative part."

LOL!  This has to be theatre just to keep the money pouring into the IGG campaign.  And they should be called out on it!

No way can they build all of these for the ridiculously discounted prices they've been offering without significant additional funding - which is almost certainly what they've realised and why they've added all the premium options.

In the end they're probably happy to just build a handful for themselves, even if slightly underspec'd, and pocket the change.
 

Offline andtfoot

  • Supporter
  • ****
  • Posts: 352
  • Country: au
Re: Soap router - HW specs too good to be true?
« Reply #147 on: July 25, 2014, 05:14:41 pm »
For example we could be looking at :
The placement of 'osc5/osc6'  6 seems most likely to clash with the metal case on the RJ45 pins after assembly, then there is the clash with the SMT on the right of osc6.( not to mention on most of the giga routing chips it specifically states to keep ALL oscillator circuits away from the data tracks
I'm curious to see how this ends up with osc6. It almost looks like the connector on the other side of the board overlaps onto the through-hole pins as well...
 

Offline dext0rb

  • Contributor
  • Posts: 49
  • Country: us
  • grind it out
Re: Soap router - HW specs too good to be true?
« Reply #148 on: July 26, 2014, 07:37:33 pm »
Quote from:
This one you have seen has the 10/100 on one of the ports and gigabyte on the rest but we are making changes in our next product run to up the speeds but did not need it in this version for our development... [long run-on sentence]

I don't understand this course of action at all. A 5-port 10/100 switch was used for "one of the ports" ??

Why would you put this part down, test it, develop with it, then rip it out and put a new part, test it, develop with it. Isn't that just a waste of time? How does doing this keep the costs down? Maybe I'm missing it, doesn't make sense to me though.
EE and Tinkerer.
 

Offline CanadianAvenger

  • Regular Contributor
  • *
  • Posts: 179
Re: Soap router - HW specs too good to be true?
« Reply #149 on: July 27, 2014, 04:07:01 am »
Not that I think these guys are legit, but GigE switch silicon seems to be hard to come by in low quantities. It's a problem I've been having for my work. None of the chips I've found appear to be available, unless you are talking quantities in the 50K range. [I have  a few more parts to chase down, but reps have been slow to respond] If anyone has a line on one that is readily available [in sub 100pc quantities], I'd love to hear from you. [Specifically I've been looking for 5 port unmanaged solutions (or 4+1)]
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf