#eevblog - 2018-12-14 - AustNet Archive Project

// #eevblog — 2018-12-14

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
« 2018-12-13 2018-12-15 »
09:04 Quit: DogWomble ([email protected]) EOF from client
58:39 <IoTPanic> "I like guys with a bit of muscle :)" @steve30 2018
23:11 <DHess_US> Ooo, https://mikrotik.com/product/crs305_1g_4s_in
23:41 <Ampera> I'd be scared to know the price.
23:51 <DHess_US> $150
24:43 <DHess_US> WTH? It even has dual +12 volt inputs on the back for redundant power.
25:04 <Ampera> 150 USD actually isn't as bad as I'd expect.
25:07 <DHess_US> Who does that?
25:19 <DHess_US> That is fabulous for a 10 gigabit switch.
25:30 <Ampera> couldn't you just stick the redundant supplies on the same line?
25:44 <DHess_US> And SPF+ is cheaper than RJ-45s
26:28 <Ampera> Indeed it is, sadly.
26:35 <Ampera> although that's not RJ-45.
26:46 <DHess_US> I think the idea was that you could use two starndard 12 volt wall adapters or a wall adapter and 12 volt batter plus charger. MicroTik does WISP stuff.
27:10 <DHess_US> I mean the 10 gigabit standard which uses RJ-45s is a lot more expensive than using SPF+
27:15 <Ampera> I guess you could, might simplify a battery switchover.
27:35 <DHess_US> Well, with two 12 volt DC inputs, you can configure it however you want.
27:44 <Ampera> Yeah, it doesn't use RJ-45, RJ-45 isn't the 1000/10000-BASE-T connector you're thinking of
28:04 <DHess_US> Oh, I know. But everybody calls it that by default.
28:05 <Ampera> 8P8C, that's the name
28:31 <DHess_US> And my crimper is marked RJ-45 and not 8p8c
29:27 <Ampera> because they are nearly identical, but (some) tech people get upset if you call 8P8C for computers RJ-45, which is a keyed connector for telephony.
29:35 <DHess_US> I've been debating between dual or quad gigabit teaming and 10 gigabit. The problem with teaming is Windows doesn't support round-robbin.
30:12 <DHess_US> In theory the Windows network driver can do it and supposedly some do but which? Nobody knows or says.
30:14 <Ampera> Just bite the bullet and run a few dedicated card to card lines on SPI.
30:32 <DHess_US> SPI?
30:34 <Ampera> You can always expand it to an entire network later on, but it depends on what your needs are
30:37 <Ampera> *SPF
30:54 <DHess_US> Yea, and when I want more than two systems?
31:35 <DHess_US> I actually did that long ago on 1 gigabit when 1 gigabit switches were really expensive.
32:50 <DHess_US> Huh, I didn't know SPF+ cables had DRM.
34:15 <DHess_US> https://www.youtube.com/watch?v=MDiiHN0MPdA
35:28 <Ampera> get a switch I guess.
35:42 <Ampera> I didn't know you could get them that cheap
36:00 <DHess_US> Or stick a bunch of cards in a PC running BSD and use a PC as a switch.
36:16 <DHess_US> It's new
37:08 <Ampera> That's what I would do, if switches weren't that cheap.
37:09 <DHess_US> Since MicroTik does WISP stuff, I'm thinking they made this for remote sites where 1 gigabit is not sufficient for some WiFi APs now.
37:30 <mattinx> oh god, not /this/ argument again
37:31 <DHess_US> That explains PoE, dual 12 volt power
37:45 <mattinx> at least there's no Jobe to weigh in on it here
38:21 <mattinx> DHess_US: I know cisco devices are damn picky about using cisco SFP+ modules
38:22 <DHess_US> And the 2.5, 4, and 5 gigabit stuff isn't common enough to have come down in price anyway.
38:35 <DHess_US> mattinx, yea, he just mentioned it in the video. Intel does the same thing.
38:48 <DHess_US> The SPF+ port is authenticating the cable.
39:01 <DHess_US> I didn't even know that was a thing.
39:26 <DHess_US> The switch being discussed doesn't care but the network card on the other end does.
39:44 <mattinx> yeah, I had a customer run into that
40:00 <mattinx> they ended up switching to fibre and throwing cisco SFPs in the cisco end
40:07 <mattinx> their SAN didn't care
40:33 <DHess_US> I laughted when the USB power delivery standard came out. The first USB 3 chipsets advertized explicitely listed DRM support for chargers.
40:55 <mattinx> hrm
41:01 <mattinx> unmanaged 10Gbe?
41:16 <mattinx> not so sure about that
41:25 <mattinx> however, I'll give the mellanox NICs a big thumbs up
41:28 <DHess_US> I mean literally the first USB 3 controllers advertised in the trade magazines listed DRM for charging as a feature.
41:28 <mattinx> big fan of their gear
42:12 <DHess_US> Before that they had a race for who could lock out foreign chargers but now it is all standardized. Woohoo!
42:42 <mattinx> don't think I've had an issue with intel cards only working with intel cables
42:50 <mattinx> but I've certainly seen cisco not play ball
42:57 <DHess_US> I haven't messed with it. If Windoes supported round-robbin link aggrigation, I wouldn't even consider 10G but since it basically doesn't.
43:27 <mattinx> check out fs.com for SFPs too
43:39 <mattinx> link aggregation is meh
43:58 <DHess_US> It would be cheaper than 10G if everythign supported it.
44:33 <DHess_US> Windows doesn't natively support round-robin. In theory some network card drivers do but figuring out which is basically impossible.
44:50 <mattinx> he didn't mention bend radius for the cables
44:51 <DHess_US> And I'm not interested in buying every make and model to test.
44:55 <mattinx> that's something that's worth a mention
45:24 <mattinx> both fibre and the copper twinax don't like sharp bends
45:38 <mattinx> and at 10Gbe speeds, it's important :)
45:52 <mattinx> Intel's drivers do
46:23 <DHess_US> You wouldn't know it from Intel's specifications or documentation.
46:24 <mattinx> Intel's Gbe drivers basically support the same modes as the linux bonding module does
47:00 <DHess_US> Some of Intel's cards might. The ones I have that I checked do not.
47:24 <DHess_US> I got the feeling that they dropped that feature for newer cards.
47:33 <mattinx> could be possible
47:38 <mattinx> I've not touched it in several years
47:55 <mattinx> in fact, I think the last box I set it up on was running server 2003
48:17 <DHess_US> Exactly
48:21 <mattinx> but it was only a year or two before it went eos, so it wasn't /that/ long ago :)
48:44 <mattinx> here we go
48:46 <mattinx> Intel ANS isn't supported on Windows Server 2016*. Microsoft Windows Server 2012 R2* is the last Windows Server operating system version that supports Intel ANS
49:03 <mattinx> but that's what they call it
49:24 <mattinx> https://www.intel.com/content/www/us/en/support/articles/000005667/network-and-i-o/ethernet-products.html
50:23 <DHess_US> The question came up when I had to copy everything off of my old NAS box which because it runs Linux supports it. The only solution I was sure of was to use another Linux box to bridge between the NAS and a 10Gbe network.
50:29 <mattinx> actually, possible that it doesn't do RR, but it does RLB and ALB
50:33 <mattinx> which are a step up
52:15 <DHess_US> "The Intel® ANS software load balances transmissions, based on Destination Address" Not the same thing
52:58 <DHess_US> None of those modes do it.
53:39 <mattinx> hrm, indeed
54:20 <DHess_US> round-robbin does eactly that. The problem and why most systems don't support it is that packets can arrive out of order which hoses the TCP/IP stack if it is not written properly.
55:08 <mattinx> you can get away with 802.3ad, so long as the devices in question support layer 3+4 hashing, and your application can open multiple streams
55:12 <DHess_US> It becomes a massive performnce problem on the receiving end. Windows does not handle it well unless the driver does it and maybe not even then.
55:39 <mattinx> cause you end up hashing based on ip and port then
55:43 <DHess_US> Yes, and that works in applications where multiple streams are being used.
55:59 <DHess_US> But if you are just doing big file transfers, that doesn't help.
56:08 <mattinx> depends what you use to do the copy :)
56:18 <mattinx> if you're using NFS/CIFS, not so much
56:36 <mattinx> but there's third party stuff out there that'll do multiple streams
56:45 <DHess_US> Anyway, Microsofts solution to their TCP/IP stack sucking was to not support it.
57:30 <DHess_US> So in the past, some network cards supported it instead. But as far as I know, that is no longer the case.
57:47 <DHess_US> Solution? SPF+
58:11 <DHess_US> But that doesn't help if your old hardware isn't 10Gbe
58:57 <DHess_US> Unless you build a briding system which would be interesting but for the cost, I could just build a new NAS box with support for 10Gbe.
00:42 <mattinx> 10Gbe really is a better solution
01:20 <mattinx> the only place where link aggregation works acceptably in my experience is the application it was designed for - uplinks between switches
01:54 <mattinx> where you're looking at many-many connections
01:56 <DHess_US> Originally it came up when I was studying my NAS in detail to see if I could speed things up. "Oh, it supports round-robbin teaming! But only with Linux systems!"
02:15 <mattinx> server uplink when you're talking to a large number of workstations is a close second
02:44 <DHess_US> Yea, like I said, the reason round-robbin isn't generally supported is that it is very hard on the receiving end's TCP/IP stack.
02:49 <mattinx> but 90% of the time, the people asking for link aggregation are wanting to do it for increased bandwidth between a small number of machines, and it doesn't work to well for that
03:58 <DHess_US> Alternatively, it makes Microsoft look bad and Linux look good so Microsoft killed it.
04:37 <mattinx> lol
05:34 <DHess_US> I'm not really joking. MS has enough influence on the standards committees to kill something like that.
05:50 <DHess_US> Why wouldn't they kill a standard which shows how crap their software is?
06:44 <DHess_US> They have done that with other things. They removed the free space indicator on the status bar because it gave users "free space anxiety". They literally said that.
07:22 <DHess_US> They removed the API which allows displaying folder sizes because of performance reasons.
08:03 <DHess_US> Which coincidently also was competition for their database as a file system which never got released because it was crap.
08:14 <mattinx> lol
08:17 <mattinx> I didn't hear about that one
10:08 <DHess_US> DMVs and hospitals and Disney removed clocks from view where people wait because it remonded people about how fucking slow their service is. Problem solved.
12:40 <DHess_US> https://imgur.com/gallery/apHJ5VZ
12:59 <mattinx> casinos do the same
13:45 <DHess_US> <mattinx> unmanaged 10Gbe?
13:51 <DHess_US> Actually the switch is managed somewhat.
14:25 <DHess_US> " If you would like the ability to use routing and other Layer 3 features in your CRS, use RouterOS."
17:01 <mattinx> ah
17:21 <mattinx> mainly I want things like VLANs, QoS and STP
17:42 <DHess_US> Performance is limited of course when used as a router but as a switch, the video did show 1GB/s when the endpoints could support it.
17:50 <DHess_US> STP?
17:56 <mattinx> spanning tree
19:54 <DHess_US> https://wiki.mikrotik.com/wiki/Manual:CRS3xx_series_switches
21:08 <mattinx> ah, it would appear they do support STP/RSTP/MSTP, VLANs, and QoS
21:11 <DHess_US> So VLAN, STP, and QoS are supported at least to some extent.
21:30 <mattinx> port mirroring, storm contro, ACLs, port security
21:45 <mattinx> those are things I care about :)
21:48 <DHess_US> It's basically just one of their router boards with 4 STP+ ports and 1 Gbe port.
21:54 <mattinx> so yeah - that looks like a nice little box
42:24 <DHess_US> https://www.shacknews.com/article/108729/what-is-the-fallout-76-refund-class-action-lawsuit-about
42:55 <DHess_US> mattinx, I didn't know MicroTik made so many switches I might use. I'll have to consider them now.
24:51 <aram> I used to be a big microtik fan, but after 1) the repeated security problems, 2) repeated GPL violations, 3) lack of desire to implement 802.11r, k, and v, I am not such a fan anymore.
24:59 <aram> I find 802.11r, k and v indispensable today
29:02 <DHess_US> I forgot abou the GPL stuff.
30:51 <DHess_US> I was just thinking of their switches. I wouldn't use their wireless stuff anyway.
31:32 <DHess_US> I haven't been completely happy with the last gigabit switch I bought, a netgear GS108T. It works but I think it has performance problems.
39:50 <IoTPanic> You would have a seizure if you saw my networking setup in my office
40:32 <aram> https://www.cnx-software.com/2018/12/13/boom-risc-v-core-amazon-ec2-f1/
40:35 <aram> interesting
40:39 <aram> albeit expensive
41:20 <IoTPanic> Im more excited for the ARM instances
41:54 <IoTPanic> But those are cool
42:54 <aram> I can't seem to find out precisely what FPGAs are supported by this thing
43:44 <IoTPanic> Oh, I didnt see the FPGAs
44:45 <IoTPanic> Huh, that is the advantage to me, put a software workload on hardware
44:53 <IoTPanic> More interesting now
18:01 <DHess_US> I took a careful look at RISC-V and the ISA would perform poorly in the application I had in mind. It is just not even cost effective compared to an older and slower ARM for me.
19:01 <DHess_US> The only reason I would use it is to build it into a SoC which of course I'm not doing and even then, changing the design to use an external ARM would be better.
39:57 Join: Polesch ([email protected]) to #EEVblog
58:40 <IoTPanic> Well, imagine you were doing something like weather calculations, offload the math to the FPGA
59:45 <DHess_US> Why would I do that when a cheap ARM core can do it?
00:19 <DHess_US> That was the problem. RISC-V is intended to be customized ... or I could use a cheap ARM microcontroller.
00:53 <DHess_US> If I was offloading the math, then how is RISC-V an advantage except if I am designing a SoC?
01:33 <IoTPanic> Because the FPGA can have an almost CPU like ALU where the instructions are specific to your application, bringing up efficiency and increasing performance. And Im not saying the RISC-V has an advantage, Im talking about The AWS service as w whole
01:55 <IoTPanic> You wouldnt have any advantage
02:29 <IoTPanic> If I could use FPGAs in AWS, I think I could do some cool stuff, I dont care a RISC processor is there.
03:08 <DHess_US> I couldn't because I would have to ship the data there before processing. That alone would kill it.
03:08 <IoTPanic> I would love to play with RISC-V and all, but if there was an ARM with FPGA service on AWS I would take it instantly
03:18 <IoTPanic> Sure, you couldnt
03:26 <IoTPanic> It is pretty application specific
03:31 <IoTPanic> All of my data is already on AWS
03:55 <DHess_US> I keep updating my design for the perfect DSO.
05:02 <IoTPanic> Okay, in the cloud, you can change everything with an update so DSO is irreverent. In a piece of hardware a consumer buys, I do not see a use case. Only for cloud applications
05:20 <IoTPanic> Then just cheap ARM processor like you said
05:39 <DHess_US> The use case was for RISC-V versus ARM.
06:25 <IoTPanic> AWS provides both, just the RISC-V instances have FPGAs, that sit
09:54 <IoTPanic> I wish they had a FPGAs as a service
15:37 <DHess_US> There was an expectation that Intel was going to do that after they bought Altera but it never came about.
16:14 <DHess_US> LIke ARM's heterogenous computing never came about either after they integrated their GPUs with their CPUs.
16:32 <DHess_US> The clostest you can get is nVidia's OpenCL stuff.
21:13 <Halcyon> Morning all
53:03 <IoTPanic> @DHess_US Huh. and morning @Halcyon
14:11 <aram> obviously at this point RISC-V is not competitive
14:21 <aram> the question is whether i can become competitve one day
14:28 <aram> s/i/it/
14:53 <aram> risc-v is clean, but TBH mips is also very clean
15:00 <aram> risc-v is very similar to mips
15:38 <IoTPanic> Dont even get DHess_US started XD He will list each and every reason it wont.
15:39 <aram> they wanted to play with new ISAs, which is fine, and something I approve of
15:55 <aram> technical monocultures are not a good thing
16:37 <IoTPanic> Indeed, and x86 is slowing down a LOT
16:49 <aram> as a kernel developer, there are some dumb things about risc-v I don't like
17:01 <aram> too many things are implementation-defined instead of being standadized
17:40 <IoTPanic> Development wise I mean. And I did not know a lot of parts are implementation defined.... That kinda sucks
18:48 <aram> they want to play it safe and allow flexibility later. but it's "design by committe" mentality. sometimes it's far more useful to make choices.
20:37 <IoTPanic> I would feel better to have instead of flexibility for the future, standardize everything and once they learn what they do or dont need do a final spec that allows anyone to interface anything as long as they fallow the specification
21:04 <aram> yeah
21:04 <IoTPanic> Obviously "anything" being a realistic category of components
22:09 <IoTPanic> But what do I know? Im not a CPU designer
24:22 <IoTPanic> Lol, when elon musk goes head to head with trump
24:24 <IoTPanic> https://www.nytimes.com/2018/12/12/us/politics/trump-texas-border-wall-.html
24:59 <IoTPanic> What even is that URL? trump texas border wall isnt the title of the article.
27:17 <DHess_US> aram just listed one of my reasons. "similar to MIPS"
28:01 <IoTPanic> Oh yeah, I just know how much you like chiming in on RISC's pitfalls, which I like. I hear all about how great it is but no one wants to throw a stone
29:05 <DHess_US> Eh, they just made the same mistakes most of the other RISC designs made. My list actually came from Linus. I"m not a CPU designer but he at least ran into these problems over at Transmeta.
29:25 <DHess_US> The Alpha guys had it right even though they made mistakes also.
29:33 <DHess_US> Throw away the first design. Then throw away the second.
30:00 <IoTPanic> What do you do for work DHess_US
30:28 <DHess_US> EE stuff. Which means I ended up writing low level code where performanc ematters.
30:45 <IoTPanic> Nice, so what I try to be XD
31:24 <IoTPanic> Throw in that linkin park song about failure
31:40 <DHess_US> Everybody kept tell me how good RISC-V is and I just see the same mistakes repeated.
21:08 Join: DogWomble ([email protected]) to #EEVblog
21:08 Mode: (by ChanOP) +o DogWomble
08:34 <IoTPanic> Has anyone used a multitech conduit?
08:55 <IoTPanic> Its been flashing mLinux for an hour and im starting to get anxious something went wron
08:56 <IoTPanic> g
09:07 <IoTPanic> With the least helpful documentation ever
01:22 Join: HexicPyth ([email protected]) to #EEVblog
25:28 Join: Chipguy ([email protected]) to #EEVblog
25:35 Quit: Chipguy ([email protected]) Quit: Connection closed by foreign host
57:36 Quit: ServalSpots ([email protected]) EOF from client
21:45 <aram> https://www.bmj.com/content/363/bmj.k5094
21:48 <aram> hilarious
37:56 <T3sl4> "it's not the speed that kills, it's stopping"
44:45 Join: MrPokeboy8 ([email protected]) to #EEVblog
50:35 Quit: HexicPyth ([email protected]) Quit: Leaving
53:14 Join: JohnSmithVII ([email protected]) to #EEVblog
56:07 <JohnSmithVII> What should the characteristic impedance of the parallel LC circuit be in relation to the two bypass capacitors in a Hartley oscillator? (i.e lower, ~7x higher, etc.)
04:09 <T3sl4> which ones?
04:28 <JohnSmithVII> Here is the schematic of the oscillator in question: https://imgur.com/a/HXxF59q. It simulates well in LTSPICE but doesn't oscillate in real life.
05:11 <T3sl4> ouch, really trying to punish the gate
05:38 <IoTPanic> Thank you for that aram
05:55 <IoTPanic> That is really good research
10:43 <JohnSmithVII> I found that the LTSpice "educational" Hartley schematic file featured an LC Circut with an Impedance of ~1.5KΩ, roughly 20% that of the gate capacitor and 5x that of the drain capacitor at it's resonant frequency.
14:19 Quit: eddie ([email protected]) EOF from client
18:08 <IoTPanic> Impedance of 1.5K?
18:36 <IoTPanic> Ω anyway
18:47 <DHess_US> T3sl4, rapid deceleration trauma, concrete poisoning
19:15 <JohnSmithVII> Zo=sqrt(30uH+30uH/250p)=1.549kΩ
20:07 <T3sl4> DHess_US: similarly with most incidences of acute lead poisoning
20:24 <T3sl4> JohnSmithVII: why has the schematic been labeled with impedances, yet no frequency is defined?
21:17 <IoTPanic> I understand how most of it is supposed to oscillate, but looking at the entire thing it doesnt make sense to me, but im really bad with oscillators so I wouldnt take my word
21:36 <JohnSmithVII> T3s14: I labeled the schematic when I was tired and did the calculations on a calculator. f=5.55Mhz
21:57 <JohnSmithVII> Which incidentally assumes the capacitor is 100pF higher than it is, oops
22:10 <IoTPanic> L4 is what is what is causing a question mark in my head.
22:11 <T3sl4> seems to be assuming the freq of 2uH + 300p? not accounting for the actual frequency then, which is probably lower due to J1 capacitances and C2 loading
22:26 <T3sl4> IoTPanic: just an RFC
24:23 <JohnSmithVII> T3sl4: Yes, I was assuming ideal capacitors and inductors, I'm just trying to get an idea of why this circuit doesn't work and the example one (presumably) does.
26:50 <DHess_US> In the real circuit you fried the 2N4391.
28:48 <DHess_US> Even with the best case, you exceeded the power dissappation of the transistor by twice. Worst case is 6 times.
29:48 <DHess_US> And I don't know what inductor you used for L4 but it could have saturated.
30:18 <JohnSmithVII> Dhess_US are you referring to the 9V power supply? It's just a 9V battery which I tested to peak at 2A for a few milliseconds, I doubt it could source that for long.
30:48 <DHess_US> 50 milliamps is best case and could fry the 2N4391.
30:50 <IoTPanic> Why the hell do you need 2 amps for this?
30:59 <JohnSmithVII> I don't
31:15 <IoTPanic> Ah. I see, just a 9V battery
31:23 <IoTPanic> Well, thats just setting you up for failure
31:47 <JohnSmithVII> I can use a linear power supply too, but that doesn't work either.
32:08 <DHess_US> Add a bypassed source resistor to limit the drain current or use a 2N4393.
35:22 <JohnSmithVII> In reality I pretty much copied that section verbose from the example(https://imgur.com/a/mK7hn6R) and changed the transistor because the one the example used is obsolete as far as I can tell. I will add a bypassed source resistor.
36:34 <DHess_US> Any JFET will work. Look for one with a low Idss (saturated drain current)
37:09 <DHess_US> Usually they are graded into three versions for high, medium, and low Idss
37:37 <DHess_US> Although the 2N4391 series is intended for switching appliations so they are all pretty high.
39:28 <JohnSmithVII> I already have the 2N4391s though. Am I reading the datasheet correctly in that I should use a resistor value which limits gate-source current to a maximum of 150mA? This datasheet is particularly unhelpful as it does not show the spelled out acronyms in the 'symbols' column. (datasheet: https://www.mouser.com/datasheet/2/68/2n4391-4393-53044.pdf)
39:59 <DHess_US> Not the gate-source current but the drain current.
40:25 <IoTPanic> I was just looking at a much better datasheet
40:29 <IoTPanic> Wait one moment
40:37 <DHess_US> The 2N4391 needs more negative bias on the gate to lower the current and the easy way is to raise the source voltage.
41:12 <IoTPanic> http://interfet.com/datasheet/2N4391/
43:57 <T3sl4> increasing V1 Rser to 100 ohms for safety, and a couple of other minor differences, i find 4.95MHz and 3.3Vpp on the drain
44:11 <T3sl4> (probably one of the differences being somewhat higher inductance)
44:33 <T3sl4> it works far better with D2 --> 47kohms or thereabouts
44:42 <T3sl4> (4.92MHz, 8.48Vpp)
46:26 <T3sl4> and yeah, datasheets assume you know what they are, you need to look up mfg's info for descriptions
46:30 <T3sl4> on semi is usually good with that
48:05 <JohnSmithVII> Should I add a series ~100 ohm resistor to V1? I don't know of any power source with a 100 ohm ESR :)
53:49 <JohnSmithVII> I find that with 100Ω ESR SPICE doesn't show oscillation
54:12 <JohnSmithVII> *SR
56:01 <JohnSmithVII> Oh okay it works with 18V DC at 100Ω
59:13 <T3sl4> ¯\_(ツ)_/¯
59:38 <T3sl4> it's your responsibility as the user, to construct a model of the circuit you're trying to analyze
59:49 <T3sl4> SPICE isn't a magic oracle
00:27 <T3sl4> "Pray tell, Sir Babbage, if one gives [the computer] incorrect numbers, will the correct results still come out?"
02:12 <JohnSmithVII> I'm very aware, I just find it helpful to simulate my circuts before I make them because SPICE gives you a free infinite-channel infinite-bandwidth sensitive to femtovolts oscilloscope, free unlimited transistors, etc,
03:17 <T3sl4> it also gives you transistors with sometimes completely wrong models (i don't know offhand how well qualified LTSpice's models are), and no package capacitance or inductance
04:56 <IoTPanic> I do not use LTSpice, but I have found to use simulators to help find a circuit that works, then figure out the correct values after. And a good point @T3sl4 makes is LTSpice assumes everything is perfect and it doesnt sometimes know your fet is gonna blow
05:35 <T3sl4> 33pF may simply be too small compared to the FET
05:44 <T3sl4> i have a PN4392 here and it's stronger with 100p there
06:41 <T3sl4> and again, better without the diode, i have no freaking clue why that was there in the first place, whoever drew it was an idiot
07:02 Quit: MrPokeboy8 ([email protected]) Quit: http://www.kiwiirc.com/ - A hand crafted IRC client
08:42 <IoTPanic> I was wondering what the diode was doing
09:25 <IoTPanic> But legal cannabis makes it so if I say everything I wonder Ill sound really stupid
09:39 <IoTPanic> Thank you maine.
10:13 <IoTPanic> Thats a cool little oscillator though
10:23 <T3sl4> you might see that in a BJT oscillator, where: 1. the base drive is supposed to be more squared-up (maybe a full-throttle blocking oscillator more than a resonant one?), 2. the coupling capacitor serves as some phase shift, because 3. the coupling cap sets base drive power so it has to be relatively small
11:09 <T3sl4> you NEVER do that with a JFET, because it needs more than -0.7V to turn off, so you're forcing class A, but you're also forcing all that rectified current into the gate, which just isn't needed, or wanted
17:15 <JohnSmithVII> I will update my physical model with a 100 ohm source resistor and omit the diode. I predict that it won't oscillate because the SPICE-perfect-component model doesn't without the diode(oscillations start at ~25uS and end ~20uS later), but who knows?
22:25 <T3sl4> try the larger cap too
23:53 <JohnSmithVII> I think I can find a larger ceramic cap from a radio in my to-be-recycled electronics bin :)
25:24 <T3sl4> yay
29:07 <JohnSmithVII> Will 220Ω work or is 4.04mA too low for gate current?
29:22 <T3sl4> high*
29:28 <T3sl4> you don't have anything >10k?
29:34 <JohnSmithVII> Oh I do
29:39 <JohnSmithVII> I thought 100Ω was suggested :)
29:39 <T3sl4> literally it'll work with megs
29:55 <T3sl4> wait, gate?
30:15 <JohnSmithVII> "Bypassed source resistor"?
30:58 <T3sl4> https://wps.prenhall.com/wps/media/objects/416/426156/fig129.gif Rs
31:10 <T3sl4> bypassed with a cap, where Xc << Rs
35:29 <JohnSmithVII> I think the "33pF"(soon to be larger) cap works out, but that Xc=800 Ohms now, and will decrease with C, so I'll try something R>800 Ohms
36:37 <JohnSmithVII> Can I do that or do I add another cap?
36:49 <DHess_US> You don't *have* to bypass the source resistor but if you don't, then the resistor controls and lowers the AC transconductance which you probably don't want.
37:30 <DHess_US> In a low noise design, the unbypassed source resistor also clobbers the noise figure.
37:39 <DHess_US> So you wouldn't do that in a low noise oscillator.
40:08 <DHess_US> Use a JFET with a lower Idss and the resistor isn't needed. Or the resistor could be replaced with a current sink.
46:48 <JohnSmithVII> I just put a 56k resistor in for the source resistor
47:02 <T3sl4> no, for gate to ground, in place of the diode
47:21 <JohnSmithVII> Oh, okay
54:25 <JohnSmithVII> Okay I substituted the diode for a 56k resistor. I didn't realize how long the diode leads were until I had to re-organize my component layout to fit the resistor :)
00:38 <JohnSmithVII> Oh no my 145pF scrap capacitor is literally split in half :(
18:46 <JohnSmithVII> With the diode and capacitor modification I now scope ~5mV oscillation at 5.405Mhz, but I'm not sure that this is not just some strong source of RFI
18:56 <JohnSmithVII> I'll check on my shortwave receiver
27:37 <JohnSmithVII> Nope, just some powerful RFI and a power supply acting as an antenna :(
32:01 Join: Crazor_ ([email protected]) to #EEVblog
33:03 Quit: Crazor ([email protected]) Ping timeout: 121 seconds
33:03 Nick: Crazor_ is now known as Crazor
35:15 Join: eddie ([email protected]) to #EEVblog
23:37 Join: ThunderSqueak ([email protected]) to #EEVblog
44:52 Quit: ThunderSqueak ([email protected]) EOF from client
45:03 Join: ThunderSqueak ([email protected]) to #EEVblog
50:35 <Ampera> you there, Halcyon?
52:24 <Halcyon> I am now
53:31 <Halcyon> What's up Ampera?
54:06 <Ampera> just wanted to say hi
54:22 <Halcyon> Hi :)
54:25 <Ampera> My SFTP share is working great for streaming video.
54:33 <Halcyon> Yay1
54:34 * steve30 waves
54:34 <Halcyon> !
54:36 <Ampera> I was able to watch Full Metal Jacket over it from 120 miles away
54:37 <Halcyon> Hey Steve
55:00 <Halcyon> You know physical distance doesn't really matter over the internet :P
55:06 <Ampera> For whatever strange reason, VLC absolutely choaks on it, but Windows' in built media players are fine.
55:14 <Halcyon> You could have been on the International Space Station and still have watched it :P
55:19 <Ampera> I never said it did, I said it because to me it's impressive.
55:29 <Ampera> Actually maybe not, their connection isn't spectacular up there.
55:38 <Halcyon> Their connection is faster than mine
55:44 <Ampera> I forget the actual numbers.
56:11 <Ampera> As of 2012, it was 10/3
56:20 <Halcyon> *nod* It's much faster now
57:12 <Halcyon> Anyway I need to get dinner started
57:29 <Halcyon> I'll be back later
57:47 <Ampera> Not sure what it is, can't find it
57:48 <steve30> https://twitter.com/eevblog/status/1073129880936562688
58:08 <Halcyon> Heh
58:34 <Halcyon> I just bought a UPS by the way
59:12 <steve30> cool
59:36 <Halcyon> I need one, we get quite severe storms here sometimes and one day it'll fry something expensive
59:56 <Ampera> I just have regular surge protectors, how well they work is beyond me, but my UPS is currently broken
00:21 <steve30> I've pondered a UPS, but i find mains good enough
00:38 <Halcyon> Anyway, back soon lovely people
00:41 <Halcyon> (and Ampera)
00:42 <Halcyon> :P
00:50 <Ampera> ciao
01:10 <steve30> Its freezing today, and I have to go out later. Fortunately its nice and sunny.
25:34 <coppice> steve30: it seems to be warming up quite quickly
30:00 <BurtyB> not quick enough imho
31:08 <ThunderSqueak> I showed this to a gay friend, he said he would... https://pbs.twimg.com/media/DuWNovAUcAAJL9d.jpg
31:17 <ThunderSqueak> ... pull he-man's pants down
31:18 <ThunderSqueak> o.o
32:54 <ThunderSqueak> and I look at it and all I can do is hear skeletor's laugh
41:20 <Halcyon> OK I'm back
41:36 <Halcyon> Frittata is in the oven :D
41:39 <ThunderSqueak> hi Halcyon ... i am on my way to being drunk
41:44 <Halcyon> The shop was out of dill though :(
41:47 <ThunderSqueak> decided to finish off a box of wine
41:55 <ThunderSqueak> its 12:40am
42:00 <Halcyon> Me too ThunderSqueak, I'm on my second glad of wine, I've had a shit day today
42:08 <ThunderSqueak> Halcyon, same!
42:09 <Halcyon> I yelled at a Senior Sergeant
42:13 <ThunderSqueak> and yesterday
42:18 <ThunderSqueak> so, drinking today
42:22 <Halcyon> He was wrong, but I did apologise later though.
42:48 <ThunderSqueak> my highlight today was a gay friend of mine said he wanted to depants he-man
42:51 <ThunderSqueak> <.<
42:57 <Halcyon> Hahaha
43:22 <ThunderSqueak> then he told me he wasnt going to be here for a planned dinner, so I said fuck it, and in a cast iron skillet cooked everything in one pan in butter
43:27 <Halcyon> I want to de-pants this guy: https://www.youtube.com/watch?v=fGXMpWSJceQ
43:44 <ThunderSqueak> https://pbs.twimg.com/media/DuXbaa-VYAEWEAC.jpg
43:59 <ThunderSqueak> couldnt let the pork chops sit longer
44:07 <Halcyon> Yummy!
44:10 <Halcyon> My kind of dinner
44:13 <ThunderSqueak> so he missed out
44:29 <Halcyon> I'm being especially good, decided to start working out
44:36 <ThunderSqueak> me too
44:38 <ThunderSqueak> :D
44:49 <ThunderSqueak> and I am drunk and very talkative now
44:54 <ThunderSqueak> and no one I know is awake :/
44:55 <Halcyon> \o/
45:01 <Halcyon> Not even T3sl4 ?
45:04 <Halcyon> He's usually lurking
45:07 <ThunderSqueak> I get talkative when drunk
45:11 <Halcyon> Don't we all
45:13 <ThunderSqueak> very very talkative
45:26 <ThunderSqueak> and ... cuddly
45:27 <ThunderSqueak> <.<
45:28 * Halcyon wonders what Ampera would be like drunk
45:30 <ThunderSqueak> yes I am one of those
45:44 <Ampera> I'd probably kill myself for real.
45:44 <Halcyon> `8ball Would Ampera be a fun drunk?
45:44 <ChanOP> 8ball: Would Ampera be a fun drunk??? Not even on a GOOD day.!!
45:51 <Halcyon> I guess that answers that
46:08 <ThunderSqueak> Ampera, over what, my eggs fried in butter?
46:11 <Ampera> I'd probably end up blurting out everything nobody should know about me and then smash my head open on a brick wall.
46:17 <ThunderSqueak> everything is better when based with butter
46:21 <Ampera> what?
46:23 <Halcyon> So once again the 8 Ball is right
46:26 <ThunderSqueak> Ampera, ah XD
46:32 <ThunderSqueak> sorry I am rambling
46:35 <Halcyon> ThunderSqueak: True, but only real butter. Not that margarine rubbish
46:44 <ThunderSqueak> who uses margarine
46:47 <ThunderSqueak> ...
46:48 <Halcyon> Not me
46:51 <ThunderSqueak> I use the real butter
46:52 <Halcyon> Not even on sammiches
46:53 <Ampera> A part of me dies inside whenever I see a recipe say margarine.
47:00 <Halcyon> Ampera: Too true
47:00 <ThunderSqueak> real butter == flavor
47:04 <Ampera> Not that you can't just use butter, but because someone even thinks it's a good idea.
47:06 <ThunderSqueak> magarine == plastic
47:07 <Halcyon> Real butter = Dairy
47:10 <Halcyon> Not processed oils
47:25 <Ampera> margarine == plastic reminds me of american cheese
47:27 <ThunderSqueak> yep
47:30 <Ampera> now I want a grilled cheese.
47:31 <Halcyon> I *never* buy margarine
47:34 <ThunderSqueak> plastic == processed oils
47:43 <ThunderSqueak> margarine == plastic
47:44 <Halcyon> I would prefer American cheese over Marg.
47:46 <Ampera> hey there we go.
48:06 <Ampera> American cheese isn't bad. It's a very cheap cheese, quite salty, but good if you put it on something.
48:09 <Halcyon> I would prefer Kraft Singles over Marg.
48:16 <Ampera> It's good on a sandwich
48:17 <Halcyon> I find it good on burgers
48:29 <steve30> I sometimes use margarine in cooking, but thats it
48:31 <Ampera> They sell American cheese in the land down blunder?
48:35 <Halcyon> I don't normally add too much salt to my burger patties, so a salty cheese works
48:36 <ThunderSqueak> ... I have real cheese in my fridge
48:38 <ThunderSqueak> <.<;;
48:40 <Halcyon> Of course
48:50 <Ampera> I don't know if I should feel bad or not.
48:55 <Halcyon> My local supermarket literally has about 100 different cheeses
49:21 <Halcyon> They have a delightful French creamy Camembert
49:38 <Ampera> I looked into how much an ex-US army M1 Garand would cost
49:41 <steve30> why do some modern computer cases only have a power switch, and no reset switch?
49:44 <Ampera> as it turns out not too too much.
49:46 <Halcyon> It's a wheel, but if you leave it out for 30-45 minutes, its almost gooey
49:49 <Halcyon> So fucking good
50:00 <Halcyon> Steve: I didn't notice that was a thing?
50:04 <ThunderSqueak> I am buying another gun tomorrow probably
50:05 <Ampera> steve30, there's a lot of things modern computer cases omit that I wish they didn't
50:10 <Ampera> like 3.5 inch drive bays
50:14 <Halcyon> Keylocks?
50:19 <Halcyon> Turbo buttons
50:20 <Halcyon> :P
50:23 <Ampera> lol
50:28 <ThunderSqueak> hey I love my turbo button
50:32 <ThunderSqueak> push it.. push it a lot
50:34 <ThunderSqueak> <.<
50:37 <steve30> My new one has both 3.5" and 5.25" :), but lacks a reset switch. It does have a Clear CMOS switch on the back though.
50:41 <Halcyon> I loved making the little LED display go from 33 to 66 MHz
50:43 <Halcyon> It was the shit
50:44 <Ampera> I think I only have one board that even has a keylock header
50:51 <Halcyon> Or before that from 18 to 25 on my 386
50:54 <ThunderSqueak> Halcyon, you know what I am refering too :P
51:00 <Ampera> However I think I actually have 4 different machines with turbo headers.
51:04 <ThunderSqueak> I have that very button on my 486
51:17 <Halcyon> ThunderSqueak: You would be weird if you didn't
51:23 <Halcyon> I think even Pentium's had it??
51:32 <steve30> I have a baby-AT case with a turbo switch
51:33 <Ampera> Real turbo buttons don't increase CPU speeds, they decrease them.
51:37 <ThunderSqueak> I built a 486 using NOS
51:39 <Halcyon> Correct
51:40 <ThunderSqueak> it was amazing
51:43 <ThunderSqueak> :D
51:46 <Halcyon> Turbo = Native clock speed
51:48 <Ampera> I'm pretty sure the SE440BX-2 has it
52:02 <Ampera> Turbo = reduced clock speed, unturbo = native
52:05 <ThunderSqueak> well actually a turbo button when on == slow speed
52:12 <Ampera> That's the default, but they have gone both ways
52:14 <Halcyon> The other way around Ampera
52:25 <Halcyon> Turbo = Faster
52:29 <ThunderSqueak> turbo slows down the machine
52:32 <ThunderSqueak> <.<;;
52:41 <Ampera> No, it's not. Turbo buttons were intended to slow down computers to support programs that were reliant on slower clock speeds.
52:42 <Halcyon> That's why the LED comes ON when Turbo is on
52:47 <ThunderSqueak> Ampera, is correct
52:50 <Ampera> Otherwise they really have no reason to exist.
52:50 <Halcyon> I know what they were intended to do
53:08 * ThunderSqueak was a PC tech during those days
53:08 <Ampera> Some cases or boards do that, Halcyon, but that's not the default/de facto standard
53:11 <Halcyon> Why on earth would anyone think that Turbo = Slower ?
53:24 <ThunderSqueak> the standard was pushing the turbo slowed the machine down
53:24 <Ampera> I don't know why anybody would think that, it's just how it is.
53:31 <Halcyon> Ampera: I've built MANY 100's of PCs, I've never come across one that was the other way around
53:38 <Ampera> That's exactly how my 486 board works, you press the turbo button, and it slows down.
53:43 <Ampera> I've never come across one that speeds up.
53:48 <ThunderSqueak> and no one listens to the girl
53:49 <Halcyon> Even AmiDiag indicated when Turbo was ON (i.e.: The CPU ran at the full clock speed)
53:50 <ThunderSqueak> <.<
53:55 <Ampera> Every single machine I have with a turbo button slows down, including that 120Mhz Pentium you gave me.
53:59 * ThunderSqueak watches two guys argue
54:07 <Ampera> did you just assume my gender, ThunderSqueak?
54:08 <Halcyon> It doesn't speed up -- Turbo ON is the normal clock rate, Turbo OFF is the reduced one
54:09 <Ampera> lol
54:13 <Ampera> he doesn't listen to me either.
54:13 <ThunderSqueak> yep I did
54:17 <ThunderSqueak> :P
54:21 <ThunderSqueak> you female?
54:24 <ThunderSqueak> or other?
54:28 <Ampera> I'm not even sure anymore.
54:30 <Halcyon> He wishes Thn
54:32 <Halcyon> ThunderSqueak:
54:58 <ThunderSqueak> anyway, just saying, no one listens to me
55:16 <Ampera> Halcyon, would the word of LGR be enough to convince you?
55:17 <Halcyon> But anyone, when you close the circuit (i.e.: The turbo switch is on), the CPU runs at normal speed
55:20 <Ampera> if so, watch his video
55:20 <ThunderSqueak> but Ampera is correct, technically on most boards of that vintage, a turbo button slows the machine down
55:27 <Halcyon> Link me
55:35 <Halcyon> Yes, Turbo OFF does
55:42 <ThunderSqueak> Halcyon, no turbo on does
55:52 <Halcyon> I need to test this for myself
55:52 <Ampera> https://www.youtube.com/watch?v=p2q02Bxtqds
55:53 <ThunderSqueak> it is a backwards thinking but that is how it worked
56:14 <Ampera> He specifically states the default is on=slower, but there were some machines that did it in reverse
56:26 <Ampera> which in my mind makes absolutely no sense as a feature, and is only there to make the name mean something
56:33 <ThunderSqueak> so ampera, what gender are you
56:42 <Ampera> Born male.
56:43 <ThunderSqueak> so I can know :D
56:46 <ThunderSqueak> ok?
56:49 <ThunderSqueak> but what gender are you
56:49 <Halcyon> I skipped through it, at no point did I see him actually demonstrate it
57:00 <Halcyon> I will pull out my 486 tomorrow to test
57:05 * ThunderSqueak differentiates
57:06 <Ampera> If you mean how to address me, I've made a point not to get bothered either way.
57:14 <ThunderSqueak> gender and sex are different
57:14 <Halcyon> If I'm wrong, I shall admit it
57:22 <ThunderSqueak> and you aid "assume your gender"
57:29 <Ampera> It's something that critically annoys me about trans people, and I endeavor to not hold people to anything they don't want to call me by.
57:31 <Halcyon> ThunderSqueak: He was being funny
57:37 <Ampera> yeah that's the joke
57:41 <Halcyon> Like what Feminists do
57:42 <ThunderSqueak> I am a drunk girl
57:46 <ThunderSqueak> so, yeah :P
57:55 * ThunderSqueak wants to know
57:56 <ThunderSqueak> :P
58:12 <Halcyon> Ooh dinner is just about ready, back later lovely people
58:13 <steve30> is assuming gender politically incorrect these days?
58:22 <ThunderSqueak> steve30, yep
58:23 <Ampera> I've not entirely decided, this particular issue has been sort of shoved onto the back burner to deal with more pressing matters
58:37 <Ampera> I do it all the time because I don't give a shit, and I don't care if it's done towards me.
58:39 <ThunderSqueak> Ampera, ah gotcha
58:57 <Ampera> I'll respect someone's choices if they specifically ask me to address them in a specific way, but I'll call people whatever
59:00 <Ampera> fite me irl
59:11 <ThunderSqueak> I generally call people what they present as
59:17 * ThunderSqueak dated trans people
59:20 <ThunderSqueak> *shrug*
59:24 <Ampera> I generally don't care a whole lot.
59:34 <Ampera> as it's something that shouldn't define a person, and shouldn't matter.
59:45 <ThunderSqueak> ok, I shall call you zer
59:49 <Ampera> The fact that it does to me is just people projecting their insecurities onto others.
59:50 <ThunderSqueak> since you wont answer
00:02 <Ampera> alright that will get me mad, but for entirely different reasons.
00:06 <Ampera> /j
00:11 <ThunderSqueak> :P
00:13 <ThunderSqueak> Ikr
00:14 <ThunderSqueak> lol
00:22 <Ampera> he, she, heck you can refer to me as an it, I don't give a damn
00:31 <ThunderSqueak> nah, it is rude
00:35 <Ampera> as long as you don't call me late to dinner.
00:45 <ThunderSqueak> Ampera, PM :>
00:53 <Ampera> you want me to message you?
00:59 <ThunderSqueak> I messaged you yes
00:59 <Ampera> or am I just particularly thick
01:01 <ThunderSqueak> oo
01:07 * steve30 proudly uses "he" as a catch-all pronoun :)
01:16 <Ampera> uhhh, let me check my other IRC client, as it didn't pop up, one second
01:50 <ThunderSqueak> I have a D&D book that says "we use he throughout this book, not to seclude females from this wonderful game but simply for conveinence"
01:51 <ThunderSqueak> ;p
02:27 <ThunderSqueak> aww it wont go through, apparently I forgot to identify?
02:29 <Ampera> My ZNC should pass through all messages to my clients, but if you just DMed me, then I didn't get it.
02:33 Quit: ThunderSqueak ([email protected]) Quit: Leaving
02:33 <Ampera> Ah, yes.
02:36 <Ampera> let me see if I can disable that.
02:41 <Ampera> damian, how do I let anybody DM me?
02:45 Join: ThunderSqueak ([email protected]) to #EEVblog
03:02 <ThunderSqueak> hmm
04:04 <Ampera> just send it over AfterNET until damian tells me how to disable that
04:15 <Ampera> austnet's auth system is really really weird, and is probably my one and only major complaint.
06:30 Quit: ThunderSqueak ([email protected]) Changing hosts
06:30 Join: ThunderSqueak ([email protected]) to #EEVblog
06:44 Mode: (by ChanOP) +o ThunderSqueak
06:46 <ThunderSqueak> hmm
06:52 <ThunderSqueak> no I am identified
06:53 <ThunderSqueak> odd
06:58 <Ampera> shouldn't give you op if you're not identified
07:02 <Ampera> maybe I'm not
04:48 Quit: BurtyB ([email protected]) Quit: Leaving
Search this day's buffer: