Windows/Driver bugged on purpose so you must buy expensive routers.

I just found that video, or one similar enough :

Your technician was taking the pragmatic approach to tuning his radio. All the other stuff, deviation, cleanliness of signal and frequency accuracy, don't really apply when tweaking the Tx for maximum 'smoke'. I'm not so sure whether tuning a NBFM Rx by ear is as accurate as using wobbulator and 'scope (FFT spectrum display or WHY) but I suspect it's a damn sight quicker for an experienced technician (and possibly just as accurate, especially when tweaking an Rx that has already been lined up 'By The Book')

The "Woktenna" is simply a pragmatic approach to boosting the maximum range on a point to point link. The rest of it is assumed to be taken care of by the USB WiFi dongle. Your only concern lies with combining a wok with a dongle to extend the range in a specific direction. You don't even need to damage the wok by drilling holes in it, you can simply fabricate a support out of coathanger wire or similar that can be clipped onto the rim of the wok to hold the dongle in the right place and orientation.

When a pair of such woktennas are used to create a point to point link, Tx/Rx asymmetry effects cancel out. The asymmetry effect might be an issue when using it to connect to a WAP serving a local population of client machines but it could be addressing an existing asymmetry within the dongle itself so it's not worth considering unless you observe such issues in the practical use of your pragmatic cheap 'n' cheerful solution. Sometimes, you can land up 'overthinking' a problem. :-)

It's been the motivation for my subscribing to sed. I was biding my time before asking for advice on how to add the required BLDC motor drive module to upgrade the Parkside inverter genset to electric start. It's such a no-brainer to utilise the flywheel three phase multi-pole PM alternator as a starter motor (the modern day version of the old Dynastart (tm) setup - a special dynamo/starter motor used with some small marine petrol engines), I'm surprised the extra circuitry isn't already a standard built in feature of the current crop of inverter modules used by the suitcase class of petrol (gasoline) inverter generators today[1].

Retro-fitting a BLDC motor drive module isn't going to be as straightforward as those youtube video demos of converting a car alternator into an electric go-kart motor since the module will either have to tolerate the 400v peak output once the engine has fired up or else utilise an extra isolation device, possibly a purpose made SS relay module.

The very same, except for it being sold by the UK division of Aldi. Interestingly, I note they don't share the review data between the Irish and English operations - crafty move. My own scathing review on the first two samples was pulled from the English site about 24 hours after it was published. TBH, considering I'd used my usenet posting nym and the invalid email address, I was surprised it was even added to the list. I don't know whether it was pulled due to my criticism of the product or whether it was because they couldn't contact me by email to verify I wasn't a "Robot". I might have another go, using my actual contact details - I'm not heavily dependant on email.

I did look for servicing inspiration after spotting 4 trimpots poking out of the encapsulation in the inverter module[2] and downloaded the workshop manual for the Powerhouse PH2700PRi which showed 4 trimpots on the inverter module in the circuit schematic, labelled as "voltage", "balance" (presumably an eco-throttle related item), "current" and "overload". The Workzone module was similar but not identical so I had to 'suck it and see' test each trimpot - not so easy when the generator is running and vibrating the whole shebang. I only managed to identify the voltage pot and gave up when I couldn't identify the next pot on my to do list, the "balance" pot.

To identify the remaining two pots would require a variable dummy load setup that I still haven't built (I've got 3750 watts' worth of 150W incandescent lamps and a 220v 180VA panel mount variac but no suitable box to assemble my continuously variable test load into). I reckon a half dozen BC22 sockets,one of which is fed from the variac, should let me go in 150W steps all the way up to 900W and fine tune an extra 0 to 150W on each step.

That and the 900W toaster will take me to 1800W and I can always employ a 2KW oil filled radiator with a built in 500W fan heater for heavier test loads. The key thing to running these gensets into test loads is having some means by which to vary the loading continuously throughout the whole range. My box of plug in 150W light bulbs and variac transformer will allow me to discover the overload limits with the required precision.

If I'd had such a setup to hand, I'd have pushed on and identified the "current" and "overload" pots but as I didn't and I couldn't get a "balance" related response from the 3 remaining pots, I decided enough was enough and high time for me to get my money back. "Three strikes and you're out" seems only appropriate in this case. :-(

I first heard the phrase on UK CB radio back in the eighties, mostly by the "Pirate" CB fraternity, running illegal American multimode CB rigs or radioham transceivers, commenting on the idiots running UK Legal FM kit through a "Burner". It was a reference to CBers using illegal "Linear Amplifiers" to boost their 4W legally limited output to 100W or more (a

14dB boost on the Tx but 0dB on the Rx - hence all mouth and no 'ears', aka, an "Alligator Station").

That seems a bit harsh? (and by my standards, a bit premature to say the least) :-)

I used to shut my desktop machine down to give it an overnight 'rest' back in the day when I was running win2k up until three years ago when I finally had to resort to Linux Mint due to lack of win2k driver support for my last bi-decadal hardware upgrade. WinXP was a sufficiently s**te version of win2k to begin with and the successor versions of windows just kept getting ever more s**te with each new release.

Shortly after I'd transferred to a Linux distro full time on my desktop machine, I subscribed to the alt.windows-10 newsgroup just to remind me why I had made the move to a Linux distro and the upheaval which that had initially entailed. It helped to remove all temptation to give up and "take the easy way out" since I had clear proof that win10 was totally the opposite of an "Easy way out" solution.

I'd finally had had enough of all the wailing and gnashing of teeth being posted to a.w-10 after just over a year, so unsubscribed myself from the group, by which time I needed no further convincing that a Unix based distro was now the only sane option.

Now, I rarely have the luxury for such overnight shut downs since I've usually got an early morning/afternoon TV programme or two scheduled for recording by Kaffeine. Plus, up to a few months back, I was running overnight batch mpg to mkv conversion jobs using Handbrake to compress my existing mpg SD off-air recordings to free up disk space on the NAS box to postpone the impending HDD upgrade I'd otherwise have required two years back.

I reckon it's gained me an extra two or three years out of the existing

14TB's worth of three HDDs (seperate disk volumes - no RAID!) which included being able to release a 3TB drive from active duty - it's still physically present, just no longer connected up, downgrading capacity from 17TB to its current 14TB configuration.

The cost of the hardware upgrade has amply paid for itself for the several thousand hours I'd had it running Handbrake jobs over an 18 month period to earn its keep in saved HDD upgrade costs. :-)

I suppose I can probably go back to an overnight shut down routine now if I can be bothered to tailor my workflow to suit but quite frankly, I'm not that bothered by the electricity costs and CBA to do so right now.

[1] I was somewhat taken aback to see the use of the old fashioned method of electric start used by that Powerhouse PH2700PRi generator when it already possessed all of the essential electromechanicals of a more than adequate BLDC 'starter motor'. Such a terrible waste of resources and the needless extra weight of a seperate starter motor (with solenoid) and starter ring gear and modified crankcase housing.

This seems to demonstrate a complete lack of imagination on the part of the designers working on portable suitcase inverter generators. You can add an electric start feature/optional extra for little more than the weight penalty of a 5AH 4 cell LiPo or Li-ion battery pack.

Over and above the presence of a starter battery of some sort, the rest is simply a bit of extra silicon real estate in the inverter module to cater for the BLDC drive and battery charging/management functions with an option to use the battery to stabilise the generator voltage during transient loading events whilst the engine speed is ramped up to meet the extra demand.

I know the basics to making a perfect lightweight electric start suitcase inverter generator. Unfortunately, I'm a little hazy on the details, in particular what sort of ready made modules might be available to save my having to re-invent the wheel. :-(

[2] The little Parkside PGI 1200 B2 1000/1200 watt inverter generator I'd bought two months back from Lidl for a mere 99 quid, didn?t have any such trimpots for me to twiddle with. I would have liked to raise the no-load voltage from 233 to 240 (it only dropped a mere 3 or 4 volts at full load, unlike the Workzone unit which dropped a whole 10 volts on half load). The idea being to raise the 1st overload threshold above the 980W mark it turned out to have actually been set at and take it just north of the claimed 1KW continuous rating. It would cheerfully run an 1160W overload for exactly 30 seconds before shutting down the inverter (the user handbook claimed a mere 5 seconds overload duration) so it didn't seem an unreasonable adjustment to attempt.

Compared to the Workzone unit, the Parkside one is perfection personified. Unlike the WZ unit the PS one shows no hint of bogging down in eco-throttle mode when hit with a 965w (98%) load, it just responds so fast that all you can hear is an increased exhaust noise level immediately followed by an increase in revs, just as you might well expect out of any properly set up inverter genset. The WZ one, otoh, leaves you in do doubt that the inverter module has been taken completely by surprise by almost stalling out before it picks the revs back up after applying the same 960w (53%) load.

The little Parkside unit is just adequate for the essentials but I'd have liked a little more margin, hence my time wasted with three Workzone generators with all three showing pretty well the same serious shortcomings you *don't* ever want to see in any small inverter genset. I guess I'll just have to make do with the Parkside one until Lidl finally get round to offering a 2KW or so rated Parkside inverter genset. I can wait. :-)

--
Johnny B Good

Jeff Liebermann wrote in news: snipped-for-privacy@4ax.com:

One word. Reflectometry.

They can tell you even how far down a line a break is located.

Sure, but the twist rate? Please elaborate.

Jeroen --The time-domain reflectometrist-- Belleman

Keep trying. It's an interesting problem even if reflectometry is not used in gigabit auto-polarity negotiation.

Since we know the twist rate (pitch) is slightly different for each pair of CAT5e wires, it should be possible to measure something that makes each pair different. You can use the impedance, which is identical for each pair at 100 ohms +/-5% at 100MHz. The wire gauge is usually 24AWG and is the same gauge in every wire in the bundle. The wire insulation thickness is a consistent 0.245mm. Capacitance is a nominal 52pF/meter for all 4 pairs.

In theory, if each pair in the bundle were square cut to identical lengths, the differences in pitch should produced slightly different lengths on a TDR. The problem here is that the actual twist rate for CAT5e is not specified or very well controlled. Still, it should be possible to compare relative pitches between pairs to find the longest, shortest, and those in between. Measured by Wiki article author: Color cm/turn turns/meter Blue 1.38 72 Green 1.53 65 Orange 1.78 56 Brown 1.94 52 It's an interesting math exercise to determine how much shorter a pair becomes when twisted at different pitches. I'll probably cheat and just cut a measured length of cat5e, strip off the jacket, untwist each pair, and measure the lengths. Later tonite when I have more time.

For a sanity check, it might be useful to see how much of a difference in length would be required to obtain a useful measurement. In 1nsec, light travels 300mm. None of my scopes can resolve 1 nsec from a TDR waveform. At best, maybe 5nsec. So, in order to see anything on a TDR, the difference in cable lengths produced by different twist rates or pitch would need to be at least 1.5 meters. For a 100 meter maximum cable run, I don't think that unraveling the wire twists is going to produce a 1.5% difference. Certainly not for shorter cable lengths and definitely not for measuring small differences in wire length between the various wire pairs in a CAT5e bundle. In short, one has to have a very good TDR and scope in order to see anything useful on a scope. While it might be possible with a TDR that works at microwave frequencies, it's unlikely that using an ethernet NIC as a TDR is going work.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Part 2: I found a partial roll of CAT5e by Commodity Cable Inc. Not exactly name brand cable. At least it's copper, not CCL (copper clad aluminum). From the numbers printed on the cable, the remaining cable is 562ft long. According to CAT5e spec, it should be

17pF/ft (52pF/meter) = 17 * 562 = 9554 pF/pair

Measuring the capacitance across each pair, I expect the blue pair to have the highest capacitance, while the brown pair, the lowest: Color pF Blue 7940 Green 8230 Orange 8090 Brown 8010 So much for that theory. Something is wrong.

After stripping the cable, I find that the Wikipedia CAT5e page has the wrong wire twist pitch. Measuring the pitch from 100mm length cuts, I get an amazing: Color turns/meter pF/562ft Green 100 8230 Orange 90 8090 Brown 80 8010 Blue 70 7940 That makes more sense. I did this measurement twice, on different cable cuts, and came out with the same numbers. So, the Wikipedia article is wrong.

Next, I chopped off a 1 meter length of CAT5, removed the jacket, untwisted each pair, straightened the wires as much as possible, and measured the lengths: Color turns/meter pF/562ft elongation mm/1000mm Green 100 8230 41 Orange 90 8090 9 Brown 80 8010 27 Blue 70 7940 0 The Blue and Green wires look right, but the Orange and Brown are wrong. Assuming a didn't screw up somewhere on my measurements, all this means is that I can't straighten twisted wires very well or measure the capacitance accurately.

If at least the Green and Blue wires were correct, then the difference in length translates to a difference of: 41mm / 1000mm * 100 meters = 4.1 meters difference in length over a maximum 1000base-T segment of 100 meters, which should be easy to detect with a TDR. So, maybe it is possible to use a TDR to identify the cable pairs. It won't work with short length cables, but sure looks like it might work with longer cables.

Stay tuned for the next episode of adventures in CAT5e (after I get some paying work done, and after I find where I buried my three TDR boards).

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Ah, exactly whom I need to answer a deceptively simple question.

The data sheet says that the Vf (velocity factor) of CAT5e cable is

72%. If I untwist and straight this pair of wires, does Vf increase to something more like 90%? More generally, do the 4 different twist rates (pitch) of CAT5e each have a different Vf? I'm trying to determine why 1000base-T NIC chips have deskewing registers.

"Apparatus and method for deskewing 1000 BASE-T Ethernet physical layer signals"

I could probably determine the answer by measurement, but my TDR's are currently hiding from me.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

In a transmission line, the energy is transferred in the insulation material between the conductors. Thus, both Vf and losses depend on the insulation material used. The conductors just limit the area, in which the field propagates.

If you untwist the pair and move the conductors so far away that there is a lot of air between the conductors, the Vf could be well over 90 %, but in this case with such large spacing, the impedance is in the order of 400 to 600 ohms, not 100 ohms any more.

If you have a tighter twist, the untwisted conductors are longer, thus it takes a longer time to propagate.

Twisted pairs are dispersive because the fraction of the field that's in the plastic dielectric doesn't propagate as fast as the part that's in air. Consequently, Vf isn't all that well defined. Untwisting a pair is not likely to change Vf all that much, unless you also change the wire spacing. Without a more precise idea of the geometries, there isn't much I can say.

Twisted pair Vf is sufficiently vague that I'm not surprised that deskewing is necessary with symbol rates of 125MBaud over up to 100m.

Jeroen Belleman

Thanks. That's what I needed. I should have mentioned that when un-twisted, the two wires remain in contact, where a change in dielectric material is part of the problem.

Therefore, if the Vf (velocity factor) does not change when the wire is un-twisted, then the change in wire length should product a measurable reflection time. Had Vf changed to something like 0.9, which has the effect of decreasing the reflection time closer to the speed-o-light limit, and effectively shrink the wire, which would not be very useful more comparing lengths.

I've had some other surprises playing with the CAT5e. (See my tinkering notes elsewhere in this thread),

1. The Wikipedia article is wrong about measured twist rate (pitch).
2. Measured capacitance across each pair is not identical, but close (worst case about +/-0.4% variation).
3. Nominal capacitance was suppose to be 17pF/ft, but is really more like 15pF/ft.
4. It takes about 45 minutes to un-twist 4 lengths, 1 meter each of CAT5e. My fingers still hurt.
5. The pair with the lowest pitch (blue) did not show any increase in length when un-twisted. The largest pitch (green) pair increased in length by 4.1%.
6. My TDR boards like to hide when I need them most. It now appears that it will be easier to build another one than to my find old boards.

I'll grind out how many cycles of 125MHz gigabit ethernet symbols are delayed by the extra 41mm of wire in a 100meter CAT5 cable after I deal with a broken laser printer. That should give me some idea of what fraction of a 125MHz signal is shifted (skewed).

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

onsdag den 11. juli 2018 kl. 00.15.25 UTC+2 skrev Jeff Liebermann:

afaict the spec is max 45ns skew for 100m

and I can't see there is any spec for the twist being different, just a spec for crosstalk between pairs

Thanks. I should have mentioned that the insulated wires in a pair remain touching after being untwisted, thus eliminating any changes which might be caused by a change in dielectric (air).

Oops. That should be Vf, not Vr.

I'm still a little confused (ok, more than a little confused). I thought I understood how it works, but now I'm not so sure.

Let me suggest an extreme example, which I can try when I get my workbench back into useable order.

Start with a 1 meter cut length of CAT5. Pull out the Blue pair, which has the lowest pitch (70 turns/meter). Measure the propagation delay. Assume a Vf (velocity factor) = 0.72. Now, take an electric drill and increase the pitch to as much as possible without destroying the wires. Measure everything again and compare.

The additional twisting will shorten the pair. From my experience with braided wire rope, reduction in length to 68% is about the maximum possible. That reduces the physical length to 0.68 meters.

Now comes the problem. If, as you say, the propagation delay is based on the wire length, and the Vr remains constant, then a TDR should show a 1 meter long twisted cable, including compensation for Vr = 0.72.

However, with the pair twisted really tight, the pair will be shorter. Yet, you suggest that the TDR will always show a 1 meter long cable, with compensation for Vf = whatever. Since the pair is now 0.68 meters long, instead of the previous 1 meter long, the only way to get this shorter length would be if Vf decreases to perhaps Vf = 0.68. In other words, Vf decreases with increasing twist rate in order to shorten the physical cable length to the electrical cable length.

Ok, where did I screw up this time?

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

Lumped inductance ?

Thanks. It's probably in 802.3ab. I can't find a bootleg copy.

At a symbol rate of 125MHz, 45 nsec would be: 125MHz = 8nsec/cycle 45nsec / 8nsec/cycle = 5.6 clock_cycles That's more than enough to require de-skewing.

My un-twisting experiment showed that in a 100 meter cable, the green wires are 41mm longer than the blue wire. Light travels at

3x10^11mm/sec, or in this Vf = 0.72 environment, at 2.2*10^11mm/sec. To travel the extra 41mm distance, it takes: 41mm / 2.7*10^11 = 15*10^-11sec = 0.15nsec or 0.15nsec / 8nsec/clock_cycle = 0.02 = 2% of a clock cycle, which probably does not need de-skewing.

So, the de-skewing is required because of added delays caused by circuitry or maybe different data paths, not by any CAT5e wire pair length differences.

Yep. NEXT (near end crosstalk), along with cable losses, are probably the main limitations as to cable length. I was rather surprised that my measured twist rates came out as even numbers. I expected some kind of randomization between the twist rates to reduce any repetitive wire contact patterns which might introduce resonances in the crosstalk for repetitive bit patterns. However, in hindsight, I guess scrambling the data would have the same effect.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

No lumps in the cable. Well, there was one tiny bump. I did a needle biopsy on it, found it to be benign, and removed it by massaging the cable.

Since I was only measuring the capacitance of each pair, it would take quite a lump to produce much of a difference. Lumpy cable was a problem in CAT6 and above, so it was decided to solvent weld the wires from each pair together. The wires could still bend and twist, but could not separate from each other. Splitting such solvent welded CAT6 wires apart in order to crimp on a connector is a bit tricky, but can be facilitated by a judiciously applied layer of profanity.

You only get 1/2 a point because you supplied a good possible answer, but failed to also provide and explanation for how it works.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

but failed to also provide and explanation for how it works. "

I thought I said a TDR but with common mode and differential, and possibly energizing a conductor or pair and seeing what it induces in the other pair (s). I didn't quite put it that way.

Now to do the math to figure out all this shit you need a REAL engineer. I am just a decent hack. But I get ideas and have invented all kinds of backd oor ways to measure shit, and now it seems - why backdoor ? That's how thin gs are done.

OK, we had a bu nch of these Samsung or Funai VCRs come in and someone had screwed up the record alignment on then with a screwdriver. According to in structions you need t his specialized equipment and all this shit. Well, no good businessman would have spent what that cost and I also concurred that it was not correct to do it. Still here is a pile of work and I am sure we can get at least $ 70 each, so I gave it some thought. My mind went back i n time to the basement with my Grandfather. Kinda like the TV show Kung Fu. But then I had it.

I took a working VCR and put a dummy tape in it on play, connected to a mon itor. Then I took the VCR that had the recording problem and put a dummy ta pe in it on record. Then I loosely coupled the record signal to the head am p in the VCR that was in play.

Once adjusted for a decent picture that way, rip it all down and the thing works. Write the bill, tell the customer to never let this happen again and thank you very much.

The other tech, VERY competent, in fact much better than me in RF, biut any way, even though this is almost like RF, he said "He is genius".

Sorry to bore you with all this but, oh well.

Lasse Langwadt Christensen wrote in news:b3f980df-e3f5- snipped-for-privacy@googlegroups.com:

The level of crosstalk is why a cat 5e link gets negotiated to a slower cap.

The cat 6 spec for the connector alone negates much of this.

Properly configured, of course.

Jeff Liebermann wrote in news: snipped-for-privacy@4ax.com:

NO, it does NOT. The electron STILL has to traverse the entire path regardless of how you configured (twisted) it.

Jeff Liebermann wrote in news: snipped-for-privacy@4ax.com:

No. It physical length will be, but its electrical length will not change.