Nearly all the cars around here are monochrome: mostly grey, with the occasional white or black for variety. Spray paint seems like a good idea.
How do people find their cars in a parking lot? They all look alike.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
AD or DA is it basically the same thing, it just done analog or digital
one thing to consider is that anything more than first order isn't unconditionally stable
-Lasse
Actually, the sorta10 thing may not work. It doesn't seem to deliver
*the* edge that I need. It doesn't obviously identify the single 155.52 MHz edge that happens once per second, time aligned with the 10M and 1PPS reference inputs. Again, lots of work has been done on frequency locking, but practically nothing on time locking. A spectrum analyzer tells you nothing about time alignment.I hacked a spectrum-shaped time-domain noise source into my bang-bang sim, to add jitter, but any useful simulation would run for weeks at least, and overflow my hard drive with .RAW data.
-- John Larkin Highland Technology, Inc picosecond timing laser drivers and controllers jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Get a big dish and put the little dish at the focal point.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Am 03.12.2014 um 16:41 schrieb Joerg:
I have good results with Samsung 830 Pro series, 512 GB. I get transfer rates of upto 500 MBytes/sec in a Dell Precision laptop. Previously I could barely reach that in a workstation with an Areca controller with local CPU/Cache and 4 striped Western Digital Raptors. (that sounded like a turbine!)
LTspice runs more than twice as fast on the Precision laptop than on my customers CAD server. :-) (Linux Mint, XP, VMware) Well, laptop may be the wrong word, it is NOT small and it is heavy. Dell sells it as a "portable workstation".
Size DOES matter with SSDs. Within the 6Gbit rate limit of the SATA, twice the size means twice the transfer rate and twice the number of transactions/sec since they can multiplex more flash chips. The number of transactions/sec is more important than raw transfer rate.
Because of the high transaction rate and the missing seeks/rot. latency, file system fragmentation is no issue. You can/should switch defragmentation off, it only costs write cycles.
Also, the partitions should start at large sector multiples (> 4 or 8KBytes IIRC) to avoid read-modify-write cycles when your OS writes sectors that are misaligned to the physical sectors of the disk. (512 byte offsets)
regards, Gerhard
Am 03.12.2014 um 19:57 schrieb Gerhard Hoffmann:
800-something Pro. I have seen it only once.
Can be worse. I arrived at a Dutch company in Hengelo and since I brought my bicycle I used that to go there. Small parking lot with a few hundred cars. "Where can I park my bike?" ... "Oh, that's the other parking lot, just cycle down this path here and you can't miss it". I could not believe what I saw. Thousands of bicycles and all were either dark blue or dark green. Luckily I had a yellow road bike.
-- Regards, Joerg http://www.analogconsultants.com/
[...]
Sure. But it would have to be on the east side of our house wherethe back yards and deck are. So the WAF would be a problem.
-- Regards, Joerg http://www.analogconsultants.com/
There are apparently a small number of different car keys, so people will occasionally go into a parking lot and drive off with the wrong grey or black car!
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Happened to my dad in the early 60's. Ford 12m, looked like the US version but shrunk. Dad opened the trunk, loaded all the shopping stuff into the car, hopped in, key would not go into the ignition. "Dad, look, the car behind us has the same license plate and is also blue" ... "Oh!"
-- Regards, Joerg http://www.analogconsultants.com/
Aha, thanks. I did not know about the "size matters" effect. So it seems it's better to buy a larger one even though I need only 2-3% of the storage space.
-- Regards, Joerg http://www.analogconsultants.com/
You laugh, but once I was using the company van to pick up supplies at Costco. On reaching the Van with the goods I opened the back only to feel like something wasn't right. Not being mine I didn't immediately spot that the items in the back were not what was there before. It took me a moment or two to realize I had unlocked the wrong vehicle.
But I don't think the key would have started the other van. My understanding is they put more wafers in the ignition lock and fewer in the doors. This also lets them have different keys for the doors, trunk, glove box and ignition so you can give the ignition key to the car park attendant and they can't get into the trunk or glove box.
-- Rick
Except if you are doing the DDR output thing you are actually using a
155.52 x 2 clock or 311.04 MHz clock (with large inherent random jitter due to the poor pulse width control). This changes a lot of the numbers you just gave without changing the overall approach.I don't know why you can't grasp that the bang-bang detector will give you horrible phase noise. Just look at your simple simulation. The output wanders with a 40 kHz period. Filter as much as you want, that noise is always there. You are limited by the need for the circuit to respond to actual phase errors.
-- Rick
Which isn't very well.
Not when compared with most examples of "complex".
But the bang-bang detector turns the sub-nanaosecond jitter on the oscillat or output (sometimes early, sometimes late) into +50nsec or -50nsec correct ions.
So it forces you to average a limited number of artificially large offsets, and only allows you to do it at 80kHz.
A linear phase-offset to voltage converter operating on a fractional-n appr oximation to a 10MHz signal produces 125 times more samples. There are buil t-in offsets but they average to exactly zero over 125 samples, and you get 125 times more information about where the VCO phase is vis-a-vis where it ought to be, and that part of the offset is linearly related to where it o ught to be rather than being a mixed string of "too fast" and "too slow" co mments.
-- Bill Sloman, Sydney
Using a 4k-byte alignment for your partitions is important on many of today's larger-capacity hard drives, as well. Most hard drive manufacturers have switched over to "Advanced Format" - the sector size on the platter is 4kB, while the interface continues to operate with 512-byte sectors.
If your O/S continues to do writes in terms of single 512-byte sectors, this can lead to a lot of read/modify/write cycles on the hard drive and cuts write performance down rather badly.
If your operating system file-system code organized sectors into larger blocks (4kB or 8kB) as many do, and you don't line up your partitions on physical-4kB boundaries, then *every* write done by the O/S is going to require read/modify/write processing.
So, wherever possible, do your formatting in terms of logical block addresses (not the obsolete cylinder/head/sector numbers), and make sure that every partition starts on a 4kB (8 512-byte sector) boundary. It makes a huge difference.
Some newer drives now operate in a "native 4kB" mode, where the drive no longer "pretends" to have a 512-byte sector size. All reads and writes *must* be done in multiples of 4kB, with the logical sector numbers done accordingly.
Many older operating systems (and PC bioses) can't use these drives.
It's funny in that I think both Larkin and myself think in similar ways. We both want to have an intuitive feel for what we are working with. However, when I am in an area I don't fully understand I am happy to admit it while Larkin seems to be unable to accept that fact. This a perfect example. He "gets" the bang-bang detector, but seems ignorant of the limitations. The other methods he clearly does not have a feel for so are "awfully complex".
It is just so clear that the bang-bang detector will have ripple in the phase corresponding to the 80 kHz ripple in the PD output when locked. Nearly any other method greatly reduces this. Yet John still wants to consider the bang-bang detector.
Now he is talking about how this is a "time" problem and not a "frequency" problem. Actually PLLs have never been about frequency even if that is what the user needs. They are about "phase" which with a given frequency is equivalent to "time". PLLs have always been about time locking even if the user is more interested in the frequency aspects of them.
One of his early ideas to PD at 10 MHz included an ADC to add a "correction" to each sample to shift the input in time. In essence a phase control sample by sample at 10 MHz. If a linear PD were used, the output could then be "corrected" on a sample by sample basis giving a linear response at 10 MHz update which would add as little noise as feasible.
-- Rick
Samsung does some nice ones in the same sort of performance market with space for two hard drives internally.
It only really matters up to 256GB on the current crop of controllers - at that point all the IO channels are populated, but check benchmarks for the latest series to be sure. These things have a habit of doubling every now and then as the chip capacities also get bigger.
The OS should do this for you on an SSD.
The sweet spot is around 256GB after that they get slightly slower. If you really want something that runs like the wind you get a matched pair and put them in a RAID0 striped array almost doubling bandwidth. But don't put anything on it that you are not prepared to lose...
I found an SSD plenty fast enough. Way faster than a whining hot SCSI.
-- Regards, Martin Brown
On a sunny day (Wed, 03 Dec 2014 08:18:46 -0800) it happened John Larkin wrote in :
I have written most of the code for my ADF4350 evaluation board control program. Powered it up last night, and really it works. The program has so far more than 30 command line flags to select different config options, counter values, output configuration, etc. phase detector modes, what not. Was late, but it 'obeys' the commands.
I did have that desire to nuke the datasheet writer, he is beyond all logic (figuratively speaking), he counts data bits from 0 to 31 allright, but then in the same line the register bits from 1 to 16. And from his examples it is clear he is just parroting some lab report but in my view never even used the chip.
So I will have to write my own (make my own) block diagram from whats-in the chip (board), and put the command line flags in it so it [and using my program] can be understood by mere mortals, and this ebay board differs from the real AD evaluation board, and even from the picture on ebay (layout is different), so more fun. But... it works. That means that maybe in the weekend I can measure some spurs, although I am aiming for integer dividers and that should be spur free.
Then put it in a box, add the AD8346 quadrature modulator, and 2 differential inputs for I and Q, maybe a PIC to do the config via serial, and then I have an universal 135 MHz to 4.4 GHz quadrature modulator. I have called this thing: 'Anything Transmitter' as it basically cam make any signal in that range.
Sorry forgot to do simulations. ;-)
A serious case was what a friend experienced. Somewhere during a trip a waiter must have accidentally switched two credit cards. When he saw lots of expenses he never made on "his" statement but none of his expenses were on there it dawned on him. A call to the card company, the two met, switched the cards back, ironed out all the financials, then had lunch together. It seems for over a month nobody ever compared the signature on the cards with the ones on the sales slips.
-- Regards, Joerg http://www.analogconsultants.com/
Maybe the better option would be to use some of the huge RAM as RAM disk instead of a SSD. 2-3GB would be enough and that still leaves me 5GB for code execution. Right now I have 2GB which suffices.
-- Regards, Joerg http://www.analogconsultants.com/
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.