EEVblog
Vložit
- čas přidán 12. 07. 2012
- Forum Topic: www.eevblog.com...
Some follow-up measurements on the Jim Williams pulse generator circuit.
With a special guest appearance by a rather expensive scope...
www.home.agilen...
The Amperes T-Shirt is available here:
www.redbubble.c...
Join the EEVblog Electronics Community Forum:
www.eevblog.com
Dave's Amazon Store:
astore.amazon.c...
Donations:
www.eevblog.com...
Projects:
www.eevblog.com...
Electronics Info Wiki:
www.eevblog.com... - Věda a technologie
Nice go with the 35 cm of coax Dave. That turns it into what I call a charged-line pulser. I learned that term in the mid-late 1970s, when I worked for an org that dealt in FAST pulse phenomena. I built many charged line pulsers and my co-ax of choice was always semi-rigid. The solid outer conductor keeps all of the E field inside the coax, whereas the braid outer conductor of flex cable is "leaky". This means semi-rigid is not susceptible to body capacitance altering the pulse characteristics.
That $140,000 dollar scope... Dave, since you used it, I demand a teardown, review and, more importantly a drop test.
Are those six decimal places of picoseconds in the measurement really meaningful... one attosecond?
My jaw dropped when you said 16Ghz. For a split second i was expecting 6Ghz, and i was already in awe.
Be nice to see what a scope like that is required for, I guess there are many reasons people require them.
Keep up the good (GREAT) work!!
I have to say, throwing scopes around is very convenient. It gets them out of the way so fast and easily :-)
Help from who? I'm director, producer, cinematographer, sound engineer, editor, grip, gopher, and on-screen talent.
You can't use the waveform as-shown to calculate scope bandwidth. You have to extrapolate the rise part of the waveform to full vertical width of display. Think of it this way; if the voltage out of generator is 1/10th taking up only one vertical division, you'd get rise time of 0.2 ns, but that doesn't mean your scope has 1.75 GHz bandwidth.
If you extrapolate, the rise time is close to 2.5 nsec, which is about 140 MHz. Yes, not as accurate as testing with sweep frequency, but close.
00:30 gave me a heart attack!
That attenuator looks like a prop from the original Star Trek. :D
That ramp up after the leading edge is caused by high frequency loss in the small coax you added. Try some bigger cable like RG-8
we need a damage report of that scope you threw away
You could add a 50Ω resistor to the end of your coax to get rid of the "step" in the rise time. That step is the signal hitting the end of the coax and reflecting back down the line due to not seeing a 50Ω termination. If you want to really get it as flat as possible you could use a pot instead of a fixed 50Ω resistor to dial it in perfectly. On an aside, if you know the velocity factor of the able you're using, you can measure its length using that "step" in the waveform and a bit of math. Cheers.
This video gave me a double heart attack followed by an envy induced seizure.
I will be sending you my medical bill, apparently you can afford it lol.
With the Rigol scope if you enable equivalent time sampling it changes something in the front end that increases the bandwidth somewhat, at least on my DS1102E. But I think the modified 1052E will probably do the same.
Yes, I actually threw it. I'm not that adept at video editing to fake that.
T-Shirt link added to description!
You'd be surprised. Anyone who does any sort of leading edge high speed design, which is a ton of stuff these days, would consider this tool essential. You cannot characterise your design without a tool like this. $140K is petty cash for a lot of large tech companies.
Why you throwing scopes though?
I suspect that stepping on your waveform (after removing the capacitor) is caused by signal reflection on your length of RG174.
If I'm right, you've essentially built something along the lines of a time-domain reflectometer (TDR) rig, which is measuring the length of that piece of RG174 cable (or possibly the path from the pulse-gen through the attenuator and to the scope).
lol you crack me up! First you throw a $10000 scope, then you forget that you had a 13 GHz scope "gathering dust" in the lab! XD
Nice to see that Rigol back again:)
Fun experiment... Send me that scope you tossed on the floor... I'll fix it..! LOL
Hi Dave, When I was testing Bandwidth I used a Tektronix Tunnel Diode Pulser on a TM 500 series, with a calibrated cable for the Tunnel Diode Pulse Generator 067-0681-01 Calibration Fixture in the UKAS Instrument Test Laboratory,
For myself I managed to get a HP 54750A 50GHz Scope & repaired that, very nice unit, for fast pulse Tektronix 284.
The question for big industry is 'without this tool what is the risk?' (i.e. measuring 5.8GHz signal with a 6GHz scope). If it is a product for aerospace then a widget going wrong could at best cost hundreds of k's to fix - for want of 10k a week lease or 140k capital...
Also remember these will last years, at least 3 year warrenty with full repair service avaliable. Use it 5 hours a week, thats $60 /hr over 10 years. And it will still have at least a 25% trade in value.
set it to manual focus, that way you wont have such dramatic focus change, and seeing you have help, your helper maybe able to stand behind and do the mini focus changes needed.
HI Dave!
I've seen this little flat bit in the rising edge at 7:00 min before, in my Signal Integrity class. The additional coax forms a transmission line of course, which is not terminated! :(
So you have to wait the propagation delay of the t-line to have the full peak due to the reflexion at the end.
I've modeled your case in LTSpice with the tline component and got a similar result.
Greetings from Germany ;)
We have a 100GHz scope at work for all of our radio and satellite work. Cost is no problem when it's an essential tool
You may want to try and terminate the add-on co-ax cable to stop reflections that may be causing the "kink" on the rise slope. A carbon variable resistor in the 200 ohm range should help.
After watching 4 GHz Tektronix scope teardown I am just wondered what kind of high-freq design should be inside this thing. And yeah, if there are 100 GHz scopes, how do they (a designers of 100 GHz scopes) debug and test their scopes on development stage?
Connected the Jim Williams pulse generator with 136ft of coax to my Tesla and the zero to 60mph times dropped to 2 seconds flat....
I don't think it likes the dark screen being in most of the shot.
Sometimes, a silly detail can ruin your sleep. For example, the video mentions the scope price in Australian Dollars. Now, But I learned that Australia used to be a British colony. Then why isn't the price mentioned in Australian pounds?? Why dollars? I've been oblivious to this for too many years.
Aargh!!! My heart stopped when you threw away that scope. :|
Please don't tell my you actually dropped the thing on the floor. *Please* tell me there where a whole bunch of extremely soft pillows all over your workshop. PLEASE tell me you edited that drop sound in afterwards!
If Dave's scope is a 4 Gsps, that's 250ps between each sample, so how can it measure a rise time at 630ps? I'm assuming it's interpolating between samples and then calculating the 10% to 90% time of the interpolation. It'd be interesting to see it with interpolation turned off. I wonder how much of that signal is actually real.
that would be excellent for a delay line memory.
"Errr boss, I just smoked that brand new scope you bought us"........
Nice!
Gotta love that humor :)
Well, they look impressive :->
wow a Tandy calculator, not seen one of them for a long time.
Yes, I laughed at that too. Basically, he's using a $140,000 scope to do the measurements, then working out the answer with a $2 calculator.
So, I didn't get it at the end... are we are going to have peek inside Infinity?
ok where and how the hell did you get a 140grand oscilloscope?
Hello Dave.
Talking about Jim Williams (RIP), could you give some inspirational insights about the so-called analog circuit design, as well as its legendary heroes (including Bob Pease, Bob Widlar etc) and the needs of analog engineers. Bob Widlar once said "every idiot can count to one".
WELL if you don't want the old one, .... ill look after it for you
Those cascaded adaptors are degrading the accuracy of the measurement!!!Btw. (cheap) BNC's are usually no good above 1Ghz unless so specified!
Did you really throw a $10k scope on the ground?
mrnuke I fucken hope he didn't. I saw red with that. Hopefully it was just a case or someone caught it.
Will sin(x)/x interpolation affects the result?
Dammit Dave, I've got a Rigol-scope like yours wich I really do like. Now you´re just making it look like some kind of a toy.
Beasty oscilloscope!!!
You must teardown the 140,000 $ scope !
Auch! It hurts to see when you throw a 10K AU$ scope like that.
Yikes!! Are you going to do a giveaway with the old one? If it still works, that is...
Yeah, it made me jump a little bit.
Australia changed from Australian pounds to Australian dollars in 1966.
You are not changing slogan, are you?
"Don't turn it on, crash it apart!"
Yes I also think, that he tricked us. The possibility that the scope would have survived a impact on the floor would have been minimal. So we maybe wouldn't have seen the Agilent 3000 in action later in the video.
What's the deal with adding open coax across the cap? What does it really do?
add capacitance.
Did you really throw that scope?
I bet that linear fall is due to the coax being rolled up. Guesssss
I honestly never heard BNC sex adapter.
A coaxial cable.
How did you get that SCOPE! Also you made me cringe when you tossed your other scope.
What kind of industries can afford a 140K (Aus$) capital expenditure like that?
Aerospace? A factory manufacturing consumer goods?
A University classroom with 20 lab benches would not use these.
They might have one in a locked closet, closely guarded.
Even a lease for that would be a business killer.
I'm betting it's on loan for a review, Dave.
How somebody who supposedly needs this bandwidth can avoid
this prohibitive expense would be a fitting topic, wouldn't it, Dave?
Greegor47 Those retain their value reasonably well. If you need that sort of a scope, you have engineers costing you $100+/hr, and prototypes costing $5k in BOM cost alone. Not having one of those scopes and having to respin an unfixable prototype, or failing conformance tests - it’s cheaper to have the scope. At such speed, the prototypes are mostly unfixable. This is not something a hobbyist would need, because it’s hard to meaningfully use it without using expensive parts (cost-wise or know-how wise - you may need plenty of expensive experience to use some chips you find on a boring PC motherboard). Note that 1GS/s scopes can do 40GHz equivalent time sampling so for repetitive waveforms you don’t need such a scope. You need it to capture transient events in high speed logic. Even an eye diagram at 50ps/div can be done with a used $3k scope. But that’s because you got a repetitive waveform - if you got one. If the event that closes the eye is rare, it’ll take lots of time to capture it in equivalent time sampling.
If you need a scope like this then generally you really really need it. And if you can not afford it then you will most likely not be able to afford to do the high speed design you need it for anyway. I actually prefer to use prebuilt boards and modules with proper quality control of material and precise dimentions for everything that runs at above 250MHz. Especially if the something is digital. Unless it is something very simple like a pulse generator or a pair of roughly impedance controlled short traces from a USB port through protection diode IC straight into an MCU it is just where it starts to become way cheaper and easier than trying to design and tune it yourself and where even for like a few hundreds a year batch manufacturing in most casrs there is just no point in rolling your own modules.
Your canon keeps losing focus ...
How do you even get this kind of equipment? Do manufacturers simply let you play around with 100k+ AUD oscilloscopes and hope they aren't broken when you give them back?
Imagine you tripped on something and you grabbed that huge attenuator: the connector would be broken and the oscilloscope would be on the floor (with a broken screen)!
50 MHz scope reading 136 MHz.
Awe yeah.
How can you afford it :) ? ... Noobquestion: What is koax cable ?
Omg... I damp!
better than a teardown just a threwdown
Is that $140,000 AUS dollars!!!!????
Sexy bit of kit :)
Teardown?
13GHz!!!!!!11!!1!1!!!11!!
man. dat 'scope
Hi, I've build two of the Jim Williams pcb's obtained from OSH Park because the picosecond-pulser, see: www.siliconvalleygarage.com/projects/picosecond-pulser.html, or other kits couldn't be bought anymore. See also: oshpark.com/shared_projects/KrhgN8JM . Sadly my transistors do not work on the OSH Park boards that provides about 90V DC from a LT1073. I have tested the transistors with part of Kerry Wong's design, See here: www.kerrywong.com/2013/05/18/avalanche-pulse-generator-build-using-2n3904/ . And apparently the OSH Park design doesn't work with the tested 2n2369(A), BSX20, BCY70 etc. It only pulses at about 2 Hz at random and so doesn't work as intended. And their design uses a 1Mohm resistor where the Picosecond pulser design uses 10Mohm in the collector of the 2n2369, plus more 2A capacitors of 4x 2pF instead of only two 1.8pF capacitors in the OSH Park circuit plus a trimmer of 5 to 27 pF in parallel. Also the OSH Park pcb circuit uses 100 ohm to power current limit the LT1073 instead of 220 ohm and only one 10uF on the 5V microusb input voltage which is a 3V lithium cell in the Picosecond pulser. Changing those from 1Mohm to 10Mohm , and adding a second 10uF didn't make any difference because all ten brandnew 2N2369A's kept refusing to oscillate in the OSH Park pcb boards. I used Kerry Wong's partial transistor circuit to test with my DY294 the voltages at which my transistors oscillated at avalanche level. And that was for a BSY at about 174V, for my 2n2369A about 114.5 V DC (on the 220kohm test circuit input! The collector voltage was lower at about 40V DC). It looks like that it is impossible to get the OSH Park design with just about 90V DC working and I wonder why the Picosecond pulser design or the here tested Jim Williams Pulser is any different and does work? But since the transistors work in Kerry's circuit but refuse to work in the OSH Park boards there must be something wrong with the design or their pcb manufacturing. And the last version 2.4 of that board was published in March 2017, and in February 2017 the one before that with the same version number!, but the board must have serious flaws! So better do not order those boards or the parts needed to build that one! Instead make your own Jim Williams Pulser as Kerry Wong and others did!
mother of god
@nangbutinang
Smells like ... eeh ...
Wrong! Smells like Teen Spirit! :-)
Link to the previous video on this topic: czcams.com/video/F-ZDiGmLvTs/video.html
I can't stand you throwing a piece of equipment, that a lot of us wish we could own. thumbs down mate.