@@Thesignalpath Another trick making life easier: when you edit the video, add markers to the timeline, and then export them to an EDL file. It is a simple text file, which can be easily converted/edited into the table of contents. You won't have to go through the video again just to write down the timestamps. Each time I edit a video, it saves me a lot of time.
It's one of the (rare!) times that RU-vid has added a feature which is actually useful. I use it on all new videos and maybe it's worth going back over some of the older ones too.
@@video99couk I don't know if there has been a subtle shift in the algorithm, but I went back and added it on some of my older videos, and engagement for those videos increased the next day for a few days.
New Signal Path upload! New Signal Path upload!!! Alas, I die just a little more inside each time I don't see Phased Locked Loop in the title. Oh well, I guess I'll Like it anyway. 😉 Seriously tho, thank you for all of your efforts put into teaching others.
Just amazing to see what's inside such a sophisticated tool. I have one of those PLL EVK from Hittite and was wondering how to use them since sometime. Its software is fancy though!
I'm really starting to DISLIKE watching TSP videos... I keep seeing amazing equipment that I would LOVE to own, but I also know that I NEVER will! P.S. Keep up the great work kind sir
As much as i absolutely love this, i cant really think of any part of my work that would benefit from most of those features. I mostly design NFC readers and some SBC (mostly for payment or ticketing applications) and in that field, those kinds of measurements are not really needed. Still it looks like a very useful and well made piece of equipment, that would be useful to people who need it. Thank you for showing stuff like that on youtube, its always really interesting to come just a little bit closer to equipment like that.
You are great. Just I start to look for an oscilloscope (Tek 5 & Keysight MXR), here I have notification with Tek 6 series. I would like to compare this with newly released Keysight MXR series (12bit Vs low noise 10 bit)
According to official specs Tektornix ENOB is 8.8 compared to 9.0 in Keysight. Tek wins in bandwidth, Key in acquisition, also Key has 8 channel model. But what really sells me to Keysight is their beautiful phosphorus-like signal imaging with smooth gradients that gives fast insight how particular signal really behaves in dense areas.
Tek ENOB is 8.2, Keysight 8.2. Tek will win on noise, ~ 30-50% at 1 mV/div. Keysight standard record length is greater, but Tek maximum is 2x greater. The MXR is a great scope, rest assured; however, the Tektronix still has the edge in most physical performance metrics and once you get into software/analysis, like Spectrum View, Tek runs away with it.
To make so many measurements simultaneously in time and frequency domain is just fascinating. Great content on PLLs too! I have always wondered how to measure phase noise in a meaningful way. Thank you so much for all your effort, great content!
Great to see the types of interference and the effect on the eye and the freq/amplitude spectra. I wasn't sure how you generated the interference? Was the overlayed "amplitude" signal generated from cross talk from the adjacent channel and the frequency was from the BERT itself? Did you modulate the clock coming into the BERT for the frequency issues? Anyway, nice to see you bring out the BERT!
Nice as always! In one of your previous videos, you mentioned that you might someday be able to make time interleave adcs video. are these plans still valid? What are the limiting factors of this technique? What should i look at to lats say, build 1gsps adc from 10 100msps? How to characterize adc and it T&H? Digging through this i found patent US7978104 on digital compensation. Are such methods still relevant?
@wiatt: if u were involved in dev of this scope pls consider making a vid on some unique features of this scope. That would be great, especially if you are as cute as the guy from tsp
At 13:38, they have two southbound differential pairs from each TEK061 chip, but one pair has no soldermask and the other has. I guess the one that has no solder mask is for high-speed differential pair since those solder masks potentially introduce extra dielectric loss. My question is, the two chips shown in the video share the same footprint, but one chip uses the left diff pair for the high-speed signaling, but the other chip uses the right diff pair. what is the purpose of the other diff pair that has a solder mask on top?
14:05 first I though that these gaps in unpopulated channels are probably just connected with a simple jumper when channels are populated in a higher end model. But then I thought, why didn't they broke traces closer to the ADC then? Why do they want traces to go under the shield? Also remember that the shield is not covered with emi absorbing material on these unpopulated channels.
The package technology of the acquisition ASIC reminds me of IBM stuff. I wonder if it might had been made by GlobalFoundries? It's possible that the reason they didn't go with HMC memory is that Micron no longer offers the product. But then DDR is easier to interface anyway.
And here I was thinking I was clever by using my PC's 'sound card' to simultaneously measure the voltage and current of a device under test to create impedance and phase charts for audio speakers, crossovers, or similar devices... o.0
Could it be those apparent trace disconnects you were musing about on the empty channels @ 14:14 or thereabouts were done to prevent any antenna effect or noise coupling in ?..with no chips installed they would seem like un-terminated lines which might impact the active channels...just musing myself...great job as always sir
Overall, it is a very well designed instrument. The DC-DC modules look very interesting. As for those Teledyne relays, I would bet they could be used for very high speed signals without any issues, albeit they will be very pricey. The first inline relay seems to be there for grounding, or just to bypass a capacitor. It is hard to tell. As for the differential pairs going from the ADCs, they are reminiscent of the ones used in USB 3.0 or PCIe (e.g. one pair for each direction). Anyway, good video, as always!
Excellent presentation. Excellent details. I subscribed for more, more of everything. That said, the cost of your equipment is out of my reach by several decades. Programs like this teaches me new techniques and test types. I can only afford 80's and 90's vintage test equipment.
Quite surprised at the apparent lack of equalization support. Do they have de-embedding and channel emulation, at least? Is it possible these are extra-cost options that just aren't included in the demo license you got from Tek?
In your first experiment for the 2.4GHz signal and PLL. You drop the sampling rate below any usual recommandations. Your frequency marker read about 1GHz, instead of 2,4GHZ. Which is expected as you sampling rate is at 3,..GSa/s why did you do that? Does it would not influence your measurement to not sample your frequency at the right sampling rate?
When the DDC is engaged, the memory access is split between the time domain data and the baseband IQ coming from the DDC. This means that the sample rate of the time-domain samples is reduced. However, the full sample-rate data is applied to the DDC input, so there is no loss of timing resolution and Mr. Nyquist is kept happy...
"You can access the o-scope anywhere in the world with an internet connection"... 2 words... Industrial Espionage Imagine watching a competitor or rival nations labs right from the inside in real time! Security on Isle Three, ... lock this up with encryptions, please!
The advantage of the DDC - long term RF capture is possible. As the acquisition BW is reduced, the IQ sample rate is reduced, thus longer data records can be acquired.
After watching the video, I have some questions about the spectrum view feature of these scopes. I wanted to ask if someone maybe can help me out. I was surprised how the spectrum view function worked, because I assumed that an FFT is used (due to the selectable window), but the behavior doesn’t match with what I would expect in this case. After some googling I found an article to the Tek webpage which shows that a digital down converter is used to move the center frequency of the spectrum to 0Hz before the FFT is processed. Is it correct that signal must be bandlimited beforehand for this to work? When I look at the block diagram on the webpage both the time domain and the frequency domain signal processing pipeline receive the ADC data at the full sample rate. The reduction of the sample rate when using spectrum view with a wide span is therefore only due to limitations of the acquisition memory? Thanks for any help. Tek Website: www.tek.com/document/application-note/spectrum-view-new-approach-frequency-domain-analysis-oscilloscopes
Is the spectrum analysis in this scope completely in digital signal processing? Or is there an RF section that was not shown on the video? My guess is it is fully software (with hardwired DSP) because I did not see any signal split around analog frontend that would feed into an RF section.
Maybe I am missing something? My understanding is that heatpipes and coolers are totally different, almost like a resistor is different from an inductor! A heatpipe is designed to MOVE heat from one place to another while a cooler is designed to DISSIPATE heat into some other medium (air / liquid etc). Therefore, if you have enough volume to dissipate the heat locally, you don't need (or even WANT) a heatpipe. However, if you cannot dissipate the heat quickly enough with a local cooler (/ radiator), then you MOVE that heat elsewhere using a heatpipe to a place that the thermal energy CAN be dissipated.
@@trevorvanbremen4718 I'm saying that test equipment usually uses gigantic aluminium heatsinks that are extremely inefficient at cooling, instead of using a properly designed heatsink with heatpipes to transport the heat to the fins. Servers and PCs moved away from the old style a long time ago, yet you still see it all the time in test equipment, and the devices thus often need way too loud fans to keep them cool. Once an aluminium fin reaches a certain length, making it longer won't make the cooler better, since the thermal gradient along the fin means it's hot at one end, and barely above room temp at the other. A finstack with heatpipes doesn't have this problem, the heat can be spread across a much larger area in a smaller footprint. I said heatpipe cooler, not just heatpipe. Don't just assume someone has no idea what the words mean that they're using, try to understand what they want to say.
@@tommihommi1 You're still missing the point. Heat pipes don't cool at all. They just move heat to something else that cools. In the case of servers, the heat pipes are conducted directly to a refrigeration system, outside the building. In test equipment, where some degree of portability is required, heat HAS to be conducted to the surrounding air, because you don't have the luxury of having a heat exchanger at every test bench.
@@BrightBlueJim Just read what I have written here again. Heatsinks that work exclusively by conduction are inefficient. The longer you make a metal fin, the less work it can achieve, since there's a large heat gradient across it, and one end will be much colder than the other. Using heatpipes, you use convection to distribute the heat across a larger area, and thus get a much more efficient heatsink. Lool at the heatsinks in this video. They utilize heatpipes, and can be very small. It would be impossible to cool these chips using just chunks of aluminium with some fins, or the chunks would have to be absolutely massive. Like it was done in older test equipment.
Ok, I just assumed the 4-pin SOP package near the capacitor in the low frequency path was a solid state relay. Keep up with the good job! Saludos desde Argentina
I am pretty sure selling MY house would not suffice. Maybe that is good, because I have no desire to move under any bridge with that baby in my lap, but no electricity to turn it on...
Definitely not (typical) hobby material, but this isn't like buying a robot where you have an option to have people do the work. With this you're basically buying a superpower. If you could bill more for work or get more types of work or fixed bid work done faster this could easily pay for itself in less than a year.
