Products > Test Equipment

DIY Logic Analyzer Probe and Pods for Siglent (and LeCroy) scopes

<< < (37/51) > >>

rogersstuart:

--- Quote from: Johnny B Good on December 30, 2021, 10:52:40 pm ---@rogersstuart

 I was prompted by oz2cpu's sneaking suspicion that the main PCB may not have been an exact replica of his original 4 layer board and the great importance he places on it meeting this requirement, to verify whether it was a 4 layer board or not. You answered the question honestly and I thank for clarifying that point.

 I sought further advice from oz2cpu before making any decision to cancel the order since your stated eBay policy is "no returns", denying me the opportunity to 'buy and try' as you've just suggested. Cancelling the purchase whilst this was still possible seemed the best option all round and I thank you for deferring to my cancellation request and the speedy refund.

 Since the Belden ribbon cable undoubtedly accounts for most of the high levels of crosstalk shown by the graph you took the trouble to publish, it's impossible to say how much of a problem the lack of the inner groundplane shielding layers would be in this case.

 Since oz2cpu had stressed the importance of a 4 layer PCB, I deferred to his judgement over your claim that this wasn't really that important after all. You may be right but there is no way to prove this, short of actually testing using twisted pair or coaxial cabling between the main PCB and the pods.

 One way around this issue with your version of the main PCB, would be to wire up a pod using twisted pairs or coaxial cables and generate a new crosstalk graph for comparison against the Belden cable you're supplying as a courtesy to those with a more modest initial performance requirement.

 Provided you've made it clear up front that the main PCB is only a two layer version of the original design and spell out your reasons for this cost cutting choice, backed up by the new graphical evidence obtained using twisted pair or coaxial cables, I should think you wouldn't have any further problems in this regard.

 Supplying only a single 'starter pack' of test clip leads isn't such a bad idea since the quality and price of these test clips can vary considerably and it's best to economise here and allow the end user to make their own choice in the price/quality equation.

--- End quote ---

No returns is just a psychological thing and part of doing business on eBay. I do not want to pay for shipping to Europe. There is no such thing as no returns on eBay once someone opens a dispute. I would prefer to leave people feeling satisfied even if that means a loss here or there.

I feel like I should not have responded because the thread has been derailed now. I saw mention of opening a dispute and you sounded upset, so I said something. I was worried that someone might act maliciously assuming that I'm a bad actor out to scam people.

I agree with you about the high level of crosstalk. I'm not sure what it should be. At best it is -15 dBm below 100 MHz, at worst it is -8-7 dBm. Though, I have to admit that I'm not sure if I measured it correctly. I soldered female DuPont connector pins to SMA connectors and measured S21 using my 75 OHM VNA that was calibrated with 50 OHM Rosenberger "standards" in a 50 OHM system impedance to the end of the cable. I had to bend the pins a little and jam them in there. It was pretty sketchy. I have been told by the designer of the 8753E that it is acceptable to mix impedances like this. Recently I have found that adding pads can clean up a trace and improve accuracy. I would measure it that way now and might see a slight improvement.

I also normalized the data from 50 ohms to 120 ohms based on the 120 ohm cabling that is specified. I'm not sure if it was correct to do so.

As someone had mentioned, the clips that I include are the cheap kind. You know, I wish I could provide good ones but I couldn't find a supplier for them at a low enough price. I say that you get 8 but you would actually receive 10 in case 2 are bad. I count out 10 for each bag and inspect them all. It's really just tacked on so someone can immediately test their build. It's just as cheap for you to buy more on Aliexpress as it is for me to. I don't have a specific markup for this product. All I can say is that I'm selling it for at least half of my calculated price so I'm trying not to add extra things or make it more complicated.

natman69:
I am interested in building the logic probe for Siglent SDS oscilloscope based on this thread project (many thanks to oz2cpu and other forum members for their amazing work  :popcorn:).
So if someone it is also interested, I suggest to group buy PCB assembled with SMD passive components from JLCPCB or other suppliers.  Maybe also 3D printed enclosures and cables.

rogersstuart:
I kept thinking about this so I decided to go back and see if I can find better answers. It appears that oz2cpu's design (and the HP design his was inspired by) are meant to form a voltage divider. I used a VNA to generate a model of the digital input to use as termination. I would have preferred to know exactly what components are used internally but I couldn't find a picture of both sides of the scope PCB. It's 10k DC which leads me to believe there's more to it than TVS diodes and the comparators.  I also used measurements of 24" of the ribbon cable I have on hand in a GSG order. All connected by co-simulations of the PCB layouts. I was able to select a reasonable set of values to get a 1/10 voltage divider action and the waveform looked okay. Since this was based on my own measurements and I only have so much confidence in my setup and skills, I admit that there is probably some error. The values provided by oz2cpu produce a squarish waveform in the simulator with a close to 1/6 voltage division which is (for me) good enough to back my assertion that this is meant to be a voltage divider.

As expected, there was some coupling between the top and bottom traces. I isolated a segment from where the 120 ohm resistor is to the card edge. At worst you can expect 6.2% voltage coupling (700 MHz) with a 10k load and near 1.3% at 1 MHz. The amount of signal that makes it through decreases with increasing frequency which affects that number (increase).

I tested swapping the 8.2p capacitor for a 20p capacitor and swapping the 120 Ohm resistors for 100 Ohm resistors. With my function generator referenced to a 10 MHz square wave being generated by my scope I saw little to no difference between the original or either modified component selections. With the 20p capacitor maybe there was a moment of jittering while handling the probe. Of course, this was best case since the sampling should have been fully synchronized with the waveform generator. All this tells us is that the waveform was not distorted enough to be unusable. I did not try swapping the 86.6k resistor. My simulations indicated that a wide range of values (as low as 20k) would work.

To further verify functionality for basic logic analyzer tasks I fed low rate (11520 bps) serial data into the probe. This signal was generated by a FTDI UART. I measured the rise and fall time to be near 10ns. I repeated "hello" 100 times, triggered off of it, and counted the trigger pulses. There were 100 triggers. I followed this by inserting a 200k potentiometer. Across the entire range the scope was able to decode the signal (less reliably at 200k) which is a point where my simulation drastically departs from reality. No glitches or false triggering. This seems to be acceptable.

Every choice you could make has a tradeoff. oz2cpu's design seems pretty flexible. If you are trying to cut costs I can't find a reason why picking a different cable, using slightly different passives, or even making some changes to the layout would prevent this from working. The only way I could break it was by selecting a wildly different set of component values. The voltage divider action or at least some compensation for the reactance seems necessary for this to work properly.

Slartibartfast:

--- Quote from: AVGresponding on December 30, 2021, 09:33:48 pm ---Calling something "compatible" implies a matching of capabilities.

--- End quote ---

Not really. Unfortunately that word is far too corrupted by marketing people to mean anything like that specifically.

gtube:
Hello oz2cpu,
Thanks a lot for your great work and sharing! I just tried to order a couple PCBs from JLC. I just noticed that POD PCB is missing in the thingiverse link below. Only scope plug in PCB is there.  Could you please upload the POD PCB?

Thank you!


--- Quote from: oz2cpu on March 05, 2021, 08:43:53 am ---
ALTIUM AND GERBER FILES ARE NOW IN THE ZIP FILES HERE:

my 3D design files : https://www.thingiverse.com/thing:4784443


--- End quote ---

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod