| ??? 10/05/10 16:20 Read: times |
#178927 - Just an XOR? Responding to: ???'s previous message |
Your post title: "None of this works with just an XOR"
Now why in the world would you ever invent such a post title? Where in this thread could you ever get the idea that anyone have claimed the use of just an XOR for an IC tester or to detect the difference between different 74xx or 4xxx chips? Richard said:
First of all, the XOR doesn't tell you anything about an unknown device. Sorry, but unprofessional statement. If a logic tester can match the logic table of the chip then gate delay measurements are possible for a large percentage of the chips. If the logic tester can't match the logic table of the chip, then it's irrelevant if a the XOR-based phase detector can measure any parameter or not. Richard said:
Let's say you have an 85L54. That's a 4-bit tristate binary counter. How many sites do you see with datasheets for a 85L54? Did you notice that I already covered the "some but not all" part? Are you scared of arguing within the boundaries of a thread? Think that jumping in random directions will give you advantages? "Such a tester would just use a bit of clever source code to identify a huge number of 74xx chips." Notice the "huge number of"? "The scope I was discussing was autodetection of most logic chips in the 74xx and 4xxx families, [...]" Notice the "most" and not "all" in the sentence? "As noted earlier, the 74xx and 4xxx families have a huge number of chips. But most of them are very uncommon, so it might be meaningful to create an IC tester that has some limitations for more uncommon chips." Noticed the "some limitations" part? "So the outcome would not be a 200 individual sets of test vectors but instead an optimized tree where the outcome of each test vector selects which subtree to continue with until finally ending with zero, one or multiple matching chips." Notice the "zero, one or multiple matching chips"? "Multiple matching chips is possible if the hardware isn't able to separate some parameter between two chips. Such as a 4xxx chip with identical logic function and pinout as a 74xx chip and the hw or test vectors don't support the testing for existence of schmitt-trigger inputs or approximate input logic levels or the voltages of logic-high/logic-low of outputs." Noticed the "isn't able to separate" part? But let's go ahead. The logic table for the 85L54 chip isn't too hard to process. And tristated outputs can be detected - as covered in my previous posts - by driving the pins with weak pull-up or pull-down and see the result. Same logic as for detecting that inputs are "not outputs". Richard said:
How and where would you apply the "square wave" and at what rate? Having decided that it looks like a 85x54, I can clear the latched values in the chip. Then drive OD1, OD2, CLEAR low and PRESET high. Input a 1MHz square wave on TE. According to the truth table, a low on TE should emit the latch values (which I did clear earlier) on the outputs. Driving TE high should force all outputs high. The datasheet claims 26-50ns propagation delay for 8554 and 90-160ns propagation delay for 85L54. 1MHz symmetric square wave means 500ns half-periods, so an XOR phase detector would have 0-500ns measurement range for 1MHz square wave. There is a bit of measurement errors from rise and fall times of tested chip and the XOR gate, capacitances in signal traces, ... but a phase detector would still be able to do a decent job of measuring the delay. Going 2MHz would give 0 to 250ns full range if I worry about maximizing my measurement range. The L version has typical clock frequency of 11MHz so nothing strange to play with 1 or 2MHz on TE. Richard said:
I'd be interested to see how you'd exercise an unknown device and how you would, with this XOR-gate method, exercise a 74AHCT193. There is simply no way you can have found any part of my previous texts claiming the use of any XOR logic for testing the truth table of a 74193. I have very clearly mentioned the use of a phase detector for checking delays to try to decide on chip family after the base model is known. The 74193 has a logic table just like most other chips. It may - or may not result in collisions with other chips in the rule database. Already covered. But let's say the database doesn't give any collisions. What's magic about a 74AHCT193? That is may not show up in the surplus stores in some parts of the world? Or that it is one more chip with very few datasheet links, and selected by you because you just are A.N.T.I. everything? Anyway - it does have outputs that can be directly affected by inputs and with timing values specified. In this case the signal /PL - parallell load - loads the counter with Dx and outputs the data on Qx. Propagation for Dn to Qn is a defined parameter for '193 chips. How easy is it to separate the full set of families are a separate issue with one important factor? If you need a HCT and the logic tester fails to separate a AHCT from a HCT then you should be fine with using that AHCT chip in your design even if you didn't know about the "advanced" character in the name. Anyway - already covered in: "Someone who wants to make a really cool tester could add a high-frequency oscillator allowing a square wave to be sent through a gate and mixing it using an XOR gate with undelayed square wave data. Combining the two signals and low-pass-filter the result would give an analog value that depends on gate delay times allowing a tester to make guesses between 7400, 74LS00, 74ALS00, 74HC00." Richard said:
A schematic would be helpful, just so we all know what you're doing and how. In your 25+ years, you haven't picked up on XOR-based phase detectors? Or didn't realized that even if the XOR has propagation delays, an extra delay on one of the inputs do affect the timing on the output? Try google and "images" and search for "xor phase detector". I'll think you'll get the idea. If not, then I think we have to do the tutoring part offline to not waste too much bandwidth in this thread. Richard said:
When you've done that, please tell us how you'd extend that to check an unknown device. Why should I? If you want to design a device to check an unknown device, please go ahead. Please check any random microcontroller with any random firmware or any CPLD or FPGA with unknown contents. Have fun. No one in this thread has claimed you can detect anything. You would also know that, if you did read the thread posts. I even mentioned that in some situations the user may have to decide manually what pins are VCC and GND since there exists logic chips with where the power isn't in the diagonal corners, and it may not be safe to try to auto-detect where to feed them. So a design may be able to autodetect "normal" chips while perform basic-level pass/fail testing of some additional chips after manual supervision. But you are basically practicing your basic hobby - makign wild statements or trying to switch the focus in random directions when you get into troubles with your debating or you just don't like a solution or idea. And your other basic hobby is to try to request others to do work to prove their claims. But you never do the same - posting proofs of your own claims. We still haven't seen you do your homework and respond with reasonable examples for your claim: Richard said:
They've made every device attached to the computer, aside from serial ports, look like a file device, so whatever is attached to a computer is infected with any and all evils that reside on that computer within 100 ms. http://www.8052.com/forum/read/178732 Most probably we will never see any, since you aren't interested in investing time just to find out that the evidence out there claims you wrong. |
