Products > Test Equipment
Some old school instruments showing how it's done (HP 3325A and Fluke 8506a)
<< < (67/91) > >>
SilverSolder:

--- Quote from: Roman oh on February 21, 2021, 11:51:28 pm ---
--- Quote from: m k on February 21, 2021, 11:30:33 am ---

I've recently wasted time with google.
Everyting seems to be plently first and then finally just junk.
I tried to find an i8080 emulator that would suite for hardware testing but found nothing.
Memory says that Altavista was better.

Obviously I had to start doing my own emulator, it's still just a shell but I'll upload it if it becomes usefull.
For a moment I thought maybe Ghidra addon can do it but then later there will be an avoidable obstacle anyway.

--- End quote ---

I couldn't find a useful 8080 emulator, but there seems to be a lot of 8085 stuff out there (100% code compatible with 8080, and just ignore the different external interrupt structure (RST 5.5, 6.5, 7.5); the 8080 interrupts RST0 - RST7 are treated as software interrupts in the 8085, but the code works fine).
I had some fun putting this code into an 8085 emulator that I downloaded from Oshon Software (no affiliation). The evaluation version is free, but closes after a one hour session, and you get 30 of those sessions for free. But the licence is only 25 euro. It allows breakpoints, single stepping, modify rom, ram and registers (very useful for shortening the start-up delay at 02E4H). I spent (won't say wasted - as SilverSolder says, it's all instructive) a couple of days trying to get my haed into the code, but it seems to the epitome of the reason why spaghetti code is frowned on. I suspect it may actually be the output of an early C compiler, or somesuch - I used one of those on my CPM system in the early 80s. The code flow threads around like <delete rude simile!>.
It transfers code into RAM, which you might do if, say you wanted to configure the code on the fly according to different models or somesuch, but the code it transfers to RAM looks like simple arithmetic operations, and they appear to be executed during start-up even before it does anything that looks like determining a model. (The Oshon simulator manages to execute these "RAM" instructions OK, but because they are not part of the loaded program (which, incidentally, it seems to disassemble OK) you can't "see" (ie trace and breakpoint) these steps. Would be simple enough to fudge, if one wanted to.
And it does other crazy things that a "human" assembly code programmer wouldn't do - eg, bunches of subroutines, 3 bytes long, clear the same memory locations twice in two different places, etc, etc, etc.
It was doing my head in... and I was only doing it "because I can".
So while I shall keenly track this thread, I don't think I'll be doing a lot more on this in the near term. But I got my 25Euro worth!

--- End quote ---

I looked at it long enough to realize the same thing - i.e. this code makes spaghetti look well organized...  they were severely space constrained and this was probably a pretty big software project in its day - they did what they could with the tools they had, I'm sure, so I'm not casting aspersions on the people that wrote it.

Do you know if there is a way to temporarily halt or suspend an 8080 in circuit?  - it might be possible to glean something by halting the processor at certain points during execution, and looking at the bus to figure out where it is...  letting it be its own emulator, if you see what I mean?

joeqsmith:
The only thing I thought about doing with the firmware was to try and adjust the timing for the serial port.   I thought I would setup the LA to trigger on the UART while monitoring the data and address buss.   Once I have the address, I would disassemble these sections.   Guessing its an interrupt routine running off some timer.   I was then going to use a ROM emulator to hack the meter. 

The low cost programmer made it in today along with the replacement printer ports.   Working on getting their software setup now.   

Odd, Motorola, Xilinx, Altera are not even listed.  No support for the small Atmel serial parts.  It does support some of the more common parts like the small Xicor ones.  Wafer Scale supported.  Too funny.    I'm impressed with the features.  It's pretty much loaded like my old one.  There is a short test for VCC and VPP.   I ran a few 4000/7400 series part tests.   Then tried some GALs.  Not sure but it looks like there may be a problem with them.   Virgin parts check out. However, after programming, the unit will come up with pin faults.   Replicated with same results.   I'll just need to spend some time with it. 



SilverSolder:
Sounds fun!  It would definitely be elite to release a new version of the software after 30 years!  :D


Meanwhile, I got my GPIB bus back in business.

I attached an interesting set of graphs:  comparing the frequency responses of two 8506A on all their different ranges, using a 3325A as the signal source.

The test setup is not "serious" because I didn't have a terminator capable of 4 Watt so had to run with "free exhausts", but the trends are nevertheless interesting.

Note that both units are to spec, as far as I could tell in a few minutes of looking, but they are definitely different!  The thermal AC modules have variable capacitors for tuning these curves, which is something I'll take a look at the next time I attempt to calibrate them - now that it's easy to perform a canned test!


joeqsmith:
Interesting data.   I don't have any way to create a stable enough waveform or to measure it to know how good the Fluke is.  I had compared it with the old HP meter and the two basically agree in all the Fluke's ranges.   Things diverge once I start getting outside the range the HP is spec'ed for.   Even at the low frequencies the HP would not tell me how good the Fluke is.   The errors from all of my sources far exceed the Fluke's specs.   This is why I decided to leave that section alone as I'm sure I would do more harm than good. 
SilverSolder:
My AC source is not stable enough to calibrate with either.   But any source is good enough to somewhat decently compare two meters triggered to sample at exactly the same time.

The AC spec gets much looser above 50KHz, so the meters being different isn't really a sign of a problem, it is just a sign of it being difficult to maintain tight production specs for the higher range attenuators, it seems?

Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod