The thing is oscillating. Try putting a few EMI supression ferrites over the cable bunch. A few of them.
Positively yes: it's oscillating a lot!
I've got no ferrite beads in my parts bin but have tried your other suggestions:
Try to use twisted wire pairs for each of the lines. Connect one wire from each pair to ground on the board side. Leave these ground wires unconnected on the remote socket side or connect them together, see if one or the other way works better.
Edit: the twisted wire should have a few twists per inch, not just loose twist like in network cables. You can use a drill to make a length of twisted wire.
Nope! No difference at all. No matter how I use twisted pairs grounded on one side or both, that thing is still oscillating.
Also what else you can try is putting small EMI chokes right on the legs of the transistor being tested.
That is a good idea but it's not practical. I pretend to use this kit to compare and match up to 4 trannies/FETs at a time from a batch and that would be a PITA.
Edit: BTW the original article posted at the beginning of this topic says connection length should not exceed 10cm, they knew long wires will create problems.
That is the key, obviously. I have reoganized things inside the box and have rewired and shortened the leads drastically and have had partial success. At least two of the four slots in the ZIF socket are now free from oscillations. Have not tested the banana connectors yet. They are intended to test bigger power transistors that wouldn't fit in the ZIF socket but I am afraid that the necessary additional leads may make things go wrong. As soon as I try it all, I'll let you know.
Thanks again for your suggestions.
My approach & understanding of this subject - your bad signals / distortion is from what I'm into - ham radio , is called common mode current - and impedance mismatch that creates CMC , that is a 3rd signal path on the outside of the shield of the coax .
Also my approach with issues like this - is to solve / eliminate the cause of the issue , rather than coverup or react to he issue .
It could also not be CMC , is could be bad coax , or the BNC connector not correctly installed .
I may look for one of these kits , besides building one of Mr. Carlsons units .
...could be bad coax , or the BNC connector not correctly installed...
I had to stop working on this project but kept thinking about it and that is exactly the same conclusion I ended up with. More on next week.
Back to the bench. Partial success
I've discarded a problem with the BNCs or coaxial cables. After much tinkering with the switch and ZIF cabling I have succeeded in making all 4 test slots to work fine but no way I can extend the first two ZIF slots to the banana sockets for the power trannies. No matter how I route the cables, the oscillations come back. Besides I'm having problems with the other cabling, i.e.: the wires from the board to the PNP/NPN external switch also make it to oscillate a lot. The "Small/Power" cabling do not seem to affect the circuit stability at all.
I'm forced to discard the current box, which I had lying 'round in my shop, and use a smaller one.
[Edit: typo]
So my turn to build one of these things.
I got it almost finished and ready to close the lid, but I seem to have one small problem.
Mine seems to trace really slowly, like one line per second or so. (Actually looking at it now, it traces the whole graph in 2 seconds or so).
I'll have to take another look tonight (I fired it up for the first time at 3am in the morning, so no time to troubleshoot yet), so can anyone suggest where I should start looking?
In all probability, I have a resistor with a wrong value installed or something silly, but if someone can tell me a likely place to poke first, that would be appreciated.
EDIT: Well, I tested all the diodes and resistors, they all seem ok and in the right spots. I'll have to check the capacitors next, then I guess go buy a new CD4024 and TL074 to see if one of those is busted somehow.
Still working on it. Had to remake all the wiring. Not yet a complete success. More on this later...
Well, I found my problem... 100nF is not interchangeable with 1nF... haha.
Turns out I had two capacitors swapped around... durrrr.
That'll teach me to construct things after midnight (no it won't
).
I also found that using shielded wire for the connections from the PCB to the front panel would distort the waveform on the scope if I had the shields grounded.
Leaving the shields completely disconnected results in a clean trace, so that's how they'll stay.
Once I have the enclosure finished, I'll post some pics (and a link to the youtube video I'm making).
I also found that using shielded wire for the connections from the PCB to the front panel would distort the waveform on the scope if I had the shields grounded. Leaving the shields completely disconnected results in a clean trace, so that's how they'll stay.
I already found what you say about grounded shielded wires but didn't try to leave the shields unconnected. Thanks for the tip!
Finally sorted it out. Everything showed right when I added ferrite beads in the Base/Gate wires, as close as possible to the transistors.
Changed the 3P4T for a 3P7T for yet more versatility.
Works like a charm!
Here's my attempt at an enclosure... It turned out ok, and works well which is a plus.
Too bad the available enclosures weren't just a few centimeters larger so I could have a little more elbow room inside to route wires more neatly..
Nice one!
I forgot to mention something:
If you see my project, two posts back, you'll notice that there are three switches instead of the two indicated by the kit itself. I have implemented a third switch for 4 / 8 curves as per the second page in the original ELEKTOR article linked early in this thread. This also limits the amount of current the DUT is going to manage, just in case... I had to cut the track going to pin 9 of 4024 and install a couple of wires to a SPST:
- Closed: 8 Curves
- Open: 4 curves
See the cited article for more details.
Ah cool. I'll check it out.
Nice, characterizing up to 4 transistors at once is pretty sweet.
I thought about making an A/B adapter for mine, but I got a Tek 577-177 for cheap to restore, so all my effort is probably going to go into that now...
Quoting myself
Finally sorted it out. Everything showed right when I added ferrite beads in the Base/Gate wires, as close as possible to the transistors.
Not really!... the graph was still somewhat fuzzy if you look closely. I had to put ferrite beads in ALL the C-B-E wiring and switching inside the box. Now the trace is
much clear. I can even use my old Tek TDS-320 crappy XY mode with a bit of success. Still much better in my old trusty analog HAMEG HM 203-6. Now the traces are crisp and clear.
Just adding these here for interest. I got a pair in to look at and they seem to be as claimed a ceramic body as they are heavy for size. One of them has an issue with a malformed contact on one position that I am not sure is correctable. Overall based on first impressions looks good but QC isn't. I will do some testing with repeated insertions on the damaged one and see how the plating is holding up.
Just an 'option' but maybe Zif sockets are better
eBay auction: #254121206657
I've been looking at these myself, I'm thinking of grabbing a couple to use in a custom jig for my Tek 277-177.
Just another project in the pile of projects.
I've got a range of curve tracers and testers ... an octopus curve tracer, and it-3132, thaikits curve tracer (jojo2520), tek 576
Here are my thoughts....
Octopus V/I tracer - cool device look it up - very cheap to build and works good but has lotta limitations but good on the cheap, I think thaikits might sell one, but an easy diy project.
Low voltage low current tracers - ie eBay kits / peak etc ... limited voltage and current makes it useful for some applications but are very limited in the voltage and current ranges they can test. might be the solution for you but know the limits...many hook to scope XY
older Commercial / KIT XY scope units - units like BK precision, leader, heathkit, that hook up to scopes. these typically have wider voltage and current rages but, are time-consuming to set up to get accurate readings. Very functional and a cheap alternative to a full blown tracer, though, you can damage the units or the device under test more easily than on a dedicated unit by setting it up wrong. great for the serious hobbyist.
Professional stand alone tracers - in this i include the older Tektronix 575,576,577,370,371, and its more modern counterparts made by Keithly and the like. typical advantages here are high voltages, high currents, and more limits and safety features over the scope XY counterparts., as well as ease of use, and speed of use. defiantly the way to go if you're using it professionally for matching components, or characterizing production components, qa testing and so on.
there are exceptions to the above, so basically i am saying that you should know the voltage and current limits of what you are looking at and be aware of what safety features it has or doesn't have.
This thread has been asleep for a long time, but I just built one of these Thai kit curve tracers. The main board and power supply board were pre-assembled. It is mounted in a metal case with short shielded cabling from board to panel mounted BNC connectors. The traces don't look right to me -- they have longer curves to peak current than I am used to seeing. I have tried short test leads to the transistors with no change to the curves. Does this look right to you? If not any ideas about how to get that shoulder more distinct?