The fast feedback mode on mine has never worked, so fA and tohm measurements are super slow, 10s of minutes sometimes. Things have to be shielded or people walking around will pin the needle.
I built my own fA parts tester before I got the Keithley, using an LMC6001 cmos opamp. It has less wiring and no switches so tends to be faster.
Maybe you guys didn't see this but why the (2) diodes in series (4 total) as shown here? Thanks.
Probably to keep the sensitivity low. Read some of the comments; a fluorescent lamp (probably electronic ballast, few kilohertz current fluctuations) reads as REAL hot.
What you usually want, is something to read a particular range of frequencies, usually in a known RF band, which means at least a grid-dip meter kind of gizmo. I'm partial to the use of AGC-type amplifiers, where one infers high RF strength from low amplifier bias.
So, what I'd find really useful, is a way to put an ADL5906 into my old grid dip meter...
but I suspect it's harder to do that, than it was to make a solid-state plugin to replace the tube for output-only GDO functions.
The components are separated by some kind of stand-off that mustn't leak a femptoamp. What material is that?
The Rat Shack binding posts were super leaky so I hacked a hole in the box and added a lexan sheet.
The terminal strip inside is a ceramic strip as used in old Tek scopes, but the critical node is a mid-air junction.
Looks like a surplus Tektronix ceramic terminal strip.
Cheers
Phil Hobbs
Yes. I got a box of unused ones at the Foothill Flea Market.
Ah, yes. Thanks Phil, for naming those. We have a drawer full of those back in the lab. Their purpose is obvious but I didn't know their origin. (We have a lot of old Tek repair parts tucked away, unless they were victims of another "5s+2 event".)
John, thanks for sharing. Your "integrator" is pretty leaky (Rf=510 R) :-) Is that Cf a 'virtual capacitor'? It seems to be hidden in the wiring photo.
The output caps, Are they special film that you chose to minimize Dielectric Absorption*? Or given that the node impedances are so low (220 R) any plastic is OK? (*Ref:
Phil, this is an excellent point, that I want to amplify. Even if accuracy is lacking, the precision and stability of a piece of equipment (no matter how old it is) is still valuable. Thanks! = RS
The opamp is a unity-gain follower. The 510r could be zero ohms, but I thought I might add an offset pot some day, which I didn't.
No, ordinary ceramics. We have a lot of RF in our place from Sutro Tower, and I wanted to keep that out. Opamps like to rectify RF.
The measurements are DC, slow, so DA is not a concern.
It worked terribly until I figured out that the RatShack nylon binding posts were super leaky, and added the lexan plate. That was messy.
I have a set of Pomona banana plugs with various resistors from 1K to
1T ohms. One of those plugs in as Z1 or Z2, and the DUT becomes the other. Then measure the two voltage drops.Another aluminum chassis is used as a shield, upside down, and can be bolted down. That avoids errors from local e-fields, like people walking and breathing in the room, or 60 Hz fields.
Keithley recommended that some of the resistors inside the 610 be replaced regularly. Mine might be 40 years old, and I doubt anyone did that.
You can buy a few high-ohm resistors and check the meter now and then. Mine seems fine.
There's probably a way to check it with capacitors too.
On a sunny day (Sun, 09 Jan 2022 10:25:32 +0100) it happened Jeroen Belleman snipped-for-privacy@nospam.please wrote in <sre9mc$prb$ snipped-for-privacy@gioia.aioe.org>:
Yes it is all not so different from the twisted wire oscillator, but mine is simpler:
When analyzing an oscillator, I always start by identifying the reactances between BE, EC and CB. (Or whatever the equivalent nodes are on the actual gain element.)The first pair should have the opposite sign from the last, i.e., if the BE and EC reactances are capacitive, the CB reactance should be inductive and we have a Colpitts. In the reverse case, we have a Hartley.
Any of the reactances could be a resonator of some sort, so what counts are the reactances at the intended oscillation frequency. I often see oscillators that use parasitic elements for one or more of these reactances. That seems to be the case in your example too. I usually try to avoid dependence on such elements when designing an oscillator.
The next step is to get the magnitudes of those reactances and calculate the loop gain. Of course, they aren't pure reactances, but I usually neglect that at first. Nine times out of ten, that turns out well enough.
Jeroen Belleman
On a sunny day (Sun, 09 Jan 2022 12:58:49 +0100) it happened Jeroen Belleman snipped-for-privacy@nospam.please wrote in <sreilq$mnp$ snipped-for-privacy@gioia.aioe.org>:
Sure, I am but a neural net and just blob these things down without much maaz. In the twisted wire oscillator case I started with a longer twisted pair resulting in a much lower frequency and then cut pieces of until I had the 2.4 GHz. Basically a wavelength related thing, twisted just to hold it together. On the PCB of that other thing you also see a feedback path. But I just think in [ parts of ] a wavelength. S parameters last time I used those was in school for the exams... One experiment is worth a thousand hours of theory, no spice in use here for this either. In the 'twisted' LOL case maybe a low-pass from the collector will also show Doppler difference frequency if that twisted pair picks up any reflections. On their PCB it looks like they somehow couple it to some bigger circular area on the other side of the PCB to get more radiation.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required