No.118 - HP 3245A Universal Source Repair, Mods & Upgrades - Part 3
HTML-код
- Опубликовано: 6 ноя 2023
- ** PCBWAY For all your PCB Services at www.pcbway.com/?from=ianjohnston **
In this 3rd video, I wanted to add a couple of banana sockets to the front of my 3245A, and I got that done easily enough, but I then noticed a problem, could be a BIG problem!
#electronicscreators
Visit my website for lots more - www.ianjohnston.com
Help me make more of these videos:
Pledge using Patreon - / iansjohnston
Donate using PayPal - www.paypal.me/IanSJohnston
Follow me on Twitter - / iansjohnston Наука
Hi, its just the range. Look for the first zero. One measure is in 10V range the other in 1V range. No problem at all. Thanks
OMG……..how did I not notice that! I need a holiday!!!
@@IanScottJohnston .. provided you did not do an ACAL DCV before making measurements.. In the inserted video, reading on the 3458A had 5 zeroes, whereas on your paper note there were 6 zeroes after the dot. So you read +10ppm in the 1V range, +6,8ppm in 10V, and +0.68ppm on your paper
Haha I think everyone missed it. Excellent spot.
Awesome catch Sir !...cheers.
Hahaha, yes, that it is! Every voltnut staring at those last 3 digits totally ignoring the most significant digits. Ian just missed the best Aprils fool joke.
"curiouser and curiouser", said Alice!
I read the pinned comment before watching the tests at the end. Not sure I’d have noticed that for a long time.
Most likely it's thermal changes or capacitance introduced in cables. Banana sockets are still better than BNC for your day to day use.
Metrology rule number one - never ever use autorange :-). Aren't those cheap brass posts? Perhaps TBP-3 would be much nicer post in this mod...
Yes, I always always always use manual range……erm, except this time!
Posts were purchased as gold plated. I’ll look at your suggestion.
Data sheet, says gold plated contact material, but index not clear……hmmm!
cdn.shopify.com/s/files/1/0695/1347/8453/files/356048.pdf
@@IanScottJohnston These dual posts look like cheap brass with heavy layer of nickel under the gold plating. Real low-thermal EMF gold plating over bare copper is almost always dull matte. But it's a special process and uncommon (read = expensive).
@@IanScottJohnston It's pretty clear to me, brass post, with zinc terminal body. Fine for generic use, but nothing to do with low thermal. Because it's rare to find pure copper+no shielding metal+gold I steer away from any gold plated stuff. It's easier to clean copper every time I need nanovolts, than settle about crappy connectors. :)
@@xDevscom_EE agree, Pomona 3770-0 and 3770-1 would also be better if you're fussy about the last couple of digits.
Is it a loading issue? Might be interesting to have two meters hooked up to the outputs, one on BNC, the other on the 4mm jacks. Unplug one set and see if the other one changes and again when it is plugged back in. (ah, OK, just read the pinned comment)
Lol,🙄 Can't see the forest for the trees.. never happened to me (tongue in cheek). Makes one wonder what triggers the range change and why it was on the borderline between ranges? Maybe it was a high inducted 50Hz mains voltage either between the instruments or cable handlerling? As was pointed out below, especially for calibration functions removing the auto range and replacing it with a fixed range would remove that possible error. And take a holiday, at least a short one.
Some stray capacitance charging somewhere?
Hi Ian, here from Nam it is very very dry out here so static buildup is always a big issue here, could it possibly be a grounding issue? And my secobd question would be, has the sheeth end only been grounde on one side or both? It should only be on one side, you could swap the jack direction between the two instruments to check for a possible grounding issue. Third question, was that auto range? It could be a ranging issue if it was, could you set it to manual if it was not manual and do the same test? If this still persists could you swap the Jack's direction between the two devices? (top to bottom and bottom to top).
Yes, I had accidentally left the 3458A in auto ranging mode……watch the MSD change!
Hello Ian, Cable capacitance ? try to short the cable before plug in.
Yep, good idea.
looks like the range on 3458a changes. I would do that test on a fixed range not auto. but I wonder why 3458A cannot decide the range and changes each time?
Yes, I am soooo used to manual range that I never realized it was in Auto range. Caught me out once before……..I never learn LOL!
Have you tried using the rear terminals on the 3458?
Yes, same problem exists.
If the cables are not shielded, what stops them from picking up emf from the surroundings?
I have tried a mix of coax, shielded and non shielded cables. Same problem.
Hi Ian, could be some spurius signal detected from the cable ?
Could be, I’ll try isolating everything in my workshop.
Your new 4mm jacks, are you sure that they are low e.m.f. ? Why haven't you assembled two POMONA CuTe in first place? I would have used properly crimped Cu lugs on the back of them. Have you used the shield of the PTFE cable, or is this used on the BNC connector side? I would try to reverse the measurement cable, and calculate probable offsets from the + / - readings of the 3458A.
We just have a similar discussion on eevblog in the thread from RAX, about his 5440A repair. I've posted a method to identify e.m.f. offsets and external noise.
Erm, there is no problem with my 3458A, it’s just jumping between 1V and 10V ranges thus the extra digit appears and disappears…….LOL!
I guess I am so used to using the 3458A via GPIB.
The banana sockets I am using will do for now, the 3245A is quite a noisy source in the first place.
The ETFE cable I fitted is screened.
@@IanScottJohnston Well, basically the reading of the 3458A in its 1V and 10V range should read the same, within about 0.3ppm, but they deviate by 3.2 ppm. That's a lot of a difference.
That looks as if you haven't done an ACAL DCV before.
@@DrFrank-xj9bc I hadn’t done an ACAL DCV, wasn’t at that point yet. It’s been about a month since the last one as I haven’t been using it.
Sidenote, it’s due for its trip down to Keysight for Cal beginning of 2024, two years goes past so fast!
Could it be a hum issue causing an offset? Is there a way to specify the number of 50/60Hz cycles in a sample?
NPLC set to 100 during my test which is high and should mask and 50Hz issues.
The 3458A always samples a whole number of mains cycles as a matter of course. It literally times itself to the zero crossing of the mains it is being fed. Usual for measurements is to ask for measurement over 100 cycles of mains
1-3uV differences?
Not sure if you have to go 100% copper with tellurium / gold plating to eliminate that issue. I know it's extremely niche but would be lovely if one of the voltnut youtubers did a deep practical/experimental dive on thermal EMFs of different metal contacts. Different metals, different platings, different solders, crimps vs nuts vs soldered connections, time to equalise, repeatability of differences
Incidentally that's one reason to use 10V as a standard number of interest over 1V, reduces the need to care about this issue by a factor of 10. Yes it'd be nice to have tellurium copper everything but it's bloody expensive 🤣
does the length of time that you disconnect the input affect the reading??
Read the pinned post Fred……..
@@IanScottJohnston you mean I have too read stuff not just watch the pictures??? 😞
@@fredflintstone1 get a family member to help Fred…….:)
@@IanScottJohnston family what is that? if the wife counts as one then no chance :-(
Stop playing up with Ian Fred or we won't get your ball back the next time it roll sunder your bench. 😂😅 Woof
Seebeck effect perhaps?
The voltage pops back n forth randomly, sometimes doesn’t change so don’t think so.
Likely right, but i'd still point a hot air gun at the junctions, at a low setting, just to rule it out entirely. @@IanScottJohnston
@@chuckvanderbildt agreed, not ruling anything out at this point.
Lol ! so obvious when you look back ! but I bet your relieved cos if that was real it would have been a pig !....cheers.
Cheese!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
SQUEAKATY SQUEAKKKKKKKKKKKK!!!!!!!!!!!!!!!!!@@fredflintstone1
Can it be stray EMF
Could be!, I will try isolating and turning equipment etc off……albeit not much has changed recently.
Oh dear, looks like the 3458 has some odd math issue or something…
This is a hardware problem most likely.
@@IanScottJohnston have you tried with a different NLPC or averaging to see if it still shifts?
@@TheDefpom yes, makes no difference.
@@IanScottJohnston could be an input buffer or ADC issue then where the loading effects it, try a parallel resistance to see if adding a constant load changes the behaviour, did you try changing the input impedance setting? In fact, could it be automatically changing the input impedance based on range, and a bad reed switch is causing it to not always switch back?
Unlikely, impedance is switched only in 100/1000V range.
But is easy to check, just set manual range and countercheck again with fixedz on