ฝัง
- เผยแพร่เมื่อ 10 ก.พ. 2023
- In this episode Shahriar takes a look at a non-functional Fluke 5450A resistance calibrator box. The unit is equipped with a wide range of resistors ranging from 1-Ohm to 100-MOhm some of which are 0.005% tolerance. The instrument powers on and shows a single LED on the front panel without any other activity.
Upon further inspection it becomes clear that there are no other activities from the main CPU despite the clock signal being present. A replacement CPU returns the unit to normal front-panel operation. The analog board also has two blown fuses which may be the result of a in-rush event. After fuse replacement the analog relay arrays also function. The unit is measured against a Keysight 3458A 8.5-digit multimeter and shows remarkable accuracy despite its age.
www.TheSignalPath.com
/ thesignalpath
www.Patreon.com/TheSignalPath - วิทยาศาสตร์และเทคโนโลยี
Love xDevs. Such a contribution to modern EE enthusiasts. Also, Ilia’s videos are like yours-a masterclass in top-notch engineering. Thanks to you both for sharing so much knowledge.
After last Kingpin lab tour, featured on Gamers Nexus channel, I tried to track down Illya "TiN" channel, with no results. Thanks for pointing me to xDevs!
@@HorochovPL Similar story here. I was so glad to find he's doing well and all the interesting things he's been up to.
One repairer I watch (I can't remember exactly who, but it may be Marco Reps) always takes out the eeproms and backs them up which seems like a very sound idea.
Yes, I should do that.
Feedback loop does it
Nice! IIRC my unit was a bit out compared with my 6.5 digit meter at the time. Might have to reevaluate it.
Shariar, it looks almost too easy. It's so refreshing to see this, it really makes us plaebeians think we could have it easy, as well.
Thank you, repair videos are the best.
The guy has a Ph.D in Electrical Engineering, works at Bell Labs, with decades of experience, I assume it should be easy for him unlike us plebeians.
Some of the PPMs escaped. Don’t tell Marco Reps.
Hi, great catch again, whenever I've bought older measuring equipment on Ebay it's been a service massacre except for the HP 427A and 209A - it suffered in shipping and the slot connector for the power supply board cracked. This catch belongs to the WOW category 😁, I would be quite interested in the process of damage to the circuit for the relay power supply. My thinking is that the cpu got damaged and sent incorrect data to control the relay and it overloaded the power circuit, I would definitely do a deep relay check.
Thank you for your wonderful videos, you taught me a lot not only about repairs, but also about equipment that I did not know.
Nice day 🙂 Tom
The missing chip is not used in the 5450A, The same CPU board was used in the 8520A and 8520A-PRT. Before you go and calibrate the unit as it has been sitting a while you should exercise the relays. Also when you do the calibration you will want to use the rear terminals.
If he only uses the front terminals, wouldn't it make sense to calibrate to those?
@@jaro6985 The front terminals are Banana jacks while the back terminals are Copper Binding posts. The binding posts allow for a much lower EMF due to thermals where Banana plugs are great for normal connections but if he wants to make sub ppm measurements on the 5450A, The copper binding posts will make a much better connection.
Leave the 3458A powered 24/7 (30w approx.) if you aren't already, and perform an ACAL after a few days, then weeks, then months whilst doing some tests in between. It worked wonders for mine.
I love when your channel shows up on my notifications. Thanks for the awesome content!
Always entertaining and informative. Thanks for a great channel!
by digital ! the sexiest machines ever made! thanks 4 educating me, now I can fix the spaceshuttle! cheers!
Mesmerizing relay action. I learned a lot in this short video!
Regarding uncovered EPROMs, I've been doing a bit of an experiment over the past year. I left a couple of them out to see what effect ambient and direct sunlight would have on the contents.
For the EPROM in direct sunlight (sitting on the window sill facing west here in the UK) it took approximately 1 month before I started noticing that some bits were erased. In the end it was about 3 months to fully erase back to all 1s.
The other EPROM is sitting on my desk exposed only to ambient light. The desk is about a metre from the west facing window, so it still receives sunlight, but generally not directly - just whatever comes in through the window as well as lamps, my monitor, etc. Ambient. After what is now about 10 months, this EPROM is still verifying 100% against the initial contents that I dumped.
I don't have a production quality programmer with which to do verification, but so far it doesn't seem like ambient light in a room is sufficient to cause any immediate harm, and I imagine some EPROMs tucked away inside a case with even less ambient light available are even less susceptible to erasure.
Adrian's Digital Basement has also done an interesting experiment where he left an EPROM with the window covered only by a piece of Post-It note in an eraser overnight, and it also verified correctly the following day.
So I kind of have a theory that the window doesn't need to be made 100% opaque, and that the top surface of an EPROM is simply the most convenient place to put a sticker, even if it is a simple paper sticker. I guess paper is sufficient to block any light with sufficient properties to cause erasure.
I'll keep doing my experiment and see how long it takes for ambient light to have an effect on my EPROM, but it doesn't seem to be the "certain/imminent death" that some people make it out to be..?
Erasure needs shortwave UV, which is blocked by glass, so any indoor location is probably fine.
I did a video about this about 5 years ago, from memory it only took 1 week to corrupt the EPROM data having it sitting on a window sill… I am in NZ where UV is more of an issue.
Interesting. Thank you.
@@mikeselectricstuff I'll have to try putting one outdoors and see what kind of difference that makes.
@@TheDefpom It's probably less of an issue now since task lighting isn't florescent. We always used the al sticky backed covers that you could indent with ver no.
Great video.
The large chip used for IEEE-488 support is a MC68488, a purpose built PIO chip designed by Motorola for the 6800 series microprocessors. It handles most of the bus handshaking and data transfer operations. Interestingly, it cannot drive the bus directly instead requiring separate bus transmitter/receiver chips. The MC3447 was the chip originally used for this purpose.
Love the repair videos!
Very good video! Might be worth saving a copy of the eproms content just in case they ever become corrupted (assuming that they are socketed)
Nice!
We still have the 5450A in operation at Fluke. The dry current relais in it are a bit troublesome as they develop instability with time. Also the high ohm reed relais in "newer" units which are plastic encased become leaky.
For relay contacts that develop high closed resistance which happens when they got old and don't switch much current. Get a say 1000uf low ESR cap charge it up to 12V, put it across the contacts and close the relay, there will be a spark, which will clean the contacts and drop its closed resistance. Good. For another 20 years.
@@universeisundernoobligatio3283 This will work on normal relays but when you do this on types which have a coating to be switched without a "wetting current" the issue can even become worse, especially when you look onto it from a point of contact repeatability which is the major concern here less the absolute contact resistance.
And this is not wild guessing but actual experience as we tested this as a last resort measure before buying new specially manufactured relays.
@@benedienst
Interesting, my experience with this has been with relays switching into logic gates which draw no current therefore no spark. So for repeatable resistance the contact materials must not oxidize over years of use?
@@universeisundernoobligatio3283 with logic gates you are working into devices which accept quite a large error margin in their levels.
The relays of those calibrators are required to achieve contact repeatabilities in the single digit ppm range or below. Especially for the workstations we do the Fluke 57xx calibrators resistance verification by transfer method, the repeatability as to be better than 0.5ppm over many measurements and also perform good on drift over a 90 day period in a similar range.
The further you go beyond the decimal sign the more impact even the tiniest imperfections get.
This is also the reason why you often find more "aged" equiipment in primary laboratories instead of brand spanking new instruments.
It's all in regards of predictability and good behavior. only then you are able to squeeze out the number of digits you need to achieve good uncertainties for highj end equipment like the Fluke 57xx Series calibrators.
@@benedienst
Thanks for the info, yes its can be a black art to get that type of precision and stability . Lots of experience plus lots of trial and error.
I have many years experience doing strain gauge measurements for gas turbine R&D, designed and built many telemetry transmitters for rotating component strain gauge measurements.
Nice job.
Actually, it's all Butch's doing, but since Butch can't talk, his master has to do the talking 🤫
6:20 interesting choice to use flex PCB material to connect that regulator to the PCB. They must have decided the labor to use individual wires would be more than the cost of that purpose built flex circuit bit.
The flex ribbon was probably a standard part from the first time it was used and became popular.
Funny, one would expect such an ancient microprocessor with µm feature size to last forever.
Same relays as in my Fluke 5440B, btw
I did a couple of repair videos on one of these, mine had issues with all of the relays, I spent ages taking them apart and cleaning them.
You can tell if the relays are not perfect by switching values back and forth to see if the reading is exactly the same each time, I highly recommend you check for it.
Awesome. Good thing old micros do not contain any firmware. What is the temp. coefficient on those nice resistors?
I wonder if you can calibrate the old Fluke manually? I'm sure Fluke themselves would still cal it too, for a fortune of course.
Proper calibration of 5450A is rather tedious process, since it has very good uncertainties as calibrator, very similar to 1 year spec of calibrated 3458A. It's not as simple hook up calibrated DMM and record values, but whole setup with external sources, null-meters, etc. Today can be simplified a bit with bridges, but not much. Fluke goes into great details about it in 5450A manual too, which is excellent.
@xDevscom_EE interesting info Illya. Sounds like a proper project for your lab full of calibrators and met gear!
Fluke should do some lab and manufacturing video tours as marketing to engineers. Would be cool to see the who, where, and how behind world class metrology boxes. Not like it could be easily copied just by watching a video...
I think there is a 60Hz flicker on the display. I am not sure that is the way it was designed to perform, unless that only shows up in the video.
Calibrating calibrator using calibrator
Are those going to be mercury wetted relays or something? I would think anything other than that would cause the resistance to go increase with time/use through the relay itself.
Looks like it's using a 4-terminal measurement, so relay resistance shouldn't be an issue, just need to ensure no thermocouple effects
Interesting repair!
You did not mention the type of processor you had to replace. Z80?
The pinouts on the schematic at 9:28 correspond to a Z80.
@@KallePihlajasaari Thanks. I am NOT surprised... It was a very common processor at the time. I had one in my TRS 80 Computer ca 1982
The relative error seems to always be close to 1E-5 ~~
Perhaps the contracts if the relays add a bit of resistance
In a 4-wire measurements the contacts are ignored.
I wonder what keeps the relays contacts from getting crusty over time? Are they sealed with argon or neon and some kind of moisture absorber?
Yes, hermetically sealed.
How do they compensate for the contact resistance of the relay contacts?
It would've been good if we got a close-up of the micro, even though it's not going to be a Z80 - lol -. But then again, it's probably just some uncommon ancient micro that I'm not going to know anyway - lol -. At least we could've got a date code - lol -. A lot of lol's going on here - lol -.
Looks like Motorola 6848
The schematic pinouts correspond to a Z80.
9:08. What is the missing option that the unpopulated chip would be for? In other words, what does that chip do? Looks like another CPU by the size of it. Could you add it back?
There is a schematic fluke 434?
Hello how are you? Do you have the fluke 435 service diagram?
Fluke 233 red ligth you now..
I think I have some of those resistors and the parts they're made of. A friend of mine's brother used to work at the Fluke factory in Washington. Let me know if you're interested and I'll mail them to you. I think you'll find them very interesting!
first
Copper tape is the worst thing you can use! 🤦🏻♂️😝
Why?