This video has been supported by jlc PCB.
Hey guys, I once made a lighthearted joke at the expense of Germany's national metrology
Institute, the PTB I should have known better Germans being as humor intolerant as they
are. Well, no, it's retribution time. I got this menacingly looking email from them,
suggesting that I should make amends by helping them dispose
of a particularly big and heavy piece of scrap. I was supposed to come and pick that up in
paradise itself, located just outside of Brunswick in northern Germany, the physikalische technischer
bundesanstalt is an interdisciplinary precision city. Unfortunately, I didn't get the filming
permissions inside yet, but I did get something at least as nice. I adopted to this old broken
fluke 5708 MultiFunction calibrator myself, who knows maybe we can even bring it back
to life, that would be a big deal. I'm told that this particular unit with parts from
1987 must be one of the earliest 5700 out there.
It's series being launched officially in 1988. By now there are countless small hardware
revisions and two major ones which have been released as separate products. First came
the 5728 with some my newly improved specifications, but a nearly indistinguishable look. That
has only just been superseded recently, by the 5738. What gets me all excited about the
old chanca in front of me is that the analog fundamentals that all of these calibrators
are based on every maintained nearly unchanged over the last 30 years. So I like to think
we are not that far away anymore from the top of the electro metrology food chain. I
think the most important property of these machines is that they can be calibrated with
only three external known References 10 kilo ohm 10 volt and one ohm. When given some alone
time with these external standards, the calibrator will soak up some of their precision, only
to redistribute it by generating whatever is needed to calibrate other devices, AC DC
milli ampere, kilo volt and a wide range of resistances
sound good. Let's see what we can do. This thing consists of more than 14 modules
and hundreds of obsolete or special custom components. To make matters worse, flukes
official product support lifecycle for the 5723, which I could have gotten a few astronomically
priced replacement parts has just ended a few months ago. So as much as I would like
to perform an initial smoke test, I think it would be wise to fix as much as possible
first. I'm not the first one with that idea, it seems on this power supply filter barge
somebody has decided to replace some of the aluminium electrolytic caps, as well as a
small section of PCB material and traces. As a rule of thumb, I wouldn't recommend administering
routine replacements to small patches of circuit boards.
Even though this 1987 four layer laminate being bleeding edge technology back then isn't
a particularly good quality. It gets kind of soft at high temperatures, likes to laminate
and lose pets. So I'm going to remove the electrolytic caps as gently as possible, all
of them this time by dual wielding soldering irons. Generally whenever dealing with very
valuable boards, I'm looking for ways to make the soldering safer. One great way to do that,
whenever possible is to dismantle components in situ, so that afterwards you can attack
the remaining leads one by one. I like the story that that patched hole in the bar details.
It confirms my deep rooted suspicions against carbon resistors.
They drift like crazy, usually upwards, but they can still damage other components in
their facilities. So at least the carbon composition power resistance will have to leave as well.
They may have a low inductance which gives them advantages when dissipating pulsed power.
But you no such thing in my precision calibrator feels good man.
Although the entire calibrator is full of these suspicious parts, so it'll be a long
and expensive journey. But hey, it'll be worth it 10 times over.
I couldn't have let the thing operate over longer periods or even 24 seven with questionable
components still cooking inside. The only way to do that to worry free is metal resistors
and newer high temperature long life caps everywhere. Oh and perfectly bend leads brought
to you by the one and only Beagle later of course. Check out my authentic line of bigger
lailah merge down below the video. Luckily, there's a service manual and schematics
available for this product. For the fluke, I mean, not for the bigger labor. That way
I know what goes where. And even more importantly, what it does there turns out to this filter
assembly as to more jobs in addition to rectification and filtering of the transformer voltages,
it has an adjustable pre regulator for the plus minus high voltage lines so that the
precision stuff when putting out one volt, for example, doesn't have to drop the remaining
599 or whatever. And then there is also overcurrent protection. There is precise current regulation
on the other side of the instrument, but I think it's good to have another safety layer
here in case something goes wrong on the way there.
These are the off into low settings of the negative high voltage supply rail, they are
working as expected. This is the high setting of the positive high voltage rail which is
on permanently for some reason. And this is the complete absence of overcurrent protection.
I mean sufficiently strong power supplies can sometimes vaporize their problems before
vaporizing themselves, but I don't think it's the way it's meant to be. In this case.
I found that two transistors were to blame for these problems, who in their old ages
had decided to become diodes. Instead, this SPS one in combination with a 5.1 ohm resistor
is the current limiting one. And it happens to be a specially selected high gain specimen
for which I couldn't find a replacement part. I ended up just increasing the 5.1 Ohm parent
shunt to seven, but to watch out for problems that that may or may not cause down the line.
But for now I got both of the high voltage selection and the overcurrent protection
working and marking every new component with a white
dot four and a half to recap this thing again in 30 or 40 years. It's not just a phase Mom,
I want to be a voltage not forever. That module supplies power only to the so called the guarded
analog section. That's like the entire 75% portion on the left.
The digital section consists of those two boards on the right
and most of the front panel. No idea what kind of front penalty this thing
has suffered, but it's not good. See those loose filament wires dangling around in there.
Those are cathodes, which emit electrons much more easily when they're heated to a few hundred
degrees C. They grow longer when they're heated, so they have to be kept under tension adjust
until they're heated too far in the filament spray tank. Obviously, I'm not going to demonstrate
that these displays are way too awesome. replacement parts with the correct pin outs are unavailable
or very expensive. Of course, I found one with the correct with at least but it wants
to talk USB so not exactly a drop in replacement. Let's see if we can find a culprit.
The digital power supply which is responsible for the vacuum fluorescent display has these
conveniently labeled test points like every other module in the instrument 75 for Wii
U r volt unregulated is reading a happy 92 VDC 75 v however, which is supposed to be
the regulated version is reading for not really sure how to connect that to the busted display.
But I like to imagine that the dead power transistor I found here just to try to do
as much damage as possible on its way out. And replacing that actually gave me back my
regulated 75 volt rail. And to not sure if one can tell in the video, but that in turn
gave me back both displays. The Data Matrix one with multiple broken filaments is so dim
it's hardly worth mentioning, but hey better than nothing. It's a miracle that this is
working at all and not just giving me magic smoke from the loose filaments making contact
with the grids or the anodes. There is a bit more to do and to see in the front panel area.
The keypad for example, is working sporadically at best. To prevent the high voltage business
on the display board from scaring the volatile ppms in
the connector area. There is a lot of shielding in place, both behind and
in front of the VFD This is a somewhat transparent but conductively coated film. It's not exactly
pretty but its job is to be invisible so that shouldn't matter. Same goes for yours truly.
These other lower thermal EMF matched material connectors, copper binding posts washers,
nuts and Cable lugs crimped onto a silver plated copper cable by the looks of it.
I think that SM a connector is the only gold plated thing in the entire
instrument or has been for the last 30 years. It is used only for the high bandwidth option
which I don't have yet. This is the over a 30 year old keyboard. conductive silicone
membrane keys are pressed onto this thick carbon footprints to short them out. This
technology predates me and I've never met it anywhere else before. But as far as I can
tell. Now, the black tarnish on the carbon path is what's preventing good operation,
and a gentle dry brushy brushy is probably the right method to restore them for another
30 years of service. What went through my head after doing this was something like Lou
it's not clean enough. So I grabbed the IPA and started wiping away.
Oops. I even tried to cover up this colossal screw
up by quickly painting them back on with conductive silver paint. But alas, my forgery left some
to be desired. Luckily, this video is sponsored jlcpcb helped
me out of this embarrassing situation quickly, and I think they even made my calibrator a
little bit better in the process, I recreated the service manual schematics
and key cat. Then I took some caliper measurements of the original boards to digitize that as
well. Because this board is outside of multiple enclosure layers and only connected to digital
circuitry. Anyway, I didn't want to copy flukes four layer stack up with
just large ground planes on the outer layers. I did
however, go for gold plating, which will hopefully make my keypads live
a lot longer than that IPA soluble carbon powder stuff. I should have suffered more
for my mistake. But jlcpcb took the catharsis away from me by delivering this beautiful
replacement part quickly and cheaply. A few friends actually expressed doubts about the
suitability of electrodes gold plating like this. For membrane keypad purposes, the gold
layer might just be too thin and we are off quickly. But I don't quite believe it. No
doubt about it. If we were talking about edge connectors, like a PCI to these soft little
rubber pads, I don't know, I guess I'll find out in a few months and so will you because
if I need another board, there'll be another sponsored video opportunity. There is another
important feature on that part, which I've just transplanted. It's one of if not the
best feeling rotary encoder I've ever laid hands on. And it's completely were free. They
made that discretely from ball bearing, laser cut steel sheet metal magnets and photo electric
fork sensors, perfect for dialing in TPMS. That pretty much gives us a usable front panel.
If the user has photomultiplier eyes or something. Oh well, I'll continue to try and find a suitable
replacement display. But this machine is the most powerful when controlled remotely via
GPI be anyway, so I'm not in a hurry.
During booting it has already executed and passed to this brief self test,
which involves mostly the digital circuitry. And that's a nice start.
But there is another built in self diagnostics program
which is a bit more thorough. We are going to consult that when we are through
with all the arcane analog modules to see if any survived my interference. Let's fast
forward through the final few mundane tasks to get to the good stuff quickly.
I started to print some handles because the thing only came with two I pulled to the CPU
board out of the instrument and the notoriously forgetful eeproms out of that, being extra
careful not to bend any leads. I dismantled to the entire instrument front to rear to
find even the last suspicious electrolytic cap or carbon composition resistor. I erased
to the handles and cured the promise in UV light. I
took the disassembly even further. The too powerful fence and the absence of an air filter
has accumulated quite a bit of dust in the chassis. A vet revealed even more of the marvels
that were hidden in this beauty like the potter to enter new metal shield it means filter
and transformer as well as the Bowden pulley main switch and the fiber optical isolation
of the analog part. I wonder what's in that mains filter?
Oh no, really, you've got to be joking. I gave the eeproms some new extra opaque labels
that are reprogrammed and re installed to them.
The original fans sounded worn out. So they were replaced with some good quality
similar airflow once. Oh yeah, that's better.
That's something oily going on in the relay area over here. Not sure if somebody tried
to use contact cleaner, or if that's just leaking relay fluid. Either way that's got
to go on just like every last little bit of dust.
Selecting the right filter material is a bit of a balancing
act. The higher the density, the more effective
a filter can be. But if it gets too dense, it can impact airflow and ultimately, the
device's thermals. I went with this stuff for the time being no idea of good but probably
better than nothing. And finally, while we are here, check out this reo connector situation
where one could connect a special fluke power amplifier to source higher currents under
the influence of the calibrator. They went with a thinnest wire as possible for the sake
of low capacitance. But Gee, does that look frighteningly fragile? Hey, don't look at
them too hard. Okay, I think that's a wrap. At least in terms of the easy stuff. Now we
are going deep, I will continue to hunt down all the parts with a questionable reputation.
aluminio meant tantalum electrolytic caps and carbon composition resistors. But the
replacement process pretty much the same as before, it's not going to be the videos main
focus anymore. Instead, in the following section, I'd like to focus on what's going on where
and to why. Because as I said at the beginning, the analog fundamentals are as relevant as
always, this is a 17 the regulator and guard crossing assembly. Can you guess which two
main functions this thing has? This board is fed by the filter assembly, which we've
already finished, it takes some of those rectified and filtered voltages and regulates them down
to no less than 13 different DC voltage rails to be
used by other modules down the line. All low noise linear Of course, either Beaufighter
Zener diodes or integrated voltage regulators like 7805. That's what it needs that huge
air duct for to keep everybody nice and comfortable. This module also contains the one and only
in guard processor. It is a special resident of the analog ivory tower, because it's the
only one with the ability to talk to the digital mainboard.
And that way to the poisonous outside world. I wanted in God the process of my own to make
uncomfortable phone calls for me. Luckily, this guy only talks via duplex fiberlink.
So risk of noise contamination is fairly low. The in guard process our controls all of the
analog modules based on what the main process are commands. And so far, it's been doing
a great job applause. Next in line would be a 16 the power amplifier. But for a coherent
narrative, I'm going to mix up the order a little bit. So next up is going to be a 11
the D AC have the heart and soul of this whole operation.
Look at its shiny armor, one can tell right away that that's a big
deal inside or it's compensating for something like small men in big cars.
Nope, that's legit All right. It has sub cans after all, and two large hermetically sealed
axial components. So the purpose of this digital to analog converter is to generate extremely
precise and stable DC voltages between zero and 22 volt. He does that like an Arduino
word with pulse width modulation. Sounds appalling, doesn't it? But with a few little tricks,
it's actually fantastic in many ways. First, it needs the absolute best voltage reference
possible. In this case and organized SSE a 263 basic module is used the ultra stable
voltage that this thing generates becomes the high level of our P w m wave from this
in other words, to generate arbitrary DC voltages lower than that it only needs to adjust to
the duty cycle and low pass filter the output.
digital electronics are very good at turning stuff on and off quickly. The Arduino comparison
wasn't even that absurd, such as super simple microcontroller along with a standard
issue of crystal oscillator can get you in the hundreds of picoseconds
total jitter range of which only a portion called deterministic jitter will have an impact
on your filter output voltage. do the maths a few Pico seconds of uncertainty on top of
millisecond pulse width with which eight 11 is working equal to spectacular deck performance.
That's all I know. There's also a special a custom fluke 22 bit ADC in here, which is
a very important actor and the calibration of this calibrator. It's used as a null detector
to compare and adjust itself to the previously mentioned three external standards. This precisely
temperature controlled hybrid module is the reference buffer into the axial hermetic capacitors
are a part of the PW m filter. Next in line is the ohms main assembly, which is a lot
more simple than it looks. It uses these three fluke branded hermetic resistor networks,
which contain resistance values between 10 ohms, and nine mega ohm, as well as an additional
90 mega ohm fixed resistor made by kaduk, which is synthesizers close approximations
to the resistance values that the user has programmed using that battalion of relays,
ie 10 ohms 100 Ohm one, K, ohms, and so forth. What's nice about this topology is that it
allows for for wire measurement on every value except for the 90 mega ohm where it doesn't
matter anyway. The next one belongs to the resistance system as well as the ohms Keller
assembly. This one adds a one Ohm made from four four Ohm wire runs in parallel, or 1.9
ohms made from two wire rounds, and zero ohms, which is probably just a PCB trace. It also
has a battalion of these obsolete yellow relays for various configuration purposes. And a
blatant isolated DC to DC converter right there in the middle
of that seems a bit out of place in the middle of analog paradise,
doesn't it? Well, I guess it exists for a good reason. It's part of a circuit that eliminates
internal voltage drop whenever calibrating an ohm meter that can only do two wire measurement.
Next up a battalion of obsolete yellow relays for a change. I heard that all of these can
start to fail quietly, lose a few ppm here and there, but nothing drastic enough to be
noticed by the self diagnostics program. Oh, boy, that sounds like a good time, troubleshooting
something like that. Because these relays are not available reliably anymore, I thought
it made sense to pick them up cheap, just to be prepared for the inevitable.
This is the switch matrix, it connects to the output terminals to the
various functions, or the various functions to one another for self testing and calibration
purposes. There's also a times 130 gain amplifier on here, which seems to be missing one of
these covers, it is used to detect and adjust to zero of most DC voltage ranges. Next, we
got a low current and high resolution oscillator assembly.
Those two functions have nothing to do with each other. I guess they were just
small enough to fit on the same board. Why do we need another current source anyway?
Couldn't we just combine the voltage output deck with a resistance synthesizer? Well,
practically Yeah, sure. But imagine your Big Short industrial customers asking
how accurate is your current calibrator? And you having no better answer than a head
scratching? Hmm, depends.
Nah, this thing needs a proper transconductance amplifier that within reasonable limits doesn't
care about the meter impedance. It takes some extra ordinary care to achieve this kind of
performance in the exclusive realm of the nano amps signal guarding for example, and
the entire transconductance amplifier as well as the various current shunt resistors are
kept in another one of these precisely controlled hybrid ovens. I have yet to see something
like that in a keathley. The high resolution oscillator on the other side of the board
generates a variable frequency squarewave 10 hertz to 1.2 megahertz, which is then reused
most importantly by a 13 the oscillator output assembly where it's used as the AC output
frequency. This module is based on a quadrature RC oscillator with relay selectable feedback
capacitors as sort of chorus frequency ranges, and custom made high precision digital potentiometers
as frequency fine adjustments within those ranges.
These are the selectable film caps right next to the precision Digi pots in the
shape of ceramic hybrid modules. This module is almost only concerned with the frequency
of its output sine wave. It detects the zero crossings and uses a phase locked loop to
synchronize with the high resolution square wave from the previous module. That's nice
and all don't get me wrong, but when calibrating A True RMS voltmeter of the amplitude is usually
much more important than the frequency which is why there isn't an Our module is dedicated
to just that the oscillator control board. Now how do you even measure the root mean
square value of an AC waveform with world class metrology precision, it's defined as
the equivalent DC value that would cause the same amount of heat in a resistor. Ordinary
solutions are either discrete or circuits that literally Root Mean and square while
hoping that none of their components brings too much error to the table. Or in more modern
icees, it's probably just a digitizer that hopes that nothing important happens in between
samples. The fluke way that is used on this assembly is a bit more direct than that. It's
all based on these super special fluke thermal RMS sensors, which also happened to be used
in fluke seven nine to a AC DC thermal transfer standards, as well as they are 5798 ac measurement
standards. More details and detailed on photos about all of these arcane devices can be found
on X f.com. I'm just going to borrow this one so that I don't have to crack open my
thermal RMS converter. These are basically just two matched pairs of resistor and transistor,
one pair is heated by the signal to be measured, which influences its transistor current. The
other pair is heated by an error amplifier, exactly to the point where there is no more
temperature difference between the two. Well, and then we are already the output of that
error amplifier happens to be the equivalent DC value that causes the same amount of heat
as the signal to be measured. Man I sure hope these guys won't break before I have a working
semiconductor wire bonding and laser trimming setup.
Hey, everybody seemed to like the longer videos no complaining now.
This is the power amplifier which I didn't want to cover earlier.
Because we didn't know about any of its input signals yet.
This module takes care of the voltage ranges that the dc dec
and the oscillator system can deliver directly to 120 volt AC or DC with some D rating at
the higher frequencies. For various reasons
this thing has been designed to be a three stage amplifier of the input stage is another
one of those temperature controlled hybrids for environmentally independent stability
and low noise. The mid stage at most of the gain if needed, its bandwidth is configured
by relays depending on what frequency range the calibrator is in. The output stage including
those heatsink power transistors has no gain at all. It's just a follower but with the
highest current capability. advantage of being that block is allowed to change temperature
because that doesn't add too much of an error. semiconductors at elevated temperatures can
actually degrade over time. So heavily heatsink power transistors are always worth checking.
This patient has flatlined. Wow, we are nearly done. second to the last is this innocently
looking high voltage assembly which is to blame. However, for a disproportionately large
portion of worries. I didn't even see this myself, hidden from xf.com spotted this crack
in the previously hermetic resistor network or know all the ppm czar going to fall out.
While the resistance values are still what they should be, according to the schematics,
so I guess it's only the coverglass and two that appears to be bonded to the alumina ceramic
substrate only by epoxy. That's a beginner hermetic seal, in my opinion, only real glass
to metal seals count. Maybe I can borrow the ultrasonic soldering method from Applied Science
and to just put a big metallic snail trail all over that thing. But since that part is
connected to such a big heatsink, I'd be a bit worried about thermal expansion coefficients
of all the different materials involved in such a scenario. Oh, well, that network is
only involved in the 2.2 MPR current range anyway,
so it could be worse. No clue what 2.2 MPR circuitry
is doing on the high voltage assembly and no clue either why they're placed to the power
transistors that are responsible for heating two of these hybrid ovens so closely together
that they have created a hotspot. It almost looks like something broke here, but I can't
find evidence for that. The heater resistances and the transistors are both fine.
But I think I'm still going to replace the ones that have cooked to the worst.
With some special extra high power bjts. I think most of the hybrid temperature controllers
in here work the same. They are configured for either 50 or 62 degrees C so that ambient
temperature never gets in their way. This one happens to be the high voltage divider
for feedback purposes. The heater film resistor covers the entire ceramic substrate,
so that there are no thermal gradients on it
or on whatever is bonded to it. The temperature is sensed by a thermistor and the heater power
is controlled proportionally by an op amp.
And to the last one Finally, this one is the only module that needs screws to be undone
before it can be pulled out. for good reason it carries a heavy new metal shielded backpack.
That is what actually generates the high voltage up to 1100 volts, a good old linear transformer,
some rectifier diodes and some relays to invert the polarity or to rewire the transformer
windings for a different function altogether. What's ostentatiously missing though is the
usual double half h bridge with which transformers are usually driven, as well as some kind of
high voltage regulation. I would have expected an entire array of high voltage transistors
cascaded to withstand the full force of 1.1 kilo volt. Well, the 5708 elegantly reuses
its existing modules, the oscillator system and the power amplifier work together to drive
the transformer. The Ultra precise high voltage control loop is closed by the organized 100
to one divide that we've seen on the last module that works both for DC and AC just
without the rectifier then he has the mind blowing explanation for the presence of the
2.2 M power range on the high voltage assembly. All those relays can actually reconfigure
the transformer windings to be either step up for high voltage generation or a step down
for high current generation. Absolutely brilliant. So anyway, how do you like the 3d printed
handles? Wow, what a treasure chest this thing is, I think my lucky black gloves have touched
everything now. And everything still seems to work somehow.
In the next video, we are going to have a special guest who will tell us how well we
are really going to counter those pbms to generally speaking, what's next? What am I
going to do with that beast? Well, I've got a number of multimeters here that haven't
been calibrated in years or even decades in some cases, when they were last adjusted and
pair were still defined as Newton's of force between two conductors, but to bring them
up to speed and tell them about cesium atoms and cable balances. The one multimeter that
I've been lusting after for years still eludes me. So I would also like to use this calibrator
to play around with ADCs myself and see if I can DIY something better.
flukes advertising material gives me the impression that a short warmup time is sufficient for
the 5700 a. But it's obvious that this machine would love to be turned on 24 seven for maximum
stability. To keep all those transistors biased and all those ovens at the temperature. This
unit draws 160 water permanently. Some of that energy is transformed into noise or air
movement in general, I'm pretty sure that 95% of that energy is turned into heat, which
was unpleasantly noticeable in the recent summer months. Unfortunately, German electricity
is not that cheap. And I'm not sure if I'm willing to invest that much in pbms yet, at
the moment, we are freshly repaired and out of kill anyway, so nothing really matters.
And I have a bit more time to think about the energy side of things.
Of course, I'm already looking hard for the holy trinity of external
standards with which to perform the artifact calibration. I think I have 10 volts and 10
kilo ohm within reach already, but no clue how to find a world class or one ohm resistor.
And if anybody happens to find one of these guys, ideally one or more 10 K, you know who
to call rewash to be negotiated. Here's a quick fast forward through that 10
minute self diagnostics program which uses the loudly audible rely matrix to connect
hundreds of different test points from all over the modules to the ADC. I was curious
and captured a little bit of that with an oscilloscope in the middle of the repair.
Didn't have its new fans yet at this point. What matters the most is that it doesn't find
anything to complain about. I'm thrilled to report that at the end of
this makeover. It didn't.
Okay, that was a massively summarized version of this super massive repair and rejuvenation
project, the longest video I've made in years and dedup being insultingly superficial somehow.
No kidding, I could probably make a half hour video just about the LT flow voltage references
in this thing. Who knows, maybe I will.
As soon as my dedicated YouTube voice has recovered, you don't have to wait for that
though. The 570 pages worth of service manual for the series, countless lovingly handmade
articles on X f.com. Enter the jungle of arcane knowledge. That is the Eevee blog forum
are out there waiting. Knock yourselves out.
Thanks for watching and see you soon.
