| Products > Test Equipment |
| Metrix MX56C / BK Precision 5390 Multimeter Teardown |
| << < (14/18) > >> |
| Vince:
... |
| retiredcaps:
--- Quote from: Vince on September 01, 2016, 07:23:43 pm ---The unlikely indeed happened ! It was the PSU's galvanometer reading garbage ! --- End quote --- Glad you got it working. Everyone on the thread contributed to make it working. |
| Assafl:
Got to love R2 in the power supply! |
| Vince:
--- Quote from: Assafl on September 04, 2016, 10:55:29 am ---Got to love R2 in the power supply! --- End quote --- Well yes, kinda stands out eh ?! I don't know why they needed to make that shunt wire sooooo long. Who knows. To help cooling/accuracy maybe.. though accuracy is hardly a concern in the design of this particular PSU for sure. So don't know. Anyway how, I have just validated my ordered at Farnell, to get the bits I need to finally put the Metrix back in one piece. However sadly, the order will be delayed because of the 28 pin LCD header connector, as it comes direct from the manufacturer's stock. They state a delay of 5 working days, so at best, if I am lucky, I will get the parts on Saturday morning. So in the meantime, I will work on trouble-shooting and hopefully fixing the PSU. Since this topic is about tearing down the Metrix, I guess the schematics I uploaded are in line with this topic and that deserve some comments as part of the tear down, no ? I see that dozen or so people have downloaded it ! So it must be of some interest to some people, more than the few of us who actually post in this topic anyway. So... what could we say about this board/schematics ? I spent a few hours reading the m again and again... Overall, it's more complex/convoluted that I thought it ought to be ! To start with, the representation of the contacts/tracks of the main rotary switch, is not so easy to understand... one can eventually figure out the broad lines, enough to get a sufficient understanding of where the signals come and go... most of them anyway, but saying that I understand easy and every little detail of it, would be rather presumptuous... Anyway. The main thing that I find interesting, is the overall architecture of the thing. You have got the big "DMM5" proprietary chip on the main board, and the little LCD board with a big chip on it, and a header connector linking the two. So, one would assume that it's all very classic an arrangement : the DMM 5 chip has got to be the main processor doing it all, and it just talks to the LCD board to control the display, and the chip on the LCD board is obviously the LCD driver. Right ? No, totally WRONG ! LOL Looking at the schematics, one quickly sees that the big DMM5 chip handles only the analog stuff, doing most of the magic. The main board actually is 100% analog. The LCD board is actually the main digital board, everything digital is all happening in there ! The LCD driver is actually not the driver, it's the main processor/controller, a modest 4 bit MCU from NEC, featuring 16KB of PROM (in my meter anyway, a 316 version, not the 308 / 8KB version pictured in the schematics), 512 bytes of RAM, and running at around 4MHz. So, it's completely upside down compared to most other meters I would guess. That little LCD board contains everything digital, and the main board and that DMM5 proprietary chip, are completely dedicated to handling all the analog stuff. So, a very clear separation between analog and digital. I guess this helps achieve the high accuracy of this meter, by keeping away the fast digital transients, from "polluting" weak but precious analog signals. Well, just a supposition... after all, there are many meters in the MX series, hence sharing this same basic architecture, yet most of them are hardly precision instruments. But maybe Metrix being a fairly small company after all, couldn't afford to design 20 different architectures for each and every of their instruments, so they invested their R&D money in something that would be suited to the most accurate of their meters, then all the less accurate ones would therefore also benefit from it. The LCD / digital board _______________ As we have just seen, it's actually not just the LCD board, but the main CPU board. Besides the LCD controller, the MCU doesn't integrate very much in terms of peripherals, just the bare minimum to be able to operate the meter : a couple 8 bit timers, a serial interface so it can handle the external Cal EEPROM but strangely it is not being used, the pins are used as general purpose pins, so chatting with the cal EEPROM has got to be implemented in S/W... why. .. and a few digital I/Os to handle the membrane keypad, RS232 IR link to access the outside world, and a few more input lines to get the status of the Fuses and position of the rotary switch (using 4 lines, " ROT0 " up to " ROT3 ". And of course, the data bus and usual associated control lines, so it can control the DM5 chip to get back the readings from the DMM5 built-in ADC, and also drive the 3 and only digital lines featured by the DMM5 chips (labeled "AUX 1/2/3" ) in order to select ranges and enable or disable the 10M input resistor on the mV range, according to what the user is asking for in this regard. The MCU actually does not feature an UART (the serial interface is just for simple SPI links and the like). The TX and RX signal are then tied to general purpose I/O pins. The link must then be implemented in software, which might be why, I would guess, the CPU runs at a rather fast almost 5 MHz, which would otherwise be unnecessary I think (?), to operate this meter. At any rate the DMM5, which is connected to the data bus of the MCU primarily to send back the results of its internal ADC, runs from a modest 32+ KHz watch crystal ! The LCD board also includes the backlight circuitry, for those meters who do have it (not my vanilla MX55 anyway). It's totally independent from the MCU, so should the MCU fail, you would still be able to turn light on and off. It uses a TLC393 comparator (comparator of choice throughout this meter) to H/W debounce the membrane button, then goes to a couple D flip-flops (usual 4000 series CMOS) to perform the toggling action. The LED backlight is then driven by a discrete transistor. Check Fuses : once again, a clear distinction between the analog part of the detection circuitry, which is on the main board, and the digital part of it. So, the digital board gets the analog feed from the main board, and adds a little bot of shaping circuitry so it can apply it to the digital IO pins of the MCU. In the case at hand, a couple comparators are sued for FUSE1 (TLC 393 as always), and a third being used for FUSE2. The Cal EEPROM : an ST 93C46 ( 461 actually on my meter). NOthing special you say ? One point worth noting though : they are not rated at the usual 10 year data retention period, but a surprisingly high, and much more suited to the task... 40 YEARS ! And the C461 I actually have in my meter, is rated at 100 years ?! Dear... didn't know such EEPROMs even existed ! So I found that interesting. Size-wise, it hold only 128 bytes, so just enough to cram in the required cal data. The main / analog board ____________________ So, that big DMM5 chip, to start with. It's main purpose is then to assist in implementing all the analog magic required to put the meter together. It embeds teh ADC and sends its readings to the digital board via the CPU(s 4 bit data bus). It houses the buzzer functionality, said buzzer is connected straight to it. Might be why it is scratchy/not latched... the chip being so analogy in its design, I guess a latch was not in order. It features a current output which goes straight to the input jack. Of course this is used when in ohms mode, or capacitance or diode test. In order to produce accurate(ish) current for the various required ranges, it uses an external 3 pins 1.25 voltage reference, along with a little cascade of carefully calculated resistors. It has a digital input dedicated to measuring frequency. Being digital, the signal passes first through some external shaping circuitry, in the form of a couple comparators, yes, TLC 393 again... The precision/potted divider network is connected straight to the DMM5 pins.... along with no less than 12 capacitors ! Boy what the hell do they all do. The 3 trimmer caps found on the board, are actually all part of this bunch. There are quite a few analog inputs and output on this DMM5, sometimes a bit hard to follow. Signals come in, then out, then in again via some other door... its a bit hard to follow I admit. From what I can gather : two signal come from the "front-end" of the meter/ input circuitry : "PR" and "10M". 10M as its name suggests, goes to the highest/10M tap of the precision network. PR attacks it from the 1M step just below. From that, the DMM5 generates several other signals : "AN0" , which is then sent to the shaping circuitry which produces a clean digital signal which is fed back to the frequency input of the DMM5. The DMM5 also output a couple more signals : AC1 and AC2. These feed a couple things : the RMS converter, and a "discrete"/improvised 10 bit DAC made out of a binary counter (a 74HC4040), coupled with a 10 stage/bits R-2R ladder, as you do... but whose goal escapes me, any idea ? These two signals are then fed to an analog switch (a classic CMOS 4052 chip), whose output, labeled AC3.. is then fed back to the DMM5 ! See ? signals just keep going in and out of this DMM5 chip ! LOL This analog switch is actually a double switch, and serves a second purpose : switching R25 on and off, the 10M input resistor, in order to be able to obtain a super high impedance mode when using the mV range. The DMM5 has 3 main analog inputs : IN1 to IN3. IN1 is the one used to read DC voltages, through the precision divider network. IN2 is the input used to measure currents. IN3 is used I think when measuring ohms or capacitance or testing diodes : the DMM5 outputs the current to the jack/into the DUT, then reads back the resulting voltage via this IN3 path. In the path of this signal, is a complex network of no less than 12 resistor... which really make up for a simple two resistor voltage divider, as you can see. But I guess it was necessary to put several resistors in series to get the high voltage protection a single tiny SMD resistor had no chance to meet. A similar arrangement is used nearby, by another circuitry in the front end : a chain of 5 resistors to make up for one high-voltage item. said resistors drive a rather complex arrangement of 5 bipolar transistors.. though their final goal is clearer : they drive a FET that shunts the input jack. Probably used to discharge capacitors once they have been tested ? Something like that... As for the section related to current measurement, it has been already described in a previous post. Basically the signal goes through a ladder of shunts, and an analog switch taps it where necessary, for the required range, then the signal is sent straight to the "IN3" in put of the DMM5, without any further processing. Switch control inputs are driven directly by AUX 1 and 2, 2 of the 3 digital I/Os the DMM5 features. Obviously, again, the DMM5 doesn't even know what it's doing... it merely drives this pins blindly, ordered by the MCU sitting on the LCD board. The DMM5 merely passes on the orders... The current ranges are switch by an analog switch bu sadly it a proprietary Metrix-ITT switch, so not data sheet available. but well, its operation doesn't seem to hold much mystery. The terminals being labeled on the schematics, and being rather self explanatory. Just a switch. 4 inputs controlled by two digital lines, "AUX"' 1 and 2, which come directly from the DMM5 chip, and which are remotely controlled by the CPU on the LCD board via the data bus. Well that's all I can think of. At any rate that give me a good enough understanding of these MX50 series meters, so I feel confident to buy some more, as I know how they work and could potentially fix them is need be. So that is it, the tear down is now complete I think ! :) |
| Vince:
Hi again ! Sorry for inundating this topic, I think I am almost done now... because today I just put my MX 55 back together, and it feels much better now, so I will call it a success. I have not yet received the 28 pin header connector for the LCD/digital board, but decided to reassemble it anyway, because I figured replacing the connector would be took great a risk : the pads on the PCB are round and really tiny, so little copper there... so I think leaving my bodge wire, which works just fine, is doing much less harm than damaging the tiny pads on the PCB, at best, or ending up bricking my meter at worst.... Farnell sent me all the other stuff I required though, so no worries : a few 10M 1% resistors, Vishay no less, a can of IPA and a couple of anti-static brushes (a soft one and a "tooth brush" stiffer one). So, I replaced that 10M resistor.. and realized stupid me, that I ordered a 500mW one but the original one was only quarter what.... luckily it still fits. Then I cleaned the golden tracks of the rotary switch... couldn't get it 100% because it appears a bit worn out so can't do much about it (is there "liquid gold" available, like there is "liquid tin" to tin PCB tracks ?? could be interesting.) However, looking at the before/after close-up pictures I took, it's undeniably much better than it was, so well worth it. Then I sprayed the entire PCB with IPA, both sides, to give it a good clean with the soft brush... a clean PCB being essential when dealing with high-impedances/low level signal. Couldn't hurt I thought. Then I soldered the RF shield back on. Then I tested it... and it works so much better, I am happy camper ! :-) In the mV range, my ears feel much better : now the alarm screams only when "expected to", in the high impedance mode that is. But in the regular 10M mode, no alarm no matter how much I play with the test leads. It reads 0.XX mV like the Fluke. The capacitance mode is back to its former self as well, very happy with that : now it behaves perfectly. After a reading/measurement (real or fake, with open or shorted leads), the display does'nt "float" randomly or erraticly displaying "Run'. No, as soon as you remove the capacitor, the display goes back instantly to a solid/steady 00.00nF ! Great ! :-D Oh, and ths week I bough a couple MX53C to complement the 55 ! :-) One of them is NEW ! Bought in 1999 by a guy at his work place, for teaching purpose... never got used. When he left work for retirement, he took it with him... and after a few years decided to put it up for sale. BRAND NEW ! He opened the case merely to take a pic to illustrate his ad, but other than this.. this meter quite literally never saw the light of day ! So the foam is like new (not discolored and still very firm, a set of spare Ferraz fuses, manual, and even a full page of calibration data from Metrix, for all ranges ! such a joy. So, with this "reference" meter at hand, I could better test/compare with my MX 55, to validate my repair. Everything is fine then. Capacitance wise, they are within 1% of each other from 47nF right up to 6800uF. Also, I checked of course the main 10A range, again by shorting my lab supply, and boy I was not disappointed.... up to 1A about, the match was perfect right down to the LSD ! And at 3.5A or so, the max deviation between the two was only 3 or 4 LSD ! I also couldn't help but make the two meters test each other : the MX 55 in ohms range, connected to the 5C3 mA range. The 55 was then measuring the value of the current shunt (the value being actually high enough to be measurable in the mA range), and the 53C measuring the precision/calibrated current output by the 55. See picture below... the two read just perfect ! On its 5mA range, the current shunt of the 53C reads the expected 100ohms, and teh current output by th 55 to measure the shunt, is bang on the the expected 1.0uA So... looks like my 55 has now a clean bill of health. Behaves well, reads true... I now consider it a reliable meter I can use with confidence in my project. The new 53C came with a holster.... at last I have one ! Well, if you don't have one.. GET one. Really makes it so much more usable : no fear of dropping or banging it on the bench, features the indispensable stand so you can actually see the display without having to get up your chair to look at it from above... and it is very rubbery-anti-slip, you can actually operate it with one hand, be it the knob or even inserting or disconnecting the leads. So, really, it makes the meter much more pleasurable to use on the bench, I find. What about my other 53C ? It stands right at the other end of the spectrum, relative to the new one. I picked it up on Ebay France/locally, starting price was only 20 Euros so I kept an eye on it... thinking that as always the bidding would skyrocket all of a sudden, 5 minutes before the end of the auction. So I was not holding my breath/ But somehow, turned out there was only one other bidder, and I won the auction ! It was mine for only 46 Euros ! :-D So even less than half the price of the new one (100 Euros). Obviously, for that price, the pics on Ebay looked like it was in a very sorry state, and until I got it in my hands, I was kinda prepared to have wasted my money. But... after I dismantled it and gave every part a thorough cleaning and loving care... and replaced the leaky battery and cleaned the corrosion on the battery terminal, and put two new ceramic fuses (bought a 10 pack for each range the other day)... it turned up actually REALLY nice, so a great buy indeed, very happy ! It took came with the holster too ! :-D So i have kinda a little collection of MX50 series now ! Love them ! I intend to buy some more... as dave said "you can never have too many multimeters ! " LOL The MX55 will be my main meter, the most accurate one, and the 53C will be there to supplement it when I need to keep an eye on several measurements at the same time.. or just as a backup. Their 0.1% is still plenty good enough for what I do or intend to do, and that's 3 to 5 times more accurate than the Fluke 70 series I am fancying too, not to mention 50,000 counts versus 4,000, a larger display, more functionnality etc.... yet they cost the same amount, at least of here ! around 80 Euros plus or minus 20 Euros depending on condition. So really, regardless of my love for theses MX50 series, the cold facts are also on my side : for 80 Euros or so, the choice between a Fluke 77 and a Metrix MX 53C is a no brainer I find. At least for bench work, which is teh goal. For a general purpose meter you put in you tool box and use to work on your car or around the house or anywhere to do whatever.... I would prefer the Fluke 77. smaller/more nimble, a bit ore rugged I feel, and easier to operate with one hand, and better protection if it does (?) have the same huge HRC fuses that the 87V has. So I might still buy for this purpose, to complement my fluke 11 which is perfect for this kind of use but.... which doesn't measure current sadly ! And also it doesn't get as much use as voltage measurement, there had been occasion where I really needed to measure current... Anyway. I tore down this well used 53C apart, and noticed a few things which might be of interest to this tear-down topic, so I will end with this. First things I notices, compared to my 55, was the the RF shield is not soldered to the back of the PCB anymore, but rather replaced by a foil glued to the back cover. So, nothing get in your way when you want to access the back side of the PCB, like I needed to. There is a little spring soldered to the ground plane of the PCB, which make contact with the foil/back cover. Then, I noticed that the LCD header connector was a different type : mine which I broke a pin off of, was the very fragile round and thin terminals. the 53C on the other hand, had the much more robust beefy square type. now way I would have broken a pin, had my 55 had that type of connector ! :-/ But the most interesting, was that.. horror... when I was presented with the back side of the PCB, I instantly noticed something very wrong ! It was unpopulated ! Alll the stuff related to RMS measurement was gone ! The AD converter itself, the "home made" 10 bit DAC, and the analog switch that feeds the RMS result back to the DMM5 chip. And the top side of the PCB... some more stuff is gone : the current shunts for the mA range, and the analog switch that taps them ! Still, the 53C is an RMS meters just like all the 54/55/56 of course, so... this tells us something new about the DMM5 chip : it must have a built-in RMS converter, as well as current shunts and analog switch for the various mA ranges.... but that the performance of the DMM5 chip in these department was not deemed good enough by Metrix to achieve the accuracy they wanted so the added their own external circuitry to remedy this. Looking at the spec sheet for the 50's series, which someone posted toward the beginning of this thread, it all become more obvious : the 53C is not just bottom of the range for it's lesser 0.1% basic accuracy, no it also has worse specs than the 54/55/56 for current measurement accuracy, which indeed confirms that it must uses some built-in circuitry from the DMM5 chip, and it also has lower RMS bandwidth than all the others in the series, yet again confirming the DMM5 having a built-in, but poor, RMS converter. So here you go, we learned a little something more about this DMM5 chip, even though it's a proprietary black box. So... at first I though well, maybe these 53C aren't such a good deal after all... but really I think they still are. I mean the RMS features is a selling point but honestly, I can't think of a use for it for lab/design work, the accuracy, crest factor, and piss poor bandwidth renders them rather useless I find, in practice, what do you think ? In practice, it's DC voltages I am interested in, as well as the other features the meter offers. If for some application I have to actually get a half decent RMS measurement of a signal, the only option I guess is to use a digital scope. It's got the bandwidth, and the crest factor is not an issue either, since it can calculate the RMS on any arbitrary waveform : if it can trigger on it, it can display, and if it can display it, then no matter what the the shape of the signal is, it can compute the RMS value equally well. I think this RMS feature on hand-held meters is very useful for measurement on the filed on electrical system/industrial stuff, motor control, what have you, but as an accurate way to measure electronic signal of any frequency and any shape and crest factor.... So.... I think these 53C meters are still well worth "collecting" to complement a more accurate 55/56 meter, for when you need more than one meter at a given time. I have spotted a couple more of them in excellent condition for 80 Euros each, in the local ads.. might be tempted ! LOL I didn't think they would (relatively) so easy to find. The 54/55/56 are much harder to find. Came across a brand new 56 the other day, for 160 euros.... so excellent value for money I find, but sadly it was already sold, stupid seller didn't bother removing his ad, like so many other sellers, who just don't care wasting prospectives buyer's time ! grrr.... drives me nuts. The 53C has one thing for it though, over my 55 ! It can measure duty-cycle ! I tested on my ancient Philips analog function gen. Duty cycle is not adjustable, but still, wanted to see it would read. No disappointment there : bang on 50.00% !!! Well, half of the time. Every now and then it would wander a bit, but always in a window of 49.9 - 50.1, so pretty good, for such a basic and old function gen ! I find this DC feature quite nice... it's simple enough to implement, that one can actually trust the value displayed by the Metrix, and I think it could come very handy when checking or designing anything using PWM signal, so quite a useful feature I find. Keeps from having to hook up a digital scope ! Much more convenient to use the hand-help meter I find. Sorry for a long post again ! But well, somehow I found it at home here, hope I didn't annoy to many people. Well at least I managed to get the schematics and share there here, so a decent contribution to this topic I reckon :-) A few piccies and that will be it. Thanks again to everybody who helped me in fixing my MX 55, now working just fine. Below pics :: - Farnell shopping : IPA, anti-static brushes, Vishay 10M 1% metal film resistors, packs of ceramic fuses suited to the MX 50 series, a couple tiny chisel tips, with built-in temperature control, for my 30+ year old 24V AC Weller iron... can't believe Weller still sells this model of tips ! That's customer support for ya ! - before/after pics of the golden tracks, using IPA and the anti-static brushes. not quite perfect but much better none the less. - The bodge wire for the header connector will remain for now... |
| Navigation |
| Message Index |
| Next page |
| Previous page |