Takeda Riken TR6878 6.5 digit m/m teardown

[Chalmml1]

So for my first post I decided to do a teardown and since I can't turn it on (no-one where I live had heard of a fuse smaller than half an amp) I'm taking it apart.

What I have for you today is a Takeda Riken TR6878 6.5 digit multimeter, brand new in 1984 (designed in 83). It's somewhat different from most multimeters in that it doesn't have any AC capability and I'm not even sure how low it measures in the dc ranges as I can't yet turn it on and there is zero documentation to be had for it online or other that I can find.

Features of note: 2-4 wire ohms with and without guard, front and rear input (rear has triax for the milliamp and below? range), GP-IB, sampling? and lastly some sort of compute functions just how complex some of you might know from looking at the front panel section for it but again no manual so its going to be see as I go.
I'm still very in the realm of amateur hobbyist so anything you can give me in advice on how to best operate this beast would be very welcome 

And now for the pictures, please forgive the bad photography, I have terrible lighting and had to flash a few shots

Starting with the reference, who doesn't like to know what makes these things accurate? yes its a zener, what type I couldn't even begin to say as its made by Takeda Riken in-house and has 3 leads.


The lovely hybrid, on its own machined aluminum block, guessing there is a bit of stuff going on inside the block but I'm not rushing to pull it apart, even for science.


The front and rear inputs to the board, the giant blue blobs are inductors for those not familiar with this style.



A touch of input switching, there interestingly enough appears to be absolutely no input protection that I can find.


The analog power supply section with a beautiful example of star grounding. They also thoughtfully have provided nicely labelled testpoints everywhere for the power rails and outputs/critical functions, quite the service techs dream.


Ohms and amps section, lots of high impedance magic and unobtainium here. Also a lot of lm339's and 74 series logic.


This mysterious black thing is actually a Takeda Riken 1 ohm precision resistor.


The A/D section, complete with scary amounts more unobtainium and high impedance hybrids. The large black things are just potted caps.


And the last section of the analogue board, control, complete with a fujitsu mb60h125, guessing a 8/16bit micro? and a pile of what seem to be hand wound pulse transformers to isolate the analogue and digital boards.


Now on to the digital side of things. Firstly the 8086 is fairly self explanatory, along with a good amount of 74 series logic and an interrupt controller and interval timer as peripherals. Oh and some more pulse transformers, they didn't all fit on the other board.


Next up we have some SRAM a whopping 8k per package, some socketed roms and a NEC 8279 keyboard and display driver. I will have to get a reader to pull the data from the roms in the off chance there is someone who wants it.
Also of interesting note the large ceramic cap is the only connection between the chassis and the internal components, literally everything else is insulated and floating.


Lastly on the internal side of things we have the GP-IB controller and a couple more ram chips socketed this time. They're 2k and slower so I guess for the GP-IB controllers use? and yes that battery is as dead as a doornail, hopefully the only thing wrong with this unit as it was bought as not working(hopefully no repair thread in the future).


Onto the outside. For those of you that have made it this far you may want to take a break, stand up/get some water we're almost done.

On the front we have happily a led segment display with a couple of led dot matrix character segments on the right (which better not break as I've never seen anything like them before). All the function selection and compute is entered on the front along with the front input jacks. There is also a button for shorting/opening the guard to neutral, this applies for both the front and back inputs at the same time.



Not much of interest on the back apart from the mAmps Triax input which I likely wont be using anytime soon as to buy a new cable for it is over twice what I paid for the instrument itself. I have been following with close interest those attempting their own triax cable and may go down that path if I need the ultra precise measurements in the future.




So there you have it, one more piece of gear from the 80's in its full glory. If anyone would like better pictures of any part of it and/or a more in depth examination do let me know.
Also if anyone can think of a reasonable use for any of the slightly more esoteric compute functions I would love to hear from you, I don't really actually need a 6.5 digit multimeter by any means, I just got it as a birthday present to myself because I wanted something other than tv's to fix.

Thank you for sitting through my first teardown. Any and all constructive criticism of my posting I would really appreciate, I have some more gear I've been given which I wouldn't mind taking a peak inside and sharing with you all and getting a better idea of how to go about it would be great.

[daqq]

Wow! Thanks for the teardown! I really like the small ceramic blocks all around the place - not standard at all. I don't think that there's necessarily something inside the big aluminum block though - it might be there just to keep the whole block at the same temperature to avoid all sorts of nasty.

[chickenHeadKnob]

Welcome to the forum. No need to apologize for your photography as it and your test subject is excellent. When ever I see an obscure (to me) mitsubishi  micro I think that is one of those TRON optimized instruction set ones, but this meter is too early for that. ITRON and TRON really took foothold in japan in the last half of the eighties along with fuzzy logic.

[TiN]

Nice score for $150, maybe take few photos of overall view.

Does it power on? Would be interesting to see it working.
As of VREF, it can be diode + heater or two diodes. Measure all terminals related to star GND point, to find out

[Chalmml1]

Had a bit of a closer look at the alu-block hybrid and I believe you are right in that its for thermal bonding, I had thought the devices on the top side were something exotic since they had 4 legs and that the forth leg went to something on the underside but it turns out they're just 2n117a's (ultra high impedance n-channel jfets). It's also glued together when you look at it closely which is a bigger giveaway

I haven't yet powered it on as I couldn't get a fuse for it locally so had to order one in and these things take forever to get to New Zealand, I could just power it up with what it has in it but I'm a bit pedantic about getting the right parts for such things. I've also ordered a new battery and will likely try to find replacement electrolytics as I had a closer look and while they haven't thrown up all over the board like is common in keithleys they are starting to get a touch of leakage on the non sealed end and they're 30 years old.

Another thing in the list of stuff to acquire is a universal programmer, something along the lines of the one Dave recommended, I've always wanted to dabble in some of the older micros 8051 etc and this gives me a good excuse

A few more photos, some of the full unit and a few more interesting bit's I've come across.
My Japanese is non existent but I think that's a cal sticker for 2004 on the left. Overall the size seems about the same as hp system meters namely the 345x series.

Inside is all nicely shielded, the top shield has slide retention as well as screws and helps control the thermal side of things on the analogue board with only slots over the power section leading back to the control IC then out the rear fan.

Overall the board is laid out very nicely, everything is in its own little defined block.

This is the first time I've come across any high impedance magic that had a power transistor tacked into the middle of it, guess its purely for switching.

And what piece of 80's equipment would be complete without a bodge, though this is the only one I could find. I haven't the faintest clue why its been done as all it does is go from one ground trace to some other ground traces all of which terminate about 5cm away.

[Mickle T.]

VREF (-10 V) is the same as in the R6871E: 1N829 zener with 600 Ohm ballast resistor (there is no heater in the tube).

[SeanB]

I like the mounting they used for that TO220 package, and the nice way it holds it down. The wire and track cut was probably used to improve the noise performance to get the last bit of noise out of the unit. It probably improved the noise of the reference a little by removing induced digital noise in the ground lead from adjacent traces on that cut trace. I like the second wire there, they needed a lower impedance than just the **one** trace.

[Vgkid]

Thanks for the upload, I wish Ve7xen teardown pics of his Advantest  were up. I wish the internals of my meter looked more exotic.

[martinr33]

I picked one of these up - such a fascinating set of guts!

Here's a pic of the analog board, which looks about 1 - 1.5 years newer (8604 was my latest date code on a quick check). Apologies for snapping handheld with flash, but I wanted to get this taken and didn't have time or space to do a proper setup.

One problem, on mine the individual Toko 609 pushbuttons have rubber domes that have hardened and no longer spring.  I don't know if I can find a sub. Any ideas? This has to be a common problem.

Also, mine has a relay bodge board.

For full res:
https://goo.gl/photos/jnbWzBJCvXTuRoRy7

[martinr33]

I finally finished replacing the front panel buttons on my TR6878. The originals switches were all suffering from hardened and cracking domes. I got new switches from Japan.

I also replaced the battery. However, that seems to have caused a problem because when I turned the meter on, it lit every indicator and displayed Err CA. A bit of button pushing got it to Err 1.

I found this bit of wisdom on an old R&S paper:

TR6878 [CE][ENTER] [LOCAL] + POWER SW ON [CLEAR][ENTER]

That actually gets you to a mode where you can poke around. I bumbled about, played with the cal switch, fed it 10V, and eventually got it to the point where it kind of works. Seems to be a bit off in the fourth digit, but I'll let it warm up and see where it goes.



 

[amspire]

I have a couple of Advantest R6871E-DC  7 1/2 digit meters that are made in the same factory, so a lot of the parts are identical. One of my meters has stopped working with Err messages ever since the 3.6V battery exploded and wiped out all the settings in RAM. These Takeda Riken meters have no schematics or service manuals available, so you get no clue.

Looks like your meter has the CAL mode switch on the rear, and it probably uses a similar entry of calibration data to the 6871E (the user manual is easy to get).

Also the internal -10V pin is probably still pretty accurate. lacking anything else, it is a good starting point for DC volts.

[martinr33]

Now that I have working buttons, I can see that the 6878 is very similar (in terms of operation) to the R6561 - which does have a manual.

Strangely, my meter was fine until I replaced the battery(which was reading a solid 0.000V). After that, every panel light came on and I got that CA error message, and occasionally I could get that to Err 1. It really felt like a dead meter, but I guess that is the idea.

I am going to see if it will cal for me tomorrow, at least on DCV.

This 6.5 digit instrument was built around 1984, and I suspect represented state of the art at the time. The industry went to 7.5 and then 8.5 digits in just a few years, then stopped. Even the HP 34401A - less than a decade later - represents a remarkable advance.

[HighVoltage]

and there is zero documentation to be had for it online or other that I can find.

Try to send an email to Advantest.
I have a few Advantest instruments and they have been kind of helpful each time, if they had anything.

Thanks for the really nice pictures, it seems to be a very impressive instrument.

[amspire]

Probably a good idea getting the R6871E manual as well. There are differences to the 6561, and since your DVM has a numeric keypad, it may use the 6871 method where you enter the correct reading in calibrate mode rather then the 6561 method where you modify the current reading.

http://www.gotroot.ca/media/Advantest%20R6871E%20Series%20Instruction%20Manual.pdf

[amspire]

and there is zero documentation to be had for it online or other that I can find.

Try to send an email to Advantest.
I have a few Advantest instruments and they have been kind of helpful each time, if they had anything.

I will have to give them another go. I did try years ago a few times and no answer.

[martinr33]

Credit for the pictures should go to Chalmml1, who did a much better job  than my single photo - which I posted for reference - and obviously influenced me to pick this unit up. The innards are surely a very fine piece of work.

Advantest made some really nice stuff for precision measurement, but now produce only more basic multimeters - they have moved on to much more complex equipment, after a couple of transitions in their business. I found a claim somewhere that they installed the first Josephson Junction in private hands in 1990.

[helius]

One of my meters has stopped working with Err messages ever since the 3.6V battery exploded and wiped out all the settings in RAM. These Takeda Riken meters have no schematics or service manuals available, so you get no clue.

The Maxell lithium batteries are notorious for exploding, causing extensive corrosion.
You can find some TR manuals from companies that specialize in that area. I didn't see one for the 6878.

[martinr33]

Well - calibration is easy. Works the same as the other Advantest equipment of that era.

Put it in cal mode - now, anything you enter on the keypad will be used as a cal value if it is appropriate. So about 10V on the 10V range. It won't take voltages out of a reasonable distance.
The manual starts you with zero, and walks you up through the ranges. Very simple.

I can now get it and my calibrator to both display numbers like 3.14159 or 2.71828 or 9.87654.

Again, the bizarre thing is, if calibration is lost, the whole front panel lights up and has you wondering how replacing an old, dead battery can cause so much trouble. Funny thing is, the device was reasonably accurate before I replaced the battery. There are some jumpers around the voltage source, I wonder if that is some form of basic calibration to refine the 10V reference.

[amspire]

Many of the tracks near the battery are rediculously fine, and it is easy to have corroded tracks near the place the old battery was. With the unit disconnected from the mains, check the ram still has a supply voltage

[martinr33]

In my case, there's no corrosion and the RAM has 3.6V when the power is off - so all looks good. I found some decent 1/2AA battery holdersthat I am using on other stuff. If the battery does ever leak, the holder should catch some of the gunk.

[Kleinstein]

There is no need to adjust the reference to an exact value. The jumpers are more likely to adjust the reference current to a value that gives the lowest TC. This way one could adjust at least the linear TC of the reference.

[martinr33]

Yes, could be - looks like there are about a dozen jumpers around that resistor array. So it is presumably a trimmable precision resistor. There's a small pot in the same area. Seems like the resistor jumpers would be a royal pain to set up, unless the reference was previously characterized outside  of the system.

[TAMHAN]

Hello,
if you need a fuse, let me know via a PM. I go to GM once a week or every two weeks, and can buy one for you. In Slovakia, fuse availability is quite good.

[picburner]

If anyone cares about the manual I put it here.
The description says TR6868 but it is a typo...
The only problem is that it is in the Japanese language, but with the help of google translate....

[kipp]

this is an old thread, so maybe this isn't helping, but normally there's no calibration data in battery-backed ram in Advantest instruments.  Calibration data is stored in an eeprom, and the battery powered ram is just there to save panel settings across power cycles.  However, the instruments are not good at realizing that the RAM contents are invalid, and have a habit of crashing on power-up when the battery fails or has been replaced.  in that case you need to find the reset key sequence for the instrument.  there is a PDF floating around with these for Advantest instruments.  in my copy it says for the TR6878 it's [CE][ENTER][LOCAL] + POWER SW ON  [CLEAR][ENTER].