Advantest R6581 Schematic Diagrams

[Mickle T.]

Schematic diagrams: https://drive.google.com/open?id=168OEKv1ZXCj0otQcDKMG1uXkSvPy_49J
All other R6581 stuff is here: https://drive.google.com/open?id=1_dyCk3kvzEaxp3uqANyhvvmRXCpOAnJo
Full archive was uploaded to KO4BB.

[eurofox]

 

[2N3055]

Fantastic work!
Thanks!

[wutieru]

I have been looking for this Schematic for many years?Thanks a lot?

[zhtoor]

@Mickle T.

i vote for awarding you the "MASTER OF VOLTNUTS" award   

best regards.

-zia

[quarks]

many thanks

[Mickle T.]

Schematic diagrams is updated to v.4 in post #1: some minor bugs was fixed, measured all of the film capacitors.

Firmware reversing is in progress. To date i have disassembled all EPROM code and identified all RTOS VRTX32/68k system calls, exception vectors, IRQ handers, internal hardware registers, reset and early initialization procedures and many other low-level things.

Problem #1 at the moment - too many chips on the digital boards with unknown address space decoding
Problem #2 - an ancient Xilinx FPGA in the middle of the main digital board 
 

[TiN]

If you need some logic analyzer measurements, let me know.
I can hook up my LA to capture the waveforms (272 channels should be enough? .

[Mickle T.]

All of the above-mentioned problems is solved by hand without any LA or scope.
Advantest R6581 reversing project is closed now. At least until I get a ACV board or full non-T R6581

[eurofox]

All of the above-mentioned problems is solved by hand without any LA or scope.
Advantest R6581 reversing project is closed now. At least until I get a ACV board or full non-T R6581

Glad that you finally manage to fix the instrument, I'm quite happy with mine, it is maybe not so good as the 3458A but I think it is in the area of a Keithley 2002.

I hope soon to be able to meet someone in a calibration LAB or private person with a recent calibrated 3458A or with a calibrated 10V reference.
I have a near 0 temco 10K resistor. Of course I'm ready to pay for this, it take one hour or more to warm up the instrument and only 10 minutes for an external calibration, I don't need any report.

If someone can help me in Belgium, France, Holland or Germany just let me know up to 600 kms from Brussels.

[TiN]

not so good as the 3458A but I think it is in the area of a Keithley 2002

Nop, it's not in 2002 league either. And 2002 resistance performance is bit better than 3458 minus the ACAL.
At least both of my 2002s show better stability on 1K-100K values than my 3458s.

I think Mickle.T meant that he reversed logic connections and mapping, but that does not mean analog issues are fixed.

[eurofox]

not so good as the 3458A but I think it is in the area of a Keithley 2002

Nop, it's not in 2002 league either. And 2002 resistance performance is bit better than 3458 minus the ACAL.
At least both of my 2002s show better stability on 1K-100K values than my 3458s.

I think Mickle.T meant that he reversed logic connections and mapping, but that does not mean analog issues are fixed.

Is your R6581 fully operational now? Last information I remember is that display was barely visible showing long power up life, the test that you ask met to perform on my system did not perform well on yours, I remembered the polarity, the resistance check ....

I would like to check mine near a calibrated 3458A or 2002 to judge. I have a 2001 100% OK but need calibration as well.

[Kleinstein]

Noise wise the 6581T might be about on par with the Keithley2002, maybe a little better. However the Ohms ranges are likely worse, especially high higher ohms ones.  The ACAL scheme needs a lot of steps for all the Ohms ranges and also for the higher currents. Here the non ACAL scheme with separate reference resistors in the K2002 has an advantage - at the price of a larger number or critical parts and the more complicated calibration (not just 10 V and 10 K, but separate Ohms ranges and currents). Similar for the currents, especially the higher currents where the test current from the ohms source is limited.

As far as I understood it, there was only a partial fix to the problems (faster guard amplifier directly from the input - which reduces the input current spikes at highly negative voltage).

Still there seem to be quite a few of the meters to show excessive INL and the higher Ohms ranges seem to have a problem. The input current when in < 10 PLC AZ mode also seems to be quite high due to the missing pre-charge phase. This is kind of a weakness by design. This also includes the additional switching related current spike with voltages beyond about -7 V.

@Eurofox: It might be interesting to see the test for input current for a seemingly good meter. That is recoding the discharge curves with a good quality (e.g PP / PS) 1-10 nF capacitor in the 10 V , 1 PLC AZ mode, and maybe without AZ.

[eurofox]

Noise wise the 6581T might be about on par with the Keithley2002, maybe a little better. However the Ohms ranges are likely worse, especially high higher ohms ones.  The ACAL scheme needs a lot of steps for all the Ohms ranges and also for the higher currents. Here the non ACAL scheme with separate reference resistors in the K2002 has an advantage - at the price of a larger number or critical parts and the more complicated calibration (not just 10 V and 10 K, but separate Ohms ranges and currents). Similar for the currents, especially the higher currents where the test current from the ohms source is limited.

As far as I understood it, there was only a partial fix to the problems (faster guard amplifier directly from the input - which reduces the input current spikes at highly negative voltage).

Still there seem to be quite a few of the meters to show excessive INL and the higher Ohms ranges seem to have a problem. The input current when in < 10 PLC AZ mode also seems to be quite high due to the missing pre-charge phase. This is kind of a weakness by design. This also includes the additional switching related current spike with voltages beyond about -7 V.

@Eurofox: It might be interesting to see the test for input current for a seemingly good meter. That is recoding the discharge curves with a good quality (e.g PP / PS) 1-10 nF capacitor in the 10 V , 1 PLC AZ mode, and maybe without AZ.

I suppose this configuration?

[TiN]

No, just cap across the 6581 terminals. Log data with meter itself, once cap is charged to 10V and then switch open.
Siglent is useless to measure bias currents of HiZ meter.

[eurofox]

No, just cap across the 6581 terminals. Log data with meter itself, once cap is charged to 10V and then switch open.
Siglent is useless to measure bias currents of HiZ meter.

I know that the curent is very low because the very impedance of the meter, I was just trying to understand how you proceed based on explanations of Kleinstein.

I don't have a high quality capacitor in my home collection, I was in the shop yesterday  , they have a full range in this shop of audio Polypropylene capacitors, I just order some on internet, it take me 2 hours to drive to the shop and back 

The problem with your procedure it is not easy to record the data with time information, it should be done by communication with GPIB and output to a PC with trending, it is one of my project on the list.

But due to the very high impedance discharging is maybe very slow ...   

[TiN]

Sorry, but anyone who is going to do anything useful with 8.5digit meter must have a setup to log timestamped data with attached environment conditions. This is the basics when you entering this realm. RPi + one of available GPIB dongles would pay for itself many times over the time, if you into this game.

[eurofox]

Sorry, but anyone who is going to do anything useful with 8.5digit meter must have a setup to log timestamped data with attached environment conditions. This is the basics when you entering this realm. RPi + one of available GPIB dongles would pay for itself many times over the time, if you into this game.

I already got a genuine USB - GPIB Keysight interface since more than 2 months, got connection cables last week.

I have now a temperature and humidity logger with PC interface.

I'm checking now on the PC side, NI, Excel or any other interface but no Python, I already know basic, vb, C, C++, assembly, Pascal and that is enough 

[eurofox]

Those strings you have to send to the R6581 to read data.

I try with Excel but it is slow 

Range 10V DC and one NPLC

    instrument.WriteString "*RST"
    instrument.WriteString "CONF:VOLT:DC"
    instrument.WriteString "VOLT:DC:RANG 10;NPLC 1"
    instrument.WriteString "ARM:SOUR IMM"
    instrument.WriteString "ARM:LAY2:SOUR IMM"
    instrument.WriteString "TRIG:SOUR IMM"
    instrument.WriteString "INIT:CONT OFF"
    instrument.WriteString "ABORT"
    instrument.WriteString "READ?"

[Mickle T.]

In response to a z01z question about R6581 accessory detection code.

R6581 extention cards interface is not as simple as it appears. It have a 8-bit local data bus (buffered from 16-bit MPU data bus with U22 74HC245), 8-bit local address bus (A1-A8 MPU address lines, buffered with U23 74HC38245), high address lines decoder (PAL U114 e.t.c.), interface enable line (bit #3 port B of PPI U108 82C55), 3 (or more) lines from magical FPGA...

[z01z]

Mickle T., thanks a lot for sharing this!

[z01z]

Regarding the address decoding table, it is said in the manual that the SRAM card can be 2MB maximum. It makes sense, since 2MB will occupy the whole address range of 800000-FFFFFF.

[eurofox]

Sorry, but anyone who is going to do anything useful with 8.5digit meter must have a setup to log timestamped data with attached environment conditions. This is the basics when you entering this realm. RPi + one of available GPIB dongles would pay for itself many times over the time, if you into this game.

Second vs Volts, zero will take several hours, capacitor is 5.6NF.

[Kleinstein]

The seconds over volts is an slightly usual plot. The more normal way would be to have the time horizontal. Still one can read the input current: for the 10-7 V range it's about 200 seconds for 3.5 V and thus around 100 pA of input current at some 7-10 V. This is not that bad - other such meters had more input current. It also depends on the more used - I would guess this to be in a non AZ mode. The input current is expected to be higher in AZ mode, especially with 1 PLC mode as there is extra switching related current.

Anyway the discussion is a bit off topic for this thread and might better move to the repair/tear-down thread.

[eurofox]

The seconds over volts is an slightly usual plot. The more normal way would be to have the time horizontal. Still one can read the input current: for the 10-7 V range it's about 200 seconds for 3.5 V and thus around 100 pA of input current at some 7-10 V. This is not that bad - other such meters had more input current. It also depends on the more used - I would guess this to be in a non AZ mode. The input current is expected to be higher in AZ mode, especially with 1 PLC mode as there is extra switching related current.

Anyway the discussion is a bit off topic for this thread and might better move to the repair/tear-down thread.

I will make a new one in "repair" but I'm convince that my DMM is OK.

Record with a good quality capacitor 100PLC and AZ mode. I agree that the graph should be time in horizontal, I just finished replacing the bindings posts by "good ones"