multimeter Metrix MX56 - development of a replacement for the ARC1 chip

[sergej]

   I bought (for 4 Euros  ) old, but still very good multimeter Metrix (now Chauvin Arnox) MX 56. More about this series: https://www.eevblog.com/forum/testgear/metrix-mx56c-bk-precision-5390-multimeter-teardown/
But, the current measurement is bad - on lower (mA) ranges indicates very low values. 10x lower than real.

   First I reversengineering schematic from PCB for some days. and at the end I found on page 3, from the link above, the original documentation... 
All shunt resistor is OK, other components too. Only (proper) function of the OEM integrated circuit Z1, with marking Metrix ARC1, is questionable.

After I saw the original diagram, everything became a little clearer. This chip is a current range switch.
   I was built simple test stand with LED as output indications and jumpers with pulldowns resistors as a signal inputs.
After a few tests, everything is quite clear - the switch for the 500 mA range is shorted and still open.
This is confirmed by manually switching the ranges using the cable when the multimeter is switched to manual current measurement.

Next I am developed schematic with possible replacement - see attached PDF.
Technically the ARC1 is a four channel, bidirectional, power switch with 3 bits (BCD) remote - see function table in schematic.
The maximum switching voltage is low, up to 500 mV under normal conditions, 2.2 V in the worst case. This, at a 1Ohm shunt with a 500 mA range, gives a maximum current at 2.2 A until the time when the 630 mA fuse blows.
Only one integrated analog switch with this current rating i find - TMUX4827. A bit expansive, in DPDT configuration (1,1A per channel) but in BGA package only. BGA is relatively okay, but 0.2mm filled and capped vias are too expensive to manufacture. Especially if I only need a few pieces of a small PCB, but the minimum order is 5 panels of 85x85mm.
In final ( ) version V2 I used one 1:3 analog switch TS5A335 and one dual N-mosfet. The gate driver is primarily used as a NOR logic gate with a higher power supply range. Unfortunately the CMOS 4000 series is not available as a single gate in a small package.

Is there any reviewers for my schematic design ? (logic errors and etc..)
Or any owner of this multimeter (only higher types MX 55 and 56. 53, 54 have a manual range switching) who can verify function and internal connection of ARC1 ?

[Vince]

Wow, only 4 euros for an MX56 ?! 
that's pretty much free then... normal / average going price here in Frogland, its home, is about 120/150 Euros.
At 4 Euros it sure is worth spending some time trying to fix it ! 
Plus, as I understand it, only the current ranges are kaput, so the meter is still 100% usable on the DC voltage range, which is of course where its high accuracy makes this meter valuable.
Current measurement you can always do with another meter.

Anyway, good job on the reverse engineering ! 
The table in the schematic looks nice, as does the reverse engineered ARC1 pinout in the lower left corner... nice job.

Looks to like you have it all already figured out, so not sure how useful it would be for people to take measurements on their meter. But if you need a specific measurement done then yes I can probably do that, when I find time.

After a few tests, everything is quite clear - the switch for the 500 mA range is shorted and still open.

Shorted...ior open, which is it ?! 

Is there any reviewers for my schematic design ? (logic errors and etc..)

As for reviewing your schematic, according to the cartouche, looks like "Jack Daniel" has already done it, no ?! 
I am no expert / engineer, but a few things caught my eye. I will probably make a fool of myself but well, I like to participate anyway ! 

1) T1A and T1B.  I guess you put them there to do some voltage translation to accommodate the 5V inputs of IC3,  the TTL logic gates / inverters that they feed ?
If so, I don't understand how it can work. The way they are wired, the inputs of the logic chip will always be at logic level high, it will never change : the base of the transistors are fixed at 5V permanently, so the output voltage, which you tap at the emitter, will always be 5V - Vbe = 4.3Volts or something.

2) The dual MOSFET that switches the 500mA range. I don't understand how the two MOSFETS are wired together. In series ?
Shouldn't they be wired in // so as to build a bi-directionnal switch and/or lower the Rdson value ?
Even if you meant to wire them in series, why are they wired front to back ?


Looks like you are 95% there... if the PCB's are two expensive to make, maybe you can make a prototype board by hand to ensure that it works (and if not, modify it 'til it does) before ordering nice PCB's.

[shakalnokturn]

I once had a modern version MX 59HD I think, with a dead ARC1 and considered this kind of project, in the end the DMM5 IC also had some damage so I let the meter go "as is" and dropped the project.
I did make some measurements and notes at the time, I don't know if I still have them around.

I haven't extensively reviewed the DMM's schematic and how Sergej's design integrates.

A couple of things that do seem important is that the analog switches should withstand high peak currents, so that in case there is severe abuse they don't blow before the fuse.
To me this could involve bidirectional clamping diodes across the switch at a value slightly above RDSon x I fuse.
I'd probably also try to make the design failsafe for the DMM5 outputs.

Vince: The series MOSFETS are necessary for switching AC because of the internal body diodes, having the FET's parallel is allowing one diode to conduct in either direction. (If forward voltage is exceeded of course.)
I'm not sure about the 500mA range switch here, usually you'd have a common Source node on the N-FET's and switching voltage would be applied between Gates and Sources. I don't see how the "high side" FET's switched correctly here.

[sergej]

Shorted...ior open, which is it ?! 

Switch is shorted and open for current flow.  But right, not a good description of states of the switch

As for reviewing your schematic, according to the cartouche, looks like "Jack Daniel" has already done it, no ?! 
I am no expert / engineer, but a few things caught my eye. I will probably make a fool of myself but well, I like to participate anyway ! 

1) T1A and T1B.  I guess you put them there to do some voltage translation to accommodate the 5V inputs of IC3,  the TTL logic gates / inverters that they feed ?
If so, I don't understand how it can work. The way they are wired, the inputs of the logic chip will always be at logic level high, it will never change : the base of the transistors are fixed at 5V permanently, so the output voltage, which you tap at the emitter, will always be 5V - Vbe = 4.3Volts or something.

Right, its only a logic translator. 0 at input will be 0 due to pulldown and 7,2V will be translated to 4,3V.
Problem is space on PCB and low current consumption. I can use diodes/zener in series for voltage drop, or another transistor for inversion, but with same or more footprint space as now.

2) The dual MOSFET that switches the 500mA range. I don't understand how the two MOSFETS are wired together. In series ?
Shouldn't they be wired in // so as to build a bi-directionnal switch and/or lower the Rdson value ?
Even if you meant to wire them in series, why are they wired front to back ?

Looks like you are 95% there... if the PCB's are two expensive to make, maybe you can make a prototype board by hand to ensure that it works (and if not, modify it 'til it does) before ordering nice PCB's.

See at substrate body diode orientation. This diode must be connected antiserial. Mosfet is ideal bidirectional switch, but his structure consist this substrate diode….
In this version V2 the pcb is cheap, because expansive vias are not used. 5 panels about 6Eur + 20Eur shipping + VAT 21%. But I maybe I will wiring it on protoboard.

[sergej]

A couple of things that do seem important is that the analog switches should withstand high peak currents, so that in case there is severe abuse they don't blow before the fuse.
To me this could involve bidirectional clamping diodes across the switch at a value slightly above RDSon x I fuse.
I'd probably also try to make the design failsafe for the DMM5 outputs.

Right. Clamping diodes are used on basic schematic of DMM. Always 2 diodes in series for one polarity. See original schematic at page 3 from Vince.
But good question is, If this (not very high speed) protection is sufficient. Its ok for slow and long time overcurrent but for high speed ESD transient voltages… But this is valid for the whole device.

I'm not sure about the 500mA range switch here, usually you'd have a common Source node on the N-FET's and switching voltage would be applied between Gates and Sources. I don't see how the "high side" FET's switched correctly here.

I think, due to low Rdson, low switching voltages and leakage currents of body diodes are functions of D and S interchangeable.

[Vince]

1) T1A and T1B.  I guess you put them there to do some voltage translation to accommodate the 5V inputs of IC3,  the TTL logic gates / inverters that they feed ?
If so, I don't understand how it can work. The way they are wired, the inputs of the logic chip will always be at logic level high, it will never change : the base of the transistors are fixed at 5V permanently, so the output voltage, which you tap at the emitter, will always be 5V - Vbe = 4.3Volts or something.

Right, its only a logic translator. 0 at input will be 0 due to pulldown and 7,2V will be translated to 4,3V.

I must be blind, I still don't see how it can work. The emitter / output is fixed at 4.3V all the time, via the base that's tied permanently to 5V.
I must be missing something sorry...

See at substrate body diode orientation. This diode must be connected antiserial. Mosfet is ideal bidirectional switch, but his structure consist this substrate diode….

Oh OK. You put the two in series so that it behaves consistently ? In both directions, the signal will go through one FET + the internal diode of the other FET.

In this version V2 the pcb is cheap, because expansive vias are not used. 5 panels about 6Eur + 20Eur shipping + VAT 21%. But I maybe I will wiring it on protoboard.

Well go for it then, and tell us what !

[Vince]

Oops, I didn't see Shakal's reply when I posted    When I clicked the link on the notification e-mail, it showed me the last unread message, not the first....

[sergej]

If so, I don't understand how it can work. The way they are wired, the inputs of the logic chip will always be at logic level high, it will never change : the base of the transistors are fixed at 5V permanently, so the output voltage, which you tap at the emitter, will always be 5V - Vbe = 4.3Volts or something.

You have right, I am stupid...
I simulated it, but without serial impedance of source square signal. With real impedance 10+ Ohm is function as you write. I must remake it, Many thanks !

[Kleinstein]

For the protection the question is if the switches and shunts are OK with the voltage set by the clamping diodes.  For a short time the current can be high and the diode voltage may reach some 1 V per diode. For a longer time heating of the diodes will reduce the voltage a little and the fuse would blow at a really high current.

For the series connected FETs the gate voltage is quite a bit larger than the maximum voltage over the FETs. So the FETs should turn on, no matter if they are connected drain to drain or source to source. It only really matters that the FETs are off, when the voltage is rather small, like 200-300 mV to stay inside the normal working burnden voltage. Under overload conditions (some 2 V at the switch) it would not really matter if there is added leakage.

[sergej]

For the protection the question is if the switches and shunts are OK with the voltage set by the clamping diodes.  For a short time the current can be high and the diode voltage may reach some 1 V per diode. For a longer time heating of the diodes will reduce the voltage a little and the fuse would blow at a really high current.

Fuse for all mA range is rated at 630mA.
Diode clamp voltage is about 2,2V.
This is 2,2A for 500mA (shunt 1R0) range - no problem for T2 switch. With Rdson = 90mR is 0,36W for some miliseconds
For 50mA range with 9R0 shunt is max current 244mA. one switch in TS5A3359 is rated at min 200, max 400mA. I hope for 400mA 
But the bigger issue is this shunt in the minimelf (0204, 0,4W)) package and the nominal power dissipation, rather than the internal switch.

[sergej]

revision 002 

[beenai2018]

Replied to upvote this as a owner of an mx57ex!!

[sergej]

Actual state: PCB are in production.

[sergej]

The operation was succesful, the patient is alive !
Current measurement is now fully functional.
Only 430nA offset i must clear, probably calibration residue ( with demaged ARC1).
Pcb layout have 2 mistakes:
wrong footprint for 5V stabilisator. Nothing nothing terrible, on pcb footprint pins 5 and 6 must be shorted.
PCB outline must be a little modified, one corner extendds into the support post.
Now i must make a communication interface and find calibration software.

[sergej]

Photos continue.

[coromonadalix]

congratulations    well done for the repair   

the cal software is available  async2 ...   

Chauvin Arnoux gave me this one    sx-asyc2cv23.rar  around 6 meg

[macboy]

Check last page of this thread
https://www.eevblog.com/forum/testgear/metrix-mx56c-bk-precision-5390-multimeter-teardown/

I posted links to the calibration software, logging software, and documentation of the communication protocol so that you could write your own software or scripts if so inclined. The links might be stale now but should point you in the right direction.

[coromonadalix]

the asyc downloads on the last page   still works

[sergej]

SW downloaded, like good.
Now I must build optic interface...

If anyone needs PCB ZRC1, please let me know. There are enough of them.