10bit binary to 3 digit BCD (fixing an old nixie tube bench multimeter)

[SK_Caterpilar_SK]

I got some very nice looking school grade Tesla manufactured nixie tube multimeters. In fact three of them. The display ooks absolutely nice and the nixie face, I can only say its pure old tek pornography. Only if it was working.

In the old days they used to use a all in one 3digit 4bit BCD ADC. The C520D. This IC came in carious flavours like the soviet german C520D, the american AD2020, or the intersil CA3161, or perhaps the newest one the NTE 2054. The thing is in all the three multimeters I got the same fault occured, the ADC fried. It still converts the analog to digital, but the multiplexing order is off. Specifficaly its most signifficant digit. Interestingly this happened in all of the meter.

So here I am making a decision. While a replacement part /any of the above mentioned/ would cost 25 dollars would work, god knows how long it would. Since all the meters have the same fault I am really hesitating to try and put a new IC into it. Maybe thats why they were shelved for 15-20years in school. But back in the days the IC was basicaly available at your local drug store....not anymore. Maybe the NTE2054 with the new processes and all that work better and longer but what is the guarantee that it wont fry the second I turn the meter on.
(no it does not appear to be a design flaw in the meter.)

So instead here is my solution to the problem:
10bit paralel ADC and convert the binary into 3 BCD digits. 10bit is the closest I can get to get the full 999 range of 3 digits (that is 25 less than what 10 bit limit is)
This all sounds fun and easy and no....I am not using an arduino or FPGA or anything like that. I am a analog nerd fighting with audio stuff a lot and analog stuff in general, and I hav absolutely no interest in copying something from the internet that I wont understand at all.

SO the real problem is conversion of 10bit binary to BCD. I have found a IC (74184 and like that) which are long obsolete and unobtanium, so I guess a built up converter it is. I just dont know how at al. If anyone could somehow point me to something expandeable or draw up somethin for me with logic gates or TTLs or CMOS I would be gratefull. And please also keep in mind that I am a total noob when dealing with digital tech. Even the basic digital tech.

Thanks.

[edavid]

The thing is in all the three multimeters I got the same fault occured, the ADC fried. It still converts the analog to digital, but the multiplexing order is off. Specifficaly its most signifficant digit.

This is not clear - what are the exact symptoms?

So here I am making a decision. While a replacement part /any of the above mentioned/ would cost 25 dollars would work, god knows how long it would. Since all the meters have the same fault I am really hesitating to try and put a new IC into it. Maybe thats why they were shelved for 15-20years in school. But back in the days the IC was basicaly available at your local drug store....not anymore. Maybe the NTE2054 with the new processes and all that work better and longer but what is the guarantee that it wont fry the second I turn the meter on.

The guarantee is that you check the voltages on all the pins to make sure there is nothing that will fry the IC.

Looks like you could get CA3162s from China for around $7.

If that's too much, it would be easier to patch in an ICL7135 than to use a binary ADC.

[SK_Caterpilar_SK]

I would be gratefull If you could send me the link for that 7dollar 3162. I have tripple measure everything and everything is fine. The meters are not flawed.

MSD is out of sinc: garbage on the pin. Out of sync pulses all over, so the digits overlap. While NSD and LSD are in proper timing. I checked this with a 4channel scope with the IC separately on a breadboard. All the faulty meters have the same problem with the same IC with the same order of MSD NSD LSD. One of the meters had the transistor (which was switching the high voltage for the MSD) was fried and replaced but the problem remained no matter how I interchanged the three ADCs between the meters.

I wanted to build a complete 10bit to BCD converter because then I could just make it bomb proof and not hope it wont die again. Also if any part of it dies I can then specifficaly point to the fault point. I have a solution how to do the multiplexing that is the easy, but even if I would buy the ADCs you said cost 7 dollars I would still like to make the full circuit. Maybe one day I would decide to make a 4 or 5 digit nixie tube multimeter with custom nixies xenon filled for fun who knows.

Thanks.

[floobydust]

I have a bench digital multimeter that blows a digit driver if it gets ESD or HV overload at the input jacks.\
I was puzzled, but it turns out the PCB is a bad layout, a trace close from the input switches to the decimal-point trace and it arcs over there to the digit driver.

Said another way, the multimeter (+) input jack goes to the range switch. The range switch also has a layer that goes to the display circuitry to move the decimal place. Any zap or arc from the two switch layers or across the pc board can kill the display circuitry. Does that make sense?

OP these IC's failing, could be due to ESD arc jumping over to the decoder-display driver. Have to see pictures of the PC boards and switches.
The A/D should have basic overload protection with clamp diodes, you could add them.

[SK_Caterpilar_SK]

I have a bench digital multimeter that blows a digit driver if it gets ESD or HV overload at the input jacks.\
I was puzzled, but it turns out the PCB is a bad layout, a trace close from the input switches to the decimal-point trace and it arcs over there to the digit driver.

Said another way, the multimeter (+) input jack goes to the range switch. The range switch also has a layer that goes to the display circuitry to move the decimal place. Any zap or arc from the two switch layers or across the pc board can kill the display circuitry. Does that make sense?

OP these IC's failing, could be due to ESD arc jumping over to the decoder-display driver. Have to see pictures of the PC boards and switches.
The A/D should have basic overload protection with clamp diodes, you could add them.

yes that is a valid option aswell. Knowing it was operated by a lot of people in scool it could have gotten zapped, but the unput curcuit and power supply and digital section have their own sections on the board. If the ic would have been fried by ESD the whole board would be toast including the input side. The range switching is done on a vertical board that has no digital components to it.. I will attach pictures of the board later today (its 2 morning). I dont think it was cause by bad layout. Experience wise Tesla (czechoslovakia) built things like tanks and that included the boards. Never had one fail because of a bonehead mistake done on the board layout and I always make sure nothing is statically charged before I begin to stick probes into something (which is not common practice in schoool...). I suspect the component choice in the high voltage portion (switching the nixies on and off) because at the time component choice was very limited by cost and availability.

[edavid]

I would be gratefull If you could send me the link for that 7dollar 3162. I have tripple measure everything and everything is fine. The meters are not flawed.

Example:

https://www.aliexpress.com/item/4000204877027.html

Or if you don't like AliExpress, you can get them from utsource.net for a little more:

https://www.utsource.net/sch/ca3162e

MSD is out of sinc: garbage on the pin. Out of sync pulses all over, so the digits overlap. While NSD and LSD are in proper timing. I checked this with a 4channel scope with the IC separately on a breadboard.

What happens if you generate the correct MSD strobe timing with a monostable?  Is the MSD data correct?

[SK_Caterpilar_SK]

I don't have a proper way to test if MSD data is correct but if I give the input 12,3V it appears to show all the characters but it's not as straight forward. I will make a video.

[free_electron]

Small cpld from altera, draw schematic with ‘TTL’ logic like 14175 74185 decimal to bcd. You can shove all logic in the cpld.

[graybeard]

https://www.aliexpress.com/item/4000204877027.html

Those are most likely "recyled" parts with an unknown number of hours on them.

[macboy]

edavid beat me to my suggestion. The TC7135, ICL7135, MAX7135, SI7135, and TC835 is a 4.5 digit converter with muxed BCD outputs. You would of course not use the outputs for the leading 1 digit, nor the least significant digit, but you might be able to find a way to make it work.  The TC14433 is a 3.5 digit similar device (+-1999 count).

[TerminalJack505]

...

10bit paralel ADC and convert the binary into 3 BCD digits. 10bit is the closest I can get to get the full 999 range of 3 digits (that is 25 less than what 10 bit limit is)
This all sounds fun and easy and no....I am not using an arduino or FPGA or anything like that. I am a analog nerd fighting with audio stuff a lot and analog stuff in general, and I hav absolutely no interest in copying something from the internet that I wont understand at all.

SO the real problem is conversion of 10bit binary to BCD. I have found a IC (74184 and like that) which are long obsolete and unobtanium, so I guess a built up converter it is. I just dont know how at al. If anyone could somehow point me to something expandeable or draw up somethin for me with logic gates or TTLs or CMOS I would be gratefull. And please also keep in mind that I am a total noob when dealing with digital tech. Even the basic digital tech.

Thanks.

BBM.

I appreciate the OP not wanting to hear a "just use a microcontroller/FPGA/CPLD" answer but doing this using just an ADC and discrete logic ICs isn't very practical.  It is doable but would likely take 10 to 15 ICs to do.  Using a microcontroller, you can likely do it with just two ICs--the MCU and a precision voltage reference IC.

There is already a project on github that uses a MCU to emulate the CA3162.  (I'm guessing you have already come across it.)  And, no, even though that is an Arduino sketch, you wouldn't have to use an entire Arduino board to do what you need to do.  You could do the early development using an Arduino and then once you are happy with the firmware you could design your own circuit using just a MCU, a voltage reference, and a handful of passives.

Just to get an idea of what it would take to do this using discrete logic ICs, watch this YouTube video (for an 8-bit binary to BCD converter) and recognize that each of those C1, C2 .. Cx blocks would need to be separate discrete logic circuits.  You would need even more since you are converting a 10-bit value.  Note that the circuit shown is just the combinational logic part.  You would still need sequential logic to buffer the output, synchronize with the the ADC, multiplex the output, etc.

You could cheat a little and use EEPROMs as a look-up table for the combinational logic.  This would reduce the number of required components but at that point you may as well just use a CPLD or FPGA.

[SK_Caterpilar_SK]

edavid beat me to my suggestion. The TC7135, ICL7135, MAX7135, SI7135, and TC835 is a 4.5 digit converter with muxed BCD outputs. You would of course not use the outputs for the leading 1 digit, nor the least significant digit, but you might be able to find a way to make it work.  The TC14433 is a 3.5 digit similar device (+-1999 count).

Thank you very much I did not even know about the TC14433. It woul be the most likely candidate to replace the C520D (AD2020 doesnt matter one of those bilions of ICs). The signal it would recieve would always be positive, and the polarity indication is done separately on the analog side of the board.

[SK_Caterpilar_SK]

I appreciate the OP not wanting to hear a "just use a microcontroller/FPGA/CPLD" answer but doing this using just an ADC and discrete logic ICs isn't very practical.  It is doable but would likely take 10 to 15 ICs to do.  Using a microcontroller, you can likely do it with just two ICs--the MCU and a precision voltage reference IC.

There is already a project on github that uses a MCU to emulate the CA3162.  (I'm guessing you have already come across it.)  And, no, even though that is an Arduino sketch, you wouldn't have to use an entire Arduino board to do what you need to do.  You could do the early development using an Arduino and then once you are happy with the firmware you could design your own circuit using just a MCU, a voltage reference, and a handful of passives.

Just to get an idea of what it would take to do this using discrete logic ICs, watch this YouTube video (for an 8-bit binary to BCD converter) and recognize that each of those C1, C2 .. Cx blocks would need to be separate discrete logic circuits.  You would need even more since you are converting a 10-bit value.  Note that the circuit shown is just the combinational logic part.  You would still need sequential logic to buffer the output, synchronize with the the ADC, multiplex the output, etc.

You could cheat a little and use EEPROMs as a look-up table for the combinational logic.  This would reduce the number of required components but at that point you may as well just use a CPLD or FPGA.

I have not seen the arduino CA3162 until now but unfortunately it does not look like it will work with this perticular meter and I dont want it to be turned into a museum piece that will be turned on once a decade. I want it the most accurate I can get it and the mod in that post has stated that there are some accuracy issues.

I would happily do a 20 IC contraption no problems I would not mind at all and I would still prefer that over emulation or replacing the same IC as before. Practicality does not matter here, neither does size. I can still make it smaller because I can make 2 layer boards and compact the whole thing quite a lot.

[TerminalJack505]

If you are wanting to go the route of using discrete logic gates then one of the first things you will probably want to do is find some software that will allow you to create and simulate digital circuits.  All of the good software will automate circuit creation from Boolean functions and truth tables as well as let you draw them by hand.

I don't have any particular suggestions so far as this goes since the software I have used is either 1) not free, or 2) not simple (and several GB to install.)  Others here on the forum might have some suggestions.

[free_electron]

If you are wanting to go the route of using discrete logic gates then one of the first things you will probably want to do is find some software that will allow you to create and simulate digital circuits.  All of the good software will automate circuit creation from Boolean functions and truth tables as well as let you draw them by hand.

I don't have any particular suggestions so far as this goes since the software I have used is either 1) not free, or 2) not simple (and several GB to install.)  Others here on the forum might have some suggestions.

that's what i was proposing : get the altera software , draw schematic using the TTL library ( almost all 74xx series is in there ) , simulate it. when you get the logic working correctly : flash it in a simple CPLD and bob's your uncle.

[TerminalJack505]

The OP should check out the Logic Friday software that they are discussing over in this thread.  It won't do simulation but it will allow you to generate schematics from a truth table.