Author Topic: ICL7135 ground loop  (Read 1625 times)

0 Members and 1 Guest are viewing this topic.

Offline davorinTopic starter

  • Supporter
  • ****
  • Posts: 939
  • Country: ch
ICL7135 ground loop
« on: June 19, 2020, 01:12:10 pm »
Good afternoon (o;

Almost finished a thermocouple design for reading out temperatures via GPIB and/or USB...

Now as I want to build linecards for a flexible measurement system a voltmeter is also a must..so that's where the ICL7135 comes in (o;

Let's say the ICL7135 is read out by a MCU which then sends data down to the host PC either through GPIB or USB....
Now when I measure a voltage from a device which is a USB device and therefore the ground is connected to the host PC as well as the measuring device...
datahseet mentions something about the floating inputs must go above +v - 0.5V and below -V + 1V....

Do I get false reading from the two floating analog inputs of the ICL7135?
Or better doing an isolated design like those good old HP 3456A voltmeters? (o;

 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: ICL7135 ground loop
« Reply #1 on: June 19, 2020, 01:47:27 pm »
When doing measurement and send to the PC, it is a good idea to have some  isolation. The usually easiest way is to have UART output from the µC , than some isolated coupler (opto or ADMU....) and than an UART to USB bridge chip. The USB can power the UART to USB chip easy.  For the rest of the circuit it may need an extra DCDC converter or an extra supply / wall wart.


The inputs of the ICL7135 are differential, but the common mode range is limited. It may work without isolation, but it can fail if there is a lot of common mode voltage, especially with a Laptop or similar that is not directly connected to ground.
 

Offline davorinTopic starter

  • Supporter
  • ****
  • Posts: 939
  • Country: ch
Re: ICL7135 ground loop
« Reply #2 on: June 19, 2020, 01:58:35 pm »
Ah okay...well I already looked up some isolated DC/DC converters from Digikey....not very expensive...
though not sure if the introduced ripple messes up with the results...but then again...the intended system is more for measuring several
DC voltages at once in the range of 100V - 400 (/me repairing old vacuum tube radio ;o).

For connecting the ICL7135 to the host interface I'll use a cheapo MCU with CAN and isolated CAN transceiver..probably NXP....
that way I can have several voltmeter cards in the same housing....or other cards doing temperature measurements....

I assume some protection diodes on the inputs might be needed as well to protect from too much floating away...


Well first I am curious how the USB K-thermocouple PCBs turned out which I should get on Monday from aisler.net....
« Last Edit: June 19, 2020, 02:06:39 pm by davorin »
 

Offline mikerj

  • Super Contributor
  • ***
  • Posts: 3382
  • Country: gb
Re: ICL7135 ground loop
« Reply #3 on: June 19, 2020, 04:23:33 pm »
For connecting the ICL7135 to the host interface I'll use a cheapo MCU with CAN and isolated CAN transceiver..probably NXP....
that way I can have several voltmeter cards in the same housing....or other cards doing temperature measurements....

The ICL7135 has mutliplexed BCD outputs which are rather annoying to interface to a micro, why wouldn't you choose an ADC with an SPI or I2C interface?
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: ICL7135 ground loop
« Reply #4 on: June 19, 2020, 04:31:29 pm »
The ICL7135 is nice for display oriented use, but a little awkward when using with a µC. With an µC I would consider a more binary ADC like MCP3421 (has a reference inside, I2C interface, internal gain, smaller and no bulky caps needed, however not so high in impedance).

With an DCDC converter the ripple is usually not the really bad part. This can be filtered relatively easy: DCDC -> filter -> LDO.  The nasty part is the common mode injected signal. Here filtering with a common mode choke is only partially effective, as the ideally is no low impedance path to ground.

The input would need a divider (for the high voltages) and some diodes for clamping. The lower voltage ranges may want an extra (possibly zero drift) amplifier, especially with the 7135.
 

Offline davorinTopic starter

  • Supporter
  • ****
  • Posts: 939
  • Country: ch
Re: ICL7135 ground loop
« Reply #5 on: June 19, 2020, 05:33:09 pm »
Hmm..interesting approach.....advantage is to have several ADCs in parallel...the ICL7135 has no tristate outputs...

Well I like the concept of dual slope as my old 3456A is using and I get a direct decimal output....
with regular ADC you get binary fractional resolution.....but then again...I want to log how the anode voltage varies...and not in 0.0001V steps (o;

Guess I order both parts for testing...the MCP3421 and the ICL7135....

 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: ICL7135 ground loop
« Reply #6 on: June 19, 2020, 06:10:00 pm »
The HP3456 is not using dual slope but multi-slope. In many respects this is a little in between the dual slope and the sigma delta ADC found in cheap high res ADC chips like MCP3421.
For the resolution the 7135 is "only" 4.5 digits,  and thus +-20000 counts, but the DNL is rather good. The MCP3421 is 18 Bits ( +- 130000 counts), but here the linearity is not so perfect - so the accuracy level is not that much higher as the 6 times higher resolution suggests.

The downsides of the 7135 are the interface, need for reference scaling (usually exactly 1 V),  the need for 2 relatively large external capacitors (some 1 µF MKS and some 200 nF PP) and for high accuracy some extra adjustment around zero. For the high voltage version the low input current is nice, so it can directly connect to a divider. For a thermocouple one may want extra amplification.

Even though it is possible to have several devices on an I2C bus, this can be tricky for the ADCs. This would link the channels to a limited common mode range and AFAIK there are only 3  I2C addresses to choose from.

With a binary output one can do the adjustment / cal factor in software instead of the trimmer usually used with the 7135. The MCP3421 is usually made with a clock suitable for 60 Hz mains - so the 50/100 Hz suppression is usually not that good.  One can get around it with using the ADC at 15 SPS and use software averaging over 3 or 6 readings - still the clock is not very accurate. With the 7135 hum suppression can be adjusted via the clock.
 

Offline mikerj

  • Super Contributor
  • ***
  • Posts: 3382
  • Country: gb
Re: ICL7135 ground loop
« Reply #7 on: June 22, 2020, 07:43:15 pm »

Well I like the concept of dual slope

You can get dual slope converters with a serial interface e.g. Microchip TC500 series
 

Offline davorinTopic starter

  • Supporter
  • ****
  • Posts: 939
  • Country: ch
Re: ICL7135 ground loop
« Reply #8 on: June 22, 2020, 07:47:27 pm »
Ah very interesting...thanks :-)

Found this weekend some other ADCs from Microchip...but not in production anymore....

 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: ICL7135 ground loop
« Reply #9 on: June 22, 2020, 08:15:46 pm »
For the 7135 one can use one output to do the counting in the µC and than only read the sign as an extra line. So one can use the 7135 with an µC, but it is no longer that attractive, as there are good alternative now.
 

Offline davorinTopic starter

  • Supporter
  • ****
  • Posts: 939
  • Country: ch
Re: ICL7135 ground loop
« Reply #10 on: June 22, 2020, 08:19:46 pm »
Well I just ordered some ICL7135 today @ Digikey for testing...but the TC500 series looks promising...just reading their appnote...

but then again...the linecard design features isolated CAN for communicating to the main controller (with GPIB/USB/Ethernet) and isolated power supply...
so no matter what the frontend chip is...it is (or should be ;o) easily replaceable...

 

Offline Dr. Frank

  • Super Contributor
  • ***
  • Posts: 2440
  • Country: de
Re: ICL7135 ground loop
« Reply #11 on: June 22, 2020, 10:08:59 pm »
I used the 7135 over 25 years ago, with a small 6502 and VIA chip, sharing 5V supply,  for different thermometer sensors, PT100, special T-diode, a AuFe/Ni thermocouple.. required a 7650 chopper amplifier for x100 gain for 1µV resolution.. as long as the sensor is isolated from the digital ground there's no problem with ground loops, worked very well.
I used the counting method, with a gate, very easy, as described in the datasheet, with IRQ and FP arithmetic, everything written in assembler.
I still have this PCB /module somewhere.

Frank
« Last Edit: June 22, 2020, 10:10:59 pm by Dr. Frank »
 

Offline bin_liu

  • Contributor
  • Posts: 38
  • Country: cn
Re: ICL7135 ground loop
« Reply #12 on: June 23, 2020, 01:09:18 am »
It is recommended to use digital isolation. The power supply can use a module like B0505S-1WR2, but I think the 7135 circuit is more complicated. There are several peripheral components that need special attention. I have about 10PCS for the 7135. It is not easy to consider using it in the future. It’s a collectible. ;D
 

Offline davorinTopic starter

  • Supporter
  • ****
  • Posts: 939
  • Country: ch
Re: ICL7135 ground loop
« Reply #13 on: June 23, 2020, 10:19:52 am »
Thought of using this one:

https://www.digikey.com/product-detail/en/murata-power-solutions-inc/NXJ1S0505MC-R7/811-3094-1-ND/5697921

SMT and 4.2kV isolation...and should arrive tomorrow for testing...

Yeah I know that the ICL family is kinda obsolete...but it fits to my other chips I am still keeping since 30 years (o;

The TC510 will be on my next digikey order (o;
 
The following users thanked this post: Dr. Frank

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 15157
  • Country: de
Re: ICL7135 ground loop
« Reply #14 on: June 23, 2020, 10:49:49 am »
In principle that power module looks good, but the ICL7135 also need some negative supply. So the more logical way would be a 12 V DCDC and than some local LDO / virtual ground to make the +-5 V.
 

Offline davorinTopic starter

  • Supporter
  • ****
  • Posts: 939
  • Country: ch
Re: ICL7135 ground loop
« Reply #15 on: June 23, 2020, 11:32:03 am »
Ah that DC/DC converter is for the TC510 design....

For the ICL7135 test setup I've already ordered the cheapo ICL7660 (o;

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf