Author Topic: Fast updating DVM module or chipset  (Read 545 times)

0 Members and 1 Guest are viewing this topic.

Offline snoozer

  • Contributor
  • Posts: 11
  • Country: ie
Fast updating DVM module or chipset
« on: August 22, 2019, 06:24:13 am »
Hello,

as I had mentioned before am looking into building a linear bench PSU. I have looked at a ton of designs and suggestions on the internet and have narrowed it down now to 2 or 3 designs that look plausible and well done to me as a novice. One of them made the effort to even build their own DVM module. The reason is that off the shelf DVM have a very low conversion rate. This design is using a CA3162 chip with a conversion rate of 96Hz. I have been looking for that chip at the usual sources but it seems nobody is stocking it. I have found many people seem to use the ICL7106 in their designs but apparently that reaches nowhere near the conversion rate and I don't even find it specified in the datasheet. Other chips I looked at either don't mention conversion rate or have something like 7 Hz. The same goes for DVM modules, even the more expensive modules in the 50-70 Euro range either don't mention conversion rate or have something low like 5-7 Hz. Now I have to admit by myself I would not have given this a thought but one of the designs had its own build DVM module because of this and the Author made the effort of pointing that out and adding the diagrams. Could anyone give advise, anyone came across fast DVM modules or chips similar to the CA3162 that are no obsolete. I would prefer an off the shelf DVM module, I would not consider myself experienced enough to do a DVM from scratch.

Kind Regards
Jan P.
 

Offline MosherIV

  • Super Contributor
  • ***
  • Posts: 1312
  • Country: gb
Re: Fast updating DVM module or chipset
« Reply #1 on: August 22, 2019, 06:38:34 am »
Hi.
You will not need a conversion rate of 96Hz!
If you are just looking at the dvm, you will not be able to see changes faster the around 15Hz, you can do low frame rate video at that rate. For looking at the voltage setting or output of a psu, it may flicker and be more anoying than useful.
The 5 to 7Hz sounds fine for dvm on a psu.
Even the fastest hand held dmm just about do 7Hz.
The only reason to go faster is for logging fast Voltage changes to computer, but now you looking at data loggers or oscilloscopes.
« Last Edit: August 22, 2019, 06:41:05 am by MosherIV »
 

Online maginnovision

  • Super Contributor
  • ***
  • Posts: 1360
  • Country: us
Re: Fast updating DVM module or chipset
« Reply #2 on: August 22, 2019, 06:49:39 am »
My 1242C and 1282A can both do 40 readings/s... But he's 100% right otherwise, even just a couple of updates a second is plenty fast for most cases.
 

Offline ledtester

  • Frequent Contributor
  • **
  • Posts: 535
  • Country: us
Re: Fast updating DVM module or chipset
« Reply #3 on: August 22, 2019, 06:58:39 am »
Having >15 updates/second can be useful. Here's a comparison of the Rigol and Korad power supplies:

(Comparison starts at the 6 minute mark.)

https://youtu.be/EpIxw90Oku8&t=6m
 

Offline MosherIV

  • Super Contributor
  • ***
  • Posts: 1312
  • Country: gb
Re: Fast updating DVM module or chipset
« Reply #4 on: August 22, 2019, 07:14:25 am »
Quote
Having >15 updates/second can be useful.
Ok, lets say it is down to personel preference then.
Personally, when setting up a psu, I do not find it useful if the 10mV digit is unreadable because it is flickering between 2 values.
In most cases, 10mV resolution on a psu is not used that often, it is only used when needing a precision Voltage maybe when you need a speific Voltage as an input to a circuit
If I am trying to do that, having it flicker to the point I cannot see it would piss me off  >:(
 

Offline snoozer

  • Contributor
  • Posts: 11
  • Country: ie
Re: Fast updating DVM module or chipset
« Reply #5 on: August 22, 2019, 07:19:54 am »
Hi,

Thanks for the input, this Forum is so quick, I love it. I will surely try some DVM modules that have a lower conversion rate, just from what I have seen on my small PSU with some no-name DVM I personally find it annoying not have a continuous up or down counting readout but rather "jumpy" slow update. In my further search I have just come across this website:

http://www.paulvdiyblogs.net/2016/09/building-6-digit-digital-milli-voltmeter.html

I have to read though and watch the videos to see how fast that is but I would not mind such a solution at all. Could maybe be adapted to a STM32F103 type MC if the Arduino type MC is to slow should that be the bottle neck.

Thanks
Jan P.
 

Offline Ian.M

  • Super Contributor
  • ***
  • Posts: 8185
Re: Fast updating DVM module or chipset
« Reply #6 on: August 22, 2019, 07:21:41 am »
The CA3162/CA3161 is a pretty low spec DVM chipset.  It only does three decimal digits, and requires manual zeroing and full range adjustment.  That's basically equivalent to a 10 bit ADC so just about any MCU with an ADC could offer comparable performance with sample rates of tens of KHz or even faster.

If you were to use an external 16 bit differential ADC, you could do four digit metering at a reasonably high update rate without much trouble.    It may be worth your while to look at the ADS1115,  which has four single ended or two differential input channels and a max conversion rate of 860 samples/s, so would be well suited to metering both voltage and current with an update rate up to around 400Hz.   Its also got a PGA with gains up to x16, so you may not even need a buffer amp for your current shunt if a burden voltage of 0.256V at full scale is acceptable.  As you can afford to throw away a bit of resolution a 100mV full scale shunt would be acceptable.  To make everything easy, Adafruit have a breakout board for it with libraries and sample code for Arduino and Raspberry Pi: https://www.adafruit.com/product/1085
so all you'd need to add would be an Arduino and a display.   If you display a bargraph, a higher update rate would be beneficial, as the human eye can easily perceive flickering in excess of 15Hz

The most difficult thing is probably arranging the logic supply biassing in relation to the output 0V terminal to keep the inputs from the current shunt and voltage potential divider within the acceptable input range for the chip.  If you want to add data-logging capabilities, you may need an optoisolated UART interface to avoid trouble with ground offsets and ground loops.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 6798
  • Country: de
Re: Fast updating DVM module or chipset
« Reply #7 on: August 22, 2019, 09:25:38 am »
With a normal numerical display 15 Hz update rate is a little fast to read. For the ICL 7106 and similar the rate depends on the clock used. The normal rates are some 2 - 5 conversions per second because this is good to read on an LCD. It could run up to some 15 SPS.

For just 3 digits like the old CA3162, one could use a µC internal ADC, though these may not have very good linearity. There are plenty of ADCs available to combine with a small µC, like the ADS1115 or MCP3421.
 

Offline ledtester

  • Frequent Contributor
  • **
  • Posts: 535
  • Country: us
Re: Fast updating DVM module or chipset
« Reply #8 on: August 22, 2019, 09:35:28 am »
I agree that 10mV resolution is sufficient for voltage, but 1 mA resolution for current is very helpful - especially for uC circuits.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 10486
  • Country: us
  • DavidH
Re: Fast updating DVM module or chipset
« Reply #9 on: August 22, 2019, 11:41:39 am »
Conversion speed is often tied to the power line frequency so that 50 or 60 Hz noise is suppressed.  This places an upper limit of 50 or 60 Hz for one full cycle or 10 Hz if both are to be rejected.  For a 3 digit converter however, this is not strictly necessary although it is still useful.

A cheap microcontroller with built in 12 bit ADC might achieve 10 bits of linearity for 3 digits of resolution.  Noise concerns will still require an integrated measurement but 50 or 60 Hz to reject 50 or 60 Hz noise still results in more than fast enough update rate for a human interface.  Alternatively, a cheap microcontroller and cheap external ADC could provide much higher resolution.  Some old timers here know how to implement a discrete ADC at these performance levels but the cost would be higher.

If i was doing it, I would use a cheap PIC or AVR with enough I/O lines to directly drive a multiplexed display and an external ADC like an LTC2463.
 

Online imo

  • Super Contributor
  • ***
  • Posts: 2381
  • Country: 00
Re: Fast updating DVM module or chipset
« Reply #10 on: August 22, 2019, 11:51:45 am »
I've been using the ADS1110, diff input, i2c, 6 pins sot32, 15+ meas/sec, 16bit, 2.048V 5ppm/C internal Vref, 1-8x PGA, mind they have got several hardwired i2c addresses, so do select different part numbers when hanging them several on a single i2c bus.
« Last Edit: August 22, 2019, 12:02:25 pm by imo »
 

Offline macboy

  • Super Contributor
  • ***
  • Posts: 2003
  • Country: ca
Re: Fast updating DVM module or chipset
« Reply #11 on: August 23, 2019, 03:32:28 pm »
Any microcontroller with a decent 12+ bit ADC can easily implement a voltmeter suitable for a bench supply type of application. Take a bunch of samples and average them (reducing noise and maybe increasing resolution by dithering), do offset and gain calculations, and display the result on the display device of your choice, anything from some multiplexed 7-seg LEDs, to a fancy color LCD or OLED depending on what cost and effort you want to invest.

The ICL7106 type devices are generally set up to sample 2 to 3 times per second. I find that to be way too slow for a bench supply. By tweaking a few components, you can easily get 10+ samples per second for a much faster reacting display, which makes adjusting a supply much easier.  You need to change the RC oscillator (reduce RC time constant) but also potentially reduce the integrating capacitor to prevent loss of accuracy (smaller integration time means you probably want a small capacitor, read the datasheet application notes). I have some commercially built supplies with these devices, and I always do these tweaks to improve usability.
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 4585
  • Country: ch
Re: Fast updating DVM module or chipset
« Reply #12 on: August 24, 2019, 10:58:35 pm »
Take a look at https://www.eevblog.com/forum/projects/ode-to-the-ina226-voltagecurrent-sensor/

In it, I discuss its use for precisely this use case: a responsive panel meter for a lab PSU. I think it’s an excellent, inexpensive solution that’s also incredibly flexible.
 

Offline snoozer

  • Contributor
  • Posts: 11
  • Country: ie
Re: Fast updating DVM module or chipset
« Reply #13 on: September 09, 2019, 09:31:07 am »
Hi,

I have been playing around for a while now with the ADS1115, both single ended and differential. The resolution is definitely not a concern at all and single ended it works quiet well on the breadboard. When I set it up as differential input I do have a lot of noise. A lot of the noise comes from the USB connection through the MC that I use. If I power the circuit from a linear PSU its a lot better but still not really usable. I have to dig a bit deeper and check how I need to filter the input for the ADC to get rid of as much noise as possible. I will be using the STM32F103C.... "Blackpill" which works with the Arduino bootloader and has a working USB-port unlike the "Bluepull". The Arduino Nano does not give me the smooth readout on the OLED I am using and at the same time do temperature control for a fan and also temp readout, its just to slow. I have 2 other MCs ordered, the ordinary Arduino Mega 2..something and a Arduino Due which is a 32bit MC as well apparently. Have not received them yet so no clue about them so far. So what I am looking into now is the filtering of the ADC input and also how I can isolate the entire ADS1115 from the MC (I2C over optocoupler). In the end "Product" the digital circuits will have an independent PSU from a separate transformer winding or a 2nd transformer with a fixed voltage regulator either 5V or 3.3V depending on the MC I end up using. Also I decided I go with a 10mV resolution on the voltage readout, I have not yet even looked into the current side of things yet.

Regards
Jan
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf