Infinite hold sample and hold circuit question

[eTech]

Hi

I need a circuit to store a voltage for an indefinite length of time.

The voltage is basically the dc voltage drop across a load resistor.

Some requirements:
1. The voltage range is 10mv to 2.0v dc and it will orginate via a closed
relay contact of a de-energized relay.
2. The relay is push button operated and normally de-energized.
3. The circuit will need to sample the voltage while the relay is de-energized,
then store (hold) the voltage when the relay is energized.
4. The circuit output (voltage sampled) should be uninterrupted during contact transition
from open to closed. or closed to open.
5. The circuit output will be read by a DPM.
(I know there are DPM's with display hold capability, but I don't want this to be a DPM requirement).

I don't have a circuit design yet but believe this requires an infinite hold sample-and-hold circut. Hoping there is an IC for this...

BTW...I'm not using Microcontroller, using cmos devices...

I'm looking for suggestions...
Any suggestions?

Thanks!

eT

 

[MikeMl]

S&H requires charging a capacitor to the voltage being sampled. The capacitor must have extremely low leakage in order to hold the voltage without sagging much. Also, the device reading the voltage must have a very-high (>20megΩ) input resistance (and low bias current) so as not to create an additional sag in the voltage. Maybe you can defer connecting the readout device until you are ready to read out.

It is all about the time-constant (RC), where R is capacitor leakage resistance in parallel with the measurement amplifier input resistance, and C is the capacitance. No sag is impossible. Do the math to see how much sag you can tolerate. 1% in 1sec? 1% in 100sec, etc?

If you really need infinite, then use an A/D and store the digital value.

 

[eTech]

S&H requires charging a capacitor to the voltage being sampled. The capacitor must have extremely low leakage in order to hold the voltage without sagging much. Also, the device reading the voltage must have a very-high (>20megΩ) input resistance (and low bias current) so as not to create an additional sag in the voltage. Maybe you can defer connecting the readout device until you are ready to read out.

It is all about the time-constant (RC), where R is capacitor leakage resistance in parallel with the measurement amplifier input resistance, and C is the capacitance. No sag is impossible. Do the math to see how much sag you can tolerate. 1% in 1sec? 1% in 100sec, etc?

If you really need infinite, then use an A/D and store the digital value.

Hi Mike

Thanks for your reply..

Yes...I really need infinite hold since the operator can energize the relay for an indefinite period of time.

What do you think of an AD7569? or Maxim DS4303?

 

[crutschow]

The AD7569 will likely do what you want but you probably need a microprocessor to control it, which is not necessarily a serious limitation.

You could also use a microprocessor with a built-in A/D and D/A converter.

 

[eTech]

The AD7569 will likely do what you want but you probably need a microprocessor to control it, which is not necessarily a serious limitation.

You could also use a microprocessor with a built-in A/D and D/A converter.

Hi c...

The datasheet indicates it has both ADC and DAC. There is a circuit shown as "Figure 29. Infinite Sample-and-Hold."
It looks like all it requires is a control pulse.

Although, after reading a little more, it looks like i'm going to need a timer or precise clock source to generate the pulse.

Opinion?

 

[crutschow]

The datasheet indicates it has both ADC and DAC. There is a circuit shown as "Figure 29. Infinite Sample-and-Hold."
It looks like all it requires is a control pulse.

Although, after reading a little more, it looks like i'm going to need a timer or precise clock source to generate the pulse.

The AD7569 Figure 29 circuit connection should do what you want. I believe you could use the PB signal that controls the relay (buffered to the correct voltage and polarity levels if needed). A logic-low input causes it to continuously sample and a logic-high input puts in the hold mode.

Note that the minimum 8-bit resolution for a full-scale analog input of 2.5V is about 10mV.

You will need a clock also. This can be generated internally as described on page 10 of the data sheet.

 

[MikeMl]

If eTech is willing to settle for 8bit resolution, note that 8bits is 1 in 256, meaning he should be willing to suffer a droop of 1/256, or 0.4%. That opens up the possibility of using an analog Sample and hold again...

Also, I don't understand the requirement that system shows you what the voltage was hours or days ago. Isn't it more important to know what the voltage is right now?

 

[ronsimpson]

Infinite hold sample and hold circuit question

who is going to be around at infinite to check if the voltage is still there?

If we are talking days or weeks (or even minutes) a digital approach sounds best.
example: SiLabs has 8051 microprocessors with 16 bit ADC and the memory is as good as your power source. IF the power is on to infinite then the micro will work for a very very (etc) long time.

 

[eTech]

Hi guys...

I know it really isn't an "infinite" hold, per se...But I dont know how long the operator will leave the relay in the hold mode..

I guess I could time it out and force the operator to press the button again...

But still, I think there's a good possibilty that it might be a hour before the relay is reset by the operator.
An example would be...if i'm testing a device under load, and want to monitor it's behavior over time...I might leave it operating
for a while...while monitoring it.

Seems like th ADC/DAC, while more complex, might be a more flexible way to go...

 

[ronsimpson]

You can use a 8 pin micro computer like the PIC12F609.
If the supply is 5V then the ADC will resolve down to 5mV. Use a 2.5V supply and it will work down to 2.5mV
Because you want to use a meter for output, the micro will use a software DAC to output a voltage.
50 cents. One IC.
I know this takes software and that is a problem.

 

[MrAl]

Hi,

There is nothing that can store a voltage forever except after converting it to another form, then back again when needed.
The easiest way is using an ADC to memory and then memory to DAC. Maybe there is a company that makes a device with both for easy use, or use a micro controller.
Other ways include magnetic tape recording.

 

[eTech]

You can use a 8 pin micro computer like the PIC12F609.
If the supply is 5V then the ADC will resolve down to 5mV. Use a 2.5V supply and it will work down to 2.5mV
Because you want to use a meter for output, the micro will use a software DAC to output a voltage.
50 cents. One IC.
I know this takes software and that is a problem.

Hi ron

I haven't developed for PIC before, but have written SW in multiple languages...
If I went this way, what would I need for development, in terms of HW/SW?

eT

 

[ronsimpson]

eT,

I an not in love with PICs but they will work well. If you are going to make millions then you need a 8 pin or 6 pin 30 cent part.
If you are going to make 10 unites the price is not important.
I have a PICKET3 programmer. (for PICs)
I would look for a low pin count development board. SparkFun.com has many different boards. https://www.sparkfun.com/categories/9
PIC has a free C compiler. I use PIC BASIC PRO. There are many other CPUs with free C and BASIC and programmers.
There are many single IC options many are under $2.00.

I have used "BASIC STAMPs". They do not use a compiler. No programmer. General they just need a notepad program and a serial port. The problem is the stamp cost from$10 to $100. If you build only two then this cost is better than getting a programmer.

I have some PCBs that I use in factory automation. LDC display (16x2), 20 pin PIC with 10 analog input and includes 6 key buttons and serial port.