Improve my ADC schematic

Thread Starter

CVMichael

Joined Aug 3, 2007
419
I am using an AD9059 to convert analog to digital.
Right now the input voltage is 2.5v, how to make it to be able to measure higher voltages?
I could add a mechanical dial with resistors that lower the voltage to 2.5v, but I was wondering if it's possible to do it digitally. like for example to change the input max voltage using a microcontroller? Let's say to make it measure up to 1000v ?

Here you can see the schematic of what I have right now:
http://www.libstock.com/projects/view/37/turn-your-logic-analyzer-into-an-oscilloscope-with-ad9059
 

colinb

Joined Jun 15, 2011
351
You could use a programmable gain amplifier (PGA) which is commonly used as a front-end to an ADC.

Alternatively, and possibly more cheaply, you could build a switched resistor voltage divider using MOSFETs. Discussion here.

You might want to switch op-amp amplifiers rather than resistor dividers if you want to extend your range to lower voltages with high precision.

Or, you could try to use a FET as a voltage-controlled resistor you can use an optoisolated switch such as a solid-state relay (SSR) to switch voltage dividers.">.

If you need to measure voltages higher than the MOSFET can handle, you can use an optoisolated switch such as a solid-state relay (SSR) to switch voltage dividers.
 

Thread Starter

CVMichael

Joined Aug 3, 2007
419
Can anyone give me a sample schematic for PGA or MOSFET?

I don't know much about either of them. I need a sample to start with.
 

colinb

Joined Jun 15, 2011
351
Can anyone give me a sample schematic for PGA or MOSFET?

I don't know much about either of them. I need a sample to start with.
Here's one example of a programmable-gain amplifier (PGA): the TI PGA112. Some ADCs have integrated PGAs, which is very convenient (space-saving, and it takes care of lots of messy layout details and you are assured of ADC performance with PGA included based on the ADC's data sheet specs).

For instance, the Microchip MCP3421 is a very accurate yet inexpensive ADC with integrated PGA supporting selectable gain of 1 up to 8 (it's accurate, but not high-speed).

Now, to measure voltages up to 1000 V, there are several concerns. First, THAT IS VERY DANGEROUS VOLTAGE. You should take care of isolating the measured voltage from the rest of your measurement device and the user. E.g., you should isolate the negative input lead from your device's ground. You could include an isolation op-amp or use optoisolation in the digital domain (SPI) with your ADC on an isolated power/ground net. I'll put the safety issues aside for a moment and move on to the basic measurement idea.

If you want to use a PGA to extend your range to higher voltages, what you really need to do is divide the input voltage down using a fixed resistor divider, so that the maximum input voltage is within the ADC measurement range. This might mean a 1/200 divider for 1000 V down to 5 V. Then you would be amplifying this divided voltage so that lower ranges achieve better precision. Overall, this would probably sacrifice accuracy of measurement for lower voltage measurements (e.g., if you are measuring a 200 mV source, it would still be divided by 200 and then scaled up by the PGA 128x so likely precision and accuracy is reduced).

As for the MOSFET-switched resistor divider, this would support you switching between several divider ratios so that you don't include a divider when measuring low voltage inputs. Instead of discrete MOSFETs, of which I think you would need at least two per range-divider, you could use a couple of analog multiplexer ICs. Yeah, the thread I linked seemed to have some old image links that no longer worked.

Here's an idea of what I imagined for a switched-divider design.
 

bertus

Joined Apr 5, 2008
20,070
Hello,

Your circuit will not work for those high voltages.
The 4066 will not be able to switch more than its powersupply, wich is 15 Volts max.

Bertus
 

colinb

Joined Jun 15, 2011
351
Hello,

Your circuit will not work for those high voltages.
The 4066 will not be able to switch more than its powersupply, wich is 15 Volts max.
Doh! Thanks, Bertus. I was originally sketching it with the resistor voltage dividers being switched in and out using MOSFETs and realized that you need more than one MOSFET per range, so I tried to simplify by using the analog switch IC.

Of course the Zener diodes should clamp the input voltage, but it's wrong to have them before the dividers.... :(

Obviously it is a complex problem to solve well. I am really curious how my Fluke multimeter works. It is quite an incredible achievement to get 4.5-digit accuracy when you actually consider all the sources of error!
 

Thread Starter

CVMichael

Joined Aug 3, 2007
419
I bought some small relays HY1E-5V (it's the only relay small enough that I could find in my local electronics store).
It's rated for 1A, 30V. It should work with higher voltages with low amps, right ?

Anyways, I made a schematic (see attached) first with a rotary switch, then below it the equivalent with relays (and other parts that I have available).

This is the table I used to get the resistor values:

#|Total KΩ|Res 1 KΩ|Res 2 KΩ|Ratio|Volts|Res Diff KΩ
1|2000|1990|10|0.0050|500|1000
2|1000|990|10|0.0100|250|400
3|600|590|10|0.0167|150|400
4|200|190|10|0.0500|50|100
5|100|90|10|0.1000|25|60
6|40|30|10|0.2500|10|20
7|20|10|10|0.5000|5|10

Do you think it will work ?
 

Attachments

colinb

Joined Jun 15, 2011
351
I bought some small relays HY1E-5V (it's the only relay small enough that I could find in my local electronics store).
It's rated for 1A, 30V. It should work with higher voltages with low amps, right ?
Don't count on it! I seriously doubt a 30 V rated relay will handle 1000 V safely.
 

SgtWookie

Joined Jul 17, 2007
22,201
CVMichael,
You should never exceed the voltage rating of a device, as the device may not be able to break the arc when the contacts separate.
 

Thread Starter

CVMichael

Joined Aug 3, 2007
419
I understand that if I switch 500volts directly with a 30v relay, it is dangerous, but the last relay "rel8" gets the voltage through an accumulated 1990k ohms resistors, and then goes to the relay. When contact is made, the voltage should be 2.5 volts because of the 10 k resistor "R10".
So, is it still dangerous?
 

colinb

Joined Jun 15, 2011
351
I understand that if I switch 500volts directly with a 30v relay, it is dangerous, but the last relay "rel8" gets the voltage through an accumulated 1990k ohms resistors, and then goes to the relay. When contact is made, the voltage should be 2.5 volts because of the 10 k resistor "R10".
So, is it still dangerous?
I think it is still a problem because the 500 V is still present across the relay contacts before the relay is closed. The resistor voltage divider you've made only divides the voltage once current is flowing through it. I'm sure SgtWookie has a more thorough answer, though.
 
Top