Chopper stabilised op-amps - is there a mystery to the noise

[Peter]

I am using the TLV2333 device; actually the cheaper Philips version of
it.

The specs, of course, are astounding. I go back to the 1970s
(precision photomultiplier power supplies, ppm specs) when the AD504
was the top precision op-amp. Then we had the OP27 etc. These are 100x
worse than these chopped ones, for DC work.

The application is measuring / amplifying 10x slow DC signals
(thermocouples etc) and feeding them to a 16-bit ADS1118 delta-signma
ADC.

The data sheet shows this thing chops at 125kHz or so, so AFAICT there
is no special noise below that frequency.

One Q is whether there are actually big spikes coming out of this
thing, at 125kHz, and the data sheet is avoiding saying anything.
Somehow I doubt it because it would be pretty useless. Most people are
not going to put an RC filter on the output.

One positive is that I have a 2.2uF ceramic cap on the input, which
prevents any fast signals (or fast noise) getting in.

I wonder if I am missing something... A google has a lot of app notes
(interestingly Analog Devices did some chopped op-amps which ran at
only 4kHz) but no concrete info.

 

[John Larkin]

On Fri, 27 Sep 2019 19:32:28 +0100, Peter <nospam@nospam9876.com>
wrote:

I am using the TLV2333 device; actually the cheaper Philips version of
it.

The specs, of course, are astounding. I go back to the 1970s
(precision photomultiplier power supplies, ppm specs) when the AD504
was the top precision op-amp. Then we had the OP27 etc. These are 100x
worse than these chopped ones, for DC work.

The application is measuring / amplifying 10x slow DC signals
(thermocouples etc) and feeding them to a 16-bit ADS1118 delta-signma
ADC.

The data sheet shows this thing chops at 125kHz or so, so AFAICT there
is no special noise below that frequency.

One Q is whether there are actually big spikes coming out of this
thing, at 125kHz, and the data sheet is avoiding saying anything.
Somehow I doubt it because it would be pretty useless. Most people are
not going to put an RC filter on the output.

One positive is that I have a 2.2uF ceramic cap on the input, which
prevents any fast signals (or fast noise) getting in.

I wonder if I am missing something... A google has a lot of app notes
(interestingly Analog Devices did some chopped op-amps which ran at
only 4kHz) but no concrete info.

Since the TI data sheet carefully avoids mentioning high frequency
noise, they probably have something to hide. But a delta-sigma ADC
mostly ignores noise, so you are probably OK. But an RC is cheap
insurance.

We use delta-sigma ADCs directly on thermocouples, no preamp.

 

[Phil Hobbs]

On 9/27/19 2:32 PM, Peter wrote:

I am using the TLV2333 device; actually the cheaper Philips version of
it.

The specs, of course, are astounding. I go back to the 1970s
(precision photomultiplier power supplies, ppm specs) when the AD504
was the top precision op-amp. Then we had the OP27 etc. These are 100x
worse than these chopped ones, for DC work.

The application is measuring / amplifying 10x slow DC signals
(thermocouples etc) and feeding them to a 16-bit ADS1118 delta-signma
ADC.

The data sheet shows this thing chops at 125kHz or so, so AFAICT there
is no special noise below that frequency.

One Q is whether there are actually big spikes coming out of this
thing, at 125kHz, and the data sheet is avoiding saying anything.
Somehow I doubt it because it would be pretty useless. Most people are
not going to put an RC filter on the output.

One positive is that I have a 2.2uF ceramic cap on the input, which
prevents any fast signals (or fast noise) getting in.

I wonder if I am missing something... A google has a lot of app notes
(interestingly Analog Devices did some chopped op-amps which ran at
only 4kHz) but no concrete info.

A bypass on the input is the ticket. Many chopamps produce nasty
current spikes out their inputs ("kickout" to old-timers).

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Hul Tytus]

John - what delta-sigma adc and what thermocouples do you use? That should
serve as a way around a collection of packages.

Hul

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

On Fri, 27 Sep 2019 19:32:28 +0100, Peter <nospam@nospam9876.com
wrote:

I am using the TLV2333 device; actually the cheaper Philips version of
it.

The specs, of course, are astounding. I go back to the 1970s
(precision photomultiplier power supplies, ppm specs) when the AD504
was the top precision op-amp. Then we had the OP27 etc. These are 100x
worse than these chopped ones, for DC work.

The application is measuring / amplifying 10x slow DC signals
(thermocouples etc) and feeding them to a 16-bit ADS1118 delta-signma
ADC.

The data sheet shows this thing chops at 125kHz or so, so AFAICT there
is no special noise below that frequency.

One Q is whether there are actually big spikes coming out of this
thing, at 125kHz, and the data sheet is avoiding saying anything.
Somehow I doubt it because it would be pretty useless. Most people are
not going to put an RC filter on the output.

One positive is that I have a 2.2uF ceramic cap on the input, which
prevents any fast signals (or fast noise) getting in.

I wonder if I am missing something... A google has a lot of app notes
(interestingly Analog Devices did some chopped op-amps which ran at
only 4kHz) but no concrete info.

Since the TI data sheet carefully avoids mentioning high frequency
noise, they probably have something to hide. But a delta-sigma ADC
mostly ignores noise, so you are probably OK. But an RC is cheap
insurance.

We use delta-sigma ADCs directly on thermocouples, no preamp.

 

On Sat, 28 Sep 2019 02:34:13 +0000 (UTC), Hul Tytus <ht@panix.com>
wrote:

John - what delta-sigma adc and what thermocouples do you use? That should
serve as a way around a collection of packages.

Hul

John Larkin <jlarkin@highland_atwork_technology.com> wrote:
On Fri, 27 Sep 2019 19:32:28 +0100, Peter <nospam@nospam9876.com
wrote:

I am using the TLV2333 device; actually the cheaper Philips version of
it.

The specs, of course, are astounding. I go back to the 1970s
(precision photomultiplier power supplies, ppm specs) when the AD504
was the top precision op-amp. Then we had the OP27 etc. These are 100x
worse than these chopped ones, for DC work.

The application is measuring / amplifying 10x slow DC signals
(thermocouples etc) and feeding them to a 16-bit ADS1118 delta-signma
ADC.

The data sheet shows this thing chops at 125kHz or so, so AFAICT there
is no special noise below that frequency.

One Q is whether there are actually big spikes coming out of this
thing, at 125kHz, and the data sheet is avoiding saying anything.
Somehow I doubt it because it would be pretty useless. Most people are
not going to put an RC filter on the output.

One positive is that I have a 2.2uF ceramic cap on the input, which
prevents any fast signals (or fast noise) getting in.

I wonder if I am missing something... A google has a lot of app notes
(interestingly Analog Devices did some chopped op-amps which ran at
only 4kHz) but no concrete info.

Since the TI data sheet carefully avoids mentioning high frequency
noise, they probably have something to hide. But a delta-sigma ADC
mostly ignores noise, so you are probably OK. But an RC is cheap
insurance.

We use delta-sigma ADCs directly on thermocouples, no preamp.

ADS1247/8 are nice parts. They have a front-end mux and a PGA with
gain from 1 to 128. 24 bits and a burnout current source, and a good
internal reference.

We short the inputs now and then to take out any input offset. You can
use the internal mux to do that. We don't use the complex internal
calibration stuff.

They can also do ratiometric resistance measurement to insane
accuracy. We generally use a platinum RTD as the reference junction
temp sensor. The ref junction sensor is usually the worst part of a
thermocouple system.



--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Peter]

John Larkin <jlarkin@highland_atwork_technology.com> wrote

>We use delta-sigma ADCs directly on thermocouples, no preamp.

The ADS1118 goes down to 200mV FS which is OK for the higher output
TCs but not for the low output ones used in hi temp apps. Well, it
will obviously work, but it won't be 16 bit More like 10-12 bit.

 

[Winfield Hill]

Peter wrote...

John Larkin <jlarkin@highland_atwork_technology.com> wrote

We use delta-sigma ADCs directly on thermocouples, no preamp.

The ADS1118 goes down to 200mV FS which is OK for the higher output
TCs but not for the low output ones used in hi temp apps. Well, it
will obviously work, but it won't be 16 bit More like 10-12 bit.

John uses ADS1247, which has a PGA gain of 128, 16x higher.
They claim a 48nV noise level at that gain. And he uses the
2nd channel to measure a shorted input for offset correction.


--
Thanks,
- Win

 

On 29 Sep 2019 07:00:29 -0700, Winfield Hill <winfieldhill@yahoo.com>
wrote:

Peter wrote...

John Larkin <jlarkin@highland_atwork_technology.com> wrote

We use delta-sigma ADCs directly on thermocouples, no preamp.

The ADS1118 goes down to 200mV FS which is OK for the higher output
TCs but not for the low output ones used in hi temp apps. Well, it
will obviously work, but it won't be 16 bit More like 10-12 bit.

John uses ADS1247, which has a PGA gain of 128, 16x higher.
They claim a 48nV noise level at that gain. And he uses the
2nd channel to measure a shorted input for offset correction.

We also use one per board to pick up platinum RTDs as the reference
junction temps. We put a good Susumu resistor in series with the RTD,
measure both voltages, and do the math.

We have a local RTD on board, and also allow the user to have his own
RTD in a remote reference junction box.

Here's one that we offer:

https://www.dropbox.com/s/5i3fph5n3ptsubq/J470.JPG?raw=1




--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Winfield Hill]

jlarkin@highlandsniptechnology.com wrote...

On 29 Sep 2019, Winfield Hill wrote:
Peter wrote...
John Larkin <jlarkin@highland_atwork_technology.com> wrote

We use delta-sigma ADCs directly on thermocouples, no preamp.

The ADS1118 goes down to 200mV FS which is OK for the higher output
TCs but not for the low output ones used in hi temp apps. Well, it
will obviously work, but it won't be 16 bit More like 10-12 bit.

John uses ADS1247, which has a PGA gain of 128, 16x higher.
They claim a 48nV noise level at that gain. And he uses the
2nd channel to measure a shorted input for offset correction.

We also use one per board to pick up platinum RTDs as the
reference junction temps. We put a good Susumu resistor in
series with the RTD, measure both voltages, and do the math.

We have a local RTD on board, and also allow the user to
have his own RTD in a remote reference junction box.

How do you deal with wire resistance for distant
remote boxes? Three-terminal or four? It'd be
interesting to see your wiring setup details,
with the ADC's multiplexing inputs.


--
Thanks,
- Win

 

On 29 Sep 2019 08:50:39 -0700, Winfield Hill <winfieldhill@yahoo.com>
wrote:

jlarkin@highlandsniptechnology.com wrote...

On 29 Sep 2019, Winfield Hill wrote:
Peter wrote...
John Larkin <jlarkin@highland_atwork_technology.com> wrote

We use delta-sigma ADCs directly on thermocouples, no preamp.

The ADS1118 goes down to 200mV FS which is OK for the higher output
TCs but not for the low output ones used in hi temp apps. Well, it
will obviously work, but it won't be 16 bit More like 10-12 bit.

John uses ADS1247, which has a PGA gain of 128, 16x higher.
They claim a 48nV noise level at that gain. And he uses the
2nd channel to measure a shorted input for offset correction.

We also use one per board to pick up platinum RTDs as the
reference junction temps. We put a good Susumu resistor in
series with the RTD, measure both voltages, and do the math.

We have a local RTD on board, and also allow the user to
have his own RTD in a remote reference junction box.

How do you deal with wire resistance for distant
remote boxes? Three-terminal or four? It'd be
interesting to see your wiring setup details,
with the ADC's multiplexing inputs.

4-wire of course.

Here's a schematic that I have handy. It combines BIST with acquiring
three RTDs.

https://www.dropbox.com/s/ttvw14wqgw55z22/P470_RTDs.jpg?raw=1

This is an old rev. We discovered that the AD7793 is flakey and
replaced it with a TI part.



--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Steve Wilson]

On Sunday, September 29, 2019 at 1:39:25 PM UTC-4, jla...@highlandsniptechnology.com wrote:

Here's a schematic that I have handy. It combines BIST with acquiring
three RTDs.

https://www.dropbox.com/s/ttvw14wqgw55z22/P470_RTDs.jpg?raw=1

This is an old rev. We discovered that the AD7793 is flakey and
replaced it with a TI part.

What is the TI part?

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

On Sun, 29 Sep 2019 17:25:22 -0700 (PDT), Steve Wilson
<9fe142ac@gmail.com> wrote:

On Sunday, September 29, 2019 at 1:39:25 PM UTC-4, jla...@highlandsniptechnology.com wrote:

Here's a schematic that I have handy. It combines BIST with acquiring
three RTDs.

https://www.dropbox.com/s/ttvw14wqgw55z22/P470_RTDs.jpg?raw=1

This is an old rev. We discovered that the AD7793 is flakey and
replaced it with a TI part.

What is the TI part?

John Larkin Highland Technology, Inc

lunatic fringe electronics

ADS1246 and ADS1247 in various places. Works great.



--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Steve Wilson]

On Sunday, September 29, 2019 at 9:03:31 PM UTC-4, jla...@highlandsniptechnology.com wrote:

This is an old rev. We discovered that the AD7793 is flakey and
replaced it with a TI part.

What is the TI part?

John Larkin Highland Technology, Inc

lunatic fringe electronics

ADS1246 and ADS1247 in various places. Works great.

Thanks. If you have the time, could you describe what made the AD7793 flakey?

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

On Sun, 29 Sep 2019 18:57:34 -0700 (PDT), Steve Wilson
<9fe142ac@gmail.com> wrote:

On Sunday, September 29, 2019 at 9:03:31 PM UTC-4, jla...@highlandsniptechnology.com wrote:
This is an old rev. We discovered that the AD7793 is flakey and
replaced it with a TI part.

What is the TI part?

John Larkin Highland Technology, Inc

lunatic fringe electronics

ADS1246 and ADS1247 in various places. Works great.

Thanks. If you have the time, could you describe what made the AD7793 flakey?

John Larkin Highland Technology, Inc

lunatic fringe electronics

It has in internal 24-bit shift register. But nothing initializes the
24-bit counter after powerup. De-asserting chip select doesn't
initialize it. So if you ever miss one count/clock, the data is
forever trashed.

And when it gets cold, it misses.



--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Steve Wilson]

On Sunday, September 29, 2019 at 10:14:45 PM UTC-4, jla...@highlandsniptechnology.com wrote:

It has in internal 24-bit shift register. But nothing initializes the
24-bit counter after powerup. De-asserting chip select doesn't
initialize it. So if you ever miss one count/clock, the data is
forever trashed.

And when it gets cold, it misses.

These are Major design flaws. So now we have to put any new device through major qualification testing!!

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

On Sun, 29 Sep 2019 20:37:34 -0700 (PDT), Steve Wilson
<9fe142ac@gmail.com> wrote:

On Sunday, September 29, 2019 at 10:14:45 PM UTC-4, jla...@highlandsniptechnology.com wrote:
It has in internal 24-bit shift register. But nothing initializes the
24-bit counter after powerup. De-asserting chip select doesn't
initialize it. So if you ever miss one count/clock, the data is
forever trashed.

And when it gets cold, it misses.

These are Major design flaws. So now we have to put any new device through major qualification testing!!

John Larkin Highland Technology, Inc

lunatic fringe electronics

ADI used to design great analog stuff with bizarre and broken SPI
interfaces. One serial ADC had a bit in a control register that
determined if the SPI interface accepted data on the rising or falling
edge of the SPI clock.

Reminds me of the old days in the USSR, when there were no road maps.
If you don't know where you are, you shouldn't be there.

Maybe they are better now. Both.





--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Tauno Voipio]

On 30.9.19 17:28, jlarkin@highlandsniptechnology.com wrote:

On Sun, 29 Sep 2019 20:37:34 -0700 (PDT), Steve Wilson
9fe142ac@gmail.com> wrote:

On Sunday, September 29, 2019 at 10:14:45 PM UTC-4, jla...@highlandsniptechnology.com wrote:
It has in internal 24-bit shift register. But nothing initializes the
24-bit counter after powerup. De-asserting chip select doesn't
initialize it. So if you ever miss one count/clock, the data is
forever trashed.

And when it gets cold, it misses.

These are Major design flaws. So now we have to put any new device through major qualification testing!!

John Larkin Highland Technology, Inc

lunatic fringe electronics

ADI used to design great analog stuff with bizarre and broken SPI
interfaces. One serial ADC had a bit in a control register that
determined if the SPI interface accepted data on the rising or falling
edge of the SPI clock.

Reminds me of the old days in the USSR, when there were no road maps.
If you don't know where you are, you shouldn't be there.

Maybe they are better now. Both.

GPS and Google Earth have spoiled the map-hiding game.

--

-TV

 

[Hul Tytus]

Thanks John. As you say, the local reference sensor is a problem. Might use the
adc to measue a 1n914 or similiarly heavily used diode and sort the diodes
as they come in the door.

Hul


jlarkin@highlandsniptechnology.com wrote:

On Sat, 28 Sep 2019 02:34:13 +0000 (UTC), Hul Tytus <ht@panix.com
wrote:

John - what delta-sigma adc and what thermocouples do you use? That should
serve as a way around a collection of packages.

Hul

John Larkin <jlarkin@highland_atwork_technology.com> wrote:
On Fri, 27 Sep 2019 19:32:28 +0100, Peter <nospam@nospam9876.com
wrote:

I am using the TLV2333 device; actually the cheaper Philips version of
it.

The specs, of course, are astounding. I go back to the 1970s
(precision photomultiplier power supplies, ppm specs) when the AD504
was the top precision op-amp. Then we had the OP27 etc. These are 100x
worse than these chopped ones, for DC work.

The application is measuring / amplifying 10x slow DC signals
(thermocouples etc) and feeding them to a 16-bit ADS1118 delta-signma
ADC.

The data sheet shows this thing chops at 125kHz or so, so AFAICT there
is no special noise below that frequency.

One Q is whether there are actually big spikes coming out of this
thing, at 125kHz, and the data sheet is avoiding saying anything.
Somehow I doubt it because it would be pretty useless. Most people are
not going to put an RC filter on the output.

One positive is that I have a 2.2uF ceramic cap on the input, which
prevents any fast signals (or fast noise) getting in.

I wonder if I am missing something... A google has a lot of app notes
(interestingly Analog Devices did some chopped op-amps which ran at
only 4kHz) but no concrete info.

Since the TI data sheet carefully avoids mentioning high frequency
noise, they probably have something to hide. But a delta-sigma ADC
mostly ignores noise, so you are probably OK. But an RC is cheap
insurance.

We use delta-sigma ADCs directly on thermocouples, no preamp.


ADS1247/8 are nice parts. They have a front-end mux and a PGA with
gain from 1 to 128. 24 bits and a burnout current source, and a good
internal reference.

We short the inputs now and then to take out any input offset. You can
use the internal mux to do that. We don't use the complex internal
calibration stuff.

They can also do ratiometric resistance measurement to insane
accuracy. We generally use a platinum RTD as the reference junction
temp sensor. The ref junction sensor is usually the worst part of a
thermocouple system.

--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[John Larkin]

On Thu, 3 Oct 2019 20:46:18 +0000 (UTC), Hul Tytus <ht@panix.com>
wrote:

Thanks John. As you say, the local reference sensor is a problem. Might use the
adc to measue a 1n914 or similiarly heavily used diode and sort the diodes
as they come in the door.

Hul

That sounds like a hassle. We use a surface-mount 1206 platinum RTD.
We put it in series with a good Susumu resistor and use the
delta-sigma ADC to measure the voltage across each. A bit of math gets
the RTD resistance, and a bit more gets the temperature. It's easy and
super accurate.

 

[John Larkin]

On 3 Oct 2019 17:52:40 -0700, Winfield Hill <winfieldhill@yahoo.com>
wrote:

John Larkin wrote...

We use a surface-mount 1206 platinum RTD. ...

Wow, sounds interesting, p/n?

VISHAY PTS1206M1B100RP100

About $2.