Author Topic: Composite amplifier (low noise + chopper) thread (diff between ltc1052 & ltc1151  (Read 17315 times)

0 Members and 1 Guest are viewing this topic.

Offline Lunasix

  • Regular Contributor
  • *
  • Posts: 142
  • Country: fr
I can't really help but I used, many years ago, a similar circuit with LT1007 stabilized by a a LTC1052. There are small diferencies but principal is the same. Design was from somebody else and the board was not working. I made corrections to have it working fine. There was errors in resistor's values and chopper amplifier was unable to correct offset.
Try to open the offset loop to verify voltages values, and verify that inputs of LTC1052 are in a valid range.
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
I can't really help but I used, many years ago, a similar circuit with LT1007 stabilized by a a LTC1052. There are small diferencies but principal is the same. Design was from somebody else and the board was not working. I made corrections to have it working fine. There was errors in resistor's values and chopper amplifier was unable to correct offset.
Try to open the offset loop to verify voltages values, and verify that inputs of LTC1052 are in a valid range.

I saw that circuit in another app note. I believe it might be in the 1052 datasheet.
 

Offline Lunasix

  • Regular Contributor
  • *
  • Posts: 142
  • Country: fr
The board was designed in 1990 and 20 to 30 boards were produced, all working fine with the right values.
I will try later to scan the part of schematic.
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
well, hmm, i just noticed something. I was trying to make heads or tails out of the LT1028 schematic and I noticed that the internal resistors in the unit are both 130 ohms. He chose 130 ohms for one resistor (same as internal resistor) and then 68 for the other resistor.

The voltage divider in the lt1028 is



Is a 68 ohm resistor not throwing it out of balance? Then again I don't really know whats going on in there.
 

Offline Lunasix

  • Regular Contributor
  • *
  • Posts: 142
  • Country: fr
Have you tried to replace output of chopper amplifier with voltage generator and verified that low noise amplifier's offset react as wanted, for all equivalent chopper output swing, without any latch (I think I have seen this problem on my board when values were incorrect) ?
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
Have you tried to replace output of chopper amplifier with voltage generator and verified that low noise amplifier's offset react as wanted, for all equivalent chopper output swing, without any latch (I think I have seen this problem on my board when values were incorrect) ?
I will also try this. This is a good idea, I only hooked up a high precision power supply to the input of the op amp.
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
ah i think i have this fucking bastard licked finally.


I figured to replace the 68 ohm resistor with a 630 ohm resistor (*just as an experiment). As I see it, with 68 ohms, the chopper amplifier is pretty much fighting the power supply, since only 68 ohms separate it from the 15V PSU. I also put a trimpot on the 130 ohm resistor. I then turned the circuit on, the offset was approximately 6 milivolts, I adjusted this trimpot to the max (1k ohm) and it went down to about 1 milivolt. I then plugged in my precision psu and found a certain voltage to dial in, I was able to get the amplifier offset to be like, almost dead on to 1 uV (though it jittered some because its low level measurement with air currents n other bullshit going on)

I believe the datasheet meant to say a 1300 ohm and a 680 ohm. Having tiny ohm values does not make much sense to me, again, because the chopper is fighting the power supply, how could a op amp stand up to 68 ohms on a 15 volt rail? With this kind of impedance difference I believe the OPamp was trying to effect the entire 15V rail. I was clued in to this fact when I connected my precison power supply, which brought down the 15V rail to almost 0 volts as soon as it was connected, then I realized oh, you need some kind of impedance separating the two supplies.

Anyway, this thing is finally behaving, it hunts though, I believe that the 100nf resistor is actually too large, leading to some kind of over shoot, but I am not sure. There is a tupperware container over this device (with about 1cm gap, highly nonideal) but the unit is pretty steady with about 5uV offset voltage, it hunts from about -10 to +10 uV now, according to my keithley 199. I believe that this is because of temperature problems due to the disgusting status of this prototype (it looks like modern art) and leakages from flux (its covered in flux again).

I had enough of this shit on chirstmas but I think replacing the 100nF with a 10nF will allow it to compensate for errors faster and I suspect that a 1300 ohm is dead on, because 1kohm (the maximum of the cheap trimpot I installed on it) gets the offset to around 1mV, 300 ohms more and it might be like 200uV... I will also try to get rid of the 1kohm resistor seperating the 100nF (currently) capacitor from the chopper amplifier and I will try replacing the 15kohm that I installed with the datasheet 30kohm before I replace the 100nF with a 10nF.



 :phew: :phew: :phew: :phew: :phew:

 :phew: :phew: :phew: :phew: :phew:

 :phew: :phew: :phew: :phew: :phew:

 :phew: :phew: :phew: :phew: :phew:


finally some progress :clap:
I wish I realized earlier that 68 ohms does not make very much sense... unless I am missing something here........ I would love to be corrected. at least one good thing happened on this holiday.

these experiment really needs a shield box, it does not even like when I move around it......
« Last Edit: December 24, 2013, 04:55:50 pm by SArepairman »
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: nl
  • Please stop pushing bullshit...
Hi SArepairman,

If you look @ the datasheet of the LT1028 you can the schematic @ page 17.
The are two resistors to make it posible to adjust the offset of the LT1028 and these are R5 and R6, both 130 Ohms.

LT dit its best to make the offset as low as posible for you a user.
The schematic of EDN witch is a from a aplication note of LT, is a automatic trim pot :-)
The two resistors, 130 and 68 Ohm reduce the control range.

The resistance of 68 Ohm creates an imbalance and is being compensated by the LTC1052 via the 30K resistor.
It is pulls down current untill the difference between the two inputs of the LT1028 are the same.

It is only possible to test this circuit @ uV level if you ground the +input.
The 30K resistor should be enough to correct the offset of the LT1028.
Measure the output level of the LTC1052, that its whitin the ouput commonmode range.
If not, make the 30K resistor lower, do not go lower than 10K.

Be aware of the bias current problems with Low Noise ApAmps, the LT1028 has max 180nA bias current!!!
The LTC1052 is not there for fixing the bias current problems :-)

I hope this helps you...

Kind regarts,
Blackdog
Necessity is not an established fact, but an interpretation.
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
Hi SArepairman,

If you look @ the datasheet of the LT1028 you can the schematic @ page 17.
The are two resistors to make it posible to adjust the offset of the LT1028 and these are R5 and R6, both 130 Ohms.

LT dit its best to make the offset as low as posible for you a user.
The schematic of EDN witch is a from a aplication note of LT, is a automatic trim pot :-)
The two resistors, 130 and 68 Ohm reduce the control range.

The resistance of 68 Ohm creates an imbalance and is being compensated by the LTC1052 via the 30K resistor.
It is pulls down current untill the difference between the two inputs of the LT1028 are the same.

It is only possible to test this circuit @ uV level if you ground the +input.
The 30K resistor should be enough to correct the offset of the LT1028.
Measure the output level of the LTC1052, that its whitin the ouput commonmode range.
If not, make the 30K resistor lower, do not go lower than 10K.

Be aware of the bias current problems with Low Noise ApAmps, the LT1028 has max 180nA bias current!!!
The LTC1052 is not there for fixing the bias current problems :-)

I hope this helps you...

Kind regarts,
Blackdog


I think that LT is just bloody wrong with the 68 and 130 ohm resistors, the thing won't bloody work with values this low. And yes, I have the input grounded, that is how I have been testing this thing the whole time. I think someone just assumed that it would work better with 68 and 130 ohms (when they tried it with a higher value in the lab). The 68:130 ratio is right, but it just does not work. I did nothing with my circuit but replace the resistors with higher values and it magically started working. With the 68 and 130 ohms it made no difference if I connected the output of the chopper amplifier.
« Last Edit: December 24, 2013, 06:41:03 pm by SArepairman »
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: nl
  • Please stop pushing bullshit...
Hi SArepairman,

What is the offset voltage whitout the Chopper correction amp.
Is the LT1028 in spec?

Kind regarts,
Blackdog
Necessity is not an established fact, but an interpretation.
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
Hi SArepairman,

What is the offset voltage whitout the Chopper correction amp.
Is the LT1028 in spec?

Kind regarts,
Blackdog

yea the LT1028 is pretty much in spec, the offset is like a few milivolts with a gain of 100, I have like 6 LT1028's they all pretty much do the same thing (and I recieved them at different times, its not from the same batch). I can't tell you the specific offset without gain, but with gain its like 2-3 mV (without any offset resistors) , which makes sense given the max offset of 40uV * roughly a gain of 100. If its important to you I can make a bare bones offset test circuit, but I'd rather not bother with it now. I am using 1% tolerance resistors but  I don't remember if I used a 10 and a 1000 or a 10 and a 1100.
« Last Edit: December 24, 2013, 07:37:58 pm by SArepairman »
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
OK, I put a 10nF instead of the 100nF, I put a 10k instead of the 30k and a 1300 ohm instead of the 1kohm. The circuit works well now, it pretty much is focused around 3-8uV offset in its deadbugged and still unwashed state.

I found that it does not like to work with a 30k ohm on the output, if I put a 30k there, along with 680 and 1300 ohms the offset sets around 400uV, I believe that a 30K is not letting enough current through.

Do you think that better performance will be offered by increasing the 1300 and 680 ohm resistors by some multiplier in order to achieve the 30k ohm current/clock feedthrough resistor or to stick with the 1300 and 680 ohm resistor and use a 10-15K clockfeedthrough/current limiting resistor?


Another theory, other the LT engineers being careless, is that the LTC1052 has more current output capability then the LTC1151 due to its design, which is why it works with 68-130 ohms.
Do you guys think that maybe the 10x increased resistors on the offset pins degrades the performance of the LT1028 significantly? Perhaps I should just stick to the LTC1052, as it is a known working design. I don't really have the setup to measure the performance of this "hack" that I made..

blackdog, can you be more specific as to why the resistor should not be made lower then 10k? just wondering about your reasoning here.

« Last Edit: December 25, 2013, 08:57:18 am by SArepairman »
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: nl
  • Please stop pushing bullshit...
Hi SArepairman, :-)


The extra 130 Ohm and 68 Ohm on the outside of the LT1028 is to make the offset controle range smaller.
There is a deliberate imbalance between pin 1 and 8, the LTC1151 corrected this imbalance by looking at the voltage difference between the inputs of the LT1028.
Because you use 1300 and 680 Ohm, there is a smaler controle range for the opamp.
If you lower the 30K to 10K you wil get a bigger controle range, but also there is a change that the chopper frequentie wil be at the output of the LT1028.
You inject more and more noise out of the chopper opamp in to the LT1028 if you make the controle rage bigger.
I hope if you use a LTC1151 you connected the "spare" the wright way? output to - and + to ground?

OK,
If you use the LTC1151, remove the zeners from it's powersupply.
If you have a scope, use a probe to measure the DC level on the output of the LTC1151 (pin-7 or 1)
Optimum is the scoop line near the center of the screen (bottom is say -15V and top of the screen is +15V)
The scoop line wil move if you move above the circuit :-)
This level of measurements @ uV levels is a bitch :-)
Light, draft, moving measuring cables, mechanical stress all make you crazy  :-DD

You have to screen the circuit to make good measurements.
Change the 130 and 68 Ohm resistor so that the scope line stay's reasonable in the middle of the oscilloscope.
Maybe you need a 130 and a 100 Ohm resistor...
The trick is to find out how big the controle range must be for your circuit.
If you make the rage to big than there wil be some chopper noise on the output of the LT1028.

The output of the chopper opamp must not be near the powersupply voltage!

Be aware that the pins 1 and 8 are verry sensitive, see them as to extra inputs of the opamp, ofcourse with a lower gain than pin 2 and 3.

Ih hope this helps.

Kind regarts,
Blackdog
Necessity is not an established fact, but an interpretation.
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
Oh, your oscilloscope visualization method is wonderful, I was using a multimeter prior. I did not realize how narrow the range for success was. I guess there are two stability points, if one output is fixed at 130 ohms then you can either make the output greater or smaller, so the op amp can either source or sink current in order to stay within bounds, like the solution of a quadratic equation.

A ten turn 200 ohm trim pot cleared this issue up. I wish I realized the merit of using a scope in this situation, it would have saved me tons of fustration....

Howver, there is still a problem. With this resistor change, to 130 and 116, the offset seems to center at 20uV, while, with the 1300/600/10000 configuration, the offset settled at like 4uV.

How come the circuit is having troubles honing down to 0 now, I figured that it should also go down to 4, like the previous circuit.   :wtf:

I am not sure if this offset is because I am adjusting the potentiometer in such a way that works with the oscilloscope bias current injection but then steps out of bounds when the scope is disconnected, leading to the 20uV offset. I will wire in a electrometer in voltmeter mode to measure the voltage output of the chopper and adjust the trimpot then, as it has a very small bias current, much smaller then my other multimeters.
« Last Edit: December 25, 2013, 12:22:17 pm by SArepairman »
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: nl
  • Please stop pushing bullshit...
Hi SArepairman, :-)


Download the datasheet of the LTC2057 ( thats my Girl ;-) )
Start reading page 18 to 22 just some basic info how to build at micro signal levels.
Also in the datasheet of the LTC1051 from page 6.
If you don't comply, you burn your signal ash.

The 130 and 68 ohm resistors are to make a deliberate imbalance, and the chopper opamp compensates for this,
by removing too much current from the 68 Ohm resistor via the 30K resistor, so that a balance is created.

Maybe start, by connecting the 130 and 68 resistors, and then make the 30K resistor 3K or 3K3.
This must give you enough controle range, but don't forget the fist four lines in this replay  ;D

Succes!

Blackdog
Necessity is not an established fact, but an interpretation.
 

Offline SArepairmanTopic starter

  • Frequent Contributor
  • **
  • Posts: 885
  • Country: 00
  • wannabee bit hunter
I am just playing with it now, but I found out that using an electrometer to measure voltage on the chopper output is ideal. The reading is much much much more stable on the electrometer (while the multimeters cause the circuit to oscillate, at least some what).... good reason to have a nice cheapo 1970's electrometer in the lab.

Out of curiosity what kind of effect does an oscilloscope hookup have for a low level circuit like this?
I noticed that now even connecting a multimeter @ 10mega ohm to the output of the lt1028 causes a 10V shift (still in bounds because I zeroed the potentiometer nicely) while a >10 Gigaohm causes a 3V shift the output of the chopper (while it is ~0.3 volts output with no multimeter connected) in order to get the output offset ~5uV.


what a treacherous beast! luckily I have another electrometer in the attic....

i am going to add a device to the output of the LT1028 and see how things behave then....

at least its getting exciting.  :-+

and then I am sure it will behave totally differently when I add guard rings and dummy thermal loads on the real pcb....  :clap:



lmfao connecting an alligator clip to this thing causes it to go bonkers... an alligator clip not connected to anything...
this is not a circuit is a circus  :-DD i guess thermocouple...


i think that everyone interested in electrons should try to build some kind of circuit like this at least once in their life.....
« Last Edit: December 26, 2013, 08:09:27 am by SArepairman »
 

Offline blackdog

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: nl
  • Please stop pushing bullshit...
Hi SArepairman, :-)

Just some remarks...

Say a measuring kabel of less dan 1M is not just a cable.
Its a resistor ~ 60 mOhm
Its a inductor 1 a 2 uH
Its a capacitor 10 a 150pF (it is one site off a capacitor, thats wy its difficut to be precisely )
Its a antenna

Be always aware of this!

Older types of chopper opamps don't like capacitive load's, the modern types are a lot better, but not as good as say a OP07 and its family.
If you want to measure the outputpin of a chopper amp, first connect say a 1K resistor directly on the output pin and connect your multimeter on the other side.
Normaly this wil help to keep the chopper opamp stable.

If you use a scoop probe in the 1:10 Mode, most of the time its a ~15pF capasitor in parallel with a 10M resistor.
If you use a scoop probe in the 1:1 Mode its about 1M and 100 - 200pF <- this is a NO NO Mode!!!
You can youse the 1:1 mode only on verry slow circuits, say a output of an powersupply, there is already a big capacitor on the output.

Your composit amplifier is NOT a slow device...

Just some 2 cent of opinion  :-DD

Kind regarts
Blackdog
« Last Edit: December 26, 2013, 09:50:37 am by blackdog »
Necessity is not an established fact, but an interpretation.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf