Noninverting op amp doesn't work on a highly resistive load

In summary, Michael's circuit doesn't work well with a load of 5000 Mega ohms. He is using an op amp with a gain of 22, which is not enough for the load. He is considering increasing the load resistance, but is worried about hitting the output rails.
  • #1
mlamont1
3
0
TL;DR Summary
My op amp setup does not amplify correctly to its full gain when I increase the load resistance to 5000 Mega ohms
Hi,

I am using an Apex PA443DF operational amplifier to drive a sinusoidal signal at 100 V amplitude across a very large resistor with resistance of about 5000 Mega ohms. The amplifiers are set up as noninverting with a gain of 22 and function very well for small resistive loads. The sinusoids being amplified are also very low frequency (approximately 1 Hz).

They work very well for a load of 3 Mega ohms, but decrease in performance I see measured on my oscilloscope to an amplitude of only about 0.1 V when the full 5000 Mega ohm load is put in place. I understand there would likely be some frequency dependent instability for capacitive loads, but I am only directly changing the resistance. I am a physicist rather than an electrical engineer, but I can't seem to think what would be the problem with increasing the load on the op amp, because less current would surely be required by ohm's law.

Any guidance you can offer would be greatly appreciated.Michael
 
Engineering news on Phys.org
  • #2
This sort of question is nearly impossible to answer. We don't know what your intended application is (i.e. load characteristics), we don't know your implementation (i.e. schematic et. al.), and we don't know how this was actually constructed.

"I built a circuit and it doesn't work, how do I fix it?" Doesn't inspire much motivation or confidence in us to answer.
 
  • Like
Likes berkeman
  • #3
New membrer has been reminded to post more information in their OP to help us help them.
There is no need to be so snarky. Here is a schematic of my circuit. The problem is that increasing the resistance of the load is causing the gain to fall to basically unity, not that the circuit does not function at all. It functions fine in exactly the same conditions with a smaller load. To test it, I am using an ohmite SM108035007FE 5000 Mohm resistor, so I believe there should be no capacitance or inductance, correct? Except perhaps some effect from the oscilloscope?
1596396571050.png

I left out RL (the resistor in parallel to the load) because it's function did not make sense to me.
 
  • #4
mlamont1 said:
View attachment 267201
I left out RL (the resistor in parallel to the load) because it's function did not make sense to me.

Hi! Would you be okay with showing which values you are using for each component?

There are a few words in the post that are having me feel a little bit twisted and I'd like to understand the problem better. I personally call the op-amp in that image as inverting, but you called it non-inverting in the OP; I also call ##R_L## the load resistance, and so when you say that you've increased the load resistance and removed ##R_L## I have a difficult time understanding which resistor you are talking about.
 
  • Like
Likes jasonRF and berkeman
  • #5
mlamont1 said:
There is no need to be so snarky.
He was NOT being snarky. He was pointing out that your question was incredibly ill-formulated and incomplete. One of the keys to getting a useful answer is to ask good questions.
 
  • Like
Likes berkeman
  • #6
mlamont1 said:
There is no need to be so snarky.
That's an inapproriate reply. Your original post (OP) was very incomplete. I clicked into your OP soon after you posted it, and had no way to intelligently respond. In the future, please post lots more information in you OP so we can help you. Thank you.
 
  • #7
mlamont1 said:
Here is a schematic of my circuit. The problem is that increasing the resistance of the load is causing the gain to fall to basically unity,
Have you checked the output voltage drive circuit compliance? My initial suspicion is that you are hitting the output rails as you increase ##R_L##
 
  • #8
The question is theoretically does anyone know of a situation where increasing the resistance of the load on an op amp would lower the gain. In theory, it should require less current.

I do not believe that more information is required to answer. If you do not have an idea for an answer, you could simply say "I don't know," or better yet say nothing at all.

There is no reason to disparage the question.
 
  • #9
mlamont1 said:
I do not believe that more information is required to answer. If you do not have an idea for an answer, you could simply say "I don't know," or better yet say nothing at all.
You didn't answer my question, which goes directly to your question...
berkeman said:
Have you checked the output voltage drive circuit compliance? My initial suspicion is that you are hitting the output rails as you increase ##R_L##
 
  • #10
mlamont1 said:
There is no need to be so snarky.
mlamont1 said:
There is no reason to disparage the question.
It's not such a bad question, it just doesn't include enough useful information to give a simple answer.

mlamont1 said:
The question is theoretically does anyone know of a situation where increasing the resistance of the load on an op amp would lower the gain.
Yes, I can think of a couple of scenarios that might look like this. But I don't know which one applies in this case.

In any case, you aren't inspiring me to get involved. Back in the day I was paid $120/hour do deal with this stuff. It's not that I'm not willing to help, but I don't want to give the wrong guidance or spend all day writing a treatise on all the possibilities.

Here are a few hypotheses:
1) It's oscillating.
2) You didn't actually build the prototype it correctly.
3) Issues related to the measurement apparatus (probes, etc.).
4) Some obscure footnote in the data sheet wasn't complied with.
 
  • #11
mlamont1 said:
There is no need to be so snarky. Here is a schematic of my circuit. The problem is that increasing the resistance of the load is causing the gain to fall to basically unity, not that the circuit does not function at all. It functions fine in exactly the same conditions with a smaller load. To test it, I am using an ohmite SM108035007FE 5000 Mohm resistor, so I believe there should be no capacitance or inductance, correct? Except perhaps some effect from the oscilloscope?
View attachment 267201
I left out RL (the resistor in parallel to the load) because it's function did not make sense to me.
If you ask for help, you should be more polite and humble even though you are a scientist.

You did not nearly give enough info even with this post. You have NO value on the components, how are we going to even think about the problem.

I designed with plenty of Apex HV opamps for spectrometer raster scan and all, I never see a situation where you raise the load resistance and the gain change. You do it right, it's just another opamp that can take high voltage.

You said you use PA443 in your first post, then you show the schematic of PA441 straight out of the datasheet. You don't show any value, the current limiting resistors Rcl. How do you want anyone to help you?

Without further info, my suspicion is you wired up bad. Something is wrong. Also this is not a slow amp, wiring and layout can be tricky. You might run into instability and cannot trust the output on the scope.

Be humble, fill in all the values in the schematic, take a picture of your circuit so we can look at how you build the circuit then we can think about it.
 
  • Like
Likes berkeman
  • #12
5000MΩ ?
 
  • Haha
Likes berkeman
  • #13
hutchphd said:
5000MΩ ?
Yeah, that's what I was thinking. It might be OK and I haven't worked with extreme high resistance circuits but I look at that and don't think it's significantly different than an open circuit. I certainly wouldn't be surprised if a PC board had less resistance between components.
 
Last edited:
  • Like
Likes berkeman
  • #14
Maybe scrupulously clean PTFE would be OK. Let the OP describe the circuit and we can assess...
 
  • #15
There is no logical reason the gain goes down when increasing load resistance. I used Apex amps, they are just like regular opamp. We drive open circuit also. Deflector plate in the spectrometers are DC open circuits. Just like any opamps, they are happy with no load.

Without knowing the component value, there might not be that critical to clean the board. Only the summing junction ( -ve input) is more sensitive, but dirt just cause offset as it creates a leakage path. Something is really wrong with his circuit.

If the Rf is say 20k, the gain setting resistor is about 1K. I doubt dirt will cause any problem as the resistance is low. We did a lot of electrometer amps, when the Rf is like 1Gohm or higher, then we worry about the cleanliness of the summing junction( -ve input),
 
  • Informative
Likes hutchphd
  • #16
I must admit I didn't bother to analyze the circuit...I just thought that was an odd choice for a test article.
 
  • #17
Even without Rload there is still Rf across the output, unless you have made a wiring error.

Check the voltage of the power supplies. What are those voltages?
Check the integrity of the power supply ground, and the link to that from the signal ground.
 
  • #18
mlamont1 said:
...5000 Mega ohms.
If that value is correct then I think you should check on some basic DIY ion chamber projects first before even attempting to ask for help.

Ps.: that resistance range requires very special knowledge and meticulousness so better not rely on general knowledge and assumptions. Ion chambers are the topic the closest to that range.
 
Last edited:
  • #19
Baluncore said:
Even without Rload there is still Rf across the output, unless you have made a wiring error.
At high frequencies, the feedback may not work, so the inverting input isn't necessarily at ground potential. The source of the input voltage could also contain an inductance, making high frequency noise or oscillations or capacitive coupling to any other part of the circuit possible.
 
  • #20
BTW, your thread title and this statement say that you are using a non-inverting opamp configuration:
mlamont1 said:
The amplifiers are set up as noninverting with a gain of 22
But the schematic you posted is of an inverting opamp circuit:

1596465686204.png
 
  • #21
I think OP is gone. Obviously he didn't want to give out his schematic. He got his answer, it's on him, not the PA441. Of cause, unless the amp is defective. Unless OP comes back, it's time to move on.
 
  • Like
Likes phinds and berkeman
  • #22
Just for fun speculation I personally think they were placing what they called the load resistance where ##R_{CL}## and was reducing the current limit.

I think because of the super low frequency only a few Hz the odds of it being a stability issue, although not completely out of the question, is a low chance. I think they would describe something different on the oscilloscope if it were stability.

I noticed the mistake with the "non-inverting" with the "inverting" schematic (post #4). After some thought even if they got it swapped up the magnitude of the gain would still be pretty close. Let's they were expecting the gain to be ##22## when they used "non-inverting" formula they looked up somewhere; this would be mean that the feedback resistance would ##21R_{I}## they would look at the oscillope and see their sinusoidal has been amplified a lot (almost 22 times their input) and say it's working without careful measurement or noticing that it has been inverted. They did say it "worked" with lower "load resistance."
 
  • #23
And I'm still wondering how he got a 5000Meg load while looking at it with a 10Meg 'scope probe! (or so he indicates in the Original Post)

Could he have confused the load resistor as the feedback resistor? That would explain a lot.

Guess we'll never know. :cry:
 
  • #24
Tom.G said:
And I'm still wondering how he got a 5000Meg load while looking at it with a 10Meg 'scope probe! (or so he indicates in the Original Post)

Could he have confused the load resistor as the feedback resistor? That would explain a lot.

Guess we'll never know. :cry:
This is a CMOS amp, input bias current is below 100pA typ. even using 5Gohm Rf is ok(only 0.5V offset X22 =11V, no big deal for HV amp easily adjust out). I've design electrometer amp using feedback resistor like this. They worked. This is not an issue.

you probe the output that is low impedance, it won't affect anything. I won't second guess him, if he cares enough, he can come back.
 
  • Informative
Likes phinds
  • #25
mlamont1 said:
Summary:: My op amp setup does not amplify correctly to its full gain when I increase the load resistance to 5000 Mega ohms

They work very well for a load of 3 Mega ohms, but decrease in performance I see measured on my oscilloscope to an amplitude of only about 0.1 V when the full 5000 Mega ohm load is put in place.

Just curious to ask, when the load resistance is very high (for example, 100 to 5000 megohms), how did you know the output signal state of the operational amplifier?

Because as far as I know, the input impedance of general-purpose multimeters and oscilloscopes is only 10 megaohms, so when you try to use these instruments to directly measure the output signal on the output port of the operational amplifier, the effective load resistance will be reduced to less than 10 megaohms.

🤔
 
Last edited:
  • #26
alan123hk said:
Just curious to ask, when the load resistance is very high (for example, 100 to 5000 megohms), how did you know the output signal state of the operational amplifier?
For both inverting and non-inverting amplifiers, the feedback resistor that sets the gain is also connected to the output.
 
  • Like
Likes alan123hk and hutchphd
  • #27
Baluncore said:
For both inverting and non-inverting amplifiers, the feedback resistor that sets the gain is also connected to the output.

This is a good discussion point, which also highlights that we need detailed information about the actual circuit when analyzing the circuit to find the problem.

If the resistance of the feedback network, which is taking effective output current from the operational amplifier, is much smaller than 5000Megohms, then it seems obviously incorrect and contradictory that the problem occurs when the load is close to 5000Megohms.

Anyway, the fault does not seem to be caused by too large load resistance, because the feedback network is also connected to the output, thereby reducing the actual effective load resistance and its rate of change, and in fact, due to the limitation of input bias current and parasitic capacitance, the resistance of feedback network cannot be too large.

I personally think that it is more likely to be the problem of high frequency oscillation. OP said that it is to amplify very low frequency signal (approximately 1 Hz ). Has OP tried to tune the oscilloscope to high frequency range to observe the output signal of the operational amplifier with gain bandwidth product of 10 Mhz ?

I am just worried that if the time base is set too long, some high frequency oscillations may not be displayed on the oscilloscope screen.

Of course, there may be many other different reasons for the failure. I really hope OP can come back and talk about the latest developments or discoveries.
 
Last edited:
  • #28
mlamont1 said:
Summary:: My op amp setup does not amplify correctly to its full gain when I increase the load resistance to 5000 Mega ohms

Hi,

I am using an Apex PA443DF operational amplifier to drive a sinusoidal signal at 100 V amplitude across a very large resistor with resistance of about 5000 Mega ohms. The amplifiers are set up as noninverting with a gain of 22 and function very well for small resistive loads. The sinusoids being amplified are also very low frequency (approximately 1 Hz).

They work very well for a load of 3 Mega ohms, but decrease in performance I see measured on my oscilloscope to an amplitude of only about 0.1 V when the full 5000 Mega ohm load is put in place. I understand there would likely be some frequency dependent instability for capacitive loads, but I am only directly changing the resistance. I am a physicist rather than an electrical engineer, but I can't seem to think what would be the problem with increasing the load on the op amp, because less current would surely be required by ohm's law.

Any guidance you can offer would be greatly appreciated.Michael
5000 megohms is an unheard-of resistor. I can't imagine electronics of any type working with this high a level of resistance.
 
  • #29
rude man said:
5000 megohms is an unheard-of resistor. I can't imagine electronics of any type working with this high a level of resistance.
I did, it's common with transimpedance amps for electrometers that measure pA or fA. There are 100G resistors on the market, we had those before, they are encapsulated in glass. It's a different mindset in design, how the pcb layout, special cleaning method for circuits. We used Freon ultrasonic bath to clean them to get rid of all the dirt that caused leakage.

Here are some 5G to 500G resistors, you might have to order them. 5G and 10G are in stock.

https://www.digikey.com/products/en/resistors/through-hole-resistors/53?k=resistors&k=&pkeyword=resistors&sv=0&pv2085=u5+GOhms&pv2085=u10+GOhms&pv2085=u15+GOhms&pv2085=u25+GOhms&pv2085=u30+GOhms&pv2085=u47+GOhms&pv2085=u100+GOhms&pv2085=u500+GOhms&sf=1&FV=-8%7C53&quantity=&ColumnSort=0&page=1&pageSize=25
 
Last edited:
  • Informative
Likes berkeman
  • #30
yungman said:
It's a different mindset in design, how the pcb layout, special cleaning method for circuits
Did you use FR4 for the boards?
 
  • #31
hutchphd said:
Did you use FR4 for the boards?
It's better to use teflon or other special material, technique using guard ring also common, BUT I actually cut groove around the critical junctions ( actually cut all the way through) on the pcb to avoid creepage. I have been using this technique for high impedance and high voltage isolation. There's similarity between the two, surface creepage is a very a important thing. This technique allow me to use cheaper material and not as critical in cleaning...I am born cheap,cheap is my middle name!
Reduce creepage.jpg

I used round can opamps at the time, I created footprints that had pins farther apart so I can cut groove around pin 2 ( this is standard pin out of opamps, 2 is -ve input, 3 is positive, 6 is output, 4 is -V, 7 is +V). I made the trace of the -ve input as short as possible as shown. I layout my own pcb, I don't want to hear complain from the pcb designers that it's too hard, too much trouble... From my experience with HV, high impedance and high speed RF, pcb layout is where you win or loss the war, I layout all my pcbs.

Some people use standoff for the -ve input of opamp too. I tend NOT to do that if I can help it. Cost money to custom assemble.
 
Last edited:
  • Like
Likes Rive, hutchphd and berkeman
  • #32
I am seeing many attempts to help answer a question that is simply bait.

The circuit given was from the application data without substantial detail.

And how is the output being used,, a "rad" meter?, a programable electrostatic device at a few hundred volts?, a smoke detector than uses a nominal 100-1000meg ohms.

And there may be confusion as to where the load resistor is being used as in the comment on not seeing RL as needed, (from the ap data circuit).

Many tried to explain that sometimes simply taking the measurement disrupts the measurement. Did someone say...

[Post edited slightly by a Mentor after a PM discussion]
 
Last edited by a moderator:

1. Why doesn't a noninverting op amp work on a highly resistive load?

A noninverting op amp relies on a feedback loop to maintain stability and accurate amplification. When a highly resistive load is connected, the feedback loop becomes unstable and the op amp may stop working altogether.

2. What is considered a highly resistive load for a noninverting op amp?

A highly resistive load is typically any load with a resistance greater than 10 times the input resistance of the op amp. This can vary depending on the specific op amp and circuit design.

3. How can I determine if my load is too resistive for a noninverting op amp?

You can calculate the input resistance of your op amp and compare it to the resistance of your load. If the load resistance is significantly higher, it may be too resistive for the op amp to handle.

4. Can I use a noninverting op amp with a highly resistive load if I increase the power supply voltage?

Increasing the power supply voltage may help in some cases, but it is not a guaranteed solution. The best option is to use a different op amp or circuit design that can handle the highly resistive load.

5. Are there any other issues that can arise from using a noninverting op amp with a highly resistive load?

In addition to instability and inaccurate amplification, using a noninverting op amp with a highly resistive load can also lead to increased noise and distortion in the output signal.

Similar threads

  • Electrical Engineering
Replies
6
Views
1K
  • Electrical Engineering
Replies
3
Views
745
  • Electrical Engineering
Replies
7
Views
1K
  • Electrical Engineering
Replies
12
Views
1K
  • Electrical Engineering
Replies
7
Views
3K
  • Electrical Engineering
Replies
9
Views
3K
  • Electrical Engineering
Replies
5
Views
1K
  • Electrical Engineering
Replies
11
Views
2K
  • Electrical Engineering
Replies
28
Views
3K
  • Electrical Engineering
Replies
8
Views
4K
Back
Top