DSO input stage question

Go To Last Post
26 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here is diagram of DSO input stage (PCS500 actually).

Anybody knows what R108, R119 and R120 are for? How were their values chosen?

Attachment(s): 

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

They look as they are there for attenuation and isolation, but I could be wrong.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

UNiXWHoRe wrote:
They look as they are there for attenuation and isolation, but I could be wrong.

How they supposed to work? Their resistance is quite small compared to following 100k resistor.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That 100K resistor (R30) only comes into play for very large signals to provide amplitude limiting.

I think you left out something important with that line that runs off the bottom of the page.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
That 100K resistor (R30) only comes into play for very large signals to provide amplitude limiting.

Yes. And this is why I dno't understand why one would need another 150 Ohm there. 100 Ohm deviation may be within tolerance of 100K resistor.

ka7ehk wrote:
I think you left out something important with that line that runs off the bottom of the page.

No. This line goes to another relay which is just grounds input.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

OK, then the real attenuator is R6, R12, R13.

The other resistors such as R89, R108, R119, and R120 are likely for transient response correction. It is hard to say, precisely, without the original designer's notes. There are components like these in many oscilloscope inputs and they tend to be there to fix something that the designer did not like. Only you cannot tell what it was that the designer did not like. And, they won't tell you because its a "trade secret" and I cannot tell just by looking at a schematic.

Speaking from experience as a former Tek oscilloscope designer.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
OK, then the real attenuator is R6, R12, R13.

Yes, they are making prefect 1MOhm input in any position with minimum components. It looks like it is the best design from what I've seen lately.

ka7ehk wrote:

Speaking from experience as a former Tek oscilloscope designer.

Then I'll just have to make a board and see what will happen. Thanks.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

alexru wrote:
Here is diagram of DSO input stage (PCS500 actually).

Anybody knows what R108, R119 and R120 are for? How were their values chosen?

The various low value resistors are there to control the pulse response and ringing due to wiring and component inductances.

They were probably arrived at by trial an error from the original prototype.

kevin

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here is second part of input stage that I designed to suit my needs.

ADC I am planning to use (AD9057-80) reqires 1Vp-p input signal centered at 2.5V and provides stable 2.5V reference.

On the diagram U2 is just buffer with high impedance input. And U1 is centering signal at 2.5V and divides it by 6 since transistors are linear only at +/- 3V region and I am planning to use this as full-scale value.

My questions are:
1. Is this correct way of doing things? I am not an analog electronics expert :)
2. Is it possible to simplify this schematic?
3. I want to measure voltages up to 75V, in ranges 0-3V, 3-7.5V, 7.5-15V, 15-30V and 30-75V (division by 1, 2.5, 5, 10 and 25). Current implementation would require to place 5 relays at input attenuator but it occurs to me that some divisions could be made after voltage follower. Will it be better to do so?

Thanks.

Attachment(s): 

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

One of the things you REALLY need to consider is dynamic range. Carefully check the max and minimum voltages at each point in your signal chain and make sure that nothing is saturated, cut-off, or exceeds swing limitations (especially op-amps).

You MIGHT consider eliminating the input FETs. There are lots of good FET input op-amps these days and they have as high an input resistance as a discrete FET when operated as a non-inverting buffer. That should give you a lot better stability, also.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Using the FETs and an extra OP amp as unity buffer is strange. Usually its FETs or OP-Amp as non inverting buffer.

The circuit is somwhat similar to the above, just for low frequencies only. At high frequencies the 100 K serie resistor is way to large. Thats why the have the capacitor in parallel in the first circuit.

The resistors from the original question are for high frequency components of all amplitude, not just high amplitudes.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Kleinstein wrote:
Using the FETs and an extra OP amp as unity buffer is strange. Usually its FETs or OP-Amp as non inverting buffer.

Yes, I'll use opamp.

Kleinstein wrote:

The circuit is somwhat similar to the above, just for low frequencies only.

The input stage will be as in first image. I just left it out from the second one for modeling.

Kleinstein wrote:

The resistors from the original question are for high frequency components of all amplitude, not just high amplitudes.

How they supposed to work?

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The resistors in the original question often compensate for board traces and such. They are very difficult to model because you often don't know what to model. As a result, they are mostly empirical.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
As a result, they are mostly empirical.

That's ok, I'll put placeholders on final board.

Engineers design scopes (and analog electronics in common) for decades already, there must be books or articles on the topic. Leaving this particular schematic, in theory, what kind of signal distortion could be corrected by such resistors?

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Tektronix had a little book on oscilloscope design. I think it was available on the open market but not through mass distribution. There might be some out there in the universe. It was, however, oriented toward design using CRTs for display, and analog, not digital.

It is not quite accurate to call it "signal distortion" that is corrected by these parts. Distortion usually implies non-linearity of the output voltage with respect to the input. Here, it the issue is more correction of the frequency response. And, on an oscilloscope, uncorrected frequency response shows up as step response over-shoot and ringing or non-gaussian rise and fall.

It is important to point out that the dielectrics used in the input capacitors, especially in the attenuator, ARE important. You get a phenomenon called "dielectric absorption" which is modeled as several RC series circuits in parallel with the primary capacitor. It can give you very strange step responses! If you use ceramic caps, use NPO or COG dielectric. If you use film caps DO NOT use polyethylene. There is a good article in Wikipedia that tells you more: http://en.wikipedia.org/wiki/Typ...

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

Tektronix had a little book on oscilloscope design. I think it was available on the open market but not through mass distribution. There might be some out there in the universe.

Part 1, which includes "INPUT CIRCUITS AND COMPENSATED ATTENUATORS":
http://radiomuseum.org/forum/tek...
http://www.radiomuseum.org/forum...

Also http://www.leftfield.org/~dd/con...

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Geeze! That is like passing through a time-warp. All those glass thingies that glow in the dark.

The good stuff starts on page 39 of the first volume. Lots of information, there. If you don't know about linear circuits, it could be a challenge, but there is enough there so you can get through it.

There is a challenge to using op-amps in oscilloscopes. The tendency is to switch or change the feedback resistor. But, you need to watch out for two things - signal (amplitude) swings at various points in your circuit at maximum and minimum gains when the trace is at the top or bottom of the screen. More insidious, however, is the fact that the -3db bandwidth changes with the gain if you do that. There is no easy way to compensate for this fact.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I continue to explore possible solutions and found another one.

Here V1 is input signal (I left all input attenuators for clarity).

R8 will be substituted by switched resistors to add few more attenuation steps.

Advantages of this implementation are:
1. No need for negative power supply.
2. Allows to have 2 channels with independent grounds.

What disadvantages it has?

Attachment(s): 

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What happens when they DO have different grounds? At some point, they are all referenced to the same point (the ground terminal of the 2.5V reference. So, you have one input common connected to measurement ground and a second somewhere else. You have TWO input amplifiers each with a negative rail that wants to be 2.5V above your amplifier "ground" . Those 50 ohm resistors will have a lot of current.

By "splitting" the supply, you also halve the dynamic range. Now, an op-amp output can only swing +/-2.5V from zero input level. That also halves your signal-noise ratio (with respect to internally generated noise).

I also do not understand why your first stage has a gain of two and the second, 1/2. With a dynamic range of +/-2.5 V, your input will only be able to handle +/-1.25V. That is assuming perfect rail-rail op-amps. And, with that max input, you will only have +/-1.25V at the output of the 2nd amp.

I would think a bit harder about your amplifier. You would not be happy with it.

Me, personally, I would never do that. A negative supply is just not worth the effort to eliminate. In my book.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have another question regarding input attenuators.
On the left image there is sinusoidal signal that swings full scale. +3.5v/-3.5v are limits of opamp linearity, signal must be considered clamped to this value. We could have scaled input signal to full +/- 3.5v but consider situation on the right image: offset on the scope is set that zero voltage is at the bottom of the screen and input signal is square wave which alternates 0 to 3.4v.

Since offset is applied after input buffer and limiter we must narrow dynamic range of the system.

Is this how all oscilloscopes work or there is different approach which I don't see? Is there any document that describe how oscilloscopes are expected to behave?

I've used oscilloscopes for years and never actually noticed how they behave. Well thought systems, just work as expected :)

Attachment(s): 

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Well, a big part is what you see on the display, If the signal limits beyond the viewing boundaries of the display, you don't care. You never want limiting behavior to be visible on the display.

So, I would define what the visible (voltage) range for the display and work forward toward the input.

I'll give some thought to your actual question.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Alexru,

I think your analysis applies to a classic, analog, CRT O'Scope.

But remember, with a DSO the input circuitry up to the ADC is totally separate from the display module. The input needs to scale and measure the input signal to match the ADC, and protect the circuitry from excessive input, (and ideally float). The output can be a thousand miles away...

The output is based upon readng the data values, while knowing the scaling parameters with which they were taken. One doesn't even have to "add a DC offset" to the signal, if one "windows" the display on the output range of interest. The "offset" becomes a virtual value to set the windowed output, what portion of "full scale" one actually wants to see, and where within the full range one is looking.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

DocJC wrote:
I think your analysis applies to a classic, analog, CRT O'Scope.

As far as I understand modern DSOs have coarse relay switched attenuator (1:1, 1:10, 1:100) and then (after buffer) fine electronically switched attenuator.

My concern is that if signal when centered at 0 will be near limits of current coarse divider, then when zero will be moved we'll need to enable next stage of coarse attenuation. This will cause change of parameters (accuracy etc).

I am not sure that it is correct to silently change critical parameters just by moving beam up and down.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The offset is just where 0V is displayed on the screen, it's handled after the signal is digitized and then gets rasterized onto the screen.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

jayjay1974 wrote:
The offset is just where 0V is displayed on the screen, it's handled after the signal is digitized and then gets rasterized onto the screen.

This is not true based on my experience with scopes.

I described situation with sine and pulse signals of the same amplitude (close to current display limits), but sinusoid is centered at 0 (so takes positive and negative values) and square swings from 0 to positive values only (twice amplitude of sine).

You should be able to see both waves on the same oscilloscope settings changing only vertical offset.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Agree wilth Alex on this one. You can do offset digitally IF you have a high resolution ADC. But, if you only have 8 or 10 bits to work with, it gives you a miserable waveform display.

Jim

 

Until Black Lives Matter, we do not have "All Lives Matter"!