Has anyone done a 10x scope amplifier?

[MrAl]

Hello there,

Do you know if anyone has done a 10x amplifier for a scope with bandwidth around 100MHz?

Alternately, does anyone want to design one maybe a group project?

[Mechatrommer]

sorry for the quick rough "Dave" CAD.. its just a matter of finding the right opamp...

[David Hess]

Hello there,

Do you know if anyone has done a 10x amplifier for a scope with bandwidth around 100MHz?

Alternately, does anyone want to design one maybe a group project?

What is the input impedance?  Are you intended to use this between a probe and the oscilloscope input?

Oscilloscope amplifier stages are typically differential transconductance amplifiers after the high impedance buffer and singled ended to differential conversion, which is where the offset or position signal is added.

An external x10 amplifier for a typical oscilloscope will require the high impedance buffer and controlled input capacitance.  There are some fast operational amplifiers now with a low enough input bias current to do this, barely.  I might use an LTC6269 dual with each amplifier configured for a gain of 3.2 and cascaded to make 10 and drive the oscilloscope input directly.  If the oscilloscope input is 50 ohms and a double termination is used, then the gain needs to be doubled to 20 and a higher power amplifier added, which suggests using a LTC6268 as a high impedance buffer, followed by the current feedback amplifier of  your choice providing the gain and high output current capability.

[Andreas]

Do you know if anyone has done a 10x amplifier for a scope with bandwidth around 100MHz?

Hello,

what is the use-case for that?

And why do you think you can do a amplifier better than that what is already in your scope.
i.e. Without increasing the noise by the same factor as the amplification?

Noise is mainly definded by bandwidth and input impedance.
So there is not much improvement possible without reducing one of these parameters.

And yes I have made a 10000 fold amplifier for a scope: but with only 10 Hz bandwidth.

with best regards

Andreas

[Karel]

Sacrificing bandwidth for gain and lower noise is a valid reason for using a preamplifier.
If the application is for audio (up to 20KHz) then I would recommend to use a pc soundcard with
appropriate software instead of a real scope.

[PartialDischarge]

The use of such amplifier would be interesting for current shunts, where amplification (10x or more) is needed and bandwidth is also appreciated. Offset and gain accuracy would be the difficult tradeoff of it.

[mk_]

The use of such amplifier would be interesting for current shunts, where amplification (10x or more) is needed and bandwidth is also appreciated. Offset and gain accuracy would be the difficult tradeoff of it.

Usualy the parasitics of current shunts limit bandwith much more then an amplifier do..

[2N3055]

The use of such amplifier would be interesting for current shunts, where amplification (10x or more) is needed and bandwidth is also appreciated. Offset and gain accuracy would be the difficult tradeoff of it.

Usualy the parasitics of current shunts limit bandwith much more then an amplifier do..

True. And also current shunt amp has to be differential and sometimes has to have rather high common voltage specs (depending what you measure). Scopes are grounded instruments.

[blackdog]

Hi,

Maybe you guys could take a look at this.

Input opamp, nice model from TI, the OPA653.
This is a very fast 2x opamp with a Fet input, so the gain is fixed at 2x.
Note, the supply voltage is only 12V in total!

Then behind this a 2x a THS3091/95 to amplify 10x.
1x a THS3091/95 just doesn't make it flat to 100MHz.

Follow all the examples, tips in the datasheet of these ICs.
These ICs are very broadband, breadboard testing is not possible if you want to keep it stable and also have a good pulse response.

On the output of the last IC is for example 2x a 100 Ohm resistor in parallel to make the 50 Ohm output impedance so that if this amplifier is terminated with 50 Ohm the gain is 10x.

Kind regards,
Bram

[macaba]

Quite a challenge, it's not as simple as a typical op-amp circuit with all the parasitics & careful PCB layout involved.

I have attached a potential circuit (image + sim) as a starting point, that I anticipate being a solder-in probe, with the 2k input resistor as close to the solder point as possible (maybe even inline resistor on the wire between probe & DUT).

[macaba]

Another topology for 200MHz

[MrAl]

Hello again,

Thanks for all the replies i have to do some reading up on this it's been a while for me.

But what is all this "0.3db" i am seeing at the start of the response curves?

The input impedance into the scope would be 1Megohm, so it would just have to drive that.
The input impedance into the circuit should be 1Meg also but if less it might be ok too.

The main idea here is that a lot of the scopes these days only go down to 10mv and i have had applications where i had to go down lower.  The idea to use it with a current shunt was a good idea too.

[Mechatrommer]

Ensure flatness within 0.3dB...

[MrAl]

Ensure flatness within 0.3dB...

Ok so that's still at a gain of 10 then right?

[tooki]

Yeah, but none of the gain curves are up at +0.3dB. The dotted curve is the phase.

[blackdog]

Hi macaba, 

Do you really think you can load a high bandwidth opamp with 120pF?

Don't trust Spice so much, it often lies to you.
120pF at 100MHz gives a large in fae shifted load of around 13 ohms.
Very few opamps can handle that at 100MHz.

So, you need as I stated in my previous post, a real impedance (resistiv) at the output of the opamp.

This 50 Ohm resistor is often made by connecting 2x 100 Ohm in parallel, this also helps with paracitic inductance.

And yes, this is also why you need 20x gain, this is because you lose 6dB of gain due to the termination of the cable by the 50 Ohm termination impedance of the scope or a BNC terminator.

Kind regards,
Bram

[MrAl]

Just one question, who sells the LTC6268 right now?

[blackdog]

Hi  mrAI,

You need 26dB of gain because you need to terminate the cable with 50 Ohms, if you don't do this, you will never have a good pulse response!

The LTC6268 is probably going to give you +-2dB at 10MHz if you build it very neatly and certainly not 100MHz as you said in your starting post.

Look in the datasheet for the LTC6268 page-8, top row last picture.
That shows the openloop gain and the phase margin of the opamp.
Also look at the 1K and 10pF conditions.

If you are a company then you can request two samples from Analog Devices.

Kind regards,
Bram

[tooki]

Blackdog: Watch out, the simulated circuits shown do NOT show the LTC6268, but the LTC6268-10, which is a different part (separate datasheet and everything) with much higher bandwidth.

[blackdog]

Hi tooki,

You are correct, my bad...

Ok, the LTC6268-10 will probably be able to achieve +-1dB with proper construction and trimming.

But again, look at page-16 on "Driving Capacitice Loads".
Keep the load below 5pF, so good pulse response will also require looking very carefully at the PCB design.

Also, this opamp is sensitive to overload at the input, so there really needs to be additional components there to protect the opamp.

The reason you see so few schematics of this type of preamp is that it is quite complex.

With the modern opamps a very large bandwidth is possible, but as soon as you start using something like a Scoop preamp or other device with a large bandwidth, the protections at the input and output play an increasingly important role and requires a lot of knowledge of the whole.

But I don't want to stop anyone's development of a scope preamp, it's a very good way to gain insight of e.g. paracitic properties in electronic circuits. 

Kind regards,
Bram

[PartialDischarge]

The use of such amplifier would be interesting for current shunts, where amplification (10x or more) is needed and bandwidth is also appreciated. Offset and gain accuracy would be the difficult tradeoff of it.

Usualy the parasitics of current shunts limit bandwith much more then an amplifier do..

True. And also current shunt amp has to be differential and sometimes has to have rather high common voltage specs (depending what you measure). Scopes are grounded instruments.

The bandwidth argument may be valid or not as there are coaxial current shunts with BW greater than that of current clamp probes. And current shunts are very cheap so they can be changed for a different current range. Still whatever the BW for the particular application, they need amplification.

Isolation is where the big problem lies and the reason why current shunts are used in designs with instrumentation op-amps but not for current measurements during development time, as they cannot be connected directly to scopes. But there are also solutions to this.

[Mechatrommer]

With the modern opamps a very large bandwidth is possible

the problem is, noise will also go with it... larger... the less noisier opamps are usually those with less BW aimed at lowest noise app. so this is a dilemma, if not, this thread would have been concluded much earlier.

[macaba]

Another circuit with OPA858.

[David Hess]

And why do you think you can do a amplifier better than that what is already in your scope.
i.e. Without increasing the noise by the same factor as the amplification?

Noise is mainly definded by bandwidth and input impedance.
So there is not much improvement possible without reducing one of these parameters.

Noise also depends on the transistor geometry so a transistor with a higher bandwidth, which depends on transconductance and capacitance, has higher noise.  Adding a 100 MHz preamplifier to a 500 MHz oscilloscope can lower the noise up to 100 MHz.

This is apparent in old modular oscilloscopes like the Tektronix mainframe models where the front end amplifier is replaceable.  Lower bandwidth front ends have lower noise than higher bandwidth front ends even when limited to the same bandwidth because the input device for the lower bandwidth front end has lower inherent noise.

[MrAl]

After looking into this a bit more it seems like it will be a lot of work to build one up and i cant find that op amp for sale anywhere.  Maybe i should be happy enough to have one with a 10MHz bandwidth.