Wideband 1Mohm to 50ohm unity gain Amplifier circuit

Am looking for a circuit to match a X10 Oscilloscope Probe with a 1Mohm input into a 50 Ohm Analyser, Scope, or similar, input; with a separate DC Power supply.

Gain ideally needs to be 0 dB, max output +10dBm/50R, with a bandwidth of 100Hz to above 100MHz (-3dB points) and able to be built to go inside of one of these BNC in-line BNC Module cases: https://uk.rs-online.com/web/p/bnc-connectors/0456201/.
Although RS list this as a Discontinued Product the PDF sheet is the correct item on the linked page.

thanks in advance

Mik

To match 1M ohms with 50 ohms with a passive network will mean a massive loss. You could use an active circuit (e.g. opamp in the box). Most RF probes aren't as high as 1M ohm. I have a couple of passive ones - one at 500 ohms, one at 1k - and an active one that is 100k I think.

Keith

Keith
thanks, have tried using an Anritsu MH648A pre-amplifier with an 'attenuated probe' to try to achieve a greater impedance and minimal circuit loading in front of other instruments.
Unfortunately this isn't too successful with the increased noise floor.

Hence asking about an active solution using 'Scope probes.

Do you have any ideas of suitable opamps or will a discrete option be simpler perhaps?

thanks
Mik

The specification isn't very clear. Normally we would want a buffer with 50 ohm output impedance for both-side impedance matching which involves a gain of 2. The output resistance has also the advantage of reducing the sensitivity to capacitive loads and can provide a ceratin degree of output overload protection.

Oscilloscope buffers usually won't be specified in dBm units. Do you mean 2 Vpp?

I agree that a fast OP is a simple solution. Most high speed OPs can easily drive +/- 20 mA.

Yup, should have better described what I'm looking for.

Really am looking for a very affordable, self build, sort of equivalent to an AGILENT 41802A www.home.agilent.com/upload/cmc_upload/All/EPSG087228.pdf with a greater 1dB Gain compression point.

Hope this better clarifies things
thanks
Mik

It is probably worthwhile looking at the selection guides from Texas Instruments, Analog Devices and Linear Technology. You need low bias current and low noise. I was thinking of the OPA2690 but I am not sure the input impedance is high enough.

Keith.

Keith
did find what looks like a single version of the OPA2690 you suggest.
This is the OPA690, until I saw that the input Impedance is too low by a factor of the order of 10.
The OPA2690 appears much the same.
Likewise for other 'possibles' which I looked at before starting this thread!

Thinking however that a suitable Fet might be able to impedance match the 1M ohm input impedance to one or other of these wideband Video opamps perhaps, given a few of these have reasonable stability at low gain

Mik

Mik

To refer to suitable FET OPs: For a lower supply/output voltage range, OP356 is fine, for higher range OPA656 or THS4631. Only the latter can achieve full 100 MHz power bandwidth, however.

For best pulse transmission and higher power bandwidth, I would probably accept the finite input impedance of a bipolar amplifier and adjust the input resistance respectively.

FvM said:

.......................
For best pulse transmission and higher power bandwidth, I would probably accept the finite input impedance of a bipolar amplifier and adjust the input resistance respectively.

Click to expand...

But then he couldn't use a standard 10:1 scope probe.

I built this amplifier for a similar purpose;

https://i42.tinypic.com/ofcyli.jpg

For 1 mV input it gives about -40 dBm into 50 ohms up to 450 MHz.

The response is quite flat except for a few "dips" at certain frequencies; I think they are caused by resonances in the housing I used but, since the device did what I wanted it to do, I haven't explored further.

It's not exactly what you specified but it might give you some ideas.

To get a smooth response and avoid oscillations at those frequencies you will need a very good layout. This means you must mount the amp on a board with a ground plane, use short connections, keep the input and output signals away from each other, and decouple the op amp directly from the power pins to the ground plane with surface mount ceramic caps.

Thanks
will have to look further into this.

Perhaps an FET input similar to krp's circuit but followed by a different Opamp.

Hmm, DS ground plane PCB with surface mount.
That brings back memories, haven't done anything like that for over 20 years

Mik

A bit like krp, I built a small oscilloscope amplifier many years ago to give me x10 and x100 gain as well as low noise. I cannot remember the bandwidth but I wanted low noise and gain so didn't have a x1 setting. I used a dual JFET source follower feeding an opamp. It worked well and I still use it occasionally. I will try to find the circuit on monday and see what the bandwidth could be at x1.

Keith

Any progress on circuit suggestions please?
Have tried a number of Searches on the 'net, but really found nothing suitable with good stability at unity gain

thanks in advance
Mik

Here's a FET input op amp with a 500MHz GBW that's unity gain stable. It has a maximum output current of 50mA so the maximum output voltage into 50Ω would be 2.5V. With that GBW you could go to a NI gain of 5 with a -3dB point of 100MHz.

Edit: Note at the end of the data sheet it says there are PCB evaluation boards available for that amp. I would strongly advise getting one of those since it will greatly simplify building a stable circuit.

crutschow
thanks, did look at the OPA656 earlier.
However the output of it didn't meet the specification I'm trying to achieve as it has +/- 3.5V into 100 Ohms specified, so +/- 1.75V into 50 Ohms with a series resistor of 50 Ohms.
However reducing my 'specification' is an option, but don't really want to until all avenues have been exhausted.

Do take your point on using Eval Boards/kits for one-off projects though, definitely can make life easier

Mik

The OPA656 is fine as an input amplifier, wired for non-inverting G = 2. But you'll need a serious output driver, such as the BUF634 with a 50-ohm series output resistor. Generally these are placed inside the op-amp's feedback loop, but for a full 100MHz bandwidth, it'd be safer to have it operate open-loop. They say its gain is 0.93 into 100 ohms, so you could setup the op-amp with G = 2.15. I'd run it from the op-amp's ±5V supplies, even tho that limits its output swing. You can only get just under 1V p-p at 100MHz anyway with the OPA656's limited 290V/us slew rate.

The BUF634 has up to 100mV of DC offset. If that's an issue there are several ways of dealing with it.

TI / Burr-Brown's OPA659 op-amp would be interesting to try. It has a 500MHz GBW at G=+2 into 100 ohms, and a 2500V/us slew rate. They show a nice smooth response (Fig 2, 2V/V) to past 200MHz without peaking. Since it has a JFET input stage there'd be no issue with 1.0M to ground for the input bias resistor. It has a 70mA output-current rating and looks like should be able to drive the 50-ohm load with a 50Ω source resistor. But if you expect to have much output swing you'd probably want to use the power-pad package version rather than the sot-23-5.

Win_Hill
thanks, have seen the BUF634 some time ago but hadn't thought of it as a VHF driver.

Anyway, have now thoroughly read the datasheet and am thinking that it might fit in place of the MAR-6 in a variation to krp's circuit linked in post #10. This then relaxes the +/- 5V constraint.
Whilst there will be a signal loss without the suggested gain from the OPA656, knowing the 'through gain/loss' can easily be added into any measurement result.

Mik

Although I hate to repeat things in a thread, I just want to remember that in contrast to OPA656 THS4631 does meet the power bandwidth requirements, as mentioned in post #8.

Sorry for the intrusion in this thread, but when I saw Win_Hill posting, I have to ask:

Will there ever be an update on "The Art of Electronics"?
https://www.amazon.com/The-Art-Electronics-Paul-Horowitz/dp/0521370957