EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: Someone on May 11, 2016, 09:35:52 am

Title: DIY 100MHz differential probe
Post by: Someone on May 11, 2016, 09:35:52 am
As previously promised attached is a (now tested and working) design for a practical differential probe. The input impedance is 100k Ohm in parallel with around 1pF single ended, and is gained for a simple 1:1 ratio into 50 Ohm termination. The interface is Agilent/Keysight auto probe over a display port cable.

To do, for those interested:
Create a layout for Tektronix probe interface
Implement the DC servo through the high speed amp
Come up with alternative values for a low noise, low speed, design not using the high speed amplifier
Title: Re: DIY 100MHz differential probe
Post by: tggzzz on May 11, 2016, 09:43:36 am
What's the predicted performance and limits?

What are your tests and the test results?
Title: Re: DIY 100MHz differential probe
Post by: Someone on May 11, 2016, 09:50:33 am
It works, and doesn't simulate well so performance limits with these or other OPAMPs are hard to define. Real performance is adequate with some slight mid range gain wobble less than +/-0.5dB and CMRR falls off beyond 10MHz. I've not got any appropriate gear on hand to test it all the way out to its corner frequency so I won't make any wild claims about it.
Title: Re: DIY 100MHz differential probe
Post by: Kalvin on May 11, 2016, 10:08:15 am
Nice project!  :-+

Out of curiosity, would it improve the high frequency performance if the high frequency path is made more symmetrical ie. adding an identical THS3062 to the path going to the U2A non-inverting input?
Title: Re: DIY 100MHz differential probe
Post by: Someone on May 11, 2016, 10:42:10 am
Out of curiosity, would it improve the high frequency performance if the high frequency path is made more symmetrical ie. adding an identical THS3062 to the path going to the U2A non-inverting input?
Probably, but you'd add noise and power consumption. Engineering trade-offs abound.
Title: Re: DIY 100MHz differential probe
Post by: JPortici on May 11, 2016, 11:52:18 am
Nice. Something i was really looking forward to build

Noob question: What if i wanted to use it on a scope without auto-probe?
Title: Re: DIY 100MHz differential probe
Post by: moffy on May 13, 2016, 12:08:48 am
Your trim pot for the HF CMRR should be replaced with a select on test resistor. Pots are not great at better than 1MHz as they can have quite a complex equivalent model.
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 25, 2016, 10:31:51 am
Using both an OPA2192 and THS3062
Gain 1x, >100MHz bandwidth, 1mVrms noise, +/-10V differential input, +/-100V common mode.

High gain using only OPA2192
Gain 100x, 500kHz bandwidth, 14uVrms noise, +/-0.1V differential input, +/-20V common mode.

For a cheap experiment they're good value, the $15 of opamps is swamped by the cost of the display port cable but it has surprisingly excellent handling properties.
Title: Re: DIY 100MHz differential probe
Post by: Marco on July 25, 2016, 10:51:05 am
Just curious, why not the AD8129 like the chipwhisperer probe uses?
Title: Re: DIY 100MHz differential probe
Post by: mrpackethead on July 25, 2016, 10:57:54 am
any chance of being able to push that further to the likes of 200Mhz?     and being able to connect to a low end Rigol scope?

I might be asking too much!
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 25, 2016, 11:47:59 am
Just curious, why not the AD8129 like the chipwhisperer probe uses?
Going for the dual packages offered lower input currents and better low frequency CMRR, and allows experimentation with other common 8 pin compatible dual amps.

any chance of being able to push that further to the likes of 200Mhz?     and being able to connect to a low end Rigol scope?
There are other amps that will push the frequency higher but 200MHz is already possible with the THS3062 it just hasn't been tested that far out. It terminates into 50 ohms, so you just need a way to supply the +/-12V power to the probe which is a useful feature of the second tier scopes. Other designs have used switching supplies powered from a USB port which could be added to the scope end of the cable to minimise interference. You'd want to design a new termination solution to match your scope but the idea is then you can cheaply make many different active probes to attach to the same cable/interface.
Title: Re: DIY 100MHz differential probe
Post by: Mechatrommer on July 25, 2016, 01:17:41 pm
Going for the dual packages offered lower input currents and better low frequency CMRR, and allows experimentation with other common 8 pin compatible dual amps.
talking about CMRR at high freq... your problem is not going to be near there. your problem is you'll need more robust attenuator front end matching network and at anywhere whereever there is resistive dividers there, those will define your CMRR, forget the opamps', except if you have nothing but the opamps.
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 25, 2016, 11:25:24 pm
Going for the dual packages offered lower input currents and better low frequency CMRR, and allows experimentation with other common 8 pin compatible dual amps.
talking about CMRR at high freq... your problem is not going to be near there. your problem is you'll need more robust attenuator front end matching network and at anywhere whereever there is resistive dividers there, those will define your CMRR, forget the opamps', except if you have nothing but the opamps.
The CMRR of this series pair of THS3062 is nothing fantastic (quite bad compared to an integrated design), but routing the DC path through a stable precision DC amplifier keeps the input bias and offset currents well controlled and lowers their impact on CMRR. At these lower frequencies the simple resistor balancing is all thats needed to get extreme CMRR.

The whole point of this project is to show how to easily use the active probe interface, any number of different probes with specific features or components could be built quickly and cheaply and interchanged.
Title: Re: DIY 100MHz differential probe
Post by: T3sl4co1l on July 26, 2016, 02:26:22 am
Measurements?  Rise time?  CMRR vs. F?

Tim
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 26, 2016, 07:49:24 am
Measurements?  Rise time?  CMRR vs. F?
Down to DC it gets over 100dB and a respectable 60dB @ 100kHz, but with the tight packaging you notice the stress sensitivity of the pots and resistors as it changes when you twist the enclosure or intentionally probe parts.
Title: Re: DIY 100MHz differential probe
Post by: mrpackethead on July 26, 2016, 10:34:54 am
I'm perhaps missing something here, but why are you using a display port cable?  why does'n the signal goe down a normal bit of 50ohm cable?  Appreciate that you need some power there as well, but It seems like a whole bunch of hassle to go to..

Also are those pins so it will interface with your Keysight scope?

Iv'e got a particular use case that i need a diff probe for which is creating ethernet eye diagrams.       This looks like it might be a really good solution that won't break the bank, i might see if i can create it with a pair of RJ45 jacks so it can sit "inline" with an ethernet connection.

What did you create your schematics / drawing in?
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 26, 2016, 10:54:55 pm
I'm perhaps missing something here, but why are you using a display port cable?  why does'n the signal goe down a normal bit of 50ohm cable?  Appreciate that you need some power there as well, but It seems like a whole bunch of hassle to go to..

Also are those pins so it will interface with your Keysight scope?
Its a handheld probe, having a single cable makes it much easier to use. The existing designs shared on the internet put the power supply on the probe which makes it bulky, moving the power supply to the scope end has advantages especially when most scopes that aren't bottom end models already provide suitable power supplies specifically for active probes.

What did you create your schematics / drawing in?
Altium Designer
Title: Re: DIY 100MHz differential probe
Post by: mrpackethead on July 27, 2016, 03:19:43 am
Its a handheld probe, having a single cable makes it much easier to use. The existing designs shared on the internet put the power supply on the probe which makes it bulky, moving the power supply to the scope end has advantages especially when most scopes that aren't bottom end models already provide suitable power supplies specifically for active probes.

Right got it.   for the reletively low cost, i think i can modify this to work on my specific problem, ( ie, ethernet ) and set it up so i can make it run as a "tap" and it will be really easy to use.   Now understand what / how you've made it.   For my application this will run just fine on bench..   If nothign else this will be a good exercise and some learning.. I'm goign to set it up to run wiht a Rigol 2000 series,  ( 500Mhz ).. the ethernet signal is 125Mhz, so i think it will be ok.

Quote
Altium Designer

Any chance of sharing those native files?  that would be really helfpul.

Title: Re: DIY 100MHz differential probe
Post by: MattHollands on July 29, 2016, 08:29:41 am
So I'm going go come in now and be a noob:
Can you explain a bit of the topology here? I'm trying to relate this to an instrumentation amplifier (which is the closest thing I know), but there's one more op amp than I would expect. Can you break down what's going on in the circuit?
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 29, 2016, 09:13:34 am
So I'm going go come in now and be a noob:
Can you explain a bit of the topology here? I'm trying to relate this to an instrumentation amplifier (which is the closest thing I know), but there's one more op amp than I would expect. Can you break down what's going on in the circuit?
Good question, think of it as two completely separate differential probes for the high frequencies (U2) and low frequencies (U1). The original plan was to use the precision opamp (U1) to DC servo the high speed amplifier (U2A) but I didn't come up with a stable solution for that so U2A is left open loop for DC parameters of which most concern is the input current drift of 2mV/degree in the 50k impedance. The two opamp instrumentation amplifier design comes from Linear Technology and their design note (#46) for current feedback amplifiers:
http://cds.linear.com/docs/en/design-note/dn46fa.pdf (http://cds.linear.com/docs/en/design-note/dn46fa.pdf)

The low speed version discussed above doesn't have U2 fitted at all, and has excellent DC stability.
Title: Re: DIY 100MHz differential probe
Post by: Berni on July 29, 2016, 12:41:23 pm
I would love to see some test results of the probes performance such as the frequency response, CMRR over frequency, step response etc.

I built a similar probe for the Agilent smart probe interface. But I se my goals a bit too high by making it differential 100k input impedance 1 Hz to 100MHz span along with 1000x gain and 100V common mode range. It was mostly designed for tracking down power supply noise and looking at tiny signals. It mostly worked but needs some tweaking that I never got around to and so the project sat.
Title: Re: DIY 100MHz differential probe
Post by: Mechatrommer on July 29, 2016, 01:17:03 pm
i reviewed THS3062, its not suitable for repetitive full swing signal application like this. anybody who wish to replicate this circuit may replace THS3062 with THS3092. fwiw.
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 29, 2016, 10:53:43 pm
i reviewed THS3062, its not suitable for repetitive full swing signal application like this. anybody who wish to replicate this circuit may replace THS3062 with THS3092. fwiw.
Why so? The THS3062 has sufficient slew rate and output current capability and goes on to provide separate small and large swing characteristics, there is some ringing/settling time limitations but they're insignificant compared to the input network mismatch. I would be more worried about the less capable capacitive drive of the THS3092.
Title: Re: DIY 100MHz differential probe
Post by: mrpackethead on July 30, 2016, 05:53:43 am
Sorry to ask twice, but any chance of sharing the native altium files?
Title: Re: DIY 100MHz differential probe
Post by: Mechatrommer on July 30, 2016, 06:01:48 am
i reviewed THS3062, its not suitable for repetitive full swing signal application like this. anybody who wish to replicate this circuit may replace THS3062 with THS3092. fwiw.
Why so?
because i burnt 2 of them already... from the datasheet...
(https://www.eevblog.com/forum/projects/diy-100mhz-differential-probe/?action=dlattach;attach=244379;image)
Title: Re: DIY 100MHz differential probe
Post by: Berni on July 30, 2016, 06:41:18 am
Wow I had no idea specialized opamps could be as easy to blow as simply feeding a high speed clock in to its input.
Title: Re: DIY 100MHz differential probe
Post by: mrpackethead on July 30, 2016, 06:50:31 am
i reviewed THS3062, its not suitable for repetitive full swing signal application like this. anybody who wish to replicate this circuit may replace THS3062 with THS3092. fwiw.
Why so?
because i burnt 2 of them already... from the datasheet...
(https://www.eevblog.com/forum/projects/diy-100mhz-differential-probe/?action=dlattach;attach=244379;image)

Is it pretty much a straight swap out?
Title: Re: DIY 100MHz differential probe
Post by: Someone on July 30, 2016, 07:12:46 am
That would require a sustained 100MHz input continuously at the majority of the working range, it doesn't seem to be a problem for realistic use of the probe. Many probes are derated for continuous HF signals.
Title: Re: DIY 100MHz differential probe
Post by: Mechatrommer on July 30, 2016, 09:57:22 am
Wow I had no idea specialized opamps could be as easy to blow as simply feeding a high speed clock in to its input.
yup those current feedback opamps are very good water heater at normal operation. even the THS3092 and THS3096 mentioned above that is rated for repetitive signal will get hot. regardless non-repetitive or repetitive applications, they all need an aluminum heatsink on top, or else a very good solder contact at the bottom pad. the non-repetitive THS3062 will burn without smoke if you violate the above rule.

Is it pretty much a straight swap out?
you mean replacement? no, not straight if you want match freq respond, but workable and close enough. its just that the THS3062 is not the right opamp for high slewrate repetitive signal. but they all are very close at specs such as input voltage offset, current bias and offset. ymmv.

That would require a sustained 100MHz input continuously at the majority of the working range.
it was for my 50MHz FG output ±10V. due to gain respond, those opamps roll off starting somewhere 20MHz iirc, or it maybe in the diff probe i designed back then, i cant seems to accurately tell which. the THS3062 burnt, replaced with THS3091 iirc until now working. i still have 3 THS3062, that i'm going to use in my tracking PSU that i current do (thats why it came to my attention), a bit overkill i think but... as they are not good for anything else anymore, i'm going to use them up first. i'm just giving you a precaution, ymmv.