Poor Man's Frequency Respond Cir-Kit.

[Mechatrommer]

fell free to discuss or argue in case i'm missing something... ruling out DSO BW... this circuit is to analyze frequency response of a DUT, its loading effect to the source (R)... and opamps matching freq respond... opamps BW also can be verified. this is my version 2 refined from version 1, which originally intended to measure DUT freq respond (for bode plot analysis) without DSO BW affecting my measurement... ie respond = 20.Log(V3/V2) reading from DSO. cheers

this is the thread and respond which lead me to this thread...
https://www.eevblog.com/forum/reviews/frequency-response-of-your-dso/

Please tell us the details.

easy! just use two 'almost matched respond' opamps as signal buffer (i used dual opamp package, BW comparable or higher than your scope's). split generator's signal to the 1st opamp and to (DUT to 2nd opamp). loading resulted by the DUT to the FG will be reflected in the 1st opamp too so DSO respond or loading effect... its taken care of, no worries but be sure to make as short traces as possible to rule out reflection and other black magic stuffs, those can be quite nasty and can cost you (me) another bald spot...

show both opamps signal on the scope and make your analysis. all we care is the difference (or ratio) between the two signals, its relativity! at particular freq, if the scope's respond is 3dB down, both signals will (should be) be 3dB down, but we already cancelled it by differentiating or rationalizing them. respond = 20.Log(V2/V1), doesnt matter if DSO is flat or not at that freq, the ratio should be the same... theoritically i kinda lazy to put it in my web (proper explanation) or here, so be pleased to ask anymore question if its not clear.

Regards,
Shafri.

[robrenz]

That makes sense to me, but I am a machinist not an EE.  It will be interesting to see what the forum adds  to this.

[mrflibble]

Interesting idea. I'm not sure (*) if I would want to use it for measuring scope BW, but you did give me an idea for something else. XD

*) That's the "haven't really thought it through so I am not sure" variety. Not the "NOOOOEZ it's a bad idea, and as such I politely tell you I'm not sure".

Out of curiosity, what opamp would you use when measuring about 1.5 GHz analog bandwidth?

A quick check for something obtainable + affordable turned up de AD8000/AD8003 (current feedback opamps), but maybe you have found the part in your searches.

Also, what kind of signal generator are you planning to use for the GHz range? Last time I checked those GHz capable DAC were oucho!

[Mechatrommer]

i havent make in-depth analysis, such as the influence of the DSO input loading and capacitance on both channels, i assumed they are all will be cancelled out during rationalization according to the law of V=IZ (linear equation) and Z1 ~= Z2.

there are limitation though.

1) this method cannot be used to measure:
  a) a measurement device (which produce its own measurement values or readout), such as verifying the BW of your DIY DSO or DMM.
  b) signal generating device (no input signal necessary), such as verifying your DIY FG's flatness.

2) this method is assumed only when you have a DSO and a function/signal generator (we also need not to bother with the FG output freq respond/flatness since the method is using "relativity"... conceptually (not Einstein's Theory of Relativity). and what you are measuring is relying on the FG as its input and the DSO as its measuring device.

[Mechatrommer]

re-thinking about it. there can be a dispute in the version 2 where 2 opamps run in serial, phase will not be equal and shifted at higher freq between them, the opamps attenuation (freq respond) will also building up in opamp2 (but still can be compensated using "analysis's flatness assurance" formula.) so FWIW, i'm also attaching my ver 1 where both opamps run in parallel, but DUT output will not be buffered, which is why i moved to ver 2 (i wanted to stick to 2 opamps only for cost saving ). YMMV, maybe there will be next revision and its still under development and more careful thought will be needed.

[Mechatrommer]

this is another version, it doesnt have the "phase and magnitude shift" dispute as ver2 does, and the DUT output is "buffered", but it cannot measure DUT loading unless we do the process twice separately (once with DUT, and once without DUT). FWIW.

[mrflibble]

I like this one better, since this way the buffer stage towards your DSO is the same. And with a DUT being the unity function (read: straight wire) you can then tune your opamp stages so that they are equal (enough).

[GK]

You need a decent detector for something like this to be more than a toy. Google AD8307.
 

[Mechatrommer]

You need a decent detector for something like this to be more than a toy. Google AD8307.

i'm not sure and not expert in what a logarithmic amp can do? http://www.analog.com/en/rfif-components/detectors/ad8307/products/product.html to see the difference/error between opamp1 and opamp2 output? but i agree, as the topic said, its poor man's method consisting of dual opamp (one chip minimum), more robust method needs more components. but "conceptually" this method should cancelled out "most" measurement error due to the non-flat DSO freq response by doing double ended measurement instead of single ended measurement which relying on DSO freq response function or profile. i use this method because i have no concrete data on what my DS1052E profile looks like. about concrete or sound proof whether this method works or how much error in the measurement? i cant because i dont have the "not toy" devices, we know what they are. it works only "conceptually" from what i have gathered, at least i have peace of mind, for now. (until someone teach me the more proper way within a reasonable budget)

[Mechatrommer]

Out of curiosity, what opamp would you use when measuring about 1.5 GHz analog bandwidth?
A quick check for something obtainable + affordable turned up de AD8000/AD8003 (current feedback opamps), but maybe you have found the part in your searches.

sorry i missed this query... i dont work with GHz so i didnt bother with GHz opamp sorry, at that range the signal will be highly attenuated in my 100MHz DSO and most likely null (worst at 1GSps digital sampling rate, aliasing will occur), so its incapable of doing that. its probably workable until 2X DSO BW more or less -6dB? where the signals (depending on the highest amplitude your FG can generate) still have enough resolution on the scope for reliable measurement. and i wont swear on the method if you try to push beyond your DSO/opamp limit, parasitics and murphy will ensure both input stage are mismatched, increasing the measurement error.

Also, what kind of signal generator are you planning to use for the GHz range? Last time I checked those GHz capable DAC were oucho!

inevitably, ouch! the complete GHz FG unit? ouch ouch ouch! so what we can do the hobbiest community? i believe sub-GHz is not in our reach yet.

[GK]

You need a decent detector for something like this to be more than a toy. Google AD8307.

i'm not sure and not expert in what a logarithmic amp can do? http://www.analog.com/en/rfif-components/detectors/ad8307/products/product.html

It makes it easy to measure very low level signals.

[(In)Sanity]

I have yet to hit the upper limit of my 400 Mhz 2465BDM let alone the 500 Mhz limit of my massive HP 54111D.   The 54111D by the way has more value in gold inside of it then I suspect a 100 Mhz Rigol is worth.    You can also drop it on a Rigol (any model) and crush it like a bug

So how do I get a 0 - 1 Ghz flat response sweep for little money ?   The old 54111D allegedly stays pretty flat beyond it's rated 500 Mhz.   Would be interesting to find out.

[PA4TIM]

You can also drop it on a Rigol (any model) and crush it like a bug

You're now official my friend