Replace 50 Ohm track with coax?

[FriedMule]

Hi:-) Just a hopefully simple question about impedance.
Top board is the original and and have a track (yellow arrow) that demand 50 Ohm impedance
I plan on soldering a RG400 coaxial 50 Ohm cable at At X1.
Is it possible to remove the track (lover image) and just connect the RG400 coaxial directly to X1 there?

[magic]

It certainly is possible and even likely to work

That being said, that's not exactly the typical footprint for soldering coax onto PCB. See how it's done in WiFi gear:
https://openwrt.org/_detail/media/tplink/tl-wdr3600/tl-wdr3600_board-top-side-wlan.jpg?id=toh%3Atp-link%3Atl-wdr3600

Not sure what's the role of R8/R9. Some kind of 50Ω termination with built-in divider/filter?

[radiolistener]

yes, if this line is 50 Ohm, then it's possible. But the solder point will be too close to R9 and if you will use inaccurate soldering with a lot of solder and a large tip of wire it may affect impedance. But it seems that this design is not intended for ultra high frequency, so the effect should not be significant and you can use both cases.

Also, make sure that this line is really 50 Ohm, because it may use different impedance. For example, this line may be used as impedance transformer, then it's impedance is not 50 Ohm and you cannot bypass it without disrupting the operation of the circuit.

[FriedMule]

It certainly is possible and even likely to work

That being said, that's not exactly the typical footprint for soldering coax onto PCB. See how it's done in WiFi gear:
https://openwrt.org/_detail/media/tplink/tl-wdr3600/tl-wdr3600_board-top-side-wlan.jpg?id=toh%3Atp-link%3Atl-wdr3600

Not sure what's the role of R8/R9. Some kind of 50Ω termination with built-in divider/filter?

You are right about the strange Coaxial footprint, did just select something in a hurry:-)
About R8/R9 Here is a part of the schematic.

[FriedMule]

yes, if this line is 50 Ohm, then it's possible. But the solder point will be too close to R9 and if you will use inaccurate soldering with a lot of solder and a large tip of wire it may affect impedance. But it seems that this design is not intended for ultra high frequency, so the effect should not be significant and you can use both cases.

Also, make sure that this line is really 50 Ohm, because it may use different impedance. For example, this line may be used as impedance transformer, then it's impedance is not 50 Ohm and you cannot bypass it without disrupting the operation of the circuit.

Sorry but I am not 100% sure I do understand all what you write, so please bare with me in my answer.
As I understand it, the track is meant to be 50 Ohm and the RG400 should be 50 Ohm. The circuit is meant to go from 0Hz to 1GHz.
Is there a way to look at the schematic to find out if the track is an impedance transformer?

New image shows what I think of using as solder pads, any comments about that?

[radiolistener]

If it has wide bandwidth from DC to 1 GHz, the likelihood that this line is used as impedance transformer is minimal, because impedance transformer on transmission line has narrow bandwidth.

So, most of all that this is usual 50 Ohm transmission line and you can replace it with 50 Ohm coax cable with no issue.

There may also be another purpose of this transmission line on PCB. It can be used as signal delay line. According to it's size something like 1-2 ns. If this is the case, you will need to replace it with proper coax cable length.

If I understand correctly, this is some kind of differential measurement circuit. So, if you want to replace one transmission delay line, you will need to replace second one to keep delay the same for both transmission lines on PCB. I think it will be hard to do with coax cable, because each millimeter of wire plays significant role here.

[FriedMule]

If it has wide bandwidth from DC to 1 GHz, the likelihood that this line is used as impedance transformer is minimal, because impedance transformer on transmission line has narrow bandwidth.

So, most of all that this is usual 50 Ohm transmission line and you can replace it with 50 Ohm coax cable with no issue.

There may also be another purpose of this transmission line on PCB. It can be used as signal delay line. According to it's size something like 1-2 ns. If this is the case, you will need to replace it with proper coax cable length.

If I understand correctly, this is some kind of differential measurement circuit. So, if you want to replace one transmission delay line, you will need to replace second one to keep delay the same for both transmission lines on PCB. I think it will be hard to do with coax cable, because each millimeter of wire plays significant role here.

So if I solder the coax direct to the X1 and just let it run along the board in same direction as the original track did, I would get about same impedance as if I did run a 50 Ohm impedance track and then solder to X1 at the end of that? (Same number - two solder joints and same 50 Ohm length).
In the article it does not talk about the importance of keeping other tracks at 50 Ohm impedance, so am not sure.

Would it be more "safe" to choose to keep the original track and maybe then use "FR-4 board, coplaner wave guide, 1.30 mm width, 0.2 mm spacing, 1oz, 1.6 mm isolation height." then try to play with coaxial cable as replacement?

[chrisl]

Since you are driving the 50 ohm cable you need to add a 50 ohm resistor between the output of the OP amp and X1 terminal.

[Mechatrommer]

attached is how its done for 6GHz BW circuit, just an exposed gnd pad and trace's end. you may as well use SMA or MCX connection there to allow easy disconnection. btw, tying + to - input on the 2nd unused THS3202 is not advisable since output may saturate to either power rail and this is not ideal condition for an opamp. you may tie the + input to gnd and output to -ve input as a dummy gnd voltage follower. about the delay line, if it matters then cable/trace length matters but i dont thing so in that differential circuit, if anything it will be on the input section on the left. ditto on chrisl, 50 ohm resistor on the output is what makes it 50 ohm transmission line, very important, without it you are screwed, unlike a little bit of trace width mismatch when it is shorter than 1/10X wavelength. ymmv.

[FriedMule]

attached is how its done for 6GHz BW circuit, just an exposed gnd pad and trace's end. you may as well use SMA or MCX connection there to allow easy disconnection. btw, tying + to - input on the 2nd unused THS3202 is not advisable since output may saturate to either power rail and this is not ideal condition for an opamp. you may tie the + input to gnd and output to -ve input as a dummy gnd voltage follower. about the delay line, if it matters then cable/trace length matters but i dont thing so in that differential circuit, if anything it will be on the input section on the left. ditto on chrisl, 50 ohm resistor on the output is what makes it 50 ohm transmission line, very important, without it you are screwed, unlike a little bit of trace width mismatch when it is shorter than 1/10X wavelength. ymmv.

I am glad for your comment but your reply is a tad above my head! :-)
Here are the two article about the probe:
https://xellers.wordpress.com/electronics/1ghz-active-differential-probe/
https://xellers.wordpress.com/2014/09/28/diy-active-differential-probe-characterization-round-2/

Hope that it may give you more knowledge to help me? :-)

[Mechatrommer]

yes some of us who are interested in building diy differential probe know that link very well. its like a chapter in the net that we must read in order to pass "build my own probe" degree. some members here too successfully built their own probe and try to sell it, check out nctnico differential probe, i'm also still finding time to populate my own differential probe, pcb ready few weeks ago, parts also ready. but i kind dont like the design you choose because its AC coupled it cannot read DC voltage. i'm not sure what you dont understand with my earlier comment or something i wrote wrong?

EDIT: ok now i look closely the link and you followed that exactly. this is why we dont follow someone's circuit blindly. with gained knowledge we know something is wrong. he tied the + and - input of 2nd opamp, that could work, the THS3202 may handle it, but it may also create extra distortion on the 1st opamp who knows? if we familiar with brand name circuit, or some advice from more knowledgable people here, thats not how its done. he also didnt put the 50 ohm output impedance, that will create reflection along the 50 ohm line from and to the 0.01 ohm output impedance of THS3202, he may be happy with the reflected result, just a little bit of ringing, the sense of increased risetime or whatever. you may spot the spike on his actual (not simulated) freq respond plot @ 60MHz or something in the 2nd report, it maybe it, but i cant be sure this is dark art if you dont have proper tool such as VNA. those spike may move around depending on cable length and where its terminated. he also missed the CMRR report, thats one part in differential probe spec, but it maybe not matter for that application, even LeCroy GHz differential probe didnt report their CMRR. anyway ymmv, happy building

EDIT again: you can check in THS3202 datasheet itself again how its done if there is transmission line involved, 75ohm transmission line will need 75ohm on the opamp output. 50ohm transmission line needs 50 ohm resistor on the opamp output, 1Kohm characteristic impedance line/trace needs 1Kohm etc etc, thats the real signal integrity.

[FriedMule]

yes some of us who are interested in building diy differential probe know that link very well. its like a chapter in the net that we must read in order to pass "build my own probe" degree. some members here too successfully built their own probe and try to sell it, check out nctnico differential probe, i'm also still finding time to populate my own differential probe, pcb ready few weeks ago, parts also ready. but i kind dont like the design you choose because its AC coupled it cannot read DC voltage. i'm not sure what you dont understand with my earlier comment or something i wrote wrong?

EDIT: ok now i look closely the link and you followed that exactly. this is why we dont follow someone's circuit blindly. with gained knowledge we know something is wrong. he tied the + and - input of 2nd opamp, that could work, the THS3202 may handle it, but it may also create extra distortion on the 1st opamp who knows? if we familiar with brand name circuit, or some advice from more knowledgable people here, thats not how its done. he also didnt put the 50 ohm output impedance, that will create reflection along the 50 ohm line from and to the 0.01 ohm output impedance of THS3202, he may be happy with the reflected result, just a little bit of ringing, the sense of increased risetime or whatever. you may spot the spike on his actual (not simulated) freq respond plot @ 60MHz or something in the 2nd report, it maybe it, but i cant be sure this is dark art if you dont have proper tool such as VNA. those spike may move around depending on cable length and where its terminated. he also missed the CMRR report, thats one part in differential probe spec, but it maybe not matter for that application, even LeCroy GHz differential probe didnt report their CMRR. anyway ymmv, happy building

EDIT again: you can check in THS3202 datasheet itself again how its done if there is transmission line involved, 75ohm transmission line will need 75ohm on the opamp output. 50ohm transmission line needs 50 ohm resistor on the opamp output, 1Kohm characteristic impedance line/trace needs 1Kohm etc etc, thats the real signal integrity.

Fantastic comment! What I am trying to do, is as you say, to build a differential probe. I do that for two reasons:
1) To be able to measure DC circuit isolated from ground
2) Save the insane ammout of money they normally costs
I have no nee for high voltage or something like that. My hope was to build a circuit I can put between my scope and my probe so I do not have to invest in a separate probes. I know that I can use two probes to do the measurement but that solution has many disadvantages beside making my 4 ch scope to a 2 ch scope. :-)

What I need is a schematic that works and help to draw the PCB. Hope I am not asking for to much here.

[Mechatrommer]

since you are building GHz range diff probe (low voltage maybe up to 30V input?), you may follow this thread, you may find usefull schematics there...
https://www.eevblog.com/forum/testgear/gt-1-ghz-diy-differential-probes/?topicscreen

nctnico came up with his product there selling at $150 iirc if maybe you are interested, but never make the circuit open, so i'm just guessing. maybe you can PM him to publish his circuit here i for so long want to build my own high voltage version maybe up to 500V input, maybe few MHz up to 100MHz should be fine, but CMRR demand is difficult as requirement need to read transistor Vbe or Vgs, so it never realized, the existing solutions are too bulky and oldish style (using 9V battery? duh) and too expensive for my taste. so still dreaming something that never realized...
https://www.eevblog.com/forum/projects/oshw-diy-1kv-100mhz-differential-probe-(dilemma-vs-hope)/?topicscreen

i have the pcb and parts ready for both my HV (ver2) and GHz diff probe (trying to immitate nctnico's topology et al in the thread i linked, i never try GHz diff opamp before so this is just a matter of luck), there is another GHz active probe (not differential) and another mini type passive hi-Z probe that i'm trying to build/explore. the problem is when project list get expanding, the time is shrinking into peanut. so, good luck to us...

[FriedMule]

since you are building GHz range diff probe (low voltage maybe up to 30V input?), you may follow this thread, you may find usefull schematics there...
https://www.eevblog.com/forum/testgear/gt-1-ghz-diy-differential-probes/?topicscreen

nctnico came up with his product there selling at $150 iirc if maybe you are interested, but never make the circuit open, so i'm just guessing. maybe you can PM him to publish his circuit here i for so long want to build my own high voltage version maybe up to 500V input, maybe few MHz up to 100MHz should be fine, but CMRR demand is difficult as requirement need to read transistor Vbe or Vgs, so it never realized, the existing solutions are too bulky and oldish style (using 9V battery? duh) and too expensive for my taste. so still dreaming something that never realized...
https://www.eevblog.com/forum/projects/oshw-diy-1kv-100mhz-differential-probe-(dilemma-vs-hope)/?topicscreen

i have the pcb and parts ready for both my HV (ver2) and GHz diff probe (trying to immitate nctnico's topology et al in the thread i linked, i never try GHz diff opamp before so this is just a matter of luck), there is another GHz active probe (not differential) and another mini type passive hi-Z probe that i'm trying to build/explore. the problem is when project list get expanding, the time is shrinking into peanut. so, good luck to us...

30V is fine, higher <70V would be better. <500MHz is what I need but saw the <1GHz so that's why I did choose that.
I hope that my needs is not so demanding, I hope to be able to build it for fairly cheap, the probe shall also function as an sort of "blow the probe before your scope" :-)