REW and Axe Port Pro(be)

In the quest for a better test signal input than  standard soundcard line level or even panel mounted instrument inputs and attendant cable , I've been trying the Centrance APP as a hand held signal probe ,
I made up a couple of  barrel jack adapters one with a 0.015 400v coupling cap and probe tip , one with a 150 ohm resistor for noise  bench marking  , and another direct coupled probe tip .

Input resistance on the axe port is 10meg ohms ,  all of a sudden now with no cableing Im getting much better results, on 44 khz Im seeing only a db or two down at 20 khz,at 96khz Im getting more or less flat out beyond 40khz . I measured my homemade  jackplug/ dc blocking cap and probe tip adapters incidental  capacitance to ground  ,66pf , its quite a lot , the jack socket on the on the interface itself will be adding  more ,
Im strongly considering getting rid of the jack input assembly  , fitting a bnc instead , if jack connection to a guitar or bass is still required a short patch cable can be used without  penalty , a  bnc should reduce the incidental capacitance a good amount and enable even better results , only thing is my nice  new probe tip I spent a few hours making is no longer applicable .
I can mock up and measure exactly what benefits will accrue using the LCR meter before I start dismantling and modifying things .

Obviously as is I wont be poking this probe anywhere near high voltages , far to much exposed metal .
The input can only handle a few volts of signal before clipping but Im unsure how safe it would be to probe the anode of a first preamp stage with its own dc blocking cap .

For anyone who'd like to experiment further with this idea theres a knock off of the behringer guitar to usb dongle ,it'll only do 48khz 16 bit  , gives 250-400kohms input depending on attenuation setting, the tear drop shape will lend itself to adding a probe tip easily enough . the extra performance you get from the probe tip arrangement will more than make up for lack of bits and khz in the usb interface ,also if you  manage to blow it up its only $4 for a new one 

https://www.ebay.com/itm/Guitar-to-USB-Interface-Link-Cable-Adapter-MAC-PC-Recording-CD-Q6P9/263025904928?epid=1380469181&hash=item3d3d90cd20:g:vCkAAOSwR11da5oB

A mated bnc plug and socket shows up around 8.5pf capacitance ,
a mated jack plug/socket is more like 30pf ,
If I take a reading from the presumably  fet input stage  of the unit I read 530pf across the input terminals , so Im guessing any difference related to the sockets would probably be too small to have much effect , unfortuately I dont have the schematic .
Suffice to say even with the simple ac coupled probe I made , I'm getting much much better results on the high frequency end with REW  ,dumping a few hundred pf's in cable capacity off the input made a big difference.
I can do a bit better with a simpler lower profile dc coupled probe , maybe about 25pf vs 60pf  on the barrel jack probe in the pic .
 

I fitted a high quality bnc input socket ,made a needle tipped probe and tested again, now Im only 0.5db  down at 20khz @44.1khz 24bit ,
I cleared a lot of rosin flux off the board where the old jack socket was , the input trace dives into the middle of a three layer board with ground plane eitherside only to re-emerge at the gate of the fet on another sub board , taking the + lead out of the BNC directly to the gate of the fet was tempting ,but Id need to surgically desolder a track to get the benefit , it does look like input capacitance could possibly be reduced another bit with extreme care and the use of smd rework tools, but for now -0.5db at 20khz before any calibration and including probe tip is highly satisfactory ,that in combination with the 10meg 'no load' input  is really handy for grid circuits ,backround noise tops out  just shy of the ton ,-99.5db with a customary 50 ohm bnc terminator shorting the input, even with the input wide open  nothing connected it only makes a handfull of db's in the difference ,of course as soon as you  have the  2cm needle probe on the end it acts like an aerial , pulls in all kinds of mush and input s/n degrades considerably . 
The gain control on the axeport works great ,I just set it to minimum as 'cal' like in a regular scope , the digital range control in REW  does all the work , it has no problem bringing the 50uV p-p  self noise of the interface iteself upto full scale in the scope window . I forgot to check exactly what point clipping occured , I think a few volts at maximum input . 

I dug out some specs on the centrance ,lines up fairly well with what I see , the bnc/probe mod saved me around 2.5db at 20khz compared to the graph in the picture .

It's strange that the AxePort has 10 Meg input - is that an input port bleed resistor as measured with a dc ohmeter, or were you able to measure that impedance across the audio range?

I use a 6.5mm to bnc assembly, and then use a standard 10:1 or 100:1 scope probe.  The 6.5 to bnc uses a common 6.5mm to rca adaptor, and a common rca to bnc adaptor, for a fairly robust and well screened interface.  I then just compensate the probe for flat response, and use a calibration file to flatten any measurement for the full soundcard bandwidth.

The MIC-Port Pro looks good, with 192kHz and hi-Z input, but their website has no specs, and it is obviously not a cheap widgit.

Im fighting with the Centrance Universal driver a bit , comes in two flavours , legacy version V7.4.19 ,works ok for me except the 'input monitor' slider in the software doesnt allow me to cut off the direct signal from input to output , so the headphone output isnt usable as a signal generator from REW because it feedsback on itself  :
with  the newer v 8.1 driver ,the direct mix slider function does work  ,but Asio is banjaxed , it defaults to 16 bit 44 or 48 and I cant get 24 bit 96khz. I'm trying to find Centrance UD V7.4.5  ,it seems to have been the most popular of the older versions of the driver used , maybe someone has old revisions archived or I might try the contacting the maker to see if they will send me other revisions of the software to try .

I have a bunch of old scope probes I can try out but I also made various patch cables /adapters ,
like you mentioned I can now make  cal files specific to  each to balance out the effect of the capacitances .
the 10 meg I measured across the inputs with a multimeter ,maybe the input is cap coupled  with 1 meg to ground at the input of the fet ,the capacitance itself looks like it might be  fabricated into the multilayered input board ,along with a certain wanted stray capacitance to make sure the response starts to drop above 20khz . I guess as a final test I can connect the axeport and the signal source directly without the small probe tip ,it adds around 5pf extra  in parrallel with the input .

I found an increase in LF noise dependant on usb powersource and depending on if I make contact with ground on the unit or not . With a battery power source running the axeport the very last trace of hum vanishes ,and theres no difference in noise  if the unit is being handled or not , just what we want . I plan to put two 18650 cells in an alloy box with  charge controlers and a 7805 reg to supply the axeport old style linear regulated power.



Mic port pro2 and R4 look really nice, R4 is actually a fully fledged recorder in its own right as well .