The traffic in this thread has all the characteristic of a bus service. You know, you wait at the bus stop for the next one due in 10 minutes and, an hour later, three arrive at once! In this case, it's a week and 11 posts suddenly materialise!
Ok, I'm not going to attempt to say who said what, just address the points raised here - as I can recall them - this, sadly, isn't usenet.
.
Firstly, in spite of the extremely low price of these AWGs (6600 and 6800 models), they're surprisingly high spec, comparable in many ways with kit costing some 7 to 50 times more. Secondly, most of their more glaring deficiencies arising from such product design and production techniques aimed at keeping the final price to the minimum possible, can be fixed with fairly trivial component upgrades as described over a year back in this thread.
Their shortcomings due to every penny pinching trick in the book being applied are to be expected. As others have pointed out, what is remarkable is just how much functionality you're actually getting at such a low price point. However, the cynic in me suggests the penny pinching was inflicted on these AWGs by careful design rather than just the more usual simple mindlessness of letting accountants dictate the design to shave their pennies off the BoM and production line costs.
When you think about it, these products are clearly aimed at a market dominated by cash strapped hobbyists rather than professionally run high tech businesses, a market demographic made up largely of people prepared to analyse and remedy the worst excesses of Feeltech's penny pinching compromises to demonstrate just how cheaply these "Sows' Ears" can be turned into "Rayon Purses" for the want of careful calibration of presets or replacement of the crappy dual opamp with THS3001s or better, or ripping out the nasty little XO chip to replace it with a TCXO board[1] well clear of the 70 deg C neighbourhood of the original XO location with options to modify or replace the PSU board and maybe even sort out the Skoolboy Howler error of the 85 ohm attenuator pad which Feeltech CBA to correct on the main board, electing instead to use a firmware fix that can only work in the Hi Z case (hopefully, sorted in the FY6800 but that assumption has, to my knowledge, yet to be tested).
Those unpopulated opamp spots on the main board are a positive incitement to upgrade the single dual opamp chip and, in so doing, neatly void the Chinese warranty, reducing warranty claims or sales returns to an absolute minimum, virtually all but eliminating Feeltech's customer support costs. It's a win win situation all round. Feeltech are supplying not only an extremely affordable item of test gear, it's also a nice little 'fixer upper' project for most of the market demographic this was squarely aimed at. Not only does the end customer get a 'nicely priced' signal generator, he gets a self improvement project all wrapped up in one neat package as well.
Let's face it, anyone even so much as contemplating the purchase of such temptingly cheap test gear must at least be of an enquiringly enough mind to further their knowledge of electronics just by virtue of their recognising the need for such an item, no matter how limited its performance and accuracy might be. Remember, even professional test gear has its own limitations which the user has to be mindful of - you just have to be
more mindful in the case of Feeltech's products is all.
Next, we come to the correct use of oscilloscopes, notably the business of 'scope probes. At audio frequencies, the problem of capacitive loading from the 'scope probe lead only becomes an issue when measuring high impedance points in the circuit, a situation more likely to be met in vintage valve (tube) gear, less so in modern solid state kit.
However, probe lead capacitance becomes a serious issue with MF and and HF frequencies used in radio equipment. The 10x probe setting on the typical standard 'scope probe is the solution to this problem as described in that video (one of the better instructional videos BTW). Even so, the use of unterminated probe leads with 1M ohm/20pF channel amplifier inputs is a bit of a puzzle to me. Unless the probe cable is some special Highish Z high loss cable, I can't see how this can work.
To explain how the 10:1 20dB lossy probe tip was originally conceived, you need to understand some basic transmission line theory. If you use even a short length of 50 or 75 ohm co-ax to directly connect the RF test point to an oscilloscope's Y channel input, you introduce not only unwanted additional capacitive loading (circa 100pF per meter for 50 ohm co-ax) but also the effects of an unterminated length of transmission line which can represent anywhere from very high impedance to almost a dead short depending on whether a half or quarter wavelength's worth of probe cable is involved.
Originally, for RF work, the scope input would always be terminated in the probe lead's characteristic impedance, ime, typically 75 ohm. The tip itself was connected to the end of the probe cable using a 675 ohm resistor with a low value trimmer in parallel. The trimmer capacitor was there to match the resistive 10:1 impedance ratio to the same 10:1 probe to cable capacitive ratio. Whilst the probe tip loading of such 10:1 (20dB loss) probes did introduce some additional loading at the test point (a mere 10%), it neatly isolated it from the undesired effects of adding an unterminated length of transmission line which could introduce unwanted ringing from reflections and, at critical frequencies, lead to an effective short circuit loading.
So, that's the theory behind the (passive) 10:1 (or 100:1 and higher) probe tip as used when probing high frequency signals to display on an oscilloscope. However, as I've already mentioned, the use of 1Mohm with circa 20pF Y amp inputs on modern 'scopes, seems at odds with that basic working principle. The probe tip 9Mohm resistor with a 4pF trimmer follows the same basic principle but it's the lack of termination for the co-axial probe lead itself that leaves me puzzled and I can only guess that the probe lead co-ax must be some special high loss cable designed to eliminate unwanted reflections.
After reading the wikipedia article here:- <https://en.wikipedia.org/wiki/Test_probe> , it seems my supposition about the use of special hi impedance lossy co-ax cable with typical passive 10:1 probes normally supplied with modern 'scopes was just about spot on. My knowledge as it related to probing RF frequency amplifier circuits refers to the use of "Lo Z probes".
You might think that a transmission line (75 ohm co-ax in this case) would need to be terminated at both ends with an impedance matching the cable's characteristic impedance but that's not actually required with a unidirectional signal flow. It's only the receiving end that needs to be matched to eliminate unwanted signal reflections and standing wave voltages. The signal source in this case can be anything from zero to infinite (constant current) ohms. the probe end of the cable will still look like a 75 ohm impedance regardless, provided the scope end remains properly terminated.
Matching the amplifier's output impedance to that of the load is purely a matter of maximising the coupling of its output energy into the load. A properly terminated cable, no matter how long, will always look like a resistive load equal in ohmic value to that of its matched characteristic impedance. Indeed, the longer the cable, with consequently greater loss, the more accurate the match to its characteristic impedance it will have until, given a long enough cable length and therefore attenuation, the termination impedance becomes totally irrelevant (eg 50dB loss becomes a 100dB attenuation of any energy reflected from an open or short circuit load at the far end.
Anyway, here's the 'thing'. Unless you're dealing with lengths of 50 ohm co-ax shorter than a tenth of a wavelength of your signal generator's output frequency between the generator and the 'scope input socket, as well as the capacitive loading effect of the cable, you're going to see the effect of unterminated transmission line reflections introducing variations in the voltage levels at the 'scope input unless you introduce a 50 ohm terminator at the scope end of the cable (either a built in option on the 'scopes Y input socket or the use of a through line terminator or a T adapter with a 50 ohm terminator plugged into it).
If you want to have some fun, checking the frequency response of your FY6800 using just a standard unterminated RG58 BNC lead, try various lengths of such leads (2, 5 and 10 metre lengths being typical offerings available) and you can observe the response peak at frequencies corresponding to odd numbers of quarter wavelengths of the test frequencies being used.
If, after using the correct methods of measuring the performance of the FY6800, you still feel it falls way short of its advertised specification, you'd be well advised to return it, unmodified in any way, as "Not as described" to get a full refund and look to spending, for example, some 600 quid on a Siglent SDG2082X Signal Generator
[Notes]
[1] This is a fine example of just how cheap the FY6600 is compared to the 120MHz (sine wave) dual channel Keysight AWG when it comes to upgrading their original XO references (admittedly, the Keysight reference XO was far superior to that of the Feeltech's to begin with) to 0.1ppm TCXO modules.
For a mere 23 quid and a modest investment of my time, I was able to do the same upgrade that would have cost 700 quid as a return to factory upgrade option for the Keysight. The cost of the Keysight TCXO upgrade alone was nine times more than what I'd originally paid for the FY6600 in the first place!
JBG