The dc waveform does weired things...
If you are in square, amplitude 5v, and after go in dc, you hear a relay clicking going from 1.000 to 1.001
If you are in square 5.0001v, in dc you not hear any relay clicking in the whole range
If you are in square 0.5000 , in dc you hear 2 times the relay clicking. 1.000/1.001 and 2.500/2.501
I've been following the discussion whilst I was getting around to doing more tests with my FY6600, notably confirming your own findings that the voltage is switched by a factor of ten when switching the attenuator pad in and out of the circuit.
After testing with 99.999% and 0.001% duty cycle square waves and DC voltages (and some sine wave testing at 1KHz thrown in for good measure), I've come to the conclusion that the 85 ohm pad should simply be replaced with a 20dB 50 ohm pad with no further questions asked.
The results weren't entirely consistent but, putting aside the crappy iterative adjustment requirements of the W3 /16 and W5 /W26 trimpots to set
both gain and the DC balance simultaneously, I think that this is the only way to proceed for now before meddling with those damned presets once more.
It had occurred to me that any dc offsets would skew the 99.999 and 0.001 % duty cycle square wave tests so had tweaked W1 and W2 to zero the 2mV offset each channel had. Unfortunately, I then accidentally tweaked W26 (CH1's gain balance) whilst checking CH2's dc offset so landed up having to go through a more truncated version of your round of adjustments, discovering in this process the excellent reason for my not wanting to meddle with any of these trimmers in the first place!
Now, Like you, I discovered some very peculiar behaviour when testing the DC voltage attenuator relay switching thresholds. CH2 uses the pad but now switches at the 100 and 101mV thresholds. CH1 is even more peculiar in that it has decided to not make any use of the attenuator pad!
Now this may have been the result of my crashing the function generator due to connecting to my GPSDO which is powered off of a 9v wallwart which has half mains live voltage on its DC output lead (basically, the FY6600's half live voltage issue the other way round). This was never an issue until now because previous connections had all been made whilst the GPSDO or the FY6600 was grounded via the DSO.
I'm pretty certain that the attenuator was being switched in and out around the 250mV mark (500mV in terms of P-P repetitive wave forms) prior to my discovering this old resurrected issue with Feeltech's poorly written firmware not properly recovering from the confusion of an ESD triggered lockup when rebooted.
The strange thing is that you might expect to see rather wild voltage outputs at the points where it's supposed to switch the attenuator in and boost the voltage tenfold (and vice versa) but it just behaves as if it has forgotten the need to do any of these actions as you work through the voltage settings.
I haven't tried to do a full system reset to clear out the corrupted settings and take it back to normal just yet. Mine, fortunately is running the later V3.3 firmware which, as far as I know , is free of the 'bricking issues' of the earlier V3.0 and V3.1 versions. However, I will do a restore to defaults after retesting this peculiar behaviour to make sure I hadn't imagined it all.
It seems the FY6900 isn't all that different to the FY6600 after all, at least as far as the shite firmware quality is concerned at any rate.
I've just verified the above behaviour but the truth of the matter is that if you've quit the "DC waveform"
without resetting the voltage to zero (press the dial or hit the "OK" button when the voltage is selected), you can land up with a dc offset which forms a component of the switching threshold voltage. If this is high enough it can even leave the final amplifier stage permanently switched in and you won't be able to hear any relay clicking
at any level This isn't the first time I've been hit by this peculiar behaviour but it's so easy to overlook the need to zero out DC offsets inherited from any "DC waveform" testing activity.
The problem is basically just down to the shite quality of Feeltech's firmware. It's not fatal, just pretty damned annoying!
Incidentally, I've noticed some really bogus circuitry between the twin (balanced line) 5 element filter and the inputs to the primary opamp (the OPA686 in the case of the FY6600). The loading on the filter's outputs looks far from balanced and nothing like the normal arrangement for a balanced to unbalanced buffer amplifier setup.
I suspect the real reason for this horrible interaction between the gain trimmers is to provide a means to compensate for any imbalance between the DAC904's constant current outputs (IOUT and /IOUT) which feed the 49.9 ohm resistors which turn the signal into a 50 ohm voltage source driving each half of the anti-aliasing filter which drives a 100 ohm load on the inverting output and a 100 ohm plus the variable resistance of a trimmer in series on the non-inverting output. I'd have expected the terminating impedance for this filter to match the 50 ohm feed impedance.
One other oddity is the choice of connecting the DC offset amplifier output via a 240 ohm resistor into the filter output connection rather than use the virtual ground summing point of the opamp's inverting input pin to the same effect sans the loading it currently adds to the filter.
I suppose this could simply be errors in transcribing the actual circuit layout to a reverse engineered diagram on DerKammi's part. Although it's been a year since I downloaded the diagrams, there's not been any amendments for the past two years and, after downloading it all yet again, the diagrams remain unmodified. It looks Like I'll have to double check the diagram against the reality of the actual hardware to verify DerKammi's very fine work.
Assuming no errors on DerKammi's part (which I'm inclined towards) then it looks like Feeltech have made some very odd design choices that go against established wisdom. If nothing else, it's as well to understand exactly how those W5/26 and W3/16 trimmers are supposed to work before going on a quest that's likely to drive you to despair (or even clean round the bend).
That's a problem I'm not going to try sorting out until
after I've fixed the attenuator issue. I've decided to use 249 and 61.2 ohm resistors from the E192 range. With a 0.5% tolerance,the resulting attenuators can range from a low of 19.96dB @ 50.28 ohms to a high of 20.1dB @ 49.94 ohms, a narrow enough tolerance range to stop worrying about hand picking components to obtain the best possible match.
If you want tighter control you can always pick by selecting on test from a larger collection, in which case you'd probably be better off going for the cheaper E48 range and buy a bagful or two of each value (249 and 61.9 ohm resistors - or 62 ohms from the E24 range) from which to assemble 'perfect attenuator pads'.
At these frequencies (60MHz max), the additional stray Ls and Cs from series or parallel combinations won't be any problem. When I've needed to source the odd smd resistor over the past year or two, I've looked through my collection of scrapped Desktop PC MoBos armed with a jeweller's loupe and a bright desklamp. Despite the values I'm normally searching for not being commonly found on PC MoBos, I've nevertheless usually succeeded in tracking down suitable resistors (even when Murphy has intervened and teleported my prize into a parallel universe, forcing me to resume my search of PC MoBos once more for more of the same).
Although I don't expect to find suitable E48 examples of 62 and 249 ohm resistors amongst my PC MoBo collection, I'll still have a look before I resort to shelling out hard cash to be certain I haven't overlooked a 'gold mine' of suitable resistors 'hiding right under my nose'. There's nothing worse than spending good money on stuff you didn't need to buy.
Ideally, I'd like to acquire a usefully sized stock of smd resistors covering a wide range of sizes and values. When I tried tracking down such supplies recently, they all seem to be a little overpriced to my mind. I get the feeling that the suppliers are targeting hobbyists with rather more money than sense rather than people like me with more sense than money. I suspect I hadn't been looking in the right places (or searching hard enough). I guess I'm just going to have to try harder next time.
JBG