Thanks for your detailed answer!
I had mixed up some information for 6600/6800/6900 in the past. Let me summarize what I now understand. Basically there are two independent problems:
Ground connection / ground loop / ESD hazard
FY6600 has a C8 mains connector without PE. That's ok because it's a class II power supply. And it has an advantage: no ground loops etc. Disadvantage: A voltage between PE and BNC ground can be measured (ESD hazard).
Possible solutions:
- Replace C8 with C14, connect PE to BNC/PCB ground
- To avoid ground loop a 100nF capacitor can be used (in serial)
- To avoid ground loop a 10k drain resistor can be used (in serial)
FY6900 has a C14 mains connector, PE is connected to PCB ground (with an inappropriate wire ...).
If I want to separate PE and BNC ground I have three possibilities (similar to FY6600):
- Remove wire between PE and PCB ground
- Add a 100nF capacitor (in serial)
- Add a 10k drain resistor (in serial)
Do you suggest to use option 3 (drain resistor)? Or is it better to use an isolation transformer?
Signal noise introduced by smps ripple/noise
It's a good idea to do a short test with a battery powered FY6900. If the signal is significantly better a new power supply (linear regulator, ...) is an option (this probably solves the ground problem as well). Currently I doubt that justifies the effort. At least for me and my oscilloscope. The ripple/noise of the function generator output is currently no problem for me (between 5 mVpp and 20 mVpp with a visible 60 kHz signal).
I'm not surprised you were mixing up the information between those three topic threads. The FY6600 one has been going for some two and a half years and is now 86 pages long with a total of 2149 replies as of the 2nd of this month!
If that 5 to 20 mV ripple with a visible 60KHz signal you mentioned is a reference to the 60KHz switching noise, you might see some reduction if you try it with a 10KR in series with the earth connection.
My own concern over the noise and ripple stems from my experience when generating a 30MHz carrier using a Kenwood TS140S HF transceiver to monitor it where there did seem to be some ripple and noise modulation on what should in theory have been a perfectly quiet carrier wave. Other than that, such ripple noise hasn't proved to be a major problem so far but then I am still working on modifications to enhance its frequency accuracy and stability so haven't been using it very much other than for testing those modifications and researching the workings of a DIY GPSDO I've been working on for almost a year along with discoveries of the GPS system's own deficiencies (nanosecond phase shift modulation due to ionospheric propagation conditions being the major problem for a basic pll driven GPSDO such as the one I'm trying to rebuild into a screened metal enclosure onto veroboard).
The FY6600 has been a project in its own right for just over a year now and was what had spawned the basic GPSDO project I'm trying to box up as a workable frequency reference. These two 'projects' feed off each other so I'm dividing my time between the two. What doesn't help is my reliance on Banggood and Ebay to supply the components needed for these and other ancillary projects which have created additional delays to these long drawn out projects. I might eventually get around to fitting the four long screws that secure the FY6600's case halves together and have my GPSDO up and running in maybe as little as 6 months time with a bit of luck.
When I was looking at the half mains live touch voltage issue with my own FY6600, I was a little reluctant to convert from the C8 connector and the very flexible 2 wire 6A mains cord to a C14 connector with its much stiffer 3 wire 10A rated mains cord in order to avoid the "Tail Wags Dog" effect with such a lightweight piece of test gear (only 700 grammes with no grippy rubber feet to stop it sliding all around the bench, especially true when propped up on its bail stand which left the rear hard rubber faced feet dangling in mid air as the rear edge grounded onto the bench top).
I tried all sorts of inventive schemes to eliminate the touch voltage including mostly ways to cleverly null it out, overlooking the fact that you had to at least detect which way round the neutral and live wires of the non-polarised cord were plugged into the C8 socket so as to connect your nulling out circuit's reference to the neutral.
Several doomed experiments with 1:1 mains transformers to create my anti-phase mains voltage source finally convinced me that, whilst in principle such a scheme could be made to work given a clever enough polarity detection system to control an automatic mains reversing switch, my "clever cure" was far more trouble than it was worth and still contained a risk of doubling the touch voltage if it went faulty for any reason.
Since I've already suffered way too much at the hands of the Lord Murphy (of "Murphy's Law" fame), I decided that discretion in this case was most definitely the better part of valour and gave up the whole idea of nulling out the problem and looked to the more prosaic solution of using a polarised 3 pin mains socket to provide a convenient earthing point for attaching a 10KR drain resistor to kill off the touch voltage without introducing a low impedance earthing loop into the circuit.
By way of a compromise in regard of the "Tail Wags Dog" effect, I chose a C6 (trefoil - clover leaf) mains inlet socket instead of the big butch C14 sockets other FY6600 owners (and Feeltech in the case of their 6800 and 6900 models) had elected to use. I'd noticed that a few of my C6 mains cables had been made up using thinner lightweight mains flex than typically used with C14 cords which would mitigate the "Tail Wags Dog" effect (as indeed this proved to be case).
It might be argued by some that by not directly connecting the protective earth to the zero volt rail (BNC grounding point), I have compromised the generator's electrical safety. However, since it uses a class II smpsu board which does not require the use of a safety earth anyway, I don't see this as an issue of safety in this (plastic) case. The mains socket upgrade is merely to provide a reliable earthing point to automatically shunt the Y cap's half mains leakage touch voltage safely to ground via a 10KR resistor.
It seems that Feeltech had been reading the FY6600 topic thread and taken note of the stream of complaints over this half mains live voltage issue and in a half assed ill thought out bout of unthinking pragmatic knee jerk response, decided to "upgrade" to a C14 mains socket, vandalising the 6 wire psu to mainboard ribbon cable to divert one of the only two ground return wires directly to the protective earth tag. Not only had they introduced an unwanted low impedance earth loop issue, they'd also aggravated the psu noise and ripple by stealing one of the ground return wires linking the psu to the mainboard into the bargain.
Judging by the fact that the ill fated FY6900 was also given the very same (identical in fact) treatment, it looks as though Feeltech must have given up monitoring the FY6600 thread once they'd sprung their "Improved FY6800" upon an unsuspecting world otherwise they'd have used a more elegant solution for the FY6900.
Basically, when it comes to fixing the earthing sins of the FY6800 and 6900 models, the optimum solution of using a 10KR (or even a 3k3R if you prefer) to link the ground rail to the protective earth as described for the FY6600 remains the same since these later models still use a cheap and cheerful class II smpsu board inside of their plastic cases which don't require a low resistance earth connection to meet electrical safety requirements.
However, for those of a nervous disposition over the use of unearthed test equipment, there are ways to provide a low impedance path for fault currents to flow to the protective earth which under normal, non-fault conditions, can still provide a high resistance drain path for the unwanted half mains touch voltage. Basically it amounts to shunting the 10KR (or 3k3R) drain resistor with a couple of high current silicon diodes in series in anti-parallel with a second such pair (four diodes in total). In this case, you can get away with just a single pair of anti-parallel high current rated silicon diodes by using a 3k3R resistor (the ground loop noise margin will only drop 10dB in this case).
This is a method I'd used some forty years ago to eliminate mains earth loop induced hum on a turntable with built in RIAA amplifier which blessed it with line out level signals that already gave it considerable immunity to such mains earth loop induced hum. In this case, since it was essentially only the 50Hz and the first few harmonics I was concerned with, I recall putting a 100nF cap across the two back to back diodes to guard against the possibility of high frequency harmonics should the diodes start to conduct and generate any distortion products (I can't remember whether or not I'd included a 1 or 10 k resistor across the diodes). I think the diodes I'd used had an Irr rating of 6A.
In this case, I wouldn't add any capacitance across the diodes, just a 10 or 3.3 k resistor. If you're only guarding against the risk of a mains fault internal to the generator itself, I should think 6A rated diodes would suffice. If you're more concerned over an attached DUT injecting mains voltage, you'll probably want to fit 20 to 50 amp rated diodes (in which case you're likely to see the mains flex burst into flames with that level of fault current). For such current ratings, it's probably best to parallel up several 6 or 10 amp rated diodes to improve the chances that at least one will fail short circuit in the event of any such catastrophic fault current events.
That, of course is a very simple solution and I'm sure there are better alternatives based on a pair of heavy duty thyristors (or triacs for some redundancy) where you can trigger them from a higher volt drop such as 10 to 30 volts to avoid any false triggering. Of course, triacs and thyristors have minimum sustaining current specs which will need to be considered in such a paranoid protection scheme. The one thing I wouldn't rely solely upon in such protection is any form of electromechanical relay to switch the high resistance drain circuit into its low resistance safety earthing state.
TBH, after proof reading this, it does seem a rather over the top solution to a problem that may never ever arise but If I'd seen the need forty odd years ago to provide a safety earth connection that would avoid the hum loop issue, there may well be purpose designed devices available for just such protection by now but it hasn't occurred to me to go searching for any. A quick internet search failed to reveal any such ground loop isolating devices. My perception of the quality of human ingenuity has just gone down yet another notch or two.
After having another go at tracking down a diode based 'hum eliminator' I came across this Youtube video
https://youtu.be/qNQX8jyxRrs where the guy mentioned a 70 dollar hum eliminator, the "Ebtech Hum X Ground Loop Hum Exterminator" which he'd suspected was also based on the two anti-parallel diodes device he was building. I had already come across this item which I'd also had the same sneaking suspicion that it was quite likely based on the same principle (possibly utilising triacs rather than diodes).
The youtuber mentioned that he'd seen the circuit he was building in other internet sources but failed to mention where. It would have been nice to track down other mentions of a circuit I'd invented for my own use over forty years ago. It seems I'm not the only person on the planet to have come up with the blindingly obvious solution to safely overcome the issue of hum loops in domestic audio systems where the short runs of screened interconnects are able to short out the modest voltage drops involved to keep the diodes in the high impedance state as in the case of the Feeltech signal generator's connections to the DUT and any other test gear located on the test bench.
Two final observations I have in regard of that youtube video and the "Ebtech Hum X Ground Loop Hum Exterminator", is that I think the safety concerns that were expressed in the comments are entirely valid. The only safe place for such a circuit is inside the device rather than as an external add on adapter. Also, the need for 500v PIV rated diodes is a spurious one where the anti-parallel connection of the two diode (strings) means they'll never have to face more than one or two diode forward volt drops worth of reverse bias anyway.
If one of the diode strings fails open circuit under a fault event, the 500v PIV rating won't help the situation and it would be far better to give a 50v PIV diode every opportunity to fail short circuit in such a circumstance. Indeed the lower the PIV rating, the better the chance of it failing short circuit rather than being blown to smithereens like a glass fuse by a high energy 300+ volts avalanche breakdown transient.
If a semiconductor manufacturer could be persuaded to integrate this diode or triac based ground loop hum eliminator into a dedicated device, it could be optimised to reliably fail safe and meet electrical safety standards so I'm a little disappointed that there seems to be no such device commercially available. Presumably there isn't sufficient demand to justify the required investment.
JBG