Feeltech FY6600, with Y-cap removed if you wish. For lower noise and no mains leakage, mod it to run on batteries.
That's actually not as mad a suggestion as you might think. Only there's no need to remove the Y cap which is there to return the HF switching noise that gets coupled to the output via the inter-winding capacitance, directly back to its source on the HT side of the transformer rather than let it take the scenic route via the rest of the AWG, its connection to mains ground via the DUT and back along the mains cable and into the AWG's rectifier circuit and the point it had originated from.
The penalty of having a 1.6M impedance connection to half mains voltage via this 1nF Y cap is easily nullified by using a 1 to 10K resistor between the floating ground and an actual safety ground provided by upgrading the cheap C8 (figure of eight) mains connector to either a Y6 (clover leaf) which allows the option to use a more flexible 6A rated mains cord or else use a big butch C14 and limit yourself to big butch 10A rated mains cords.
The later FY6800 and 6900 models already have a big butch C14 socket but you need to undo the earthing vandalism perpetrated by the Feeltech Arsehole (chief bean counter) who allowed one of only two ground return wires in the link cable between psu and mainboard to be stolen as a cheap way to effect a grounding connection without the slightest thought for safety (a wire not rated to carry the potentially several hundred ampere transient fault current with insulation unsuited for mains voltage).
All Feeltech had achieved was to convert this class II AWG into a substandard Class I fire hazard whilst nullifying the charm possessed of the 6600 in keeping out stray DC offsets, mains frequency hum and switching pollution noise from the countless number of other smpsu powered gadgets and devices plugged into the supply.
Undoing this grounding vandalism simply restores the later models back to their erstwhile class II state endowed upon them by use of the same class II psu and a plastic case requiring only a 1 to 10K resistor to link the the BNC gound to the PE tag on the C14 socket to suppress the half mains voltage ESD hazard down to less than half a volt rms whilst providing some 40 to 80dB attenuation to unwanted mains earth wiring borne noise interference to the DUT. Simples!
I recently purchased an SDG1032X which I returned as "Unfit for purpose" due to its cheap smd XO chip suffering a (seemingly) rare instability similar to a low frequency jitter, not unlike that, which in retrospect, was somewhat similar to my FY6600-60M's frequency stability (lack of) until I consigned the ten cent piece of shite to the waste bin and replaced it with a 50MHz 0.1ppm TCXO oscillator board mounted clear of the 60 deg hot spot I'd extracted the original XO chip from with prejudice.
That oscillator upgrade has now long since been replaced with a 10MHz OCXO feeding a 3N502 clock multiplier chip mounted in the original smd XO location. I fitted an external 10MHz reference socket to injection lock the OCXO so as to avoid switching transients of the more basic change-over switching arrangement possibly crashing the FPGA's processes. In retrospect, I could have used a simpler phase locking circuit to achieve the same automatic glitchless change-over between the external and internal 10MHz clock references.
It might come as a surprise to most but an SDG1032X is more of a downgrade than an upgrade over the humble FY6600, modified or not. Even the much higher specced SDG2042X has some rather unexpected shortcomings that are absent in that rather underrated FY6600.
My recent experience with the Siglent AWGs has given me a new found respect for Feeltech's design efforts that they'd put into the FY6600. Don't get me wrong, I have no respect for their "Customer Support" nor for their sheer bloody minded bean counter led stupidity.
The problem with the SDG2000X range seems to be entirely down to piss poor firmware crippling the potential of hardware and build quality that is at least an order of magnitude better than the Feeltech FY66/68/6900 products to below what Feeltech had managed to achieve in both usability and performance.
Whilst the SDG1032X had dashed all hopes of my retiring the FY6600 to just occasional use, the SDG2042X I'd purchased to achieve this happy state still falls short of the mark, leaving me only the hope that Siglent take note of my scathing criticism, follow my suggestion of getting hold of a properly functioning FY6600-60M to run comparative tests against their SDG1000X and 2000X models and realise that Rigol have not only been no competition to worry about for several years now but that they really should have kept a closer eye on the likes of Feeltech instead and be duly shamed into completely revamping the firmware into something more worthy of all that expensive hardware that's being crippled by the current firmware version.
You might think I'm exaggerating the 250MSa/s 14 bit FY6600's capabilities against even the 1.2GSa/s 16 bit performance of the SDG2000X series but, believe me, I'm not. The SDG's UI is best described as "User Hostile" with the frequency input interface being the epitome of this user hostility.
I think the difference is largely down to the existence of a keypad in the SDG making the UI developer take the lazy way out whilst its absence in the Feeltech case had forced their UI developer to make a real effort in getting the most usability out of just a rotary encoder and a pair of digit select buttons for the 14 digit frequency setting display.
In the case of the Feeltech, this has resulted in a sticky digit selection of a full 14 digit input range, allowing the use of a one uHz resolution over the full frequency range (DC to 60MHz in my case), versus the 8 digit limit in the rotary encoder with digit select buttons mode where a switch to another parameter takes the selected digit back to square one every time or else, a maximum of just 11 digits from the keyboard entry system meaning the one uHz resolution only applies for frequencies DC to 100KHz
Even then, once you hit the multipler button, the selected multiplier over-writes the final three digits and the display obligingly rounds the
displayed result even though it has in actual fact accepted all eleven digits. Again
The net result of Siglent's poorly implemented UI is that the FY6600's frequency selection is a positive joy by comparison. The other stand out defect of the SDGs is in regard of their poorly implemented Sinc pulse function. Sine wave output on both the Feeltech and Siglent AWGs is devoid of the infamous DAC clock jitter effect, as is the Sinc Pulse output of the FY6600 at any selected frequency. Unfortunately, the same cannot be said of the SDG2042X which can show a jitter free output at 10MHz but collapses into a horrible jittery mess at just 1Hz offset from this "Magic" frequency.
Other of the SDG's arbitrary waves which contain an 'instant' transition suffer what seems to be a 3.3ns jitter (300MSa/s in arbitrary mode - True Arb drops this to just 75MSa/s) so little better than the Feeltech's 4ns jitter on all such arbitrary waveforms (including square waves). It's only the jitter free square wave of the SDGs which sets them apart from the Feeltech's jittery output which, for a 1.2GSa/s AWG, is curiously restricted to an upper frequency limit of just 25MHz which the FY6600 can also do at a push and was the reason why I'd initially opted for the SDG1032X with its 30MHz jitter free fast rise and fall times, maxing out to the same Sine wave frequency limit of 60MHz in its 1062X guise.
Anyway, all that aside, if you do choose an AWG that's cursed by a mains earth connected BNC ground return, there is a way to break this undesired ground loop. It involves the use of a 50 to 100v 50A bridge rectifier pack. Short out the + and - tags and place a 470 to 1K ohm resistor across the ac terminals with the ac tags in series with the earth wire connection. This should provide some 34 to 40dB of attenuation against mains loop induced interference. You need to use a lower resistance value to cope with the fivefold increased leakage current, compared to that of a 10W rated class II smpsu as used in the FY6600 and its successors, typical of the higher rated smpsu with its pair of 2.2nF Y caps in series across the mains in the emc filter providing an effective 4.4nF connection to half mains voltage.
With a bridge rectifier's two diode volt drops, you should get away with using a 1KR instead to prevent the rectifier diodes from conducting in normal use and generating high frequency harmonics. In any case, once you have connected to the grounded DUT or any other grounded test instruments (eg a DSO), this resistor will be shorted out by the half to one ohm ground return connection via the BNC shield wire anyway.
If I do decide to keep hold of the SDG2042X, I'll be assembling a plug in ground loop breaker box to put in line with the mains connection to save any question of voiding the warranty by opening it up and damaging the warranty seal.
John
Home
Help
Search
Login
Register
