Re: FeelElec New Arrival FY-6900 Signal Generator

[bdunham7]

So I got my FY6900 today.  There's actually not much to tell and the video above shows enough. I'll just mention the issues that are probably of greatest concern to most.

It's the same guts in a slightly nicer case.  However, the parts of the case that stick out don't protect the BNC jacks nor do they allow you to stand the unit on its back with a right angle power cord.  It has a 3-prong power inlet, but it appears very cheap and my cords slide on and off very easily. The whole thing is comically light.

Amplitude and frequency controls oddly allow me to select up to 60MHz on any waveform, and the amplitude available is up to 24.0V at 20MHz and below.  Scope shows that it indeed is putting out at least that, if not a bit more, into a 1M scope input.  Above 20MHz it drops to 5.00 volts, which fades away a bit at 60MHz.  It's worth noting that a square wave at 10 MHz looks much worse (rounded, distorted) at amplitudes above 5.00V.  The op amps that kick in above 5.00V obviously are struggling.

The outputs are all connected and grounded to the IEC connector ground, but the measured resistance between the front BNC shell and the power cord ground is 2 ohms or so.

The DDS-inherent 4ns issue has been altered by interpolation--just look at the square wave inching its way across the screen at the end of the above video.  A quick check shows that the issue of a 10MHz square wave having a 48ns upper and 52ns lower section has been replaced by a 10MHz square wave that has equal periods measured at the zero-crossover point, but now has a 5ns rise and 8ns fall.  It seems pretty clear that they are using an interpolation method where instead of 12 high (+) samples followed by 13 low ones (-), they use 12 high, a zero and then 12 low.  Cleverly simple, but now a 10.000001MHz square wave shows three different rise times--so that the jitter is only low if you are using the zero crossover as your trigger point--and even then, there seems to be an occasional 4ns stray waveform.  My guess is that "Magic Pulse" is a half-baked simple linear interpolation method and that it will create as many issues as it solves.  We'll see.



Anyway, for a cheap-ish signal generator for hobby use, it's probably worth the US$ 86 that I paid.  There's a bit more room to build a linear PS if you want, the grounding can be undone to make it floating again fairly easily (even with the stock PS) and there's a place to install an optional fan if you like.  That's right--what looks like a fan on the back panel is actually just a place where one could be installed by you.  And while you are at it, trim the excess plastic flash out of the fake vents along the sides at the bottom....

[StillTrying]

So the only slight problem with Magic Pulse is that it doesn't work?

[SteveyG]

Here's my quick overview. Let me know if anyone needs anything specific covering in more detail

[SteveyG]

Hi, there..
Is FY6900 equipped with Fan in the back panel ??

No, it could possibly use one though - it gets quite warm when driving 50R loads.

[Johnny B Good]

Here's my quick overview. Let me know if anyone needs anything specific covering in more detail

 Hi Steve and thanks for that revealing  teardown video.

 It didn't quite hit the spot (at least as far us seasoned FY66/68 hundred users and fettlers are concerned ). However, it did reveal the same earthing trick used on the FY6800, namely the stolen ribbon cable wire in the PS to Main board DC rails link, one of the two ground wires snipped off at the main board end and dragged over to the earth pin on the C14 connector - you can actually see the cut end in your video.

 You were rightly concerned over the issue of the glass fuse and the inadequate insulation of the stolen ribbon cable wire in the face of its trailing over bare conductors carrying full mains voltage but you missed the fact that this 'hard' connection to the safety earth isn't actually required with a class II smpsu and should have been replaced with a 1 to 10 KR 'static drain' resistor to effectively suppress the half live mains leakage without introducing the troublesome earthing loop issues you so eloquently described.

 Who knew, given the presence of a C14 mains socket to provide access to the protective earth (a feature lacking in the FY6600 and its predecessors), that the answer to this half live mains leakage issue was as simple as a 1 to 10k 'drain' resistor to link the main board ground to the protective earth?

 The absence of an actual fan in the rear panel fan housing location is but a minor impediment to all of us hardy "Fixers of things broken by design", at least all the hard work has already been done for us (no need for gross mechanical modding involving the use of drills or hole saws in this case  ).

 One burning question remains in my mind at least and that is whether Feeltech have finally gotten round to fixing their "Skoolboy Howler" in the BOM selection for the attenuator pad that gets switched in at output voltage levels of 500mVpp and below. The BOM used for this attenuator in the FY6600 and FY6800 models is still resolutely creating a 20dBish attenuator with a nominal impedance not of 50 ohms but one that's curiously of 86ohm impedance. Therefore, I'm wondering if you can test whether this remains the case with this model?

[EDIT 2020-03-30 :- Answer: They didn't! I just spotted the evidence in this YT review video

 https://youtu.be/kA6uxUDg55M?t=368

 which is paused just a second or two before the evidence hoves into view. You may have to jump on the pause button straight away or else back up if you fail to pause it before it cuts to the psu board. You'll be looking at the half dozen resistors which make up the two pads right alongside of their switching relays. These are numbered RS1 to RS6. Perversely, the odd numbered trio does CH2, the even numbered trio CH1. These make up Pi attenuators with 100 ohm shunt elements and a 510 ohm series element exactly as they've used with the 66 and 68 hundred models.

[END of EDIT]
==========================================================================================

 It's fairly straightforward to check. Set the output to 500mVpp at a convenient frequency and change to a 1mV amplitude step size. Having set it up this way, check the magnitude of the signal with a 'scope or LMS as you adjust between 500 and 501mV settings (or possibly it's between 499 and 500mV - around the point at which the relay kicks in and out at any rate).

 What I, and others, had discovered was that without a terminating impedance load, the change in level was as expected but when terminated with a 50 ohm load, the change in level was much greater than the expected circa 1mV (tens of mV in fact). You can either calculate the impedance by measuring the voltage drop between no load and 50 ohm load at a fixed 499mV setting in the time honoured way or else cheat by turning the amplitude down to zero or just a millivolt or two at say a frequency of 100 to 10000 Hz and literally do a resistance measurement with a DMM (after fine tuning out any DC offset, of course!).

 Being familiar with Feeltech's track record in missing every golden opportunity to put right their various "Skoolboy Howlers" to date, I'd be very much surprised if you don't see the exact same behaviour we 66/68 owners have come to see and loathe.

 Even the error in the specifications section of all the manuals concerning the PP voltage level limits versus frequency ranges has been preserved (the only difference in this case being the 24Vpp for sine and square waves being maintained right up to 20MHz as per the previous 20Vpp limit of the 66/68 models).

 One curious anomaly appears to be the ability of this new generator to output 24Vpp sine waves out of a +/-12 supply when, quite clearly in the face of the 1.2 volt worst case "headroom" figure for the THS3495 (it'll be thereabouts for the original THS3002i dual opamp used and the THS3091/3095 opamps that have been used to upgrade the 20Vpp performance at the 20MHz limit in the 66/68 models), it should be clipping at around the 22Vpp mark.

 This begs the question as to how they've achieved such a magical result. Perhaps the PSU board has been upgraded to a nominal +/-15v unit with the silk screen printing unaltered to reflect this, or else perhaps, they've incorporated a couple of boost converters on the main board to feed the THS opamp rails. Measuring the actual voltages on the rails marked as +/-12v should answer this question.

 I've no doubt there'll be other test requests from others but those are the ones that stand out in my mind. Anything else I might think up will merely be icing on the cake and I think these will be enough to burden you with for now.

JBG

[Johnny B Good]

Hi, there..
Is FY6900 equipped with Fan in the back panel ??

No, it could possibly use one though - it gets quite warm when driving 50R loads.

 Thanks for saying that, Steve. I've had my own observation on the surprisingly high temperatures inside of these signal generators (gently) questioned by some of the contributors to the FY6600 thread (which now includes discussion of this and the FY6800 since they're essentially identical machines with just minor variations - mostly cosmetic).

 To my mind, fitting a cooling fan, whatever the model, should be one of the very first modifications to be contemplated before anything else (along with the 1 to 10K "static drain" earthing resistor which in the case of the FY6600 involves an upgrade from the C8 two pole mains connector to a three pole C6 or C14 before this can be completed - it's a trivial mod in the case of the FY6800 and this 6900 model).

 For reference, I was measuring temperatures with an IR thermometer (therefore of necessity with the lid off) of 70 deg C from the three LDO regulators with a 50 deg reading from the SMD XO only 10mm away and ditto for the FPGA and the base of the heatsink.

 Mind you, this was whilst I had modded the PSU to increase its "5V" rail from 4.96 to 5.49 volts in order to raise the 11.7v on the "+/-12v" to a less marginal 12.7v so those LDO ics were probably running some 5 degrees or so hotter than standard. Even so, I'd expect the innards to possibly be running another 15 to 20 degree warmer with the lid in place.

Plastic isn't a particularly good thermal conductor so the two hot spots on top of the case are a pale reflection of the true temperature level inside. I don't as yet possess a thermal probe to test with the lid fitted but if and when I get hold of such a probe, I'll run a fan disabled fan enabled test to confirm the efficacy of such cooling. In the meantime, the fact that the two hotspots have disappeared without trace is enough confirmation of my fan cooling upgrade's efficacy.

 My main concern isn't over the effect on the 'Silicon' but rather that of the life shortening effect on all those electrolytic caps inside. I'd rather not have to do a recapping exercise after a mere two to three years of hard use (nor for that matter, risk BGA failure) so for me, "Cool" is good, "Hot" is  (literally!) bad in this case - I'd much rather have "Cool" than "Hot". 

JBG

[ed512]

I have received my FY6900 but am unable to get the supplied software to work.  I installed the software on a Win7 machine and it opens OK but does not seem to connect the the FY6900.  I am also looking for instructions on how to use the software and specifically how to create custom arbitrary waveforms.  Of course connecting to the FY6900 is the most important thing.

[TD-er]

I also received my FY6900, but apart from typical coil whining from the power supply, it does not work at all. If you switch off the power switch on the back, you hear the pitch change and after switch this switch again you don't hear anything.
Only after a few minutes the (annoying) coil whine is back when switching this switch on again.

I am waiting for a reply from either Banggood or Feelelec, but so far it is not the best first impression here.

Not sure what issues the older versions had with their power supply.
I wish I read this thread before buying it.

[Johnny B Good]

I also received my FY6900, but apart from typical coil whining from the power supply, it does not work at all. If you switch off the power switch on the back, you hear the pitch change and after switch this switch again you don't hear anything.
Only after a few minutes the (annoying) coil whine is back when switching this switch on again.

I am waiting for a reply from either Banggood or Feelelec, but so far it is not the best first impression here.

Not sure what issues the older versions had with their power supply.
I wish I read this thread before buying it.

 Sounds like it might be a loose connector. You've seen the video so go take a look! You never know, the 'cure' might be something as simple as reseating a connector. 

 Regarding your wishing to have read this thread before buying it, that's exactly what I did some nine months ago with the FY6600 thread when it was only 69 pages long. It took me almost three days to thoroughly read every post, followed by another day skimming through them again before ordering an FY6600-60M from a UK based seller.

 The time spent reading the thread was time well spent since I was fully prepared to deal with whatever faults it may have suffered from bad assembly through to shipping damage (as well as all those "Faulty by Design" problems I was planning on putting right).

 By the time I received mine, just over a week later, I was more than thoroughly prepared for what I expected it to be, a fun filled DIY "Fixer Upper" project.

JBG

[excitedbox]

Mine is sitting at customs in Frankfurt right now (since the 27th) waiting to be released. It spend 10 days sitting at the Airport in China waiting on it´s flight. But with any luck it will be here on Monday.

I hate always having to wait so long for shipping. I get so impatient 

[TD-er]

I did wait with opening until I got a reply via the mail.
They asked me to look inside and indeed it was a loose connector.
The connector between mainboard and power supply.

I also looked at the 'ground cable' and apart from the mentioned lack of insulation level, it is also dangling on a single thread of the internal strings.
So that's going to be replaced like in the video and I will also add some shrink tube on the other mains joints just to be sure.

This ground wire is really a safety issue, since it can come loose quite ease during transport and touch one of the mains points.

The manufacturer also wrote this about the power supply:

Our power supply is + & - 13.5V and 5V, the current is 200mA.

I assume it is for all voltages.

[Howardlong]

Pro tip: a coat of clear Plasti Dip rubber coating on the knob works wonders.

I masked off the back with tape before dunking it into the tin (it might work with a spray too, I just have a small 250ml tin of it). After removing it a few second later, I left it for about three hours to dry, removed the mask and popped it back on. Like night and day.

[Johnny B Good]

Mine is sitting at customs in Frankfurt right now (since the 27th) waiting to be released. It spend 10 days sitting at the Airport in China waiting on it´s flight. But with any luck it will be here on Monday.

I hate always having to wait so long for shipping. I get so impatient 

 I presume that you meant the 27th of this month. That's hardly any delay at all!    How about a 99p tracked order for ten 74hc14s that's been sitting in Hong Kong customs since the 1st of June (just two days after being shipped out of China by the seller) and is, afaik but no longer care about, still stuck in that black hole.

 I'd finally had enough of all the waiting so got Ebay to 'resolve the issue' and the seller has reimbursed my 99 pence. I'd ordered half a dozen of these ICs from a UK based Chinaman at the exorbitant price of almost three quid about three weeks ago and they'd arrived just 8 or 9 days later.

 I'd have insisted on the seller making another delivery attempt (at just 9.9 pence each, one can never have too many 74HC14s in the parts bin) except for the fact that Ebay have a peculiar definition of the phrase "21 business days lack of activity" in determining when to arbitrarily close a case I'd opened in regard of a cheap (£40.41) FY6600-60M I'd ordered from a Hong Kong seller who had unsuccessfully tried to send me a second unit when the first had become overdue by a month.

 Luckily for me, the seller in this case did actually refund my account via the paypal system, no thanks to ebay's flawed disputes handling system. The strange thing in this second case was that the seller had chosen the cheapest possible shipping option which lacked any tracking options.. both times! A 99p order gets the tracked treatment yet a 40 quid one doesn't - go figure!

[EDIT 20191017-1]
 That pack of ten 74HC14s finally popped through my letterbox on the 14th, only 4 1/2 months after submitting the order!!! I told the seller and reimbursed him the payment he'd refunded. At a mere 99 pence, it was well worth such a tiny cost to salve my conscience with such 'honesty' in this case.

 Although it seems highly improbable that the FY6600-60M I'd ordered around the same time will ever turn up, I'm not so sure I'll be quite so honest when it comes to a matter of £40.41 on an untracked parcel. [END_EDIT]

 Both purchase attempts had become disappointing failures but at least I got my money back.      I'm now looking to Banggood as my main supplier now, ever since having had my attention drawn to the absolutely fabulous KSGER T12 oled soldering station they sell which is designed to drive the Hakko T12/T15 70 watt rated direct drive (with thermocouple) soldering iron tips far far more accurately than the original Hakko soldering stations ever could.

 However, as is usual with Chinese made mains powered kit, the 24v smpsu has potentially lethal issues requiring some gross mechanical remedial work on the diode heatsink (which bridges the nice isolation slots milled into the PCB) to improve on the use of solder resist mask as the sole method of insulation against a 170 to 340 vdc carrying circuit trace. Not a serious problem for anyone who understands this lackadaisical attitude to mains voltage safety by the Chinese and regards such safety checking and basic remedial work as SOP with such Chinese product.

 This is one soldering station I can't recommend as a first soldering iron for a newbie to purchase even though it's an excellent choice for the seasoned hobbyist upgrading from years of frustration with classic mains powered soldering irons such as the Antex 15, 18 and 25 watt models (or worse[1]) whom one can reasonably assume would know how to recognise Chinese death traps when they see them and, importantly, know just how to remove them.

 It was Banggood simply because they were the only seller who were offering the right combination of soldering iron handle (the under-rated, even unjustly disparaged, T12-9501) and a T12-K 'starter tip' to go with the oled mains powered and extruded aluminium cased station[2] at a sensible price (Ebay sellers were only offering rubbish combos at inflated prices and my recent "Ebay Experience" had rather put me off dealing with any more Ebay sellers anyway).

 The soldering station had cost me just £39.05 but by the time I'd added a soldering iron stand, a wire wool tip cleaner, a set of assorted T12 tips and a Daniu FG-100 calibration thermometer, the whole order just topped the 76 quid mark. The order arrived on my doorstep just 8 days later from their Chinese warehouse (cheapest price option) and I have to say that it was money well spent, in spite of having to fettle not only the soldering station controller box itself but also that of the wiring in the 9501 handle

[EDIT 20191017-2]
 That 8 days delivery was either a lucky fluke or else the result of special 'Grooming' treatment to lull 1st time buyers into a false sense of optimism. Delivery times for BG typically range from a fortnight (if you're very lucky) through to 6 weeks or more (via express air delivery (in a cargo glider towed from the stern rail of a container ship???)). [END_EDIT]

 For anyone here who has ever hankered after an original Hakko T12/T15 soldering station but couldn't justify blowing some 250 to 300 dollars (nor even the 150 dollars of a 'cheap clone'), these KSGER units (all faults aside) represent a superior alternative at just a fraction the cost (around the 40 quid/60 dollar mark). I'd recommend a look at the recent youtube review and teardown videos that have been made over the past year on the KSGER T12 soldering station models if this has re-awakened anyone's interest in upgrading from a classic soldering iron setup.

[1] Worse in this case being those Weller soldering station units with a totally unprotected mains transformer inside - no user serviceable fuse of any sort, not even a thermal fuse embedded within the windings!  As dangerous as the KSGER unit had been (mine is perfectly safe now!), that has not only one but two fuses! 

[2] KSGER have made a bewildering array of variants of their T12 soldering station including ones with just a three digit seven segment display both in external psu/battery pack powered 'mini' versions along with different hardware/firmware versions of same, hence the distinction I made.

 Also, these can be bought in the form of self build kits at varying levels of complexity falling just short of requiring the DIY enthusiast to populate the controller board with smd parts and solder them up - they, afaik, are always supplied as a completed module ready to be wired up to the PSU and  five pin din soldering iron handle socket on the front panel. Incidentally, the version 3 firmware/hardware seems to be the hallmark of these kit versions rather than a later upgrade. The latest versions for the ready built units is currently at HW ver 2.1S and SW ver 2.10 as of this posting.

 Surprisingly, most of these kits are often sold at a higher price than the ready built units  Even when they happen to be a little cheaper than a ready built unit, the tiny cost saving just isn't worth the hassle. You'll have enough DIY reassembly work making the PSU safe and rectifying the shoddy workmanship in the handle and plug with a ready built unit in any case but once you've put in that work, you'll have a soldering station to be proud of.

JBG

[Johnny B Good]

I did wait with opening until I got a reply via the mail.
They asked me to look inside and indeed it was a loose connector.
The connector between mainboard and power supply.

I also looked at the 'ground cable' and apart from the mentioned lack of insulation level, it is also dangling on a single thread of the internal strings.
So that's going to be replaced like in the video and I will also add some shrink tube on the other mains joints just to be sure.

This ground wire is really a safety issue, since it can come loose quite ease during transport and touch one of the mains points.

The manufacturer also wrote this about the power supply:

Our power supply is + & - 13.5V and 5V, the current is 200mA.

I assume it is for all voltages.

 It did sound like a loose psu connector. I'm glad you were able to fix it at the behest of Feeltech support (presumably, no warranty voiding issues in this case).

 As far as the earth wire goes, that had been stolen out of the psu to main board ribbon cable by Feeltech's kludgery in the first place. If it were me, I'd be be reinstating it and looking to solder a 4K7R onto a convenient ground point on the main board and wiring the other end to the PE contact in the C14 socket in place of the bodged earth connection. This will drop the 90vac earth leakage on a 230v supply down to 250mV ac and attenuate the effect of the mains earth loop by some 40 to 50dB as well as suppress the HF switching noise that Feeltech's bodge had introduced onto the mains earth.

 The PSU is still a Class II smpsu which doesn't actually require such a safety earth connection anyway. However, unless Feeltech's customer support agree to such modification work being done, you might want to hold off to avoid immediately voiding your warranty with them in case it stops functioning due to a manufacturing defect outside of all the defects it already has. Give it time to  get beyond the "Infant Mortality" phase of the reliability curve before doing anything that will void the warranty (for what such Chinese Warranties are worth).

 Regarding that statement (from Feeltech themselves?) that the plus and minus twelve volt rails are now actually +/-13.5 volt rails, that rather suggests they've modified the transformer, possibly by reducing the turns count on the five volt secondary winding by one turn, to raise the 12 volt rails by another volt, neglecting to alter the silk screen printing by the connector to reflect this change. The extra volt or so would just about suffice to allow the THS3002i opamp to output a 24Vpp signal. It would neatly explain how they've managed to get an opamp with at least a 1.2v headroom specification to output a full 24Vpp out of a "+/-12 volt" supply.

 The original, unmodified psu in the FY6600 (and almost certainly that of the FY6800) were only managing 11.5 to 11.7 volts on the +/-12v rails. A common psu mod was to replace the cheap diodes on the 12v rails with proper high speed switching, low forward volt drop Shottky types and bump the 5v up by 5 to 10 percent for good measure  to raise the 11.5v to something like 12.7 volts (makes the LDO's on the main board run a little hotter though). In this case, it looks like Feeltech have done most, if not all of the work already.

JBG

[TD-er]

Well, I did already replace the ground wire, since I'm not letting a flatcable wire hang on 1 thread of copper (??) next to mains power solder joints.
Warranty is maybe worth something, but not that much.
Also added shrink tube around the mains solder joints, just to be sure.

I did test the unit a bit last night and it does seem to work fine.
There is however a strange signal added to the output wave form, which can be seen best at the "DC waveform".
It does seem to be at the begin/end of the memory-cycle of the wave form.

Also there is quite a distinct noise pattern on top of the wave form, which does seem to come from the power supply itself.
Maybe later I will make a board using a traditional transformer and just the old fashioned diode bridge, capacitor and 78xx and 79xx.
Just like I did when I was 13 (30 years ago...)
Or could I add a 3D-printed cap over the power supply to support some aluminium foil to isolate the power supply (or the main board) ?
The noise is just strong enough to let my scope's edge trigger to swap the wave form every now and then (at 1V pp output)

Just have to clean up the desk a bit and attach ethernet to the scope to get some screen grabs.

[FeelElec]

I have received my FY6900 but am unable to get the supplied software to work.  I installed the software on a Win7 machine and it opens OK but does not seem to connect the the FY6900.  I am also looking for instructions on how to use the software and specifically how to create custom arbitrary waveforms.  Of course connecting to the FY6900 is the most important thing.

Dear Customer,
Can you connect to the software now? If it is still not possible, please check if the CH340 driver installation is normal.

Best regards,
FeelElec

[Johnny B Good]

Well, I did already replace the ground wire, since I'm not letting a flatcable wire hang on 1 thread of copper (??) next to mains power solder joints.
Warranty is maybe worth something, but not that much.
Also added shrink tube around the mains solder joints, just to be sure.

 I know exactly where you're coming from - "Discretion (in this case) being the better part of valour." and all that. 
 

I did test the unit a bit last night and it does seem to work fine.
There is however a strange signal added to the output wave form, which can be seen best at the "DC waveform".
It does seem to be at the begin/end of the memory-cycle of the wave form.

 A good candidate for which seems, imo, to be the result of the earth loop introduced by that direct connection to the protective earth.

Also there is quite a distinct noise pattern on top of the wave form, which does seem to come from the power supply itself.

 The high voltage switching noise in such smpsu boards can be very hard to keep out of the rest of the circuit boards being powered.
 

Maybe later I will make a board using a traditional transformer and just the old fashioned diode bridge, capacitor and 78xx and 79xx.
Just like I did when I was 13 (30 years ago...)

 You and me both. Mind you, I was never a great fan of this idea once I started building my own IBM desktop PCs in the mid 80s and could handle the very lightweight 200W PSU boxes with an ease impossible with an Old Skool analogue unit weighing some 15 Kg or more (mostly due to the required 300VA mains transformer).
 

Or could I add a 3D-printed cap over the power supply to support some aluminium foil to isolate the power supply (or the main board) ?
The noise is just strong enough to let my scope's edge trigger to swap the wave form every now and then (at 1V pp output)

 At best, that would only screen against the electric field component. It wouldn't do much, if anything at all, for the magnetic field component which will couple interference currents into any wiring or circuit traces. Containing such electromagnetic energy within the confines of a shielding box around such smpsu boards is a lot harder than it looks. 
 

Just have to clean up the desk a bit and attach ethernet to the scope to get some screen grabs.

 Those will be interesting to see and a handy "Before" reference for when you grab another bunch of "After" screen shots after trying out my suggestion of inserting a "Drain Resistor" between the main board ground rail and the PE connection in the C14 mains socket.

 Regarding the "Analogue" psu idea, the major downside is the need to warm up all those 78/79 regulators and deal with the extra heat load in the box. If you use a good quality "R" transformer to all but eliminate the significant contribution you'd otherwise see with a cheap 'n' cheerful Chinese transformer, you're left with (essentially) just the rectifier losses and the regulator losses. The rectifier losses are trivial in this case (an amp's worth accounting for around 1.5W in a bridge rectifier at 50/60Hz).

 It's all those pesky 78/79 regulators that are the main problem in this case. However, thanks to the wonder of modern switching converter technology, these can be replaced with DC-DC converter based drop-in replacements which typically offer conversion efficiencies ranging from a (very) low 70% to a more reasonable 90 to 95 percent, even as high as 97 to 98 percent with the very best DC-DC converter modules.

 You still have some reduced level of switching noise to deal with (typically 30mV ripple at a fixed 1.2MHz on a 5vdc output dc-dc converter module) but almost totally absent the troublesome common mode switching noise of the classic mains voltage smpsu.

 Ripple on the supply rail at a fixed 1.2MHz (or even higher frequency) is a lot easier to filter out, especially when you don't have any significant levels of common mode noise to deal with as well. That's the theory at any rate but substituting all the Old Skool 78/79 regulators with dc-dc converter modules looks a very promising way to go (almost all of the benefits of an analogue PSU but without the worst excesses of waste heat such supply technology is infamous for).

 I've got a couple of these mini dc-dc converter modules on a ten quid Banggod order that's due to finally arrive in just over a week's time (it didn't qualify for the more expedited £76- order I'd put in for the KSGER T12 soldering station and friends which arrived just over a fortnight ago).

 I'm planning on using the 5v only output one (with a <30mV ripple specification) in my basic GPSDO that I'm currently very slowly assembling into a neat little aluminium extruded enclosure. I have high hopes that this will not only avoid the need to clamp the heatsink tab of a 7805 regulator to said enclosure but also allow me to power it off not just 9v wallwarts but also 6 to 24 volt wallwarts with complete impunity.

 Also, of course, I will be testing out how easy (or hard) it will be to attenuate the residual <30mV ripple to a vanishingly small <10mV or even lower with simple LPF filtering using tiny Ls and Cs (and some Rs - unwanted energy has to go somewhere - better it's dissipated by a resistor in the filter itself in a conversion to low temperature, thermal energy rather than escape "into the wild" unaltered  ). I rather think this will be the optimum solution to a source of "clean" (if not green) energy to power all these FY66/68/6900 signal generators . 

JBG

[TD-er]

Not entirely sure yet what may make the difference here.
Last night I tested with the included cables and had not connected the FY6900 to the computer via USB.

Now I did use some proper (well, less flimsy/cheap) coax cables and the FY6900 is connected via USB.
The noise does seem a lot less apparent.
For the screenshots I did enable "persistent colors" to color the noise.

Please note that I could not set the probe as being 1x, so you have to divide the voltages yourself by 10.   (edit: I did need to have a better look, the 1x was off screen and I just had to take a better look and some coffee)
The screenshots are with 1 MHz frequency set on the FY6900 and 0.5V Ampl.
There is no offset or phase set. The pattern for the square wave apparently does not start at the same position as the sine.
At 1 MHz the phase shift is 9 degree (18 degree at 2 MHz, 27 deg at 3 MHz), so there is some strange (fixed delay) offset.

I also added a screengrab of the strange spikes when showing "DC waveform"
This does look like some 50'ish kHz signal, so it may very well be from the power supply.


By the way, I also used the FY6900 PC Software.
Apart from the most horrible GUI design ever and really bad choice of fonts, is it quite useful.
See the screenshots and a hand drawn waveform.

[Johnny B Good]

Not entirely sure yet what may make the difference here.
Last night I tested with the included cables and had not connected the FY6900 to the computer via USB.

Now I did use some proper (well, less flimsy/cheap) coax cables and the FY6900 is connected via USB.
The noise does seem a lot less apparent.

BNC connectors can be notoriously unreliable with regard to their ground contact resistance. The amount of common mode interference being injected into the signal generator's ground which appears on the centre pin connection can vary dramatically as you manipulate the poorer quality connectors.

 The usb ground connection, whilst still relying on an earth loop between the PC and the 'scope, may well be providing a more consistent, if still a little polluted with ground loop induced noise, connection of lower impedance to shunt the bnc ground return impedance which may be helping to reduce the noise in this case.
 

For the screenshots I did enable "persistent colors" to color the noise.

Please note that I could not set the probe as being 1x, so you have to divide the voltages yourself by 10.   (edit: I did need to have a better look, the 1x was off screen and I just had to take a better look and some coffee)
The screenshots are with 1 MHz frequency set on the FY6900 and 0.5V Ampl.

 A 0.5V amplitude setting isn't the best choice with this signal generator since I'm pretty certain this is the threshold voltage level where the attenuator is switched in. In the 66 and 68 models, this is an 86 ohm rather than the 50 ohm impedance attenuator it's supposed to be. Whether this same Skoolboy Howler has been perpetrated in the 69 model, no one has as yet completed the test to determine this.

 If you're testing without the 50 ohm cable terminating resistor (which, from your description appears to be the case), it won't make much difference with a short half metre cable at 1MHz since Feeltech's "fix" was simply to compensate for this in the firmware (but only for the Hi Z case - it can't possibly compensate the error in both cases).

 You need to select an amplitude greater than 0.5V if you wish to avoid having this questionable attenuator switched into circuit -  the 5.01V setting is sufficient to ensure this happy state.

[EDIT 20200815]

 I meant to write 501mV or 0.501v, not 5.01v which invokes the THS opamps for frequencies no greater than 20MHz

 Also, I attributed a cleverness to FeelTech in modifying the firmware to compensate for the "cheap" mismatched attenuator under Hi-Z loadings - there was never any such cleverness other than to create a "20dB drop" for the Hi-Z case only using the cheapest possible resistor combination to satisfy the insane insistence by their Bean counter in chief to "Cut costs at any price, even if this results in a big FU to the more serious users expecting to see a consistent 50 ohm output impedance at all voltage output levels.

 The fix for this issue is to simply replace those resistors with a set calculated for a 20dB at 50 ohm impedance attenuator network - no second guessing required

 

There is no offset or phase set. The pattern for the square wave apparently does not start at the same position as the sine.
At 1 MHz the phase shift is 9 degree (18 degree at 2 MHz, 27 deg at 3 MHz), so there is some strange (fixed delay) offset.

I also added a screengrab of the strange spikes when showing "DC waveform"
This does look like some 50'ish kHz signal, so it may very well be from the power supply.

  Such switching spikes are difficult to observe on a 'scope on account of the drifting frequency due to variations in the load on such smpsus. You normally need to capture it in a single manually triggered sweep (a feature at which modern DSOs excel). If you're examining such artefacts on a DC output rail, this is also where the 'scope's AC coupling feature comes to the fore.

By the way, I also used the FY6900 PC Software.
Apart from the most horrible GUI design ever and really bad choice of fonts, is it quite useful.
See the screenshots and a hand drawn waveform.

 I haven't got my 'scope set up on the bench due to lack of working space at the moment otherwise I'd have tried the same tests to compare. However, istr seeing similar levels of smpsu switching interference at sub 1 volt settings some time back so I think what you're seeing is typical of Feeltech's "Finest". 

 As for the freehand waveforms, I've never tried using the included software in the past nine months since I purchased mine from a UK based seller last November. I haven't bothered largely because it can be a bit of a faff setting up the USB pass-through on my winXP VM for something that's been a low priority so far.

 As well as some unfinished business with that 86 ohm attenuator (now a 45 ohm one since my last attempt to put it right) I've still got one remaining feature to add to my 6600 before I'm ready to experiment with the software and the usb interface (I'm going to add a socket for an external 10MHz reference to be supplied by an as yet unbuilt GPSDO).

 Also, I may replace the existing PSU board with an "Analogue PSU" based on dc-dc converters in place of the classic 78/79 regulator ICs to hopefully eliminate that troublesome common mode switching noise injected by the high voltage switching transients on the primary of the transformer into the low voltage secondary windings.

 The residual ripple on the dc-dc converter modules' outputs should be a lot easier to suppress with a bunch of small LPFs than the inescapable high voltage transients inside of a typical smpsu which not only conduct common mode interference into the main board via the wired connections but also directly irradiate it  as well. 

 However, all that is for a future project. I really, really should have bought those cheap 6v carbon zinc lantern batteries that I spotted in Wilkinsons for just a couple of quid each yesterday to make up the negative 12v rail to complement the 12AH SLA I'll be using to power the positive 12v rail and a 7805 regulator by which to test the generator sans that pesky smpsu board before I start ordering parts for my analogue/switching chimera of a power supply. It's something I've had a mind to do for at least the past six months but, like many things I have plans for, it's been on a rather crowded 'back burner' since like almost  'forever'.

JBG

[excitedbox]

Mine is sitting at customs in Frankfurt right now (since the 27th) waiting to be released. It spend 10 days sitting at the Airport in China waiting on it´s flight. But with any luck it will be here on Monday.

I hate always having to wait so long for shipping. I get so impatient 

 I presume that you meant the 27th of this month. That's hardly any delay at all!    How about a 99p tracked order for ten 74hc14s that's been sitting in Hong Kong customs since the 1st of June (just two days after being shipped out of China by the seller) and is, afaik but no longer care about, still stuck in that black hole.

 I'd finally had enough of all the waiting so got Ebay to 'resolve the issue' and the seller has reimbursed my 99 pence. I'd ordered half a dozen of these ICs from a UK based Chinaman at the exorbitant price of almost three quid about three weeks ago and they'd arrived just 8 or 9 days later.

JBG

Yes I am still waiting on a 4 Euro order from the beginning of June. Just a couple of i2c and spi expanders and an h11 load cell pcb. and it stopped updating tracking after reaching the shipping center in china.

I don´t recommend using Banggood after all the problems I have had with them. Only good thing is that if you complain they refund the money and let you keep the item in most cases. I got a Aneng Q1 multimeter with a cracked screen for free that way (27 Euros saved). But A usb microscope I paid 29 Euros for was a really cheap model with the wrong specs in the description. I ended up not complaining because at some point the time spent is worth more than the refund.

Banggood also has listings for colorful plant seeds (neon blue strawberries etc.) that are obviously fake and photo shopped and even after being reported they don´t remove them.

[TD-er]

[...]
 A 0.5V amplitude setting isn't the best choice with this signal generator since I'm pretty certain this is the threshold voltage level where the attenuator is switched in. In the 66 and 68 models, this is an 86 ohm rather than the 50 ohm impedance attenuator it's supposed to be. Whether this same Skoolboy Howler has been perpetrated in the 69 model, no one has as yet completed the test to determine this.

I did set it to a low value on purpose to see if the board could pick up the noise I saw before.

When switching from 0.5V to a higher amplitude, you hear a relais clicking, so it is very well possible an extra attenuator is switched in/on.
Noise levels are not better noticable when set at an amplitude of 1V.

If you're testing without the 50 ohm cable terminating resistor (which, from your description appears to be the case), it won't make much difference with a short half metre cable at 1MHz since Feeltech's "fix" was simply to compensate for this in the firmware (but only for the Hi Z case - it can't possibly compensate the error in both cases).

The first test I did was both with the included BNC-BNC cable and with the included BNC->crocodile to my normal probe.
Both showed significant noise.

You need to select an amplitude greater than 0.5V if you wish to avoid having this questionable attenuator switched into circuit -  the 5.01V setting is sufficient to ensure this happy state.

I can test again with higher voltages.

There is no offset or phase set. The pattern for the square wave apparently does not start at the same position as the sine.
At 1 MHz the phase shift is 9 degree (18 degree at 2 MHz, 27 deg at 3 MHz), so there is some strange (fixed delay) offset.

I also added a screengrab of the strange spikes when showing "DC waveform"
This does look like some 50'ish kHz signal, so it may very well be from the power supply.

Such switching spikes are difficult to observe on a 'scope on account of the drifting frequency due to variations in the load on such smpsus. You normally need to capture it in a single manually triggered sweep (a feature at which modern DSOs excel). If you're examining such artefacts on a DC output rail, this is also where the 'scope's AC coupling feature comes to the fore.

Will do, grab a single run.

As for the freehand waveforms, I've never tried using the included software in the past nine months since I purchased mine from a UK based seller last November. I haven't bothered largely because it can be a bit of a faff setting up the USB pass-through on my winXP VM for something that's been a low priority so far.

I thought that maybe such a user definable pattern can be useful in making some error state reproducible.
You can define up-to 32 of those patterns (8k per pattern)
So for that it is quite useful to have the software.
Only nasty thing is, as long as the USB cable is plugged in, my ESP flash tool keeps trying to flash to that port (and fails). So I have to remove the cable when flashing to ESP nodes.

Also, I may replace the existing PSU board with an "Analogue PSU" based on dc-dc converters in place of the classic 78/79 regulator ICs to hopefully eliminate that troublesome common mode switching noise injected by the high voltage switching transients on the primary of the transformer into the low voltage secondary windings.

 The residual ripple on the dc-dc converter modules' outputs should be a lot easier to suppress with a bunch of small LPFs than the inescapable high voltage transients inside of a typical smpsu which not only conduct common mode interference into the main board via the wired connections but also directly irradiate it  as well. 

The reason I mentioned those linear ones was because of the effect of a switching DC/DC converter which I thought of replacing.
But I guess the spikes from a switching power supply from mains to DC may be more present indeed than those generated on little DC/DC units.

[Andreas]

Here's my quick overview. Let me know if anyone needs anything specific covering in more detail

For me it would be interesting if 2 annoying "features" regarding AM-Modulation from FY6800 are still there on the FY6900.

- When AM-Modulation is switched on and the carrier frequency (main channel) is changed by USB-Interface (with delivered software or by terminal program) then on the FY6800 the AM modulation is switched off on each frequency step.

- AM modulation % is done wrongly on the FY6800:
  For 80% AM I would expect 20% minimum amplitude and 180% maximum amplitude compared to AM switched off = 100% amplitude.
  The FY6800 outputs 100% and 20% amplitude when 80% AM is dialed which is in fact only 66% AM =(100%-20%)/(100%+20%)
  instead of (180%-20%)/(180% + 20%) = 80%

see also here:
https://www.eevblog.com/forum/testgear/fy6800-dds-signal-generator-questions/msg2502990/#msg2502990

with best regards

Andreas

[bdunham7]

Here's my quick overview. Let me know if anyone needs anything specific covering in more detail

For me it would be interesting if 2 annoying "features" regarding AM-Modulation from FY6800 are still there on the FY6900.

- When AM-Modulation is switched on and the carrier frequency (main channel) is changed by USB-Interface (with delivered software or by terminal program) then on the FY6800 the AM modulation is switched off on each frequency step.

- AM modulation % is done wrongly on the FY6800:
  For 80% AM I would expect 20% minimum amplitude and 180% maximum amplitude compared to AM switched off = 100% amplitude.
  The FY6800 outputs 100% and 20% amplitude when 80% AM is dialed which is in fact only 66% AM =(100%-20%)/(100%+20%)
  instead of (180%-20%)/(180% + 20%) = 80%

see also here:
https://www.eevblog.com/forum/testgear/fy6800-dds-signal-generator-questions/msg2502990/#msg2502990

with best regards

Andreas

I didn't check the first issue, but I can verify that the second is unchanged.  Modulation leaves the high level unchanged and the lower level is (100% - mod%).  Easy enough to understand, but not in accordance with convention. 

[thomastheo]

I just got my fy6900 in the mail. After skimming forum posts on previous fy6600 and fy6800 models, am I right to assume they use a 50 MHz clock for their FPGA? This one uses a 10 MHz oscillator, which would make a tcxo/ocxo upgrade or external reference input much easier to add!

[Johnny B Good]

I just got my fy6900 in the mail. After skimming forum posts on previous fy6600 and fy6800 models, am I right to assume they use a 50 MHz clock for their FPGA? This one uses a 10 MHz oscillator, which would make a tcxo/ocxo upgrade or external reference input much easier to add!

 That's an interesting twist to the tale.

 I wonder if they're using an NB3N502 14MHz to 190MHz PLL Clock Multiplier chip with a 10MHz xtal resonator or another cheap smd XO chip? The 3N502 can be driven from an external clock source ranging from as low as 2MHz to as high as 50MHz or else generate its own clock with XTAL resonators from 5 to 27MHz which in both cases can be multiplied up by factors of 2, 2.5, 3, 3.333, 4 and 5 to provide an ultra low jitter clock output so it's certainly a contender as a 50MHz clock source which I believe is a fairly common FPGA clock frequency.

 However they've made the change from a 50MHz smd XO clock to a 10MHz XO, it should simplify an OCXO or TCXO upgrade. No need to add a 3N502 in place of the shitty little 50MHz smd XO chip as we've had to do with the preceding models in this case.

JBG