Apologies for the tardy follow up but I decided to try ordering the 'scope from Labtronics the following day to see whether I could pay without having to revive or create a paypal account and discovered the Guest Paypal option which I chose to complete my order.
The 'scope arrived two days later and I've been a little preoccupied in the six days since then gaining familiarity with 'My New Best Friend' (that and dithering over whether to order a Siglent SDG1032X for
360 quid or go for the much cheaper FY6600-60M at a more disposable
As you can probably imagine, such choices don't magically resolve themselves these days without spending hours and hours checking out Youtube review and tear down videos ad-nauseum until you can't find any more new ones, no matter how hard you look, created by people with a proper understanding of things like 'mains tingle voltage' from every commodity SMPSU ever used without benefit of safety earth connection and the importance of terminating a 50 ohm BNC cable with a 50 ohm resistive terminator and so on and so forth.
When I have to read between the lines (more a case of looking past the misunderstood misgivings which seem to pointlessly pad out a 10 or 15 minute video into an hour long treatise on how to modify this or that aspect of the kit in question), that can be quite a few evenings (even afternoons - really, most of the day but with natural breaks and lunch, tea and supper breaks plus the odd household chore and food shopping trip) worth of my time before I'm finally convinced I made the best cost/ performance trade off choice I possibly could (in the circumstances... I reassure myself - I finally plumped for the FY6600-60M).
Said FY6600-60M is promised to reach me any day between Saturday and Monday - Sunday seems unlikely, it's via Royal Mail so even Saturday seems optimistic.
After spotting that the only difference in spec between the 30, 50 and
60 MHz versions lay in the upper sine wave frequency limit (all else being identical), it did occur to me that the differences might simply have been a matter of a firmware upgrade and possibly a couple of capacitors or some such so I tried searching for a 60MHz hack for the slightly cheaper 30MHz model.
Surprisingly, in view of the various mods referenced in the EEVblog thread, none were forthcoming. However, with typically less than a tenner price difference between the 30 and the 60MHz models, I wasn't too put out.
An 8 to 10 quid price premium on the 60MHz model over the 30MHz one seemed a fair enough price point but that hadn't stopped me from looking for a possibly even cheaper option. Now that I'm satisfied I hadn't overlooked an opportunity to pay even less for a 60MHz sig gen, I feel a little happier over splurging that little extra on the 60MHz model. At least it's ready to go without the faff of hacking a 30MHz unit, leaving me to address just the other perceived shortcomings so roundly discussed in the EEVblog here:
Apologies if Pan has line wrapped the url. You can try the Tinyurl below if it has.
I've got as far as the eighth page in that 69 page thread. I did have a look at the last few pages to get a hint of just how long it had lasted. However, "cutting to the chase" in a long thread like this isn't a sensible idea when there seems to be some worthwhile discussion about improving the product. I just might learn some useful things about my next 'new toy'. :-)
I'm not sure whether your thanks were directed to both speff's and my posts combined or just mine alone but, in either case, your thanks are appreciated anyway. :-)
How recently? Are we talking days or weeks ago? Mine turned up just after 10am Friday morning with the regular mail delivery, just over 50 hours after placing my order (it beat the deadline for earliest ETA by two days and RM's tracking by several hours). For anyone interested, it's sporting the 3.3 version of firmware (and the latest revision mainboard as mentioned in the EEVblog).
I've been familiarising myself with its interface and discovering more ways to become confused with my Siglent SDS1202X-E's features as well as those of the FY6600-60M, hence my tardy response, not helped by my plodding through the last dozen or so pages of that EEVblog thread.
I've added a 47K resistor across the low side resistor in the 10K+10K potential divider network which drives the ref input of the adjustable shunt reference chip which drives the opto-isolator in the PSU to raise the 4.95v on the +5V line to 5.49v primarily so as to raise the +/-12v rails by another volt to get them to circa 12.7v (+ & - rails) rather than circa 11.7 volt.
It seemed a reasonable first mod to do since it was just a case of soldering a pin through hole 1/8th watt resistor onto the underside of the PSU board to bring the + & - 12v rails onto the better side of the tolerance range (normally +/-10%). Total power consumption went up by less than half a watt on its idling consumption of 5W ("Powered Down" using its front panel "On/Off" button).
Even at maximum output with both channels driving 50 ohm dummy loads at
10MHz 20v p2p square wave, it draws just under 9W maximum off the mains supply. Actually, after testing with and without the 50 ohm termination, it looks like the output Z is more like 79 ohms than the claimed 50 ohms (a claim also backed up by the pairs of 100 ohm resistors discovered by DerKammi's reverse engineering of the circuit boards).
Open circuit versus 50 ohm loading of a 50 ohm impedance source should only result in a drop down to 50% of the open circuit voltage, not a drop down to 37.1% as I've just discovered. I've a sneaking suspicion that all those 100 ohm resistors may turn out to be 150 ohm as per the TI reference 150 ohm units of loading mentioned in their data sheets.
However, the fixing of such 'errors' (in this case, a simple matter of shunting the 150 ohm resistors with 300 ohm resistors) is probably best left to when the dual current feedback op-amp is upgraded to a pair of seperate faster op-amp chips by those who feel the need to go against the ill conceived 'wisdumb'[1] of "If it ain't broke, don't fix it!".
Adding a 47Kohm resistor to the PSU is a simple and easy to reverse mod to correct for the voltages being on the wrong side of the tolerance range. In this case, a little more voltage is "A Good Thing" in that it gives the amps that extra margin against clipping.
I wouldn't bother about the other suggested mods to the PSU since the
10?F 450v HT smoothing cap is fine as it stands without replacing it with a 68?F 400v one. After looking at several 'scrapped for spares' 'Universal' 100-240vac smpsus that I had to hand, the design rule appears to be "Specify 1?F per watt of rated output power and no more" (and this is for 'universal' smpsus designed to work off as low a mains voltage as
90vac rms!). For those of us blessed with 200 volt plus mains supplies, the need to 'upgrade' that ht smoothing cap simply doesn't exist. :-).
Replacing 220?F 16v caps on the 12v lines with 25v versions is just pointless. A much better option would have been to use 16v 470?F caps instead if you'd really thought some improvement had been required (that should half the 300mV or so of HF ripple I saw when I looked at the 12v rails with my 'scope probes - the 16v rating simply isn't an issue, not even if you push the rails to 15v!).
For the money, I think these signal generators are exceptional value, warts and all. The nearest competing products by Siglent are five times the price and more (and Siglent are at the cheap end of this market). Most of the perceived shortcomings can be overcome using relatively inexpensive mods.
However, the one shortcoming highlighted in the EEVblog forum that was deemed the most seriously difficult to deal with was the matter of timing jitter (in this case, a mere +/- 4ns). Right now, I'm looking at the example traces in Siglent's features pdf of their sdg1000 series generators showing how they've overcome this issue of jitter with their 'innovative EasyPulse Technology' and it seems to me that all they've done is sacrifice 8ns of resolution by averaging two or three clocks' worth of sampling to smear out that 8ns rise time over a nicely displayed
16 to 24 ns slope.
It strikes me that this 'innovative EasyPulse Technology' is being oversold. For most hobbyist users, I think this feature may be largely cosmetic. Anyone who finds such +/- 4ns of jitter troublesome would be far better off building their own custom clock generators rather than rely on a cheap general purpose sig generator anyway.
BTW, since I'm posting back into this thread, I think I aught to mention that since my acquisition of a 'scope blessed by a DC coupling option (both my boat anchor 'scope and the sound card I used to to examine our
50Hz 240vac supply wave-form had no such option), I have to take back my earlier disparaging remarks about the Rigol 'scope I saw being used to show the mains wave-form in a Youtube video demonstrating the features of a Parkside inverter genset.
It turns out that the slope on the truncated tops of the 50Hz mains supply voltage I'd witnessed had simply been an artefact of using AC coupling with a significantly high (as far as 50Hz voltages are concerned) HPF turnover frequency characteristic (circa 8Hz for the sound card I believe and a rather unfortunate coincidence that a 50 years older
5MHz B/W boat anchor 'scope should seemingly share a similar HPF cut off effect).
I saw exactly the same wave-form using my recently purchased Siglent SDS1202X-E in the AC coupled setting but as soon as I selected the DC coupled setting, this morphed into a flat topped sine wave with no hint of any slope whatsoever. I feel so stupid for making such a sweeping statement about the distortion on our UK mains supply. However in my defence, until a week ago I hadn't the means to eliminate the HPF effect on my "'scope plots". It just goes to show that it's never too late to learn (given the right tools).
[1] There's seemingly no shortage of those so blind to the glaring yet easy to fix defects in manufactured kit as to willingly subscribe to this "If it ain't broke, don't fix it!" mentality to annoy those of us who can see all to clearly such deficiencies. They're clearly related to those who just keep getting in the way of those busily fixing all those "impossible (or too expensive) to fix problems".
I ordered it on the 7th, seems it will come on the 13th, so effectively onl y 3 business days :-)
I will test mine, when it arrives, and write back
with
this
s
A lot of good info, must dig deeper into it. I guess I will only really do what is nessesary, is busy at work, so cannot spend much time on instrument tweeking
Just today I was doing a test on the HP 33120A, and it has 25ns rise time, so couldn't complete my test. The FY6600 is about 7ns, so a lot better, 10 times cheaper :-)
Well, seeing as it's now the morning of the 14th (I've been pondering cost effective ways and means of improving my new toy - rereading the EEVblog from start to finish... again! - and I had to get up early to put our wheelie-bin out 'cos I forget to do so last night, therefore my being up at what I usually regard as "Silly O'clock in the morning.", hence this delayed response), I suspect you may well have had a chance to gain some hands on experience by now. :-)
One handy "shortcut setting" that's well worth remembering is the, "One long press of the "OK" button (the rotary knob) until it beeps", one which stores the current settings of the two channels (freq/amplitude etc) to use as subsequent power on defaults. That can be a real time saver at the start of your next session.
One of the things I was having trouble with was my tendency to press the on screen icons of the soft menu rather than the actual buttons. A strange affliction considering I had no such problem with the soft menu buttons on the Siglent 'scope - I guess the tiny screen on the FY6600 must be the main factor in this case.
One of my "cost effective mods" I've researched since then has been going through my collection of scrapped smpsu parts, notably the high speed low Vf rectifier diodes as replacements of the mis-identified by ebel0410 "LT7506"s which are actually FR107s, a cheap high speed version of the classic 1N4007 1A rectifier diode. His replacements, 1N5819s, were hardly any improvement despite being high speed 3A rated diodes.
I picked out all the likely candidate shotky rectifier diodes from my collection and checked them out on my DVM's diode/continuity buzzer setting. this passes a constant 1mA test current through the DUT and shows a reading corresponding to the diode's Vf in millivolts. Standard silicon diodes show, as one might well expect, values of 550 to 600, depending on their current ratings.
My candidate diodes ranged from a low of 141 (ESAD83-004R) to a high of
247 with one of a candidate pair (MBR3045CT) at 201. The FR107s I'd removed from the PSU board measured a whopping 518/520 on my DVM. Looking at all the diode data sheets I downloaded, I thought I might see an extra half volt improvement by shoehorning a couple of these TO220 dual diodes onto the PSU's circuit board but I got a little more than I bargained for with the positive rail going up again from the previously boosted 12.78 to 13.8 volts with the slightly less loaded negative rail going up by a similar amount to just over 14 volts.
Subsequent testing after the diode upgrade revealed a very slight but definite reduction in this maximum power consumption (about a 100mW's worth) which is a rather reassuring result after applying such a modification to an SMPSU. :-)
I've since done some more testing and it seems the output Z is close to the 50 ohm mark after all. The big problem here is the use of a dual high voltage CFB opamp (THS3002I which Texas Instruments refused to acknowledge the existence of the variant used by Feeltech!) which might have been fine in the first 15MHz model but which has now been pushed to its limits and way beyond with the 25, 30 and 60MHz models.
The presence of unpopulated chip locations (U21 and U22) indicates that Feeltech must have considered their frequency upgrade options and the need to upgrade that dual high voltage CFB opamp to a higher performance pair of single opamp chips, notably the THS3001 or THS3095 (or, now that TI have produced an even higher spec version of the 3095, the THS3491).
The PSU mod is all well and good (any mod that improves an SMPSU's efficiency can only be "A Good Thing"(tm) imo) but in this case, it's rather wasted on the THS3002I dual opamp chip that Feeltech neglected to upgrade to match the ever increasing demands of the later -25MHz to
-60MHz models.
The next obvious no-brainer modification (after all, you did choose a
60MHz model to at least provide 40 to 60 MHz 20V pk2pk *sine* waves... didn't you?) is to replace that dual opamp chip with a pair of THS3001s (or THS3095s or even THS3495s). Unfortunately, unlike the PSU diodes mod, this one is unlikely to be satisfied by a change of BoMs using recycled
China) and upwards to 40 quid or so in an order for a pair of opamp chips from Digikey or Mouser.
A pair of 3091s will do nicely (the 3495s are a little OTT for 60MHz 20v Pk2Pk sine wave duty and have the same minor issue of dealing with the surplus to requirements Vref and /PD pins which also curse the 3095 chip).
Either of those three chip options will provide a much overdue performance upgrade but try as I might, I can't see me satisfying my requirement for under 33 quid which represents an extra 43.62% of the price I'd paid for the FY6600 in the first place. I think that's one mod best left on hold for the time being (at least until I've exhausted all other possibly cheaper alternatives that might become available with a little more research).
=====snip=====
And there you've discovered its attraction to the cost conscious purchaser. A better performing (with some caveats) and cheaper version of all those hideously expensive mainstream contenders. :-)
JOOI, I took a look at the data sheet for that part. I thought a search on that part number might lead me to a cheaper alternative (the best
(including the VAT and free delivery) from Farnell/Element14).
Obviously, as far as upgrading the FY6600 to a better CFA high voltage opamp is concerned, its low voltage rail (+/-6v max) precludes it from further consideration. If you're thinking of using the THS3491 as a replacement to an existing LMH6702, you'll have to increase the supply rail voltages to a minimum of +/-7v (14v single rail).
In the meantime, I'm still pondering the wisdom of upgrading a cheap
looking 60MHz sine waves at 20v Pk2Pk.
sine wave output is still bargain basement territory. I'd just like to be certain that I hadn't overlooked a cheaper source for those opamp chips before placing an order with Farnell/Element14 is all. :-(
Ah, to hell with the expense! I've just reviewed my previous post and Farnell/Element14 have been the cheapest source I'd been able to track down - there simply aren't any cheap Chinese sources showing up at all, so I've no further choice in the matter of supplier anyway, it's either Farnell/Element14 or not at all.
It's a question of which parameter is the most consistent and the need for brevity. :-( It doesn't take long to adjust to this abuse of the definition of "Amplitude". :-)
I don't see such a fast drop off, maybe 3dB down at 60MHz versus 10MHz down to 5Hz at the 5V Pk2Pk (AMPL) setting. I expect this to improve once I've installed the THS3491 opamp chips.
I'm testing with a Siglent SDS1202X-E DSO using the 'scopes own 200MHz rated probes on the 10 to 1 setting, plugged straight into the output sockets via the BNC probe tip adapters. If you're using quality 50 ohm BNC patch leads of any length, you need to terminate the 'scope or LMS end to suppress reflections which would otherwise mess up your measurements.
It's simply physically impossible to generate "Perfect Square Waves". The best you can hope for when pushing right up to the upper frequency limit, is a reasonable approximation to trapezoidal with minimal 'ringing' artefacts.
The frequency response is the same for sine, square and triangle waves (50% duty cycle of course) at the 5v AMPL setting. For other, more complex waves, all bets are off on their frequency plots once you go above 10MHz (at least with the incumbent THS3002 opamp in residence).
Yeah, that's the problem with modern measuring instruments these days, no gravitas whatsoever. Even the one time champion of "Gravitas", Tektronix, is no better than any of its much cheaper Chinese made rivals these days - I suspect that's because even Tektronix kit is now also of Chinese manufacture (yet they still get away with charging an order of magnitude more for what is essentially rebadged commodity DSO and ARW generator kit of Chinese manufacture).
For those obsessed by the very real risk of zapping the chips in a DUT [1] from the half mains Hi Z voltage that's ever present should you neglect to bond the BNC shields to the ground reference of the DUT in question, there's considerable merit in swapping out the original smpsu for a case filling analogue PSU with its massive transformer and smoothing caps and regulator chip heatsinks just for the extra gravitas (weight) alone that this would bestow upon the FY6600[2].
Personally speaking, if I want to lend my generator additional gravitas, I've got a roll or two of roofing lead I can use to fill in the gaping void behind the display panel (and remove the useless tilt bar - the rear rubberised feet are set so far from the back edge that tilting the front up raises the rear feet clear of the work surface as it sits on the back edge of the case and the non-rubberised tilt bar - slippy hard surfaces all round!).
In fact, I'm rather inclined to unclip that tilt bar except for the fact that it'll reduce its gravitas just a tad - I'll leave it until I've added the missing ballast. :-)
OTOH, I might simply fit extra rubberised feet on the very back edge of the case and rubberise the tilting bar to defeat its current tendency to skate all around the worktop whenever it's propped up like this.
Yes, a Hell of a lot! :-( It wouldn't be so bad if the select buttons weren't so awkward to reach and operate (they need a lot of pressure in just the right spot before they register the button press).
You must have younger eyes than mine (or a better prescription of bi/tri/ vari focals). I find it's a little too bijou for my taste. However, ignoring the need to be able to accurately focus on the screen, it does have sufficient pixels for the job (coincidentally, the same number as in the Tektronix TPS2014B's physically larger display[3]). It's just a pity that they couldn't have been spread out over a larger screen area, say a
20 to 25 percent larger diagonal.
There's just room to accommodate such a larger display option with a little rearrangement of the LHS panel buttons and the RHS soft menu buttons (or better yet, a touch screen to eliminate those five pesky physical Function buttons). Of course, this is all added cost so little chance of seeing such "cosmetic" improvements - what you do expect at this price?
Ergonomics aside, the performance is remarkably high for its price even before its target market demographic succumb to the urge to "Fix It" despite it not being seriously broken so as to improve it even further (out of all recognition of its original OOTB performance in fact).
[NOTES]
[1] The 1nF switching hash suppression cap might represent a very Hi Z at
50/60Hz sine but the problem isn't the microamps of mains voltage leakage, it's the high amperage transient discharge of a peak voltage that can range from 85 to 170 volts from said capacitor into a poor defenceless gate ESD protection diode (or worse still, a gate bereft of any such diode protection).
[2] I think the best way to solve this issue of half live mains is to use a simple 1:1 15VA 240vac isolating transformer between the incoming mains socket and the original smpsu. If possible, use a split bobbin type of transformer to minimise the stray leakage capacitance between windings to just a few dozen pF.
However, even the classic single bobbin construction with insulation tape layers separating the windings should keep the stray capacitance to less than 100pF. If you pick a good quality transformer, you can keep the additional transformer losses below the half watt mark. Alternatively, you could contrive an external floating mains supply using a pair of 30VA or larger transformers wired "Back to Back" via their secondary windings. That gives you half the capacitive stray leakage of a single transformer arrangement.
For instance, use a pair of 48VA rated 12-0-12 secondary with split 120v
120v primary windings transformers wired back to back via the hot ends of the 12-0-12 secondary connections, earthing the centre tap of the second transformer's secondary and series up the 2nd (output) transformer's split primary to provide the floating 240v mains feed to the FY6600 with a 2M high voltage resistor linking the centre tap to ground to act as a static drain. This will give you a low 'half mains' leakage voltage worst case 60v ac with a maximum of 30 microamps and just a few pF of leakage with a very low ESD risk.
Of course, all this could have been neatly sidestepped if the smpsu transformer had incorporate an electrostatic screen between the primary and secondary windings which could have been connected to the virtual ground of the HT rectifier circuit, leaving the secondary side totally disconnected from the half mains voltage via that mandated 1 to 5 nF EMI suppression capacitor.
There'd still be some capacitive coupling to the half mains voltage of the virtual ground but instead of a 1000pF's worth or more, it would have been just a few pF's worth via the shield foil and the secondary winding, reducing the coupling by around two orders of magnitude.
[3] I was quite shocked at the recent sight of a (presumably vintage - there's a CF slot under the display FFS!) Tektronix TPS2014B DSO shown here:
It looked like a less 'polished' version of a Siglent SDS1204X-E DSO
times that of the Siglent's price!). I find it hard to believe that anyone would want to blow that much money on Tektronix kit these days.
Thanks! That's a fascinating opamp. I imagine it's not too hard to get carried away, and put it into thermal limiting! I take it the Adoner FY6600 ARB uses that for its output driver?
I'm not sure where the limit is now, but it used to be that if you were dra wing more than 100W from the grid you had to build "power factor correction " into the front end of your power supply, which extracted a more or less s inusoidal current into a capacitor which fed your high-frequency inverter w ith a more or less constant high DC voltage - it obviously had to end up wi th enough 100Hz ripple to cope with delivering a more or less constant curr ent while getting something like a rectified mains sine wave current.
Smaller stuff - like the ubiquitous wall warts - could get away with clippi ng the top off the sine wave.
A proper modified sine wave has no third harmonic component, not a lot of f ifth harmonic component but everything above that is just as bad as it woul d be in a square wave. You could filter quite a bit of that out, but a fanc ier inverter might be cheaper.
Higher harmonics are good for warming up iron cores.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.