| Products > Test Equipment |
| FeelTech FY6600 60MHz 2-Ch VCO Function Arbitrary Waveform Signal Generator |
| << < (366/450) > >> |
| rhb:
Absolutely fantastic post! With regard to the frequency response of the factory units, my FY6600 will produce a clean 10 nS pulse at a 10 MHz repetition rate. I've not looked at it on my LeCroy DDA-125yet, just an Owon XDS2102A I bought for the sake of the 12 bit 500 MSa/S ADC and 20 Mpt buffer depth. The Owon showed a bit of ringing after the impulse, but until I do some more testing I don't know which is responsible, the DSO or the AWG. The Owon does not have a 50 ohm input, so the thru terminator may be a factor. I've got a Tek 11801 supposed to arrive Friday. I received a pair of 12.5 GHz, 23 pS rise time SD-22 sampling heads today. So with a bit of luck and hard work I might have some good data on the F***Tech next week. Then I have to contend with how to test things like the SD-22, or worse yet, an SD-32 with a 7 pS rise time. I started a thread in Metrology on generating <3 pS rise time steps. I swore I'd stop at 3 GHz, but TEA got me anyway. |
| Johnny B Good:
--- Quote from: rhb on March 13, 2019, 11:08:43 pm ---Absolutely fantastic post! With regard to the frequency response of the factory units, my FY6600 will produce a clean 10 nS pulse at a 10 MHz repetition rate. I've not looked at it on my LeCroy DDA-125yet, just an Owon XDS2102A I bought for the sake of the 12 bit 500 MSa/S ADC and 20 Mpt buffer depth. The Owon showed a bit of ringing after the impulse, but until I do some more testing I don't know which is responsible, the DSO or the AWG. The Owon does not have a 50 ohm input, so the thru terminator may be a factor. I've got a Tek 11801 supposed to arrive Friday. I received a pair of 12.5 GHz, 23 pS rise time SD-22 sampling heads today. So with a bit of luck and hard work I might have some good data on the F***Tech next week. Then I have to contend with how to test things like the SD-22, or worse yet, an SD-32 with a 7 pS rise time. I started a thread in Metrology on generating <3 pS rise time steps. I swore I'd stop at 3 GHz, but TEA got me anyway. --- End quote --- @rhb It's very kind of you to offer your thanks for that post. The more usual response I've had to similar posts made in usenet groups in the past have tended to be of the "TL:DNR" kind, leaving me feeling as though I'd just been casting pearls before swine. :( What I miss in posting into a web forum is usenet's ease of use. However, what I definitely don't miss in posting to this web forum is all the back biting, personality clashes and outrageous displays of personality disorders and the out and out troll postings so typical of usenet. It's such a refreshing change not to have one's efforts at making a positive contribution to a discussion being disparaged for its "full and comprehensive coverage". To some extent, I do see their point when in retrospect, I attempt to read out my paragraph length sentences out loud without turning blue from lack of breath. >:D I guess the folks here must have much longer attention spans (and not have to move their lips when reading). ;) In any case, I post not to seek glowing testimonials for my prose so much as simply to try and contribute something useful or interesting to the discussion so no further thanks are sought nor desired. However, do feel free to pull me up whenever I make any errors or commit omissions. I'm not seeking to lay down ground rules to insulate me from receiving any well deserved criticisms for such errors and/or omissions whenever I might be so careless. :) Having dealt with the "Thank you" for the "Thank you" note, it's back to the business in hand, namely the pulse performance of the FY6600. I'd assumed you'd been referring to the 'impulse' waveform and so tried it with my AWG for comparison. I must have misinterpreted your meaning of the waveform but this is one that's always been problematical for me (presumably because I failed to take note of the much lower upper frequency limit that applies). However, at exactly 2MHz, the impulse waveform disappears. It was only after trying lower frequencies such as 10 and 100 KHz that I realised that there seems to be a bug in that at exact multiples of 100KHz, the impulse disappears completely! :-// Further investigation reveals that it stops producing predictable pulse repetition rates above the 32KHz setting. After checking the specifications given in the user guide appendices, I see that the upper frequency given for pulse waveforms is quoted as 10MHz. However, the minimum pulse width is specified as 20nS which seems to be at the heart of the 32KHz limit for the impulse waveform since it reduces below this limit at a mere 8KHz, growing fatter as the frequency is reduced until at 10Hz, it becomes a 120 microsecond wide impulse, reducing in proportion to the period of the pulse repetition rate before it hits the 20ns limit at 8KHz dropping to 10ns at 15KHz before finally dropping to a reduced amplitude pulse at the limiting width of 8ns at 32KHz after which, the repetition rate just goes to pot. I guess that's simply the consequence of trying to provide an impulse waveform from an arbitrary one loaded into the predefined memory slot labelled "Impulse" which makes it one of the least useful of the preloaded waveform options in the machine if you're regularly working with frequencies beyond the audio range. I ended up using the Sinc waveform at 2MHz to trigger my scope when I was comparing the 2MHz PPS signal from my GPS module. Having run all those tests on the impulse waveform, it has now become quite clear that you weren't referring to this one but most likely the square waveform, possibly at the 10% duty cycle, assuming you were referring to the width and not the Tr and Tf times which on a 20% duty cycle on mine, gives a clean 20ns wide pulse with Tr and Tf times of 6.8ns according to my Siglent SDS 1202X-E. My ambitions to owning fast rise and fall time square wave/pulse generators are rather more modest than yours. For my immediate needs, sub nanosecond will do me for the time being since I'm only interested right now in verifying Siglent's claim of Tr and Tf figures of 1.8ns for the 1202X-E. The FY6600 can only offer at best a mere 4ns or so with a following wind so nowhere near anything to tax the scope. The 1ns rating for the NB3N502 clock multiplier chip and the quad clock buffer chips is just quick enough to verify the validity of Siglent's claim which is why I mentioned it. One nanosecond is just the beginning of my possible adventures into the world of picosecond timings. Regarding the 3N502 chips, I might have goofed with that choice of clock multiplier since, deep within the pages of the data sheet, there is a reference to a minimum lower frequency limit of just 14MHz on its output. I'd overlooked this fact when seeing the minimum multiplier option of 2 along with a minimum clock input frequency of 2MHz so I could find myself having to multiply the 4MHz PPS output to 20MHz and follow it with a divide by two flip flop chip if it turns out that 10MHz is actually too far beyond the range of the 3n502's lower frequency output limit to work or work reliably. I've got some more testing to do. In the meantime, in case you've not read my latest postings to the u-blox GPS module thread, I've managed to successfully divorce the ornamental Patch antenna from the module and repurpose it as a plug in passive antenna, proving that there wasn't anything wrong with it, other than the very strong possibility that it should never have been glued to a groundplane containing some signal carrying tracks in the first place. Indeed, it works noticeably better than that 3/4 wave wire antenna I'd made up to test the module whilst awaiting delivery of an active mag mount patch antenna with a 5 metre SMA male plug terminated cable (bargain of the year at a mere 3 quid from a UK supplier apparently, since I couldn't find anything as good for less than a fiver - even passive patch antennas from China were only a quid cheaper!). Anyway, now that I've sorted out the mystery of the useless patch antenna, I can procrastinate no longer over the question of the 3N502 chips and 10MHz output. I'm finally going to assemble one into a test jig on a plug in breadboard and face the consequences. JBG |
| rhb:
JBG As a practical test instrument one of Leo's 40 ps square wave units is the way to go unless you insist on building it yourself. Leo is using a Maxim 3949 LED driver with 22-36 ps Tr and Tf. There is a moral hazard in buying one though. After you look at the plot of your unit made with Leo's CAS803 using and SD-30 head you may want one for yourself. Interestingly, my quest for a really fast edge has brought me full circle to the problem of reverberation due to a source embedded in a layer with a perfect reflector at one boundary. Which also turns out to be the same problem as constructing an ultra wide band antenna using an unbalanced feedline. I'm using the F***Tech FY6600 V 3.1 "Adj Pulse". The settings for the attached figures are: 10 MHz frequency 2 V amplitude 10 ns pulse I've not observed any repetition rate issue other than the BW limitation of the output. I was able to raise the repetition rate to 60 MHz. I think it worth noting that CH1 stopped producing output when set to "Adj Pulse". I then tried CH2, but it does not have "Adj Pulse". However, when I went back to CH1 and changed the waveform settings the 10 ns pulse returned. I tried the "Impulse" and it is very poor relative to the Adj Pulse. I'm using an Instek MSO2204EA with a 50 ohm thru termination taking the outputs with a 2 ft piece of RG-58. DS0004 & DS0006 are the FY6600 going as fast as it will go. DS0005 is a Keysight 33622A going as fast as it will go. DS0004 is with the amplitude at 2 V and DS0006 is with it set to 0.5 V. DS0007 is my 100 ps pulser from Leo. Leo-spike is the CSA803 & SD-30 (9 ps Tr, 40 GHz BW) results. The Keysight is clearly *very* good, but at a list of 60x the F***Tech. The apparent ringing on Leo's impulse generator is the result of Instek using a zero phase sinc(t) interpolator instead of the correct minimum phase interpolator. Setting dot mode turns off the interpolation and with infinite persistence you get the result shown in DS0008. BY digitizing the plots Leo provided and DS0008 one could derive the transfer function of the Instek AFE. I think it worth noting that for testing scopes, Leo's square wave unit is a better choice. I have both. The impulse unit is much more convenient for TDR work. Have Fun! Reg |
| Johnny B Good:
--- Quote from: rhb on March 19, 2019, 02:27:31 pm ---JBG As a practical test instrument one of Leo's 40 ps square wave units is the way to go unless you insist on building it yourself. Leo is using a Maxim 3949 LED driver with 22-36 ps Tr and Tf. There is a moral hazard in buying one though. After you look at the plot of your unit made with Leo's CAS803 using and SD-30 head you may want one for yourself. Interestingly, my quest for a really fast edge has brought me full circle to the problem of reverberation due to a source embedded in a layer with a perfect reflector at one boundary. Which also turns out to be the same problem as constructing an ultra wide band antenna using an unbalanced feedline. I'm using the F***Tech FY6600 V 3.1 "Adj Pulse". The settings for the attached figures are: 10 MHz frequency 2 V amplitude 10 ns pulse I've not observed any repetition rate issue other than the BW limitation of the output. I was able to raise the repetition rate to 60 MHz. I think it worth noting that CH1 stopped producing output when set to "Adj Pulse". I then tried CH2, but it does not have "Adj Pulse". However, when I went back to CH1 and changed the waveform settings the 10 ns pulse returned. I tried the "Impulse" and it is very poor relative to the Adj Pulse. I'm using an Instek MSO2204EA with a 50 ohm thru termination taking the outputs with a 2 ft piece of RG-58. DS0004 & DS0006 are the FY6600 going as fast as it will go. DS0005 is a Keysight 33622A going as fast as it will go. DS0004 is with the amplitude at 2 V and DS0006 is with it set to 0.5 V. DS0007 is my 100 ps pulser from Leo. Leo-spike is the CSA803 & SD-30 (9 ps Tr, 40 GHz BW) results. The Keysight is clearly *very* good, but at a list of 60x the F***Tech. The apparent ringing on Leo's impulse generator is the result of Instek using a zero phase sinc(t) interpolator instead of the correct minimum phase interpolator. Setting dot mode turns off the interpolation and with infinite persistence you get the result shown in DS0008. BY digitizing the plots Leo provided and DS0008 one could derive the transfer function of the Instek AFE. I think it worth noting that for testing scopes, Leo's square wave unit is a better choice. I have both. The impulse unit is much more convenient for TDR work. Have Fun! Reg --- End quote --- @rhb Thanks... I think. :) It took me several minutes to tie the descriptions to the scope trace captures before I could appreciate what they were showing me. Not a criticism, just a reflection on the complexity of describing the results even with "pictures that (are supposed to) say a thousand words" and the fact that I wasn't at my sharpest at five in the morning. I'm a Night Owl but even I have my limits which I all too often exceed. :( I happened to be using CH2 of the Feeltech (it's less of a stretch for my half metre RG58 BNC lead to reach the scope's CH1 input) to trigger the scope against the 2MHz square wave on the PPS line being displayed on CH2 so I too fell foul of the 'missing adj pulse' option until I re-read your experience and repeated those same steps to discover, as you did, that extra option that curiously only exists in the Feeltech's CH1 wave menu. I didn't go all the way to 60MHz, just the 10MHz point but I could see no difference in the pulse shape between 10 and 2 MHz so figured I wasn't going to discover anything new at the 60Mz setting. I got similar rise and fall times (just over 4.5 and 4.6 ns) which didn't alter with pulse width (I tried 10 and 20ns - the latter looks like a nice approximation to a square edged pulse on the Siglent). I didn't see any change either in the timings with amplitude settings right up to the 20vp-p limit (I guess I have the THS3491 opamps to thank for that) so set it to 5vp-p for the rest of my tests. Rather intriguingly, the GPS module's drop/add of a 20.8333ns pulse from the 48MHz TCXO clock to keep it in sync with GPS time displayed a hiccup effect on its once per half to five second correction interval I'd been observing over the past few days that I've been assessing the module's suitability to directly generate a 10MHz reference even with the help of that 3N502 clock multiplier chip (if it'll function at such a low output frequency) to get rid of the horrendous jitter that arises out of trying to directly divide down to 10MHz from a 48MHz clock. Going back to the Sinc waveform eliminated this hiccup effect, taking it back to its normal synchronising behaviour. Presumably, there's something going on with the adj pulse that's introducing this hiccup effect but I haven't had a chance to investigate this oddity any further. As for the impulse wave option, that's barely good enough for use in the audio frequency range being somewhat useless at any frequencies beyond the absolute limit of 32KHz. You can get output in the MHz range but you need to be aware of it's random repetition rate and the fact that it disappears entirely at exact multiples of 100KHz. The NEO M8N module I'm experimenting with is endowed with a 48MHz TCXO rather than a simple XO but it's not adjusted in frequency to keep it locked to the GPS clock signal as I'd originally and rather naively thought had been the case. It's quite clearly left to run at whatever frequency the TC happens to adjust it to and the control is merely a matter of dropping or adding an extra cycle to keep the PPS synced to GPS time regardless of the chosen frequency it's been programmed to run at. The lower the frequency, the smaller these phase shift adjustment steps become. I programmed the PPS to 12MHz (another jitter free division ratio) and saw the expected 90 degree jumps, confirming my estimated 15 degree jumps at 2MHz (I plan to multiply this back up to a jitter free 10MHz with the 3N502 chip - or failing that, 20MHz followed by a divide by two flip flop). Depending on the time constant used in the 3N502's PLL circuit, this might give me a sufficiently sanitised 10MHz clock reference suitable for use by T&M and communications kit, possibly helped by filtering to a pure sine wave output. This is something else for me to investigate once I've steeled myself to set up the clock multiplier chip add-on. I suspect the only way to overcome these phase shift jumps will be to throw a VCOCXO module into the mix but I'd like to justify this expense for myself before shelling out on one. I don't mind the expense of a component when I know without any shadow of a doubt why the component in question is absolutely indispensable to the whole project. The mention of my GPSDO project above might look like 'topic drift' but it is germane to my plans to modify the FY6600 to accept an external 10MHz frequency reference as per Arthur Dent's contribution to this thread over a year ago. At least one of the three 3N502 chips I bought will be getting put to good use in this planned modification. :) Obviously, there's no point in adding a 10MHz external clock option without there being a realistic chance of acquiring or fabricating a GPSDO to drive it. I could run the two jobs side by side but I'd prefer to get the GPSDO sorted out (if not fully completed) first before attacking the FY6600 with yet another modding effort. As dramatic an improvement to the frequency stability and accuracy as I've made to my FY6600 with that 50MHz 0.1ppm oscillator board, that's just been a taster of what can be achieved with a GPSDO reference. That basic oscillator upgrade had allowed me to run tests with the current GPS module that would have been virtually impossible with the original oscillator chip which had turned the business of trying to compare frequencies with the 'scope into a game of "Chase Will 'o the Whisp". As good as it now is, I just want the even better frequency accuracy and stability that only a GPSDO can provide. I completely understand your own TEA issues. I might think I'm not going to be going to the same "extreme" right now but, after reading, back in November, Arthur Dent's OCXO mod with the intention of adding a 10MHz external reference socket and thinking it was rather an OTT modification and electing to go the cheaper TCXO route instead yet now finding myself going to the same "extreme" of adding an external 10MHz reference socket, I'm not quite so sure I'm as immune to TEA as I think I am after all. :-\ JBG |
| rhb:
There were 3 signal sources, The 6600, the 33622A and Leo's 100 ps impulse. The DSO sinc(f) issue is *only* visible on the 100 ps signal when the DSO is sampling at 1/10th of the pulse duration. I have one of Leo's two channel GPSDOs which has been extremely satisfactory. The single channel version is cheaper. I bought a 6 output 10 MHz GPSDO from Roadrunner, but have not yet gotten around to testing it. |
| Navigation |
| Message Index |
| Next page |
| Previous page |