FeelTech FY6600 60MHz 2-Ch VCO Function Arbitrary Waveform Signal Generator

[SMB784]

I do wish though he would answer the 200MHz/5ns jitter question for his Rigol.

I actually have an answer for your question: its in the datasheet!

Look at page 3 of the datasheet for the DG1022.

It states that the jitter for non-sinusoidal waveforms is 6 ns!  So we are actually better than the jitter for the Rigol 1022G instrument, which has a retail of $250-$300 with a maximum sine frequency of 20 MHz, and a 100 MS/s sample rate @ 14 bits of vertical resolution.

The DG1022Z has an RMS jitter of 200 PS, and it costs $300-$360 with a maximum sine frequency of 25 MHz, a 200 MS/s sample rate, and a 14 bit vertical resolution.

For reference, our generator has a 60 MHz (90 with bluepill, if I remember correctly) maximum sine frequency, a 250 MS/s sample rate, 14 bit vertical resolution, and 4 ns jitter, and with upgrades it only costs ~$150.

That makes it better than Rigol's lowest cost generator on a performance per dollar level, and competitive with their mid range generators in terms of specs while significantly beating the price point.

I am certainly no longer worried about the jitter, as it appears to be a feature consistent with other generators in its class, and comes with a greatly reduced price point.

EDIT: Just as a comparison, the siglent generators are as follows:

SDG830 has 500 ps jitter, 30 MHz max sine freq, 14 bit vertical resolution, and 125 MS/s for $330

SDG1050 claims 40 ps jitter @ 25 MHz in the datasheet (I dont believe that, because the jitter is 8 ns for pulses), 25 MHz max sine freq, 14 bit vertical resolution, and 125 MS/s for $509

Still winning against Rigol & Siglent.

[soundtec]

Thanks for that comparison SMB , so really theres not much to say other than the Feeltech is great value for money ,either stock from the factory or with your own chosen upgrades .I can understand people might try to knock the Feeltech after paying several times the price for units which on paper at least seem to have worse specs .To be honest even if money was no object the feeltech would still be top of my list ,its kinda forced us all to take account and learn more about whats going on inside ,where if we had top of the line, no expense spared gear we might not have learned nearly so much about ARB waveform synthesis.

Sure if your time is worth 130 bucks an hour ,why not just go out and buy the latest and greatest from a prestige mark ,but Ill bet most of us on lower wages or retired with time on hand ,wouldnt trade back the time for money, lessons learned and friends made here, I know I wouldnt . Must get my head stuck into blue pill soon to access the extra 'goodies' Fremen has programmed in for us .If someone can point me the location of the blue pill set up/wiring for the FY6600 that would be great.
Cheers all once again and many thanks for the good spirited contributions that made this project come together.

[nctnico]

Sure if your time is worth 130 bucks an hour ,why not just go out and buy the latest and greatest from a prestige mark

Why bother with building something at all if you make that kind of money? For $130/h you can hire someone to build the projects for you!

[cybermaus]

Thanks SMB784. I guess I could have looked there myself.

So nothing special / smart in the Rigol.

The Siglents datasheet have 8ns for arbitrary, matches exactly their 125MS/s spec. But square seems very good with 500ps. Are they doing something smart?
It is possible the square and pulse are non-clock based, somehow. Maybe on an internal FPGA timer based (which is then internally a higher clock of course, but you cannot call that MS/s)

But no: I think no, just some spec trickery:
SGD800: 500ps + 0.001% (seems good) but 24ns rise/fall and 20ns asymmetric
So how do you measure a 500ps jitter if you have 24ns rise/fall as well as a 20ns PWM variation?


The timer based idea is a good one though. If I were to build a FPGA AWG from scratch, I may use that for square/pulse signals. Internal 1GHz (or higher) timer that drives the DAC with variable duration, so that even though only a 250MHz DAC, you can still have 1ns or better resolution.


----

Anyway, that means this is a good device, even with the tinker necessities. Except for the absolute and utter lack of support and warranty.
And to be honest, I think there is only one *must do* tinker necessity, and that is the grounding. The rest (crystal, opamp, psu) is only if you want to.
But to repeat: absolutely no warranty whatsoever.  If you want warrenty, go somewhere else.

[nctnico]

The Siglents datasheet have 8ns for arbitrary, matches exactly their 125MS/s spec. But square seems very good with 500ps. Are they doing something smart?

The first generation SDG1000 used a comparator fed with a sine wave to make a square wave. I doubt the newer models will be much different.

[cybermaus]

Ah. Yes, that is also pretty smart. Yes, that may be it.
still makes the 500ps a bit of a statistical thing though.

[SMB784]

Ah. Yes, that is also pretty smart. Yes, that may be it.
still makes the 500ps a bit of a statistical thing though.

Do you think that is something that we could implement within the existing FPGA fabric?  Does anyone have any idea how much FPGA resources are left in the FY6600 generator's stock configuration?

[cybermaus]

I am assuming we cannot do anything to the FPGA, other then program completely it from scratch.
But then again, I know pretty much nothing about FPGA programming. If you look at the start of this thread, I already made several FPGA assumptions that turned out to be false. (at least I learned a little)

What I do know is we can easily make a sine-into-gate wave ourselves. In fact, I think it is already there, on the TTL and/or SYNC-OUT ports.
We may need to replace the standard gate with a faster one, but after that we could add an extra OpAmp amplifier. Or even feed the gate output back into present opamp by means of an extra relay (so Fremen67 can control it with a spare GPIO, and the amplitude setting works)

[fremen67]

A bug report for the 0.6 firmware/software package:

Nothing wrong with the sweep module, but I used a different blue pill board for the 0.6 firmware and had to redo the calibration process.  Ch1 Amplitude went ok, but the control software crashed repeatedly while trying to do the Ch1 Offsets (see screen1.jpg).  Restarting the control software brought it back in the configuration in screen2.jpg, and resetting the blue pill brought it back in normal default mode.  I gave up on Ch1 and did Ch2 instead, both amplitude and offsets being set without a problem, then went back to Ch1 offsets, which also completed without further hiccups.  However, pressing either "End Calibration" button then causes another crash as in screen3.jpg (clicking "OK" shuts down the software).

Dave

Thank you for the feedback. I think I got it. It should not happen when you go first to the configuration tab and then only to the calibration tab. I wanted to speed-up the startup and only read configuration data when needed.... which I need in the calibration tab . I will correct it in the next release 

[fremen67]

Anyone with a V3.2.1 able to dump the flash memory and post it?

[Bob Sava]

With all talk about upgrades I've been thinking why I should upgrade power supply and why?  From what I've been reading it seems to me that replacing power supply would only make visible difference after opamp upgrade for higher output current and rail voltages.

I understand that smps it pretty barebone with undersized or missing capacitors but what is the manifestation of it's shortcomings? 

[Candid]

Replacing the smpsu by a linear one has the main reason in getting rid of the phantom voltage by keeping the floating ground. With the original setup you can measure high ac voltage from bnc ground to mains earth. This is because of the Y capacitors in a smpsu.

[DaveR]

If someone can point me the location of the blue pill set up/wiring for the FY6600 that would be great.

The wiring setup is here, soundtec - https://www.eevblog.com/forum/testgear/feeltech-fy6600-60mhz-2-ch-vco-function-arbitrary-waveform-signal-generator/msg1441974/#msg1441974

I assume you've got one of the aluminium shell USB dongles, and have downloaded the programming utility from the ST website?  A word about those dongles for the unwary - there are at least two types on sale, and I've got versions marked 'ST' and 'XTW' (as well as two other non-shelled variants).  At first sight the ST and XTW dongles look identical, but the pinouts are different, so any wiring diagram you see on the web may not apply to the one you've got.  Fundamentally, four connections are needed from the dongle to the blue pill board for the programming part: IO, CLK, 3.3v and GND, so just connect like to like and upload fremen67's firmware to the blue pill (five seconds to complete).  If you get any error messages, check your wiring, then check it again   Once the blue pill is programmed you don't need the dongle, so disconnect it and follow the wiring in the photos to connect the blue pill to the FY6600 main board - a single piece of 12 way M-F ribbon cable makes a nice tidy job and gives good connections (no need for breadboarding or veroboard with v3.2 boxes).  Lastly, don't forget the standard USB cable between the FY6600 and your PC, as I have on a couple of occasions     (Photos of my BP cable attached for reference.)

[Bob Sava]

Is Stm32f103c8t6 what works with bluepill?

[DaveR]

That's indeed the processor chip that's used on the blue pill boards.

[plb]

With all talk about upgrades I've been thinking why I should upgrade power supply and why? 

Those were my thoughts too. Then I noticed the distortion of smaller signals caused by the SMPS interference. It radiates a lot, and maybe there are ground bounce or ground loops. A linear supply should eliminate the interference. The first attached photo shows the FY6600 output signal when the amplitude is set to zero. The second shows the distortion when outputting 1V p-p.
(Edit. Photos attached)

Question for those who have replaced the SMPS - what is an acceptable ripple voltage on the +/- 12V and +5V outputs?

[DC1MC]

@plb, holly crap,  this is really bad !!!

Not even with the original POS mine didn't have that bad output, something is VERY wrong, at 10KHz on a 50ohm terminator with a T and connected to the scope, the signal looks perfect, no distortion is visible to the naked eye.


 DC1MC
 

[cybermaus]

Yeah, that is not a normal output, not even for a unmodified FY6600
You have an extra special bad one.

I never replaced my PSU, only moved Y-cap to ground and added 1M resistor, and my 1Vpp - 50 Ohm looks much better.
From memory, it looked much better before the Y-cap mod too.

[bugi]

Note, plb used apparently 10x probe, not 50ohm connection. (Also, if it indeed was a probe, how was it connected? With the ground clip "loop antenna" or with the small coil+spike?) That might explain some of the difference seen on the scope view. At least that is something I remember to be aware of from the videos about measuring power supply noise/ripple, and while this isn't a PSU, I think the idea still applies in this case.

[soundtec]

Hi PLB ,If the amplitude is set to zero then the unit shorts the bnc connector to ground via relay , so what your measuring in any case is whats induced into the loop .
On the highest amplitude range ie 5 volts and above and frequency set to zero I was measuring 150 microvolts into hi z at the bnc connector, this measurment was taken after the psu was replaced .

[plb]

From memory, it looked much better before the Y-cap mod too.

Thanks Cybermaus. This made me go back to basics. I had grounded the common as soon as I got the unit, and that was the configuration for those photos. When I removed the ground, the waveform improved significantly. I now have a 1K resistor in there that leaves the signal as clean as an open circuit.

Note, plb used apparently 10x probe, not 50ohm connection. (Also, if it indeed was a probe, how was it connected? With the ground clip "loop antenna" or with the small coil+spike?) That might explain some of the difference seen on the scope view.

Correct about the probe Bugi. However the SMPS frequency is 50KHz. Even considering harmonics, that's not really "loop antenna" territory.

Hi PLB ,If the amplitude is set to zero then the unit shorts the bnc connector to ground via relay , so what your measuring in any case is whats induced into the loop .
On the highest amplitude range ie 5 volts and above and frequency set to zero I was measuring 150 microvolts into hi z at the bnc connector, this measurment was taken after the psu was replaced .

Not quite, Soundtec. There is zero pickup by the "loop antenna" until I move it adjacent to the SMPS . Good point about the relay though. I hadn't noticed. Obviously what I'm seeing is being induced into the PCB and relay. Your result looks excellent though. Thanks.

Net result is that I'll stick to the Cybermaus solution, as modified, at least until I've finished my current project and am looking for something to do. Then I might consider a linear supply if someone can advise on what ripple is acceptable...

[cybermaus]

Ok, but note I put a 1M resistor, not a 1K one.

The reason is that I want to be able to use the output as float if needed. I read somewhere even brand devices with floating output really do have a 1M~10M resistor in there.
I guess an unforeseen but pleasant extra advantage is it will indeed reduce any ground loop induction.

[fremen67]

Hi PLB ,If the amplitude is set to zero then the unit shorts the bnc connector to ground via relay , so what your measuring in any case is whats induced into the loop .
On the highest amplitude range ie 5 volts and above and frequency set to zero I was measuring 150 microvolts into hi z at the bnc connector, this measurment was taken after the psu was replaced .

Not quite, Soundtec. There is zero pickup by the "loop antenna" until I move it adjacent to the SMPS . Good point about the relay though. I hadn't noticed. Obviously what I'm seeing is being induced into the PCB and relay. Your result looks excellent though. Thanks.

Well, not sure there are relays for shorting the BNC connectors to ground. It does not show up on the schematics... When amplitude is set to 0, opamps have only a 0 setpoint...

[plb]

Well, not sure there are relays for shorting the BNC connectors to ground. It does not show up on the schematics... When amplitude is set to 0, opamps have only a 0 setpoint...

There is a schematic available? Can you point me to it?
Somewhere back in the 54 pages of this thread there was a start made on a main board schematic. Did I miss the final result?

[soundtec]

Here it is PLB ,
might as well just post it again ,

if you look up github FY6600 Der Kami's page is there with most of the info