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

[xzswq21]

it's a very interesting option.
due to the 7ns rise time it's mandatory to use 50R terminated devices
consider the device use RG174 or RG316 cable, so the velocity factor is 0.7 now if you want to not encounter some problem, use 18cm Cable not anymore
if you could set the rise/fall time to 38nsec and use a 1m Cable you wouldn't have any problems

Thanks

[cybermaus]

Well, I guess you could always upload a arbitrary wave with such a shape that it is effectively a square with 38ns risetime

Maybe one such wave for 1MHZ, one for 10MHz etc. After all, you got a shitload of 64 ARB positions. One for each of your desired risetimes.


What I did find is that if you take a square, and modulate it with PSK with another square of the same frequency, I get a nice 4.4ns risetime square wave!
I think that is really impressive. I was playing with time-domain reflections to measure length the other day. with a <5ns device I should retry that play.

Edit: In fact, all the vertical edge waveforms (pulse, cmos, reverse sawtooth, all of them), *except* the square seems to be 4.4ns. Only the square has 7.8ns.
And indeed, much cleaner to see cable reflections with 4.4ns. Now I know my pieces of BNC-coax are about 1mtr in length 

[Ebel0410]

Is it possible to adjust the rise/fall time of square waves with this generator? I really need that capability.

Nope

[Ebel0410]

For those who are interested in the spectral behaviour above 1 MHz, I've made some tests with my spectrum analyzer.

Have a look at this :

30M Sine purity, better than 44dB
1M Sine 0dBm, 2nd and 3rd orders are below 52dB, not too bad indeed !
Bandwidth greater than 80MHz with a 10MHz multitone signal test, quite impressive

For <100$ , I've to admit that the results are pretty good !

[Mr. Scram]

Lots of antivirus programs will see an unknown .EXE file as a potential threat and flag it as suspicious. I doubt you have anything to worry about, nobody else here has reported any problems apart from the odd difficulty reading the CD.

While it's true that some antivirus solutions will flag certain types of code as suspicious, it's also true that many people ignore reported malware on account of it likely being a false positive. If malware is somewhat decently constructed and depending on the purpose, the user won't notice anything being wrong after getting infected.

[xzswq21]

I'm wondered HOW the designer designed the setup?!
for 1MHz the SFDR is 54dBc :| for a cheap generator is good.
But I can not admit it! they have used a 14-Bit 250MSPS DAC, the frequency is only 1MHz! the SFDR should be better than 80dBc
above 70MHz they should use back to back topology (by using two Baluns) to have the better Harmonic Distortion suppression.

*which type of capacitors they have used?!
*which OP-AMPs they have used?

I'm familiar with designing such devices

I think it's a good AFG to tune the circuits (without checking the THD and ...) and checking the step response.

[Ebel0410]

Frankly it is a mistery.
Unfortunately it will be difficult to make some reverse engineering as a lot of ICs have no names (information scratched by Feeltech)

[cybermaus]

I am still a bit unhappy about those 4ns runt edges though. Not being a signal generator specialist, it seems to me like a unavoidable problem if you have a 250MSa/s DAC, but I am wondering how my DS1074Z-S internal generator is able to work without such edges when it has an even lower DAC of 200MSa/s


Also, I think I see a JTAG port, so while not officially updateable, it should be possible if we have the firmware and correct procedure.

[xzswq21]

the Baluns are used for the frequency above 5MHz. when you have back-to-back topology the bandwidth is divided by sqrt(2).
it's recommended for the frequency above 70MHz. I think they have used some schematics from some datasheets.
there is no Baluns to support the frequency below 15KHz.

now let me show you an interesting curve about the SFDR and use of a Balun:


even I suspect the capacitor's type you have used in the design!
now let me show you an interesting curve about THD+N vs frequency and Vrms:





Best Regards

[Marco]

I'm pretty sure the two chips next to the FPGA are overclocked DAC904s, to the right of the FPGA are a couple of buffer amps and discrete anti-aliasing filters.

[Mr. Scram]

so the higher freqency models are just moer overclocked?  I just happen to be looking at this on ebay last night, local seller for best price I've seen so far. Almost...

You say it like it's a bad thing. Intel sells the same silicon at different clock speeds for different amoutns of money. Dave has done a video about factory overclocking. If you have tested and can guarantee operation, or even binned the chips, there little wrong with the practice.

[xzswq21]

Dear
I couldn't read any part numbers at that video!
DAC904 is belong to Burr-Brown!
even I see two white components near BNC jacks. can you read the part number?

My brother told me to not buy this Function generator yesterday, he told me to buy a Rigol AFG at-least.
I think the best version of this AFG is 30MHZ bcoz DAC904 is a 160MSPS DAC.

it's worthy to note that some pics I uploaded above are the performance of a 12-Bit DAC!
even the Feeltech's designer couldn't keep the performance of 14Bit DAC904!
bad layout, bad capacitor type and ... will ruin the performance. he could have achieved very better performance even with 10 or 12Bits DAC!!!


Thanks

[Ebel0410]

Some photos are following.
Your comments are welcome.

The high res images are stored on onedrive here :
https://1drv.ms/f/s!Alr1uExd0Z7lleYbxrcIoIomOrVprA

There is 14 DAC pins connected to the FGPA, photo FY6600-05 !
It seems to be a DAC904, the pinouts are corresponding to the photo.
The DAC904 is a 14b 165Msps converter produced by Burr Brown or TI.

DAC904 => LPFilter => OPA686

[ArthurDent]

A couple of comments: the power input connector can easily be changed but the back panel is thin fiberglass so care must be taken when enlarging the opening so you don't crack the panel. Also the power supply already has a unused ground pad that can be used by just soldering the ground wire to it. That ground is carried on the 2 middle pins of the power supply interconnect cable so all the BNC shells on the panel are at zero potential. I've (hopefully) attached a photo showing what I did. 

Also the 50 Mhz oscillator in my unit isn't that great although it may meet the specs. If you have a function generator capable of many digits of resolution but the timebase is only accurate to 5 places, that is a waste. There are unused pads marked "S1" on the main board near the oscillator that may be for an external timebase that I'll look into.

[bitseeker]

Welcome to the forum, Arthur and thanks for the info. Looking forward to your findings.

[mk_]

Some photos are following.
Your comments are welcome.

The high res images are stored on onedrive here :
https://1drv.ms/f/s!Alr1uExd0Z7lleYbxrcIoIomOrVprA

Thanks for detailed pictures

Well, at least it shows how not to design passive filter-gnd- areas.
In FY6600 -07.jpg there is only one (!) GND-via visible in the LPF1- filter-area (betweeen C36 and C37) so if this stuff will see the hf this filter will do a lot of jiggel but nothing predictable...
me thinks so the same mistake in LPF2.

michael

[xzswq21]

even I think the LPF should be placed after the Buffer to reduce the noise effectively not before!
the Buffer is actually a difference Amp.
I see two cap trimmers or 2 res pots. the CMRR of such designs is not good!
L10+C4+C27 and L12+C5+C16 are power supply filters to reduce the noise for the buffer.

[cybermaus]

Also the 50 Mhz oscillator in my unit isn't that great although it may meet the specs. If you have a function generator capable of many digits of resolution but the timebase is only accurate to 5 places, that is a waste. There are unused pads marked "S1" on the main board near the oscillator that may be for an external timebase that I'll look into.

Is that a HHttG reference, or are you actually named Arthur Dent?

Anyway, if we could hook this up to an external 10MHz GPS disciplined reference oscillator, that would be marvelous.

Or even better, directly build in one of the cheap uBlox GPS modules. So we do not need to mess with amps and drivers, but could use the FY6600 itself as an adjustable reference frequency. That would be a nice project. Would make it into a pretty good frequency counter too then.

[Marco]

bad layout, bad capacitor type and ... will ruin the performance. he could have achieved very better performance even with 10 or 12Bits DAC!!!

There is a ridiculous lack of competition in even medium speed DACs/ADCs, higher speed is worse, either they are completely integrated without general purpose use (mobile chipsets, cable modems etc) or you pay a ransom. There are only a couple of chips with really low prices, the DAC904 is one of them ... and it's not all that good to begin with regardless how you use it. Which is why some designs just stick with resistor ladders in this day and age still.

[ArthurDent]

You asked: "Is that a HHttG reference, or are you actually named Arthur Dent?"

This is a screen shot of the EEVBlog screen when I just registered. I had to save it because of what the answer to the last question was going to be. Certainly not a coincidence and I almost couldn't believe it! My 'real' name is Arthur Phillip Dent.   

[cybermaus]

That is hilarious 

But it has to be a coincidence. If people making HH references was normal enough for the web-admin to program such a Easter egg, your username would have been gone already.
Anyway, let me go and make up some towel jokes to include in the next couple of posts...

Edit: waaait a minute: It's actually more likely you decided on that user handle after you read the question



More seriously: I was assuming that S1 is a JTAG connector, but admittedly for no other reason that I would expect an unpopulated  JTAG on such a board, I did not follow the traces. But I am still thinking if you want to replace the timebase, you'd have to desolder the xtal module.

Edit: Yes, the chip between the S1 and the 50MHz is a Winbond 25Q16BVSIG : 16MBit Serial Flash with Dual SPI
It has at least 6 connections with the S1, so that makes it *very* likely some sort of SPI based programming header, like JTAG

[cybermaus]

While following the JTAG, I noticed 4 configuration solder islands labeled J10 J11 and J12 J6
Toward the back of the device, 2 on each side.

Not sure what they configure, my 60M board is coded like this:
J10 - closed by thin pcb trace
J11 - open, no pcb trace and no solder drop

J12 - closed by thin pcb trace
J6  - open, no pcb trace and no solder drop

So no solder drops, nor any scratched traces, meaning default PCB hardware.
I guess it would be too easy if this was the 60M 30M 15M frequency selection, but you guys check anyway.

[cybermaus]

By PM request, some sine pics.

Mind you, not sure why you want these, if you think you can "eyeball" sine quality, you would be better off checking the spectrum pics a page back, they are much more relevant. The only thing you can get from these pics is a slight attenuation toward higher frequencies, which may as well be my scope rather then the AWG.

Also, please note the scope measurement at the bottom cannot be trusted on decimals. It is constantly jumping from 19.8 to 20.0 to 20.2, showing it lacks resolution.

20Mhz to 60MHz 1Vpp 50 Ohm

[Diabolo]

Hello,

https://www.eevblog.com/forum/testgear/feeltech-fy6600-60mhz-2-ch-vco-function-arbitrary-waveform-signal-generator/?action=dlattach;attach=366988;image


The integrated circuit on each side of U302 is MCP 4822E Dual 12-Bit Voltage Output DAC.

Regards,
Diabolo

[xzswq21]

By PM request, some sine pics.

Mind you, not sure why you want these, if you think you can "eyeball" sine quality, you would be better off checking the spectrum pics a page back, they are much more relevant. The only thing you can get from these pics is a slight attenuation toward higher frequencies, which may as well be my scope rather then the AWG.

Also, please note the scope measurement at the bottom cannot be trusted on decimals. It is constantly jumping from 19.8 to 20.0 to 20.2, showing it lacks resolution.

20Mhz to 60MHz 1Vpp 50 Ohm

good job I see 0.5dB  attenuation at 20MHz and 2dB attenuation at 60MHz.

when the input impedance of your Oscope is 1M || 15pF and use a 50R capline, the VSWR<1.3 for up to 55MHz. 
but it depends on the quality of BNC T Adapter and other parasitic inductance, it could be 110MHz. 
so it's better to use an Active Adapter.

Thanks