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

[soundtec]

Here's Fremens last few versions ,

[soundtec]

Ive just downloaded an old version of picoscope for free to try and see if it will allows waveform selection from a library usable with the FY6600.

https://www.picotech.com/downloads

drawing a waveform could take ages ,surely theres a repository for such waveforms and you could have an endless variety 

[precaud]

I downloaded the software again (this time from Google Drive), reinstalled, ran, and got the same error as before. It loads into the "Extraction" Window, but when loading it into the Waveform Window, it bombs (see attached). IIRC, this software is Visual Basic code. Maybe it needs a later version of VBRUN than I have?

I then installed the 0.7 version of fremen67's software (thanks, soundtec). It loaded the file directly into the Waveform Window and uploaded it to the 6800 fine (I think, I haven't looked at the output yet).

To verify the data is correct, I tried loading the file into the Text Window (using Loading Area 2) so I could examine the data and see if it was being converted correctly. This window does not work correctly, it displays all values as 0, 1, or -1. I turned off "auto-resize data", with the same results. I then tried sending the data to the 6800 (loading Area 2 goes to ARB2 waveform). The waveform sent was the same array of 0, 1, and -1 values. So loading into the Text Window destroys the data, and there's no way to verify that the data values are being converted correctly.

[DaveR]

To verify the data is correct, I tried loading the file into the Text Window (using Loading Area 2) so I could examine the data and see if it was being converted correctly. This window does not work correctly, it displays all values as 0, 1, or -1. I turned off "auto-resize data", with the same results. I then tried sending the data to the 6800 (loading Area 2 goes to ARB2 waveform). The waveform sent was the same array of 0, 1, and -1 values. So loading into the Text Window destroys the data, and there's no way to verify that the data values are being converted correctly.

I never could understand what the Text Window was supposed to be doing - even fremen's version seems to just convert all the data points to integers (try loading the sinc-pulse file I uploaded earlier, and you'll see the waveform in block form with values -1 to 5). 

If you upload to Loading Area 2, won't you overwrite the factory-stored square wave there?  The empty loading areas are from 64 back to ?? (can't remember offhand).

[precaud]

 I looked at the output waveform (with an 8-bit scope) and the data transferred from the Waveform Window using fremen67's program appears to correct. None of the duplicate time data points that I saw before. So this can be a realistic basis to my comparison tests. Thanks for your help in gettin g this sorted out!   

[precaud]

I never could understand what the Text Window was supposed to be doing - even fremen's version seems to just convert all the data points to integers (try loading the sinc-pulse file I uploaded earlier, and you'll see the waveform in block form with values -1 to 5). 

With ver 6 SW, it showed the data values scaled to signed 14-bit integer values. I would think that is what it is supposed to show.

If you upload to Loading Area 2, won't you overwrite the factory-stored square wave there?  The empty loading areas are from 64 back to ?? (can't remember offhand).

Could be - I wasn't paying attention to that    I assume there's a way to reset the unit and restore the original factory waveforms?

[rhb]

Theres a guy I know who runs a reseach project for one of the universities here ,he's invloved in siesmic data aquisition  from sea bottom tranducers around the coast ,I bet he could save massively on his overheads by creating his own supercomputer from PI's.

Who is it?  I'd like to look into it.

FWIW Intel and AMD CPUs are more cost effective than any of the alternatives.  The basic problem is tuning the software to particular architectures.  Most algorithms are trivially parallel so everyone just runs 10-20,000 cores for a week or two for industrial work.

Seismic is both CPU intensive *and* I/O intensive.  For example, Kirchoff migration requires summing 10e5-6 points into 10e12-14 points.  And that's the *fast* algorithm.  The serious stuff is an order of magnitude more work. I have an MPI based Kirchoff code I'd be happy to give to him.  I wrote it as a commercial project when I got laid off in '98, but faced with the cost of selling it vs a job I was offered, I took the job.

If he's using the CWP SU package then he will know who I am. I did a lot of work on it for many years and also arranged for the SLT SU version to be donated to MInes. 

[DaveR]

If you upload to Loading Area 2, won't you overwrite the factory-stored square wave there?  The empty loading areas are from 64 back to ?? (can't remember offhand).

It's ok - just my memory fading again!  There are 97 slots in all, 33 for presets and 64 for User / ARB.  The first 32 ARB slots already contain copies of most of the presets, which can be overwtitten, and slots 33 - 64 are empty.

[soundtec]

Thanks for that Rhb,

I'll certainly tell him next time I see him ,he was supposed to be in touch about his music projects but I didnt hear back ,I dont have a number for him but I'm likely to see him or one of his brothers in the neighbourhood at some point . Like all research he has to justify his budgets ,something neat and tidy like PI/Bitscope data aquisition farm might allow him to get very good number crunching horsepower for small money , I think it probably takes time to interpret the data the way he's set up now ,but the possibility of realtime or near realtime processing sounds very exciting . A portion of the work involves the use of gps synchrous clocking and triangulation  to work out very precisely the locations of the under sea Hydrophones ,then precise sum and difference calculations can be done to determine the depth as well as latitude and longitude of siesmic events. Its easy to see why such a system could be of value for coastal patrol/protection too, but they do say even the ocean has had ears for years and years

[rhb]

Sounds very cool.  I'd love to do some  seismic work again even if I didn't make any money as long as it presented a new and different problem.  It sounds as if your friend has one.

I got 3 HP Z400s with 6 GB of ECC DRAM and quad core 2.9 GHz Xeons for $300 plus shipping via eBay.  That's probably several million times the compute that I shared with the entire department in grad school in Austin.  There's no way $300 worth of Pis will touch that for seismic processing.

Ask your friend if he uses Seismic Unix from CWP.

[soundtec]

Ive been trying out a few of the proprietary wave editors again ,no success as far as getting V6 SW to read waveforms produced by either easywave or TTI waveman, it either stalls or crashes the Feeltech software . I only managed to reload a previously stored waveform created on V6 . Maybe Feeltec need to re-look at the compatibillity here as well .I still cant find a library of waveforms that I can easily slot into the 66/68 hundred

[rhb]

Post  the available info on the formats.  I lost count of the number of format convertors I've written 20 years ago.  There were times I felt as if that was all I ever did.

[DaveR]

I don't think we need a format converter, rhb, the waveform files are just lists of numbers in plain text, but Feeltech appear to have introduced a bug into the latest version of their PC Control software which causes a crash when the files are loaded.  It's not a great problem because fremen67's software does a better job of waveform uploading than Feeltech's anyway; either software can be used for PC Control, and Feeltech's is only really required for occasional firmware upgrading of the 6800.  They might get the bug sorted out eventually, but it's hardly a showstopper in the meantime.

[DaveR]

Ive been trying out a few of the proprietary wave editors again ,no success as far as getting V6 SW to read waveforms produced by either easywave or TTI waveman, it either stalls or crashes the Feeltech software . I only managed to reload a previously stored waveform created on V6 . Maybe Feeltec need to re-look at the compatibillity here as well .I still cant find a library of waveforms that I can easily slot into the 66/68 hundred

The editor in Picoscope is the best one I've tried so far, Soundtec - very easy and accurate to draw with, nothing complicated about it, all the controls are obvious and straight to hand, and output in csv format goes straight into fremen's software as is.  If I was going to be doing any freehand waveform design, I think this is the one I'd use.  Is there anything in particular you are looking for?  The only waveform library I've come across is Picotech's, but I guess car engine fault tracing isn't what you had in mind?

[soundtec]

I got picoscope alright ,but didnt get stuck into it yet ,
Im not sure I'll be free handing waveforms ,but now you mention automotive scopes and such things it seems like a good place to try .There seems to be some kind of wave format management in labview ,but thats way to heavyweight a program for me to get stuck into.

I might try copy and pasting the contents of a file created in pico or wave into a .FY style container , I think I did find one way of transfering the contents earlier in the day but forgot again when I changed to different programs ,I'll have another look .

[soundtec]

I did manage to get Fremens program to make a waveform and it displayed in SW V6 although the phase was reversed . When I tried a waveform created in SW V6 on Fremens again it displayed out of phase also. I havent been able to get a single waveform from any of the other programs to work on the feeltech .

[soundtec]

With TTi wavemanger I was able to produce a waveform as .asc which would load into Fremens 0.7 software ,then I 'saved as'  and it sucessfully loaded and displayed in the Feeltech software . The waveforms in the TTI are based on .wfm files so maybe some kind of library exists already ,but I could only find a few extra examples of waveforms .I might try filling some of the blank spaces in the memory of the FY later on .

[GregDunn]

Does anyone have anecdotal experience on where is currently the best place to buy a FY6800?  Amazon US doesn't have any yet, and something tells me this is not the time for me to start randomly selecting *Bay vendors and trusting to luck.   

[Miti]

Banggood.

https://www.banggood.com/

[precaud]

Does anyone have anecdotal experience on where is currently the best place to buy a FY6800?  Amazon US doesn't have any yet, and something tells me this is not the time for me to start randomly selecting *Bay vendors and trusting to luck.   

I'll sell you the one I just got. I only bought it to test 14-bit arb vs 12. PM me if interested.

[soundtec]

I finally got the 5 volt regulator down to a proper working temperature ,I needed to heatsink the ic's and the bridge. Excess heat is directed into to metal screen between powersupply and signal board .I wound the +/- rails up 13.7 volts for a bit more headroom at high output amplitudes ,where clipping was apparent with the original Feeltech power supply of 11.5-12 volts, and as little as 100mV of DC offset caused hard clipping of the output op amps .

I was hoping I could simply swap over the main board and front panel from 6800 to the powersupply and enclosure of the 6600 with the linear supply ,but the recessed screen on the 6800 makes that impossible .

I'll most likely try a dual switchers one for op amp rails and another for 5volts ,for simplicity and size I might reuse the second Feeltech switcher and maybe try adjusting resistor values to increase the dual rail section while leaving the 5volt unloaded completely tends to raise the other rails to some degree as well ,any ideas on that would be greatly appreciated . I was thinking of trying an expanded aluminium mesh as a RF cage for the powersupply to prevent any Rf gash .

I also noticed on my 6600 with the linear that the negative rail measures just 2.2mV ac, the positive rail measures 103mV ac and the 5 volt measures 80mV ac ,is it because of other supplies being derived from the 5volt and 12 volt + that this extra noise exists?
Exactly the same readings from the unmodified 6800 ,so it has to be coming from the signal board


I decided to try the 6600 (15mhz) front panel on the 6800(60mhz) and vice/versa .
Both machines loaded fine ,appart from the waveforms preview being garbled ,the mhz on the 6600 main board now went up to 60 mhz with the 6800 V3.7 front panel and board of the 6800 only went as far as 15mhz with 6600 front panel . One difference I noticed in the 6800  was now when you change between channels if you were previously on say amplitude on ch 1 if you select ch 2 and go back to ch1 it retains the relevant page on both channels  ,where in the 6600 as soon as you change channels it reverts to frequency adjustment by default ,very anoying , Its a minor detail but it does make it nicer to use .

[Scratch.HTF]

Good idea, but the filter should be able to be switched in and out via a DPDT telecom relay (it can be controlled via a mechanical switch).

[soundtec]

So I tried using two feeltech switch modes to power the 6800 ,one for 5 volt rail the other for +/- rails .
When I drew no current from the 5 volt rail ,the +/- rails oscillated up and down and were way below the volts they should have been .
I tried adding 40 ohms across the 5 volt rail to draw some current ,and this stabilised the +/- , up around 12 volts but only with a 20 ohm resistor ,it got quite hot . Just goes to show if theres any moments of high current draw on the 5 volt line it can modulate the +/- rails ,not great really .

I guess I could split the 5 volts supply on the main board ,and run some current from both switchers 5 volt lines to stabilise things ,but then my 12 volt rails are too low . Ac milivolts on the supply rails stayed around the same in any case around 100 mV on the positive 12volts around 4 mV on the neg 12 volts line and around 80 mV on the 5 volts rail . The noise must be coming back through the 1117 regulators from somewhere else down the line . Maybe 80-100 mV is nothing to worry about in this situation.

ln the end I ended up reverting to a 2x12volts +  5 volts supply from an old satelite tv box ,bit more juice than the original item .

[GregDunn]

I guess we're still using this thread to comment on and possibly offer feedback to the functionality of the FY6600 and FY6800 units?

I've been playing around with my new FY6800 (thanks to precaud!) and it seems to be able to be coerced into doing everything the manual says; bottom line, it's a great value and will get a lot of use on my bench.  But the frequency units display is rather tedious to decode while trying to quickly enter values.  I was trying to generate an AM modulated signal and it took a couple of tries to get carrier and modulation frequencies which were appropriate to generate the display I was looking for because I was trying to parse uHz on one channel and MHz on the other during the setting procedure.

It seems that the units suffix (Hz, MHz, etc.) is usually updated to a more sensible value after you complete entering the frequency and hit OK.  That's good, and desirable; although I've been able to trick it into displaying 100 KHz as either 0.1 MHz or 100000000 uHz depending on the order in which I entered the data and hit OK.  That's not so good, and could be improved if FeelTech is interested. 

What would make sense to me would be if, as you selected the appropriate frequency digit to change, the unit suffix were to change color, or blink, or something, and indicate the range of the decade you were manipulating.  Ex: as you moved into the digit group setting kHz it could blink "kHz", and if you moved past that into the group covering Hz it could blink "Hz" and then return to a steady indication of the appropriate units after you hit OK and the software was able to determine the value of the highest digit.  As it stands, you could be looking at something silly like "uHz" even while you're in the process of setting 10 MHz, or "MHz" while you're trying to set a frequency of 10 Hz.  More mental effort required on what should be a simple setting.

It's kind of like my soldering stations - I have one which is fixed to display in F and another which is only in C. I can do the math in my head while setting values, but it's something which could be made a little easier for the user.  It would be easier still for the FY6800 user if the software dynamically shifted the frequency display left or right and calculated units as you set the digits, but that would take considerably more effort for the coder to implement - I've done such in the past, so I have a hint of the effort.

I don't know if FeelTech would consider a similar change for future revisions, but I think it would substantially improve the usability of what is already a superb value in test equipment.

[soundtec]

Hi Greg,
Thats great you managed to pick up an unloved Fy6800 from Precaud. There really isnt much to be said against these units in terms of value for money ,and of course you can do many simple mods ,which make it even better .

On the 6600 the encoder has a push switch button ,this allows you to change the display to  uHz Hz Mhz etc , its the OK button on the 6800 does that ,I wasnt sure from your post if you had figured out how to toggle Mhz/hz in the display or not .
And yes do keep posting your thoughts or possible improvements here .