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

[Candid]

My favorite trafo:
https://www.ebay.de/itm/Netztransformator-230V-8V-1-7A-4-2V-0-15A-13-6Volt-2x0-6A-Mittenanzapfung-NEU/152934971288

With 6cm x 5,8cm x 5cm it will be a challenge to fit this beast in the housing. I decided to use two small trafos and it works fine. I added a fan. All parts running well inside their thermal specs and I used standard 7805, 7812 and 7912.

[DaveR]

That's a nice, neat job there, Candid.  I had something similar in mind before I did mine a few weeks ago, but I couldn't find the right transformers or decide whether to go for 12v or 15v outputs.  In the end I went for variable outputs to cover all eventualities (including using the whole power supply for another project).  Do you think the fan is really needed , or have you just fitted it for "insurance"?

[Insatman]

Nice work Candid...clean prototype construction.

[cybermaus]

Back to the THS3095 topic.

My cheap Chinese "he promised they were genuine" THS3095 arrived and it was but a moments work to solder them in.
WOW, what a difference that made! I guess they are genuine. Or as good as.

I guess I caused some extra sales:


I hope they will not burn up after a few days, and people blame me.

[Candid]

Do you think the fan is really needed , or have you just fitted it for "insurance"?

The fan is not a must it's more for insurance. Without it the housing is getting warm but not hot and I did not do much measurement over longer time without the fan. With the fan I get about 32°C air temperature at 21°C room temperature directly at the fan after hours with both channels on at 20Vpp / 20MHz / sine. All the components stay inside their temperature specs. The bridge rectifier I used ist not the best for this scenario but I had them in stock. It stays under 50°C and so I don't care.

[texaspyro]

I hope they will not burn up after a few days, and people blame me.

It wouldn't matter... we already blame you for all the world's woes...   

[fremen67]

For added info, I attach a scope view of the SPI bus signals. Not sure if it helps in any way.
The first 2 are that I captured between BP and FPGA. The last 2 between FP and FPGA. While there are some overshoots on vertical edges, there is no noise, and there is a good clear difference between 0 and 1.

When BP is connected, these signals are shown over and over again. Hence the they are captured in RUN and nicely bright. My guess is the BP is trying to get the FPGA to say something, but it stays silent.

I think I got it. Having a closer look at the scope views, yellow beeing the clock, purple MOSI and blue MISO, the FPGA is answering.
The first picture should be Reg18 request (Flash operation status request) and picture 2 the second part of the answer (LSW) with "1" that is pending operation. There should be another SPI transmission in between with "0" as answer (MSW).
As there is no time-out for flash read, it is stucked when FPGA keep answering "1"
You could confirm by connecting also Regsel as fourth signal and trigger on falling edge of RegSel. You should then see the 3 transactions: picture 1, a missing picture and then picture 2.

I will add a time-out on flash read so it can't get stucked and then we will see what happens after.

[cybermaus]

Ok, many thanks, however I also reached some progress:

With your v0.5 on the BP, I have flashed the Winbond from 3.1 to 3.2, and suddenly both vero-BP and loom-BP worked.
I flashed back to Winbond 3.1, and they both stopped working.
I next flashed your V0.4 onto BP and it still was not working
And I ended again with your V0.5 with Winbond 3.2, and it is working again.

In other words, it is working. But also, why was the combination of V0.4 and V3.1 working a few weeks ago, and not now?
It must be a hair trigger difference somewhere in the timing. Better weather? I did move my desk to the other side of the room.

So, if you feel you still want me to do these test with V0.5 and V3.1 combination, so that we may use my "on the edge instable" setup to improve general stability, then please say so and I will do them.

[fremen67]

Ok, many thanks, however I also reached some progress:

With your v0.5 on the BP, I have flashed the Winbond from 3.1 to 3.2, and suddenly both vero-BP and loom-BP worked.
I flashed back to Winbond 3.1, and they both stopped working.
I next flashed your V0.4 onto BP and it still was not working
And I ended again with your V0.5 with Winbond 3.2, and it is working again.

In other words, it is working. But also, why was the combination of V0.4 and V3.1 working a few weeks ago, and not now?
It must be a hair trigger difference somewhere in the timing. Better weather? I did move my desk to the other side of the room.

So, if you feel you still want me to do these test with V0.5 and V3.1 combination, so that we may use my "on the edge instable" setup to improve general stability, then please say so and I will do them.

That's good news. So it may rather be related to v3.1 than to your system.
For the moment beeing I would prefer finishing the sweep functions as it is almost done and I will improve startup diag and flash functions later on as you are not stucked.
Did you use the v3.1 that you already posted (to go back from v3.2 to v3.1)or was is a recent backup (corrupted eeprom...)?
This will be easier for me to test things with one on my FY6600 downgraded to v3.1. That would be interresting if I could reproduce the problem later on. So if it is a different backup than the one you posted, I would be interrested in having it.

If you have some time, I would appreciate when you can test the last modifications (calibration and ZLoad mode) on your system.

[cybermaus]

Did you use the v3.1 that you already posted (to go back from v3.2 to v3.1)or was is a recent backup (corrupted eeprom...)?

Went back to November backup. I realized the corruption might be the case, so I did save the current 3.1 before I did the other things.
When comparing though, the only differences are in the custom waveforms.

[TechieTX]

Does anyone know offhand if the parts in the pic have the PowerPAD on the bottom?  The seller doesn't know, but at $2.85 each it's a deal either way when Mouser and DigiKey are quoting 9 bucks or more.

Given a choice, I'd rather have the ones with the PowerPAD; I can deal with 2 tiny pieces of Kapton on the FY6600 upgrade.  A quick check at TI shows the '3491 is only sampling in VQFN, and none of the vendors (including TI) are showing stock on SO-8 yet.

[cybermaus]

Yes, powerpad

or at least, I assume this is the same seller as I used, or at least they are using the exact same specific picture, and mine came with powerpad.
Note however the picture shows 3095 57C TATD G4, but the ones delivered are 3095 42T CBZK G4

Indeed, I put two small bits of kapton below it.

[soundtec]

I got out my machine gun and added some ventilation holes along the edges of the top panel ,just to get slightly better airflow over the the electronics , I measured around 27.7C air temp on the signal board side of my unit and around 32C near the psu, I noticed that the mains rectifier of the Switcher psu unit generates the majority of the heat from the power supply area , its easily detecable as a hot spot underneath .
Maybe a packaged bridge instead of individual diodes might be a good plan for longevity as all that heat getting conducted through to the first filter cap cant play out well for for its lifespan.

I tried the new version software via usb ,there seems to be some kind of issue for me with it, its kinda slow to load the UI and tends to lock up , anyway no big deal about that for the moment ,Im still waiting on the blue pill board and st link .I did get one cp2102 type programer ,is that any good for connecting to the fy signal board ? just tx rx and ground .

It seems to me a higher voltage op amp supply is probably a good thing .The extra headroom would for instance allow you output  20 volts while dialing in a dc offset without any clipping ,where with a 2x11.5 volt supply the moment you dialed in any offset at 20pp the signal was clipped ,some manufacturers run op amps at +/-18 volts although I saw +/-15 was the limit for the 3095 type.I would imagine the extra volts would play out well for THD figures at the higher output voltage settings in any case
So the 12 volt rails are in turn passed through two linear regulators to supply the OPA686N's ,how will these cope with having to loose the extra volts ,better heatsinking might be needed on the 7805 7905 ,the 16 volt caps on the signal board will need uprating to 25volts . So if we made the supply voltage +/-16.5 we'd get the abillity to set in a 5 volt offset on top of our 20 volts output and at the end of the day be able to swing 30 volts at the outputs .Seems like a worthwhile plan if you've already gone to the trouble of installing the 3095's and linear supplies .

Nice work Candid ,the seperate transformers for analog and digital rails looks like a good plan with maximum attenuation of any unwanted interactions and much quieter supply for the op amps than the standard switchmode unit.

powerpad ,Mine had the heat sinking on the underside of the 3095's, I just went ahead and soldered them on ,no kapton tape ,i think there is a small gap between the board and the chip normally ,but no problems so far in any case without extra insulation added.
 

[cybermaus]

If you have some time, I would appreciate when you can test the last modifications (calibration and ZLoad mode) on your system.

Oh, I forgot to respond: I did do the calibration and Z-mode yesterday.
Very nice, that low .1Hz nearly DC routine to calibrate using multimeter. Very nice and simple.


After calibration, when I next checked by putting out a 50Hz 1.41Vpp and 14.1Vpp sine on a TrueRMS multimeter, the voltages were slightly off, but better then before. From memory, I think the DDM gave 0.498 and 4.89Vrms, but I lost the piece of paper I wrote it down on. I think being within 2.5% is good enough for a device like this.


But still, While I appreciate the simplicity of the .1Hz nearly DC routine, I wonder you should not build the routine on a 50Hz sine.
The .1Hz routine only being usable for offset. And a 50Hz signal for amplitude. (Note that for a good sine, Vac and Vrms are the same, so a TrueRMS DMM is not even needed)


The DSO (DS1070Z-S) disagreed quite a bit with the DMM. Forgot the numbers, but at 50Hz at least 10% off. I assume the DMM is correct. But the DMM cannot be trusted above 400Hz or so, for 1MHz or 10Mhz signal it probably does not even see a signal at all.


So I am not sure if you plan to extent the calibration toward attenuation, but in that case, we'd have to come up with something smarter.
By lack of a smart idea, the only thing I can come up with is to read and use the error of the DSO at 50 Hz and next read the DSO at 1MHz, and first apply the 50Hz multiplier.
But not sure if that is reliable. Maybe we should for now not bother with attenuation correction.
I suspect you already went through this reasoning anyway.

[Ebel0410]

Thanks Cybermaus for sharing your positive experience with THS3095.
https://www.eevblog.com/forum/testgear/feeltech-fy6600-60mhz-2-ch-vco-function-arbitrary-waveform-signal-generator/msg1471111/#msg1471111

I ordered 2 pieces on eBay, same seller.
Regards

[paulca]

I finally got around to seeing about upgrading the PSU to a linear one. 

The transformer fits fine, the two regulator boards fit fine, the fused 3 pin IEC socket will fit fine.  There is no chance of them all fitting together in the case.

What I'm currently considering is putting the PSU into an external enclosure and using some 6 core cable to run a lead to a socket on the back. 

Of course the other option is to just use the existing switch/2 pin plug, not earth the transformer chassis and even then it will be highly cramped.  Maybe if I made my own regulator board combining the two I bought it would work.

Even going external, the largest enclosure I could find in maplin closing sale is not big enough   Not without things like the heat sinks being put against the insulated spade connectors for the mains IEC plug.

[bdivi]

I finally got around to seeing about upgrading the PSU to a linear one. 

The transformer fits fine, the two regulator boards fit fine, the fused 3 pin IEC socket will fit fine.  There is no chance of them all fitting together in the case.

What I'm currently considering is putting the PSU into an external enclosure and using some 6 core cable to run a lead to a socket on the back. 

Of course the other option is to just use the existing switch/2 pin plug, not earth the transformer chassis and even then it will be highly cramped.  Maybe if I made my own regulator board combining the two I bought it would work.

Even going external, the largest enclosure I could find in maplin closing sale is not big enough   Not without things like the heat sinks being put against the insulated spade connectors for the mains IEC plug.

In this situation I'd go with everything inside without the IEC power plug.
Linear transformer/supply will be good with the original non-grounded plug and the outputs will be floating. You can always ground through the BNC.

[Candid]

If you use an external housing for the PSU I would look after a similar sized housing that can be used as a stand for the FY6600. Using the screw holes of the FY6600 housings' feet for mounting the FY6600 on the PSU housing and you can directly connect the wires. If the housing is beveled you also have a good viewing angle on the display of the FY6600. I would have made it like this if my solution would not have fit inside.

[cybermaus]

How about just putting a hard-wired power cable (with a strain relief)?
After-all, in reality we always put in a cable anyway. Not really needed to make the device cable-less unless it has a battery.

[soundtec]

It might be a tight squeeze for me to fit everything in too ,once I change over to transformer/linear setup psu.
One thing I did think of was the possibillity of mounting some of the psu components to the lid of the box ,that way it might make some more room on the ground floor so to speak.
I got two different kinds of bluepill boards and a USB to TTL convertor in ,still no sign of the St link unit though.
I had a wee look at the specs for the 3095's ,absolute max volts is 33 across the rails ,normally 30 volts is the recomended max.The two regs feeding the OPA686n's wont have a huge amount of current through them ,so Im guessing they'll be ok scrubbing off the extra volts .Another good thing I noticed was that THD rises with frequency ,so for audio which will be my main usage distortion comes in at around -80db below 1mhz,and probably even better down at lower frequencies. I did also get a chance to look at the noise floor which showed some spikes ,most likely it'll be a lot quieter once the linear goes in .

I did end up putting in for the competition ,but I have no twitter account ,so I probably dont qualify ,Id prefer to see the people with defective V3.1 units properly sorted out instead though. Im happy with my FY6600 ,it does blow the competition out of the water despite the few short comings , and the learning experience of upgrading the unit is a good thing in itself as well.

If anyone has found a good tutorial on blue pill and St link stuff they might post it here as Im more or less brand new to all of it .
I am very much looking forward to exploring all the functions and tweeks that Fremen has added , I think we'll have a very high spec machine that will compare favourably with units costing many times more money once we get to where were going .
Keep up the good work .

[paulca]

I know this is off topic, but I can reuse the board mounting holes in the case to secure the important transformer, but is it acceptible to hot snot the regulator boards to the case?

[DaveR]

The transformer fits fine, the two regulator boards fit fine, the fused 3 pin IEC socket will fit fine.  There is no chance of them all fitting together in the case.

Try thinking laterally, people: there's no rule that says all the parts have to fit on the same side as the existing power supply!  There's plenty of space above the signal board for one of the modules, and enough space down the side of it for a couple of spacer legs to secure it to the case.  With one module out of the way there'll be loads of room to fit the rest in the free space - see my layout here:

https://www.eevblog.com/forum/testgear/feeltech-fy6600-60mhz-2-ch-vco-function-arbitrary-waveform-signal-generator/msg1430394/#msg1430394

It looks from the photo as if the 5v module is sitting directly on top of the signal board, but it's actually 20mm above it, with 25mm hex spacers fixing it to the case at the left side, as viewed.  It doesn't really need anything else to secure it, but I put a couple of 20mm spacers at the other side of the module, just sitting on the signal board to stop it from possibly vibrating loose.

[DaveR]

I know this is off topic, but I can reuse the board mounting holes in the case to secure the important transformer, but is it acceptible to hot snot the regulator boards to the case?

Nothing to stop you, but spacers and screws are better.  Use nylon instead of metal and you won't have any risk of accidental shorts or shocks, and there'll be no problem with drilling the case for new holes as the insulation will be fully preserved. 

[cybermaus]

..absolute max volts is 33 across the rails ,normally 30 volts is the recomended max...

Absolute max rating is where the device supposed to not yet burn, but there is no guaranteed or even documented behavior or fidelity at that voltage. It just means that if you bring it back from 33V to 30V, it should work again, whereas is you bring it back from 33.1V, it is allowed to be damaged. It will probably still work, but you are walking on a sharp edge with 33V.

[cybermaus]

Thanks Cybermaus for sharing your positive experience with THS3095.
https://www.eevblog.com/forum/testgear/feeltech-fy6600-60mhz-2-ch-vco-function-arbitrary-waveform-signal-generator/msg1471111/#msg1471111

I ordered 2 pieces on eBay, same seller.
Regards

The seller noticed the sudden boost in sales, and nearly doubled the price from $2.85 to $4.75

Interestingly, he did it in such a way that the listing appears to be down-priced: from $5 to $4.75