FeelTech FY3224S 24MHz 2-Channel DDS AW Function Signal Generator

[LogicalDave]

I agree: the power supply used in this generator was designed to be earth grounded.  I would be very surprised if this passed FCC or CE EMC compliance testing and there are likely safety issues too.  I think the two choices with this signal generator are to replace the power supply  (+12, -12, +5, +3.3) with one that properly isolates the DC outputs or replace the two-conductor power connector with a three-conductor connector and earth reference the DC ground.  Thanks again!

MODERATOR: This issue should probably be something flagged at the top of this topic; please consider it.

Those are the Y-rated safety caps.  There is a third that bridges the HV and LV sides of the transformer.  I'd rather not connect the DC outputs to earth ground because that can introduce other problems; the DC ground should float.

This generator isn't designed to have the DC ground floating. For EMC and safety reasons it is better to ground the outputs.

[jwm_]

So I just did some tests and the generator works fine down to 27v DC applied directly to the AC input. I think I will go ahead and make it fully battery operated with a few lithium ions and a boost converter.

[leonerd]

Just got myself one of these and had a quick look around inside it. On a look over the board, I see that there's a buffer op-amp just before the final (under the heatsink) output driver, which is a 4558D. I looked up the specs of that - a terrible 3MHz unity gain bandwidth and 1.7V/µs slew rate. That may go some way to explaining why this thing doesn't work so well up beyond a few MHz.

I'll have a more careful trace of it in a few days time and see if I can work out enough of the circuit diagram around it to post here with a suggestion of what it's doing in the signal path, and some scope traces before and after it. I might then see if I can have a go replacing that chip with something a bit faster and see if it improves matters.

[Aruck]

Just got myself one of these and had a quick look around inside it. On a look over the board, I see that there's a buffer op-amp just before the final (under the heatsink) output driver, which is a 4558D. I looked up the specs of that - a terrible 3MHz unity gain bandwidth and 1.7V/µs slew rate. That may go some way to explaining why this thing doesn't work so well up beyond a few MHz.

I'll have a more careful trace of it in a few days time and see if I can work out enough of the circuit diagram around it to post here with a suggestion of what it's doing in the signal path, and some scope traces before and after it. I might then see if I can have a go replacing that chip with something a bit faster and see if it improves matters.

Would it be possible to put an oscilloscope probe on the input of that buffering op amp?

[leonerd]

Would it be possible to put an oscilloscope probe on the input of that buffering op amp?

Absolutely, yes. I'm planning quite an indepth review/look over various bits of it, just as soon as I've rebuilt my workshop again. It's currently quite literally in pieces while I upgrade the mains power supply arrangements, so all my equipment is sitting around on the floor in piles of boxes.

[Aruck]

I just received my FY3324S that I ordered off of banggood. It did come branded as "FeelTech", even though the pictures do not show that branding.

Please help me understand the grounding situation.
-The front panel clearly has ground symbols with lines pointing to the BNC connectors
-The BNC connectors all share the same reference (I verified this with a meter)
-The BNC reference is NOT the mains earth ground or neutral
-The BNC reference is not completely floating
-The BNC reference is somewhere around 20VRMS above earth ground (based on people's reports)

So what happens when I connect this function generator to a grounded circuit?
-Would the voltage difference between the BNC reference and a real ground cause a measurable current?
-Would the function generator go from "kind of floating" to "grounded"?
-If there is a difference in ground reference between a circuit and the function generator, could this cause equipment damage?

[erpalma]

I just did a quick test on the 4558D. It seems to be used only for the DC offset. I see no signal on any pin.

[uChip]

Sigh.  I'm almost sorry I found this thread.  I was looking at building the FG from AoE3 (www.artofelectronics.com/synthesizer) as my next project.  I had already figured to cost reduce it by replacing the two AD9954's ( 2 @ $34.40) with an FPGA (XC3S50A $8.05).  I've even gone so far as to write and simulate a good chunk of the FPGA code.

Then you folks show me that FeelTech has already implemented it and is selling it for seventy-five bucks.  Even DIY I could not beat that price.

Of course, if I just buy the FeelTech I don't get the "experience" of creating it myself.  Or the headaches either.  Some features might be different.  I'm more interested in FSK/PSK modulation than with arbitrary waveforms, though there's no reason AWs can't be done.  The timing analysis of my FPGA design says that it should be able to do 200 Msps so that is about the same.  I had not even started to design the UI yet and only just started simulating the filters in LTSpice.

Another difference would be that I would OSHW the whole thing.  I'd likely have to lean on you folks some to smooth the kinks out of the analog circuitry since I'm stronger in digital.  What might make it worthwhile would be to turn it into a community project and get more folks to contribute feature and design ideas.  More eyes on the design would likely improve utility and stability.

So what do you think?  Start a significant community project (would anyone really contribute?) or just buy a FeelTech and be done with it? 

Thanks,
  - Chip

[pascal_sweden]

Buy the FeelTech. Take it apart and identify the weaknesses in the design. Come up with DIY solutions to improve the design of the FeelTech. Or you could also pass the improvements on to the engineers of FeelTech, so that they come up with an improved version. This way you learned a lot about it, and at the same time helped the other existing FeelTech users. Win win

[Aruck]

https://www.youtube.com/watch?v=BqxA4HRYjVg&feature=youtu.be


Not sure how to turn this into an embedded video.

[FlyingHacker]

https://www.youtube.com/watch?v=BqxA4HRYjVg&feature=youtu.be

Wow! Comments seem to be disabled on YouTube, which is a shame. Nice video, though. Now I am going to have to have a look at mine.

I wonder if that is just ghost voltage, or what. A 20,000 ohm per volt analog meter should let you easily tell the difference.

[JoeO]

Review of FY3224S.  There is a coupon code to get $4.44 off the price.

[FlyingHacker]

I found a similar 165V peak to peak waveform on my BNC outside contacts. Indeed the polarity of the outlet did make a difference. Reading it with analog meter seems to indicate it is a ghost voltage, without much current. I was using a 20,000 opv Simpson. You probably need a 1000opv.

So do these guys need to be tied to ground?

[Aruck]

I've been working on creating a drawing of the power supply board.

I looked up the datasheet of the smps chip, and noticed something odd. The "typical application" has a cap connecting the smoothed DC output of the bridge rectifier to the output ground reference.

Am I correct in thinking that is a mistake, and that they probably meant to couple the two ground references together?

edit: I also don't understand C1, depending on the capacitance value wouldn't that effectively just be a short circuit?

[nctnico]

C10 is not a mistake. It is a common way to reduce emissions. It also causes the leakage from mains to the generator's outputs.

[SeanB]

C1 L1 form a low pass filter, to reduce noise from the power supply, from both the switching transistor and the incoming bridge rectifier diodes, from being passed back to the power supply and radiated out to other mains devices. Added to pass EMC compliance, as otherwise a SMPS can be a very broadband RF noise source.

[med6753]

Just got this similar Feel Tech FY2200S delivered a half hour ago. In process of figuring out all it's capabilities.

[med6753]

The FY2200 PC software installed OK on a Win7 x64 laptop but I found a bug. When selecting triangle wave on Channel 1 it actually outputs a square wave. And of course that makes square selection reversed too. Channel 2 is correct.

D'OH!!! 

[ledtester]

The FY2200 PC software installed OK on a Win7 x64 laptop but I found a bug. When selecting triangle wave on Channel 1 it actually outputs a square wave. And of course that makes square selection reversed too. Channel 2 is correct.

D'OH!!! 

When I try to install the software I get a Launcher Error - "Unable to open archive file" message. Any ideas?

Also - is the UI still all in Chinese? Perhaps it would be worthwhile to develop a browser UI since much of the serial protocol has been uncovered.

[nowlan]

Is that 115v on the bnc something to be concerned about? I cannot figure out why rotating a plug makes any difference to an ac input (mains).

[med6753]

The FY2200 PC software installed OK on a Win7 x64 laptop but I found a bug. When selecting triangle wave on Channel 1 it actually outputs a square wave. And of course that makes square selection reversed too. Channel 2 is correct.

D'OH!!! 

When I try to install the software I get a Launcher Error - "Unable to open archive file" message. Any ideas?

Also - is the UI still all in Chinese? Perhaps it would be worthwhile to develop a browser UI since much of the serial protocol has been uncovered.

No idea why you get that error message. The GUI in my copy is in English. I upgraded my laptop today to Win10 and it still works fine other than that dumb bug on Channel 1. Haven't found any other bugs yet.

If you want a copy send me a PM. It's 6.5 MB. But I'll have to change the extension on the file to something else or put in a zip file because Gmail doesn't allow you to send files with an .exe extension.

[med6753]

Is that 115v on the bnc something to be concerned about? I cannot figure out why rotating a plug makes any difference to an ac input (mains).

If you're talking U.S. mains one side is hot (115V - 120V) and the other side is neutral (tied to ground at the breaker box). The outlets have one slot larger than the other to accept a "polarized" 2 prong plug that insures hot to hot and neutral to neutral. However, the line cord on these units have both prongs the same size (small) which allow it to be plugged either way. Therefore, if that noise capacitor is now connected to the "hot" side it allows a leakage current to the internals.

I wonder if it would trip a GFCI protected circuit? 

[FlyingHacker]

I wonder if it would trip a GFCI protected circuit?

It has never tripped my GFCI, which it is plugged into.

[med6753]

I wonder if it would trip a GFCI protected circuit?

It has never tripped my GFCI, which it is plugged into.

I think for safety's sake I'm going to install a polarized plug to insure that cap is always on the neutral side.

[German_EE]

Looking on Ebay with FY3224S as the search term I can see many signal generators with different specifications. Bandwidth appears to be either 2 MHz, 5 MHz, 6 MHz, 8 MHz, 10 MHz, 12 MHz, 20 MHz or 24 MHz. Please could someone explain?