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

[masterx81]

I've noticed from the photos that my 24mhz have the cooler over the opamps.
What opamps have the 2mhz version?

[bitseeker]

Has it been confirmed that the 2 MHz version is identical to the 24 MHz version in terms of hardware?

At one point, it was noted that all speeds were the same hardware.

How to perform the software upgrade?

An earlier firmware revision could be upgraded. That "feature" has since been eliminated.

Is there a big difference in price between the entry model and the top of the line model?
Maybe it's not worth the trouble, if the prices are very similar anyhow.

The last time I checked, the price difference between models was OK. However, that interpretation is subject to where each person draws the line between time and money. I don't know if there even is a way to upgrade it now, anyway.

[JBourke]

I have been wondering about that upgrade feature. From a production point of view it makes sense to have it included in the firmware as a way to quickly differentiate products. So in other words the feature may not have been eliminated but following a different coding scheme.

I take it the firmware is safely locked away from user view?

[bitseeker]

As I recall, the bug/loophole/method that was used to switch the maximum frequency was fixed so that it could not be exploited anymore.

[all_repair]

Has it been confirmed that the 2 MHz version is identical to the 24 MHz version in terms of hardware?

How to perform the software upgrade?

Is there a big difference in price between the entry model and the top of the line model?
Maybe it's not worth the trouble, if the prices are very similar anyhow.

Since you are in communication with the OEM, why don't you ask them and also ask why are they not performing the improvement you have asked. 

[pascal_sweden]

I contacted them already 2 times, without any response whatsoever.

[masterx81]

I've ended the build of the linear psu and tried to change the opamp (THS 3002) with a more powerful one (THS 3092), and i've traced the circuits around the opamp.
On the "IMG_20160731_190120.jpg" attachments you can see the schematic Of the CH1 with some components identified.
The output of the 12 bit R-2R if fed on a filter (LL2 1.2uH, C39 4pF, LL4 800nH, C43 8.6pF), then directly on the output opamp. The W6 trimmer adjust the gain of the output.
Then i've not fully understood what do the second opamp in the THS3002. Is fed directly by a second 12bit R-2R, then enter in an analog mux (the 'common' input of an hc4053). The non inverting input is 'adjusted' by W3. Maybe one do the positive part of the wave and the other one the negative, and the switch is done with the analog mux? There are in total 8 anaog mux (4 per channel)...
The 4558 sets the offset, seem that the mcu fed it via a pwm signal, filtered by C54/R108 and C55/R109. W4 i think that adjust manually the offset.
The change from the THS3002 to the 3092 gave me some strange results... I've tested it with 50 ohm termination. The waves don't reduce anymore amplitude increasing the frequency (image "8mhz 20vpp.png"), but i've got some nasty oscillations and overshots (image "4mhz 10vpp.png" and "8mhz 10vpp.png"), that maybe can be cancelled with some tweaks. The signal rise/fall a bit faster and the edges are a bit better. With this opamp there is the need of a bigger cooler as it heats quite fast.

Some thoughts?

[masterx81]

Noticed also that the r/2r network is done with 01B (1k 1%) resistors and 68A (499 ohm 1%) resistors, so the voltage divider created by the r-2r have a little error... Maybe using 2 1k resistor in parallel in place of the 499 one can increase a bit the accuracy, and also doing some "matching" for the resistors can increase the precision... Actually measuring them with a precise ohmmeter i get constantly a bit less than 1k... Not know how this reflect with the accuracy.

[maxiq4]

deleted..
   

[bianchifan]

Im sorry, this question some different your generator model ... a FY2300 6mhz..

FY2300 thread 

[maxiq4]

thank you
Im re-write correct post

[nctnico]

@Classical: this solution doesn't solve potential issues with noise from the AC power supply being pushed into your circuit. The best and easiest thing to do is to ground it.

[classical]

@Classical: this solution doesn't solve potential issues with noise from the AC power supply...

Yes, you are right and I am aware of that. But it reduces the leakage current and residual voltage enough for most of my applicactions.
And I am also aware that the solution of JBourke has the potential to reduce capacitive coupling about 4 times better than my solution. But that doubles the volume and cost if you do not the appropriate transformers on stock anyway.
In general I try to avoid PE on my measiring equipment but of course there are also temporary exceptions which can be wired easily if necessary. 
Many years ago I had to deal with fA and I experienced the benefits of active shielding and battery powered equipment.
But for my hobby purpose and tinkering I have to find an appropriate compromise.

[masterx81]

I've tryied to understand what was the purpose of the analog switches, and seem that the first 12bit r-2r network (24 resistors, R38-R48+R100, R85-R96) if fed via the analog switches, that switch every bit from ground to a negative voltage set by W3 via the second opamp of the output amplifier (see schematics of my previous post). The second r-2r network is fed directly by the altera cyclone FPGA (switch from 0 to +5v of vcc).
So actually there are 2x 12 bit R-2R DAC, for a total of 48 resistors and 24 bit for a full wave. Why the resolution is only 12bit? Maybe for user waveform is 12 bit and for example for a sine wave it's full 24-bit?
I've not understood why using a more powerful output opamp i've got oscillations on a square wave (as previous post)...

[masterx81]

With futher analysis i've undestood that the signal is generated by the r-2r network directly connected to the altera FPGA (so i get a 12bit 3.3v wave), then goes in the second opamp of the THS chip, got amplified and have the right negative bias, then enter in the analog mux and second r-2r, that attenuates it (via the amplitude setting). Then goes to the LC filter, and finally got amplified on the first opamp in the THS chip for the output. The offset is set by a pwm output on the FPGA that goes in a rc filter for DAC conversion, the get amplified by the 4558 opamp.
Probably i get worst response on square wave with the more powerful THS3092 because with the sharper edges (faster rise/fall times)  the filter does not work well, next thing is to try to simulate the filter and try to get it work better with the more powerful output opamp. I've also changed the resistors with some 0.1% 25ppm precision resistors (all 1k, i get the r value paralleling 2 resistors). Now i have a more precise and linear output (but i still have some non linear response as i can't set perfectly the voltage output that matches the setting over the wole 0-20vpp amplitude)

[bitseeker]

masterx81, thanks for continuing to share your findings. Has the change to precision resistors fixed the anomaly in the waveform near the zero-crossing that was attributed to the R-2R ladder?

https://www.eevblog.com/forum/testgear/feeltech-fy3224s-24mhz-2-channel-dds-aw-function-signal-generator/msg930730/

[masterx81]

With 2.3khz setting and 5v amplitude triangular wave like your test (not using 50 ohm termination, scope probe directly on output connector), i still get a 12mv "jump" between the 2 waveforms, that's the same that i have with the untouched CH2 channel.
I was expecting to have that error almost erased using precision resistors...

[masterx81]

Same test conditions, 50 ohm termination, 5mv div
RED is CH1 modified (also with THS3092, so the overshot caused by the filter),  YELLOW is CH2 unmodified
(Sorry for my noisy OWON scope I have also an old hameg analog scope but for this owon was sufficient)

[FlyingHacker]

What if you make a custom wave that is all positive, and then use offset to center it? Obviously you lose half the amplitude, but perhaps fix this crossover issue? Or is the offset before the "DAC" ?

[masterx81]

I've also tryied to upload a custom triangular waveform with 0  4  8 ...  4092  4095  4091 ... 11  7  3 sequence (with the upload software, preset waveforms) so that i'm sure that in the middle of the waveform there isn't any interruption (as the conjunction is between the 0 of the start and the 3 of the end of the waveform, so near -V and not around 0v) and it do the same thing.
The error is already present at the output of the first opamp (the one that "center" the waveform so that the 2047 digital output of the r-2r match with 0v).
Maybe i can try to scope the direct output of the r-2r but i need to desolder the opamp (with the opamp in place the output of the r-2r is so small that i can't neither scope it, as it enters directly on the negative input of the opamp and the feedback resistor attenuate it, it's something like 2mvpp) but i have no time now for this, the only thing that i've done is to scope the 11bit node of the r-2r (that's not attenuated) and the error is already here. I not know why there is this imprecision around the 2047 but i suspect a firmware bug on the FPGA. I not know anything about FPGA (know only about MCU) so are only supposition...

[bitseeker]

Thanks for checking the output with your mod. Bummer that it didn't fix it. I was hopeful.

[TheGiolly]

Hi, what is the lastest version of the PC software? And firmware?

[masterx81]

I'm using the sw version 1.4 (as the 2.0 seem only in chinese), and firmware/revision 2.0.

In my linear psu i've changed the 5v stabilizer from 7805 to a little smps module pin to pin compatible with the 78xx, and i have so much less heat on the heatsink... The only negative part is that the +5v rail feed also the +v of the analog switches (the -v is produced with a negative linear regulator on the board that is fed with -12v), but the ripple of few 10 mv is not giving any problem as after the first opamp the signal is near 7vpp.

I'm trying also to linearize the output so that the voltage set is the same that i get. For the offset i've replaced the R113 (3k9) with a 5k multiturn pot for have the correct amplification of the offset signal.

For testing the signal output i've uploaded 3 waveforms, one of all 0's (so i get a negative half of the amplitude setting), one of all 2047's (so that i have a plan 0v output - like setting the frequency to 0hz) and the last one of all 4095 (so i get a positive half of the amplitude setting). In theory all stages works in a linear way (the first amplification is linear via the opamp, then there is the attenuation via r/2r with precision resistors that must be quite precise and linear, and the last is the power opamp), but if i adjust the amplitude that matches with 10vpp setting it not matches with 20vpp, need to investigate a bit more.
The filter used to smooth the waveform seem an elliptic one, with capacitors to ground removed. Tried to simulate it but got notinhg usable. I will try a bessel one as soon as arrive a kit of 0603 inductors/capacitors...

[GigaJoe]

latest software version 2.2  + manual +  communication protocol spec ;  2016MAY11
http://en.feeltech.net/index.php?case=attachment&act=down&aid=47

CH 340 USB drivers
http://en.feeltech.net/index.php?case=attachment&act=down&aid=25

[masterx81]

Thanks for the link of the new sw! Seem identical to the other one that i had (1.4)...

I've tried to find where are the non-linearities in the amplitude, and are in the 4053 analog switches used for attenuate the signal amplified by the first opamp... i've bypassed the final amplification stage and the filter (for unloading the r-2r attenuation network), set a constant voltage out of the the dac, then set the amplification setting to 20v (maximum, so that all the analog switches was selecting the input signal - so no attenuation), measured the voltage in input of the analog switches and at the output (at the resistors), and at the output the voltage is not the same as the input. Nice to notiche that some only some chip are out of spec (as much as 1% error), so with a good selection of 4053 is possible to select the most precise one's and use them (thig that i'll do as they are quite cheap)


Got it: The error is due to the on resistance of the analog switches. I've cheked and seem that all the 4053 from various makers have almost the same specs. How much is bad paralleling 2 analog switches? Or eventually change the r2r from 500ohm/1k to 1k/2k?

Also the ADG733 is an interesting alternative, the low on resistance will eliminate the problem. But need to convert supply from +/-5v to +3.3v/-2.5v, not too complicated but then there is the need to reduce the amplification of the first opamp (that amplify the signal up to 7vpp) to max 5vpp.
I was not able to find an analog switch with same voltage rating and lower on resistance other than this... Also valuated to use an unity gain buffer but i not think that's a good solution.

Found also what's causing the error at around half of the 12bit scale, it's the switch of the msb. The problem (at least, this one) isn't the precision of the resistors, but that around the lsb the load on the altera chip caused by 500/1k resistor is too much and the output of the chip is not 3.3v but a bit less (Also here, due to on resistance). Need to try to switch from 500/1k to 1k/2k, or 5k/10k or 10k/20k. I not know if this have negative sides, but for sure change the impedance of the network, increase noise and increase error due to input bias current of the opamp.