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

[Vytautas]

By the way, this is the schematic of my +-15V rail. It is, sure, an overkill. But I like overkills 
16V AC enough to drive it. Or, alternatyvely, two 15 AC windings could be used for full wave rectification.
Cheers!

[cybermaus]

  I now have additional questions.

You seem to have it working, so I must be missing some trick.
How are you getting 16V peak out of 9.8Vrms?
What are those 100R R1,R2, C5,C6 doing there, are they not dropping voltage even more?
Also, what is that 10R R7 doing?

[Vytautas]

Not out of 9.8V but out of 19.6V. As I mentioned the double 9.8V windings are connected in series to get 19.6V. Both transformers you see in the image have DUAL 9.8V windings (and other windings which I use for the +5V rail).
Concerning the components, if you want to read more about a similar supply design, welcome to see here:
http://sound.whsites.net/project05b.htm

[cybermaus]

Ah! I stupidly assumed the other one was the 5V trafo.

I read you had them in series, but even that I had read as in series and centre tapped.
I guess your unique transformer selection means they all come from the leftover bin.

[Vytautas]

Of course, left over. If I were richer, I might have well purchased a fancy Agilent or Tektronix generator 

[beanflying]

My x1 y1 Vishays turned up today. Same esr for the stock and vishay and in the end the same sorts of voltages to mains earth.

I went back over the 1M, 470k, and the 100k resistors from virtual ground to mains earth and the 100k as before gave the lowest result at under 3Vac again.

So nothing new I guess. Must look back at the better quality mods now 

[beanflying]

Firstly I know ZERO about Cyclone chips but has anyone got any thoughts on the possibility of downloading and or upgrading the firmware without this ridiculous 'must be sent back to the factory for upgrade'?

My reason is this seems this version 2.9 is consistantly higher on output than indicated on both DC offset and AC waveforms at least. I haven't looked more comprehensively at the other waves and such or pushed the limits of the unit.

My Hantek Scope backs up my Agilent and double checked against one of my Victors on this.

5Vac p-p (1.768 RMS) 50hZ indicated (Victor 86B 1.842 49.98Hz) (Agilent 1.8396 50.001Hz)
5Vac p-p (1.768 RMS) 1kHz indicated (Victor 86B 1.826 0.999kHz) (Agilent 1.8394 1.00004kHz)
5Vac p-p (1.768 RMS) 10kHz indicated (Victor 86B 1.128** 9.99kHz) (Agilent 1.835 10.0004kHz)

The real problem becomes when you add a small DC offset to it as per the picture below.

Same series of Voltages and Frequencies as above with a "2Vdc" offset just with the Agilent. Not sure if the small drop off in AC voltage is power supply or maybe meter waveform integration related?

50Hz  1.798ac 50.002Hz 2.2215Vdc
1K 1.797ac 10.00004kHz 2.2212Vdc
10k 1.793ac 10.00004kHz 2.2206Vdc

Even when I drop the Amplitude and Frequency off to Zero it still reads 2.178Vdc (ac circa 0.06mV) apparently this is a firmware calibration issue hence the first question. 

[cybermaus]

Measured my values to compare, on a 3.1 version
Note that if these devices are calibrated, that would also be in the flash, and more relevant then the version.

I do not have such a nice bench meter, but here is what I measured:

5Vpp + 0Vdc, pretty consistent for 10KHz 1KHz 50Hz
5.3Vpp (scope)
1.77Vrms on my trueRMS BXM9BT
1.77Vac on my VC9808+

5Vpp + 2Vdc and 0Vpp+2Vdc
2.15Vmid (scope)
2.14Vdc on my trueRMS BXM9BT
2.13Vdc on my VC9808+

0Vpp + 4Vdc
4.34Vmid (scope)
4.33Vdc on my trueRMS BXM9BT
4.30Vdc on my VC9808+

So, over the board +7% on DC offset, but Vpp was pretty good.
Slightly better I guess, but is it version or calibration? They may have skipped calibration in mass production, and offered you calibration (on top of new flash)

[beanflying]

Not sure if I feel better or not now knowing mine is not the only one out of step with accuracy 

Interesting to see if anyone with 3.9 or more recent firmware gets better or similar. If similar then I guess it's time to go chasing a hardware solution?

As a fallback not that it is ideal there is several regular waveforms needed I planned to program anyway so I could always preset "known" voltages for these against  my Agilent or CRO.

For the sorts of work I am doing which is low voltage and sub 1Mhz DC the timebase issue doesn't concern me at this time but cool creative stuff still

[ArthurDent]

(Note: I edited this post after don.r pointed out I had misread the part number mentioned as a future replacement IC. The rest of what I posted regarding possible noise imposed on the +/- supplies should be correct.)

On replacing the mystery dual op amp with 2 THS3095D single op amps as Vytautas has done, I just ordered 2 from Mouser Electronics. They still aren't real cheap at $10.64 US each but still worth the cost if you want to upgrade the output to get lower distortion at the higher voltage the FY6600 is capable of.

I'm not sure if these faster chips will improve the rise time on square waves or not but they won't hurt. Also I'd assume the THS3095D chips draw a little more supply current and may impose some added load and noise on the +/- supplies feeding them and means you'll want to use a +/-15V supply with higher output current capacity instead of the existing +/-12V supply to get the full benefit of the new chips. I'll check the a.c. voltages on the bypass capacitors next to these chip with the scope once they are installed to see if there is any noise cause by the higher current drain of the THS3095D chips necessitating more capacitance needed across the existing capacitors and report what I find, probably in about 1 week.   

[don.r]

The THS3091 is an old chip. The new ones are the THS3491.

[ArthurDent]

Oops, I misread. Should pay more attention next time. Thanks for pointing it out.

[beanflying]

Output Voltage Calibration

Seems the output voltages are not in fact firmware set. After taking a careful photo and tracing the board a bit the following pots can be twirled 

As labeled. They are very twitchy so I might look at hacking some 10 turn pots to replace them.

W5 CH1 DC offset adjustment
W3 CH2 DC offset adjustment

Easiest cal method. Set Frequency to 0 and ampl to 0. Set offset to 10V and tweak. There is still a bit of non linearity so some more looking at resistors or for voltage references is in order.

W26 CH1 AMPL P-P adjustment
W16 CH2 AMPL P-P adjustment

I found setting these to 10V P-P gave me about the best linearity across the range. At 20V P-P low Amplitudes were out. So like the DC a bit more looking required. Tested from 50Hz to 100kHz and no real amplitude change.

This gets it within 1-2% across the range instead of the up to 10% I was reading 

So what does it mean - less than ideal calibration and testing in the factory I suspect?

[beanflying]

W1 changes the CH2 low end DC calibration. Set 100mV and tweak seems to work best

W2 is the same but for CH1.

They do also effect the top end DC calibration by a small margin so a little to and fro with W3 and W5 is needed to get the best result across the range.

I have exhausted what coffee can do for today. Time for a BEER 

[Diabolo]

Hello,

I own the FY2300H 14 bit 250 MSa / s and had to open the generator to adjust the output levels when I received it. Under the radiator FY2300H there are IC 30021 and MC45558, and also the free tracks U21 and U22 for soldering of 2 IC TH3091 after remove IC 30021.
The THS3095 IC has a power-down spindle that is not useful since not connected on FY6600 et FY2300H Feeltech.
 
Buy THS3091 low price :
- https://fr.aliexpress.com/item/1pcs-THS3091DR-THS3091-SOP-8/32824185282.html?spm=a2g0s.9042311.0.0.SpCn6H


Regards,
Diabolo

[cybermaus]

I have exhausted what coffee can do for today. Time for a BEER 

Well earned.

[beanflying]

And to Feeltech, you shouldn't take our criticisms of the current generation of product to heart to much. There is a reason we are working on our units the basic ideas and features we like are worth persisting with and if it helps you to improve future models then great 

So far mine for some time and under $4 of components is a much improved item. Some of the others upgrades would certainly add a bit to your costs but for most users not required I feel.

Jump back in we don't bite to much 

[sorin]

Output Voltage Calibration

Seems the output voltages are not in fact firmware set. After taking a careful photo and tracing the board a bit the following pots can be twirled 

As labeled. They are very twitchy so I might look at hacking some 10 turn pots to replace them.

W5 CH1 DC offset adjustment
W3 CH2 DC offset adjustment

Easiest cal method. Set Frequency to 0 and ampl to 0. Set offset to 10V and tweak. There is still a bit of non linearity so some more looking at resistors or for voltage references is in order.

W26 CH1 AMPL P-P adjustment
W16 CH2 AMPL P-P adjustment

I found setting these to 10V P-P gave me about the best linearity across the range. At 20V P-P low Amplitudes were out. So like the DC a bit more looking required. Tested from 50Hz to 100kHz and no real amplitude change.

This gets it within 1-2% across the range instead of the up to 10% I was reading 

So what does it mean - less than ideal calibration and testing in the factory I suspect?

I dont think so.
In my opinion W5, W3, W26, W16 are used for Differential OpAmp compensation.
More info here.

The responsible components for the DC offset are IC7 & IC8, 12bit DAC [MCP4822E].


Thanks to Ebel0410 for the image.

[beanflying]

And if we continue your logic the DAC's are actually controlled by the Cyclone so that is what actually controls the DC offsets not the DAC's.

As we currently have no way into the micro we have no control over the DAC's and their output calibration so the only control over the outputs is by controlling the opamps or additional analog circuitry after the DAC's to my way of thinking.

[cybermaus]

In my opinion W5, W3, W26, W16 are used for Differential OpAmp compensation.
More info here.

The responsible components for the DC offset are IC7 & IC8, 12bit DAC [MCP4822E]

My original thinking was also that those low-speed DAC's do slow amplification and offset levels. However, one cannot argue with results, and beanflying seemed to got that.
It may have been a side-effect of something else, but can you explain what in your view the differential opamp actually does, and what would have been de-tuned when beanflying changed those pots?

[Diabolo]

Hello,
W1 and W2 are the settings that I had to adjust equalize the amplitudes when receiving my FY2300H
I ordered 2 IC THS3091 to improve (?) the signals and also to spread the heat dissipation on 2 IC and not on 1 IC alone.

Regards,
Diabolo

[beanflying]

Coffee Morning Sunshine and a pleasant 22 degrees in the shack 

Just ran a series of tests on Ch1 and 2 after yesterdays tweaking. A few of the 1Mhz voltages must be pushing either the Feeltech or the Agilent a bit so they are out. Will boot up the scope later and have a look.

Just checked the Agilent is rated +-4% from 100-300KHz and Typical error at 1MHz of 30% 

[don.r]

Coffee Morning Sunshine and a pleasant 22 degrees in the shack 

My lab in the garage is 2 degrees this evening. 

[Mr. Scram]

My lab in the garage is 2 degrees this evening. 

Just give your office chair a push and you'll see the other 358 degrees swirls by in no time.

[beanflying]

Coffee Morning Sunshine and a pleasant 22 degrees in the shack 

My lab in the garage is 2 degrees this evening. 

I remember being in Chicago in early November once when the NW wind came in out of Canada wasn't pleasant at all.

Building Voltage references in shorts and a Tshirt at present