Author Topic: FeelTech FY6600 60MHz 2-Ch VCO Function Arbitrary Waveform Signal Generator (Read 557268 times)

rhb · « **Reply #1525 on:** June 22, 2018, 01:21:53 pm »

Thanks. All clear except the last photo. What does that show?

sorin · « **Reply #1526 on:** June 22, 2018, 01:32:36 pm »

See the 2 smd inductors on the on the left.

Sent from my Redmi Note 4 using Tapatalk

Scratch.HTF · « **Reply #1527 on:** June 23, 2018, 07:50:40 am »

With regard to the high speed DACs where their markings have been sanded off, they correspond to the pinout of the DAC904, but are overclocked given that the DAC904 has a typical output update rate of 165 MSPS @ 3.3V (these DACs are powered by 3.3V in the FY6600) and 200 MSPS @ 5V and that I cannot find a 250 MSPS (or faster) DAC with the same pinout (and interface) as the DAC904 on the Texas Instruments website.

Mozee · « **Reply #1528 on:** July 02, 2018, 03:37:55 pm »

Thanks a lot Val , but what about calibrating the frequency on FY6800 also the DC offset is way off!!!! how to calibrate those?

alexvg · « **Reply #1529 on:** July 12, 2018, 09:19:40 am »

Hi (from France),

I've bought the FY6600 some weeks ago.
Big thanks to all members for very usefull information about this product.

I've done lot's of update :
- Build linear power supply
- Change output amplifiers with a couple of THS3091
- Removed 5th order filter (currently designing a new 11th order filter)
- Added some ESR capacitors

The used DAC seems to be a custom or modified DAC904 or a special "low-cost" and high speed version of the AD9744.
On the original DAC904, adding a 100nF on BW pin limits the badwidth at 1.3MHz (I've tryied this without any succes on my FY6600).

The buffer (LNS1/LNS2) after DAC904 seems not be OPA686 (marking "A86") on my FY6600 but a LMH6702 (marking is "A83A")
https://www.eevblog.com/forum/testgear/feeltech-fy6600-60mhz-2-ch-vco-function-arbitrary-waveform-signal-generator/msg1340574/#msg1340574

pa3hfu · « **Reply #1530 on:** July 12, 2018, 10:32:33 am »

I cut a peace of tube (bicycle) of 5 mm wide. It goes straight on the knob and it turns much better now!

Scratch.HTF · « **Reply #1531 on:** July 12, 2018, 12:02:33 pm »

My tests reveal that the THS3002 has approximately 36 ohms of dynamic resistance with a +/- 12V supply driving a 1 kHz 10V pulse into 100 ohms (50 ohms/50 ohms source/termination) while the AD8009 (HKJ marking) has practically no dynamic resistance driving a 1 kHz 2.5V pulse under the same source/termination conditions.

I also measured losses in the sine shaping filter:
Positive and negative peaks within 2.5V from zero (dB loss/MHz) - AD8009 output used: 0.066/1, 0.132/5, 0.233/10, 0.472/15, 0.648/20, 0.755/25, 0.864/30, 1.159/40, 1.427/50, 1.583/60
Positive and/or negative peaks over 2.5V from zero (dB loss/MHz) - THS3002 output used: 0.067/1, 0.326/5, 0.789/10, 1.339/15, 2.052/20 - firmware limits frequency to 20 MHz under this condition

rhb · « **Reply #1532 on:** July 13, 2018, 11:57:54 am »

I bought a Keysight 33622A just before I received a replacement front panel for my 6600. Here are a couple of photos of my 8560A with both.

7146 - 6600
7147 - 33622A

Both were set to the same output level using a scope (640 mV after 50 thru terminator). Mostly the spurs are the output amp in the 6600. At lower output levels it cleans up.

VE2UM · « **Reply #1533 on:** July 14, 2018, 05:43:49 pm »

This forum is totally CRAZY !!!!!

I just discovered that the main proble with my FY6600 was .... the 50 MHz reference oscillator. I came here and you already found and published it !!!!!

So, my next step is to find a better 50 MHz reference !

The reference provided with the 6600 is so awful, just puting my finger on the epoxy body causes a huge amount of 60 (50) Hz FM humm !

Furthermore, full of infos about modifications (i seriously thought about a three-prong IEC C14 connector..

Also, i've ordered some stick-on heatsinks for a Raspberry-Pi. I'll put-on one on the ALTERA FPGA.

But with over 60 pages to read, i'll have full of infos !

Again, thanks !

alexvg · « **Reply #1534 on:** July 15, 2018, 09:04:58 pm »

After doing lots of hardware tests and updates, I’ve found a bug.
If a set an output frequency of 5MHz (250MHz div 50), the signal is perfect.
If the output frequency is 6MHz, the signal is awfull and totally unstable.
Duty cycle and phase work less and less as frequency increase.

Why ? Because the output sampling does not use a shift phase sampling algorithm (oversampling/undersampling technique).

I don’t know if the Altera Cyclone IV could provides enough power to implement a shift phase sampling algorithm.

Personnally I’ve always implement this type of algorithm in many audio engine. Feeltech has done a mistake and some functions does not work at high frequency !

Eeerkk !!!

cybermaus · « **Reply #1535 on:** July 16, 2018, 06:57:08 am »

That would be square wave or sampled wave with hard vertical edges right?

Yes, the sample jitter has been discussed quite a bit already.
But Sine is OK, showing they are in fact shifting the phase during the waveform construct.

alexvg · « **Reply #1536 on:** July 18, 2018, 07:11:50 am »

I've tested the output with a spectrum analyser and finishing the design of the new output filter.
I'm going to implement the filter in the next few day.
I also found that my current version of the FY6600 generates lots of harmonics beyond 2Vpp.

Here's theorical frequency response and 10MHz square signal.

Linearity between 1M and 120M

soundtec · « **Reply #1537 on:** July 18, 2018, 10:20:35 am »

I guess all the good weather has the people minds on other things ,either way its seems to have gone very quiet lately ,
I had trouble even accessing the post as its gone from the top of the test list ,and the search didnt find it .I needed to browse to the webpage on another tab to find the post while already logged in another tab

Anyway good to see some progress Alexvg,
Will you implement the new filter on the existing pads on the board?
designing a small smd based daughter board might be a nice idea if your design works out ,
Presumably you intend better roll off of the out of band frequency ripples and also more linear behaviour over the wanted range .
I'll be very interested in the results of your tests ,how does the new filter design look in the modeling software?

alexvg · « **Reply #1538 on:** July 19, 2018, 05:41:01 pm »

Here is a simple test with 2 capacitors of 69pF (to validate the 50R output impedance of the DAC)

alexvg · « **Reply #1539 on:** July 19, 2018, 10:49:42 pm »

Some measurements...

Channel 1 without 5nd order filter at 2Vpp and 8Vpp
Channel 2 with 1st order filter at 46MHz instead of 5th order filter, tested at 2Vpp and 8Vpp

alexvg · « **Reply #1540 on:** July 22, 2018, 12:53:33 am »

I've done the output 11th order filter with SMD components.

alexvg · « **Reply #1541 on:** July 22, 2018, 12:55:18 am »

All tests have been done with 50R load.

Sine sweep 0.5-30MHz - Spectrum 1-500MHz lin - 10dB/div - No filter (yellow) vs 11th order filter (magenta)

Sine sweep 0.5-30MHz - Spectrum 1-50MHz lin - 1dB/div - No filter (yellow) vs 11th order filter (magenta)

Sine sweep 0.5-30MHz - Spectrum 0.1-50MHz log - 1dB/div - No filter (yellow) vs 11th order filter (magenta)

Multitone sweep 0.5-12.45MHz - Spectrum 0.1-50MHz log - 1dB/div - No filter (yellow) vs 11th order filter (magenta)

Multitone sweep 0.5-12.45MHz - Spectrum 0.1-500MHz lin - 10dB/div - No filter (yellow) vs 11th order filter (magenta)

alexvg · « **Reply #1542 on:** July 22, 2018, 12:57:28 am »

All tests have been done with 50R load and 20dB att.

Yellow : no filter
Magenta : 11th order filter

1MHz square output

5MHz square output

10MHz square output

25MHz square output

30MHz square output

alexvg · « **Reply #1543 on:** July 22, 2018, 02:43:13 am »

Final measurement... with 50R load.

1MHz signal (blue = no filter, green = 11th order filter)

1MHz signal without any filter (yellow)

1MHz signal with 11th order filter (magenta)

alexvg · « **Reply #1544 on:** July 22, 2018, 02:55:14 am »

Result
The filter is not shielded some DAC harmonics could not be correctly filtered. These harmonics and parasitics frequency cound be measured on the output.

Next time ?
Replace 50MHz oscillator with a new one (+/-1ppm).

bugi · « **Reply #1545 on:** July 22, 2018, 07:09:02 am »

In the first results measurement, with sine and multitone with linear and log graphs, I think you accidentally swapped the markings between filter and no filter sine curves between the linear and log graphs. At least they would match each other correctly if either screenshot swaps those two. I'd guess the linear should do the swap as it currently claims that the 11th order filter has less loss than no filter.

alexvg · « **Reply #1546 on:** July 22, 2018, 09:29:26 am »

Edit: Previous measurements are corrected.

Crosstalk measurement
The crosstalk occurs after the filter, between LNS1/LNS2 and final output.

Yellow curve - CH1 with multitone 0.5-12.45MHz (no filter)
Magenta curve - CH2 off (11th order filter)

Yellow curve - CH1 off (no filter)
Magenta curve - CH2 with multitone 0.5-12.45MHz (11th order filter)

alexvg · « **Reply #1547 on:** July 23, 2018, 12:45:00 pm »

I'm working on a highly optimized 5th order filter and simple to build to replace the complex 11th order filter which could be hard to build for the beginner.

or

alexvg · « **Reply #1548 on:** July 24, 2018, 09:03:14 am »

I've done the schematic of the new filter at build it... With 5% capacitors and inductances (without any sort).

I've currently working on the same filter but with selected capacitors and inductances.

The schematic

The circuit

Multitone sweep -11th (magenta) vs 5th (yellow) order filter, 1dB/div, 100MHz lin

Multitone sweep -11th (magenta) vs 5th (yellow) order filter, 5dB/div, 100MHz lin

Multitone sweep -11th (magenta) vs 5th (yellow) order filter, 3dB/div, 125MHz lin

Multitone sweep -11th (magenta) vs 5th (yellow) order filter, 6dB/div, 125MHz lin

Multitone sweep - 11th (magenta) vs 5th (yellow) order filter, 10dB/div, 500MHz lin

Sine sweep - 11th (magenta) vs 5th (yellow) order filter, 1dB/div, 50MHz lin

Sine sweep - 11th (magenta) vs 5th (yellow) order filter, 10dB/div, 500MHz lin

Square signal at 2.5MHz - 11th (magenta) vs 5th (yellow) order filter

Square signal at 5MHz - 11th (magenta) vs 5th (yellow) order filter

Square signal at 10MHz - 11th (magenta) vs 5th (yellow) order filter

Square signal at 25MHz - 11th (magenta) vs 5th (yellow) order filter

Channel crosstalk
Yellow : measurement on 5th order filter channel at off state, 11th order filter channel with multitone sweep.
Green : measurement on 5th order filter channel at off state, 11th order filter channel with multitone sweep at off-state !!!
Magenta : measurement on 11th order filter channel at off state, 5th order filter channel with multitone sweep.
Blue : measurement on 11th order filter channel at off state, 5th order filter channel with multitone sweep at off-state !!!

Pro and cons of this 5th order filter "1st version"
- Need to select with high precision the capacitors and inductances
- Very simple and fast to install
- The compensation filter provides excellent linearity
- The square signal is more raw with lot of harmonics than soft square signal with the 11th order filter

Ebel0410 · « **Reply #1549 on:** July 25, 2018, 08:58:34 am »

Hi Alex,

That's a wonderful and helpful hack you did here, many thanks.
Work in progress I would say...

I didn't see the schematics of the 11th order filter published, could you please provide it us with de components values ?
(what filter designing soft are you using, Qucs?)


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: FeelTech FY6600 60MHz 2-Ch VCO Function Arbitrary Waveform Signal Generator (Read 557268 times)

Share me