Re: FeelElec New Arrival FY-6900 Signal Generator

[battlecoder]

@ battlecoder
I would not take the voltage up to 15+_ as the output ops are rated for 30v duel voltage MAX . Nom 28v.
 also take into account the working voltages of some of the caps .
I would not take the voltage above 12.5_+ . duel voltage 25v .
 I think JBG will agree . As we both have been to see Alice . 

Thanks for the warning. Any way of reducing the sine wave distortion at the highest voltage setting then? Or that's completely unavoidable?

[Labrat101]

@ battlecoder
I would not take the voltage up to 15+_ as the output ops are rated for 30v duel voltage MAX . Nom 28v.
 also take into account the working voltages of some of the caps .
I would not take the voltage above 12.5_+ . duel voltage 25v .
 I think JBG will agree . As we both have been to see Alice . 

Thanks for the warning. Any way of reducing the sine wave distortion at the highest voltage setting then? Or that's completely unavoidable?

I don't remember exactly what I did .
 The sine wave as I recall was ok ish upto 60Mhz . But I did change the ops and I put a
small ceramic cap 8pf on the input of the relay contact . as the relays on my one had a funny
contact impedance . which made the sine cleaner & the square look square.
 But the Jitter was like watching mountains during an earth quake .
The cap was on the op amp just after the dual filter there are 2 small pots ether side .
if you look there is a 49.9 resistor that goes to the relay centre .
along that track is a ground plain just put a 10pf or smaller . between the track and ground.
 this might help you .
 I did load up some pictures on another section about 6 months ago . with some stuff
JBG was working on his injection locking . which did not work on my one . I did something
slightly different . but it did not solve the problem of the  jitter .
 That's when I hit the bottom of the hole and had tea with Alice .
 And ordered the UTG 962 which just works great out the box .
 And wondering why I went as far as I did .
 It killed 9 months of Retirement boredom . Meet interesting folks on this forum .
 And Learned only one think how to waste time & money .
My  20Mhz 2ppb OCXO  + PLL was over $20 .
As I said earlier back I converted it to a Med Pulser 40hz cmos .
 I did use a proto type with a XR2206cp which actually goes to 1mhz and
No Jitter,   just 2 ceramic caps no xtal  it had a drift of about 0.02 hz an hr .
 & very clean cmos pulse . for a $1 chip. found in the scrap box .
 But for the med Pulse I needed more stable . now its ok..  40hz 10^-7

UPDATE Circuit  Diag of the FY66

[battlecoder]

I don't remember exactly what I did .
 The sine wave as I recall was ok ish upto 60Mhz . But I did change the ops and I put a
small ceramic cap 8pf on the input of the relay contact . as the relays on my one had a funny
contact impedance . which made the sine cleaner & the square look square.

I have the 30Mhz version. Didn't feel like going higher than 30Mhz back then, but still, the biggest distortion the sine wave suffers is near max amplitude. Frequency doesn't change much of it from what I've seen.

The cap was on the op amp just after the dual filter there are 2 small pots ether side .
if you look there is a 49.9 resistor that goes to the relay centre .
along that track is a ground plain just put a 10pf or smaller . between the track and ground.
 this might help you .

Will definitely try that, thanks.

And ordered the UTG 962 which just works great out the box .

I'm actually considering getting one. I just don't like the rather lacking options for modulation and external-supplied control signals.

And wondering why I went as far as I did .
 It killed 9 months of Retirement boredom . Meet interesting folks on this forum .
 And Learned only one think how to waste time & money .

Exactly my concern with the upgrades. That's why I want to keep the effort I put into modding it as low as possible. It already works reasonably well, so I'll probably just fix the frequency stability, and will probably add a fan (It already gets kinda warm. Putting an OCXO inside is only going to make it worse, so a fan is pretty much mandatory).

As I said earlier back I converted it to a Med Pulser 40hz cmos .
 I did use a proto type with a XR2206cp which actually goes to 1mhz and
No Jitter,   just 2 ceramic caps no xtal  it had a drift of about 0.02 hz an hr .
 & very clean cmos pulse . for a $1 chip. found in the scrap box .
 But for the med Pulse I needed more stable . now its ok..  40hz 10^-7

UPDATE Circuit  Diag of the FY66

That's a pretty neat use for a frequency gen.
Also, thanks for the schematic! I'm sure it'll come in handy.

[masterx81]

Hi!
What parameters did you use to check the jitter? Yesterday i've checked with my old owon 7102v, with 1mhz square, 50ohms termination on scope, enabled persistance, triggered some mv over negative part, i've moved the horizontal cursor for check the rising edge after the trigger rising edge and zoomed up to 2ns/div. I've measured less than 2ns of jitter. With lower frequency (khz region) got worst results, but i'm not sure if it's the scope or the measuring method.
Interesting the contact impedance of the relays, need to measure mine.

[Labrat101]

Your Scope is just Fine .
 The only real important thing is to set your probe @ x10 .
 If set to x1 the probe is 6Mhz BW @ 1M .
This will be on the scope labels or enclosed docs that came with the probes .  also X10 will but less load on the DUT . 
 If you already knew this ignore the above . 
 Depending which AWG you are using and they added Magic jitter software which I have no idea if its meant to
stop jitter or just give an allusion or a BS gimmick .
But I guess 2ns is better than 4ns .  Less than 1ns good standards .
 

[masterx81]

I'm using direct bnc to bnc rg58, with a T on the scope and 50ohm termination. The scope is set to 10x as i was too lazy to change the setting 
I not understand why at 1Mhz seem pretty stable, while at lower frequencies it goes almost everywhere. Maybe triggering at 50% of the edge will be better.
This evening i take few shots.

[Labrat101]

RG316 50cm cable 50 though termination.
Persistent set @ 1 Seconds both Channels
Display set to DOTs
Channel 1 yellow 11Mhz  Channel 2 Purple 19.23Mhz
Real time setting . Equal time would give a flatter top.
 jitter 200ps
PK scope (Siglent SGS1150 150Mhs)) 
Aglient 200Mhz (bottom Pic)

[masterx81]

In order:
1.000.000,000 hz

1.000.000,001 hz

10.000,000 hz

10.000,001 hz

100,000 hz

100,001 hz


Persistance is set to infinite, almost 30secs.
As this is a 200Msps device, any multiple of 5ns must be almost free of jitter. If i select 0.001hz more than the multiple, the fun begins.
But... why at 100hz i get a so bad result?
Note at 1.000.000,001 hz seem that what's changing is only the rise time. I not know how is possible.

[Labrat101]

@ masterx81
I ran those frequencies with the UTG962 with your setting could not replicate  .
 I think the problem is as I have said earlier .  Fake Cyclone and the software is
over compensating .
This is why I gave this unit up as a Dead duck in a cute Box.
 Its as good as a multimeter that can't measure volts .

edit my scope did not show frequency in image . 1.000,001hz
I can't figure out how your rise time is so slow ..
  its a 1 second pulse . on/off relay contact would show faster .
100hz is showing a double sinc . hardware error . FY fuzzy logic

[masterx81]

My tests are at the rising edge after the trigger point. You are showing the trigger point

[Labrat101]

sorry .
I still could not replicate that .
I used your setting still could not replicate that @ 100hz

[masterx81]

The FY6900 have a 0.9ppm TCXO, maybe the clock source for the scope is higly unstable?

[Labrat101]

The FY6900 have a 0.9ppm TCXO, maybe the clock source for the scope is higly unstable?

I doubt that very much more like the cyclone is playing up .
Do a recal on the scope and reset but the scope is fine.

[masterx81]

Interesting that at 1Mhz is quite stable. I'll reset the scope, and try to scope the 50hz mains to see if it's more stable, and maybe the 1khz probe compensation output.

[Labrat101]

Stop trying to finding things wrong with the scope .
 Its the FY69 that is faulty .. and it won't mater what you do . will not change that fact.
 All you will succeed in is messing up a perfectly good working scope .
 This was the same crap I was getting from my FY . Its full of Fake chips crap software .etc
 Its hard to except it .. sorry .. I had the same thing . same outputs ..
   This is why I don't use it any more. .
Sorry no hard feelings .

[Johnny B Good]

 Try swapping the dots (periods/full stops) for commas and replace the comma with a decimal point (period/full stop).

 You've just fallen into the trap known as "Mathematic convention". Most, if not all, of Europe use a comma as the decimal point and the full stop (period) as the digit formatting separator.

 The first frequency shown is actually 1MHz. The key is in recognising the anomalies in the use of commas and full stops or periods. For Anglo-Americans the oddities are a comma followed by one or more full stops or periods or else a single comma preceded by one or more full stops or periods.

 Once you keep in mind the international nature of the EEVBlog membership and take a closer look at any numbers that don't look "quite right", it'll become obvious enough as to which convention has been applied.

John

[StillTrying]

Most, if not all, of Europe use a comma as the decimal point and the full stop (period) as the digit formatting separator.

Most of Europe is wrong then,   no wonder they all drive on the wrong side of the road as well.

[nctnico]

My tests are at the rising edge after the trigger point. You are showing the trigger point

Set the triggering to normal mode instead of auto. At low frequencies the auto trigger likely fires before the signal which results in drawing random traces.

[battlecoder]

Most, if not all, of Europe use a comma as the decimal point and the full stop (period) as the digit formatting separator.

Most of Europe is wrong then,   no wonder they all drive on the wrong side of the road as well.

Not only Europe. We use comma as decimal in many countries here in South America. Definitely not a fan of that.

[masterx81]

Set the triggering to normal mode instead of auto. At low frequencies the auto trigger likely fires before the signal which results in drawing random traces.

I will try it, thanks!

And sorry for the decimal point thing, also here often cause problems...

[StillTrying]

By my rough guestimate that X smudging is not too bad. The +Ve trigger point is 133,000  (133 thousand ) screen widths away. Roughly +/- 1ppm of jitter shared between the FG's and scope's clocks.
- I think.

[TurboTom]

Just for comparison : This is what the same measurement looks like on a "liberated" DG811AWG.

[rf-loop]

Just for comparison : This is what the same measurement looks like on a "liberated" DG811AWG.

I am not sure now, so many images and bit poor explanations how they are exactly done.
But is it so that this previous Owon scope image about FY square, or was is pulse, is 100Hz, signal, period 10ms and in display next rising edge after trigger. So it is exactly what mean cycle to cycle jitter.

When do higher freq and then look over same time period 10ms situation is or is perhaps different depending... Now it least in some cases may affect lot of or nearly nonsense. It depends how square/rectangle  or pulse  is produced in system. Some times it can also see on data sheet where is jitter defined as just imagined example 100ps + 1% period.

So because it is somehow unknown how different generators produce sqr, rectangle and pulse, mu opinion is that it is best to use same square/rectangle period for compare.  Pulse is one special case when generator (most of) have separate pulse and rectangle mode.  Some simple generators may use even DDS arb procedure for produce pulses.
In history there was example one simply Siglent SDG1000 (no X)   where square/rectangle was internally produced using classic comparator method where square was derived from internal sine using fast comparator leading to quite good cycle to cycle jitter in high frequencies and then very extremely poor jitter in very low frequencies.  If example with this test over 10ms period using  tens of kHz or MHz range cycle cycle jitter is amazing low and if test with 100Hz it is nightmare and with 1Hz no need tell... yes it can read from data sheet, they did not lie. But how many understand it - well, I have seen some wondering after purchase when not even read or think what is told in data sheet.

Long to short... better to use same setup for compare.   Even if then can see difference is nonsense, but this can see only afterwards.

FYI  one parcel is on the road...

[masterx81]

As i've specified few posts before the screenshots, on the fy i've only selected square wave and changed frequency, default 5vpp and other settings.
The square is geerated  like every other waveform, through the DAC.
I'm pretty sure that the dac run always at the same 200Msps sampling frequency

[Labrat101]

Just for comparison : This is what the same measurement looks like on a "liberated" DG811AWG.

Dictionary   lib·er·at·ed 
1) (of a person) showing freedom from social conventions or traditional ideas, especially with regard to sexual roles.
"the modern image of the independent, liberated woman"
2)(of a place or people) freed from imprisonment, slavery, or enemy occupation.
"liberated areas of the country"

All my tests were Done on a UTG962  . Un modded in any way.
 I don't think my UTG962 or my scope . come under this category . ( liberated )
  Mind you The UTG962 has a very slim Body and small foot print .