Re: FeelElec New Arrival FY-6900 Signal Generator

[Bad_Driver]

I have only little base knowledge of it. It's far better to read the last postings of Johnny B. Good.
Yes very long but very helpful. He describes the impact and usage of the VCO-calibration.

At the end it is simple, 5...10 k multiturn R between Voltage and ground and the variable output (I do not know the english words)
to VCO but Voltage must be precise. Ask JBG he has worked on it for a long time! And it takes time.

[Bad_Driver]

I found under the heat sink this two non populated IC pads. Does someone has a clue what this could be for?
Different OPs with other footprint? Or another pair of THS3002 to increase load?
Has someone used it? I apologize when this was part of the ongoing disussions and I missed it.

Thanks in advance!

[Labrat101]

I found under the heat sink this two non populated IC pads. Does someone has a clue what this could be for?
Different OPs with other footprint? Or another pair of THS3002 to increase load?
Has someone used it? I apologize when this was part of the ongoing disussions and I missed it.

Thanks in advance!

O yes these were discust in great detail .
They are for adding 2 singe op amps instead of the dual op amp . There is a circuit diagram I will have to dig it out
its also on github  don't have all the details on hand . I did this upgrade about a year ago .
I will try and find them .  JBG may have a better memory were all this was posted .. way back.
I'm a bit involved in rebuilding 3  HP 5385a Frequency counters got cheap as not working got 2 out of 3 working so far.  ..
RNS
 

[Bad_Driver]

Thanks for your reply! I found the schematics of the 6600 but I was not aware that there are spare pads
for other OPAs. I read all the stuff about changing the old THS3002 against the better single one but
I didn‘d find it in the FY6900 topic. Have you done it?
And btw nice counters! Hope you get all three running!

[Johnny B Good]

I found under the heat sink this two non populated IC pads. Does someone has a clue what this could be for?
Different OPs with other footprint? Or another pair of THS3002 to increase load?
Has someone used it? I apologize when this was part of the ongoing disussions and I missed it.

Thanks in advance!

 That's an interesting photo you just posted. It looks like someone will have to reverse engineer this main board on account it is radically different from the FY6600 and the FY6800 models as far as the final 12dB boost amp circuit goes.

 What's different in the case of the FY6900 is that they've used a pair of THS3002i dual opamps in place of the single dual opamp used in the earlier FGs. It was a little tricky to make out what that IC in the top left hand corner of the heatsink's keep out area was (I'd been able to positively identify its cousin in the bottom LH corner, a THS3002i). I was surprised to see a second THS3002i chip but the 100 ohm resistors and the FB and Gain resistors rather give the game away.

 The unpopulated U21 and U22 are for single higher spec opamps THS 3001/3491/3495 etc. If you have a go at this upgrade, you'll need a pair of the higher spec single opamps to fill the unpopulated spots and remove both THS3002i opamps and strap the 100 ohm resistors now separated by the removal of the 3002i opamps (each one of a pair used these resistors to combine their individual outputs so the load saw a 50ohm source but the outputs were in no danger of fighting each other due to slight gain mismatches - it's a common way to use more than one opamp in parallel to improve performance at higher frequencies).

 TBH, I'm rather surprised they'd bothered providing the unpopulated U21 U22 locations on the main board since with a 20MHz upper limit before the 24v pk-pk limit is dropped to 5v (and these opamps switched out of circuit), I'd have thought just doubling up on the 3002i opamps alone would have addressed the horrible distortion on sine wave output at 20v pk-pk at the 20MHz limit exhibited by its predecessors.

 Check the waveform at the same 20MHz limit for the 24v p-p option driving 50 ohm loads which should result in a 12v p-p waveform. If it looks more like the 2nd attached image SDS00227 (THS3491 upgrade) rather than the 1st image SDS00163 (the original THS3002i opamps), I wouldn't go to all the bother and the expense of another pair of THS3491 opamps just for a marginal, if any, improvement. There might not be any real need in this case to duplicate the opamp upgrade efforts of the FY6600 and 6800 community. 

JBG

P.S. I'm still busy testing my MK II GPSDO (after an initially promising but, ultimately disastrous start - I somehow blew up a 74HC14 and two of my 5v RRO cmos opamps, culminating in the tiny 1.3A mini 360 5v buck converter starting to let out its magic smoke.

 I had modded the buck inverter to give me 5.340v (the cmos opamps are rated for 5.5v working but it was the first one that went short circuit and cooked the buck inverter's goose). I landed up trying out my one and only example low noise 5v buck converter with 5v LDO in place of the burnt out one simply on account it had the same pin out footprint and it seemed a convenient way to test the claimed "Best of both technologies" most of the efficiency benefit of a switching regulator with an LDO to get rid of the switch noise (yeah, like that's gonna happen ).

 I haven't gotten round to checking this low noise buck/LDO chimera's noise and ripple yet - I've been rather busy fettling my now functioning GPSDO board. The latest experiment involved a test with ten times the 500 seconds TC (replace the 100K and 1M resistors with 1M and 10M - I couldn't find another 5.6M in my collection but I had a few 10M resistors ).

 As you could imagine, it took way longer to lock the OCXO - around 3 1/2 hours versus the 25 minutes with the 500 seconds TC !!! Fortunately, I have a simple circuit add on which should shave a couple or more hours off that startup time which I hope to put to the test later today.

[EDIT 2020-07-09]

 That "Low Noise best of both worlds buck/ldo" module exceeded my expectations of decrepitude, largely I suspect due to the use of an unshielded inductor. Also, my simple two diode (one for each of the 470 and 47uF TC caps) plus a 5K trimpot accelerator circuit did solve the slow startup issue, shaving some 3 hours off the original 3 1/2 hour lock up time.

 I also finally got round to matching up a set of 5 1K resistors to create a selection of 5 voltmeter offsets in 1v steps from zero up to 4 volts to let me use the mV range on a cheap (13 quid) Mestek 9999 counts DMM to monitor the VFC in tenths of a mV using the OCXO's 5.127v Vref to provide an extremely stable voltage source (far better than the TL431 I used to create a 3v meter offset in the MK I).

 The motivation to finally get off my backside and install this reference source came about as a result of my using a part used but well rested AA alkaline cell as an initial temporary measure to monitor the MK II's VFC in 100uV increments after the apparent VFC started bucking the trend of a rising voltage with age exhibited by the 13MHz OCXO used in the MK I which proved to be the AA cell starting to increase its voltage output about 1 mV per day

 I hadn't taken note of the AA cell's initial voltage so wasn't able to work this out until after directly monitoring the 2.282v VFC for a couple of days to confirm it wasn't actually dropping at all (at least at nowhere near the rate implied by my AA cell voltage offset lash-up).

 Lithium coin cells are a much better bet on voltage stability for such a stunt provided the 3.16 or so volts gets you close enough to allow a cheap 3 1/2 digit DMM to use its mV range and show readings to tenths of a millivolt. Here, absolute accuracy whilst nice to have, is not essential - it's the weeks long stability of a CR2032's open circuit voltage that counts (worst case drain with a 10M DMM input is just 100nA, in my case it had been just 15nA or so with the MK I and a CR2032).

[Labrat101]

Hi JBG
that looks good .
I managed to pick up 3 .HP 5385 frequency counters cheap . All had the same faults
On the power supply bad connections.
I did some tests on my 6800 . And they all say to the 11th digit . I am now at 10.000000001Mhz.
 
And using  my oven 20mhz ocxo divided by 2 to give me my 10mhz reference signal .
Is well good to any standard.
..
About that photo of the 6900 unpopulated pads I think this is so it can be upgraded to us 2 ths 3491 in parallel as reccomend by Texas instruments.
Don't us the 3495 as they have the heat pad on the bottom.
From what I could see from the photo .
You will have to trace out the tracks on paper
As it dual 3495 will need a couple of smd resistors for load matching..
..
It not that much different from the fy6800
Putting two 3491s in parallel will allow for
Better load output .
This may cause a relay problem as it will amplifier any errors .
The relays are contact type and not Reed relays that don't have the same noise problems.
All the good equipment use Reed relays .
On high frequency.
That's why put the 10pf on the relays it not as good as a reed but it make a pore man's alternative. 
..
Try to do a trace out of those spare pads.
Most of it should be pin to pin  ..
Have fun
RNS.

[Bad_Driver]

HI John,

good to hear from you! I done some tests with your advice, attached you find the screenshots. Up to 20 Vpp (what is 7.6 V measured)
sine stays sine as you can see. I recognize a little deformation at 24 Vpp but related to your picture the both OPAs in parallel doing a better job.
I also run some FFTs do get an idea of the THD. You find this with the steps from 24Vpp-20Vpp-15Vpp-5Vpp-0.5Vpp (to get the magic clicks of
the relays @ 5V and 0.5V).

I see no reason to change the OPAs now (may be later). I'm a bit careful with this. I started my "OPA-career" in the late 70s with the east german clone of the uA709 (named A109) and you may remember that this circuit was really sensitive to compensation. I learned during this time a lot about gain, bandwith (and the product of  both) but also about "swinging" of badly compensated OPAs.

I'm always shaking my head  when I read in audiophile forums from members, which
explain how good the change of an standard audio OPA against a fancy high performance RF model works and what they now
on new details from their sources can hear....  And this without any lab experience, without Osci-control. What they now hear are
the arte faktes of high frequency swinging.... but it makes no sense to discuss with this kind of people which spend 500 € for SPDIF optical cables.

 btw: I follow your advice to rethink the idea of using a laptop PSU and the little 2x15V converter. I have done some measurement and the
switching noise is hugh! So I went thru the whole FY6600 forum again and found in the beginning the first PSU-solution of Arthur Dent with
an industrial grade PSU (ARTESYN NFS40-7610J). It's expensive (around 50€) but I got a new one today from a UK based reseller for 15€.
This is a little OTT but with this I get enough power for my planned OCXO mod. I also read a lot (your advice as well) about SMPSU-filtering and ran
some tests with ferrite beads, rings and capacitors. I keep you informed!

[Johnny B Good]

 The first two thumbnails don't match the images that pop up when they're clicked to view them (the same result when downloading them).

JBG

[Bad_Driver]

Fixed! Something went wrong. 

[Labrat101]

Fixed! Something went wrong. 

With the ops I would leave alone as well. to get 24vpp
 as the SMP is 12v plus/minus  (24vpp) the ops require about 2v head room .
So if your getting 24v this maybe why the sine is funny above 20v .
The max safe voltage supply of these output ops is 30v on a dual rail.

Noise off the SMPS 0.01pf Ceramic  & a  2 Large 470uf with a small choke between them and
and a ferrite bead works wonders. ( caps can be Larger upto about 2200uf )
  If you want all out quite go for 10uf Tautanim Caps that's what i used. A liitle more cost but well worth it . I had a whole box of them so I used what I had.

Always Remember if it aint Broke don't fix it 
It also depends on what you will be using it for .
  Myself  I am a very precision Nerdy pensioner and if not perfect it bugs me     

[Bad_Driver]

The 6900 runs with the „old“ PSU at 2x 13,5 Volts. So 23 Vpp can be possible.

I have done today the mechanical work to bring the new PSU into the box
and startet with the filter connection. (choke, tantalium and ceramic Cs, I also have a great choice by the hands)
As Arthur Dent suggested I have 2 diodes in serial with the +/- 15 volts rails to stay a little bit more away from the OPAs
limits.

The best thing is that the SG now is much heavier and doesn‘t much longer has this „toy weight“.

[Labrat101]

I am now put my FY6800 to the full
Test .
The bottom scope is the oscillator sync
The 10Mhz out of fy6800 from the ocxo & the TCxo from Hp 5385 .
Top scope is 10mhz sine sinked to the 2 difference references .

BTW   ''Bad_Driver'' Try using your probes set @ x10  or your bandwidth will be out.
          also it will reduce the in circuit loading .

 My cables are M/Grade and are 50ohm inductance .So you won't see the termination.
 in the photos . but its there..   The square ish is from a buffered JK Divider 0.01PPb
 10Mhz Reference signal.  un terminated .
 Update:: I forgot to add a resistor to my divider output .    It only affected the scope wave not reading   

[Bad_Driver]

To make the measurements comparable to the results of JBG I used H.Q. coax with 50 ohms termination.
Therefore Vpp is "only" around 8 volts. I tried internal and external termination, results were the same.
Since my FG is under reconstruction I cannot repeat the measurments without load.

After some strange experience I learned that the weak BNC connector of the FY FG and cheap coax cables doesn't fit together.
So I upgraded my cables and connectors withe much better one and couln'd find my old crimp tool what to a new one lead.
I'm sure the old one will show up the next days....

[Labrat101]

Yup the connectors are not over good
I found by closing the small center pin on the BNC solved 99% of looseness.  Also if you look at the grounding on the connecting plug the are 4 very thin cuts an can be opened carefully to make the plugs fit snug .
The problem is that a lot of these cheap BNC
Don't seem to be standard sizes  .. check with a micrometer .
There are different only a few thousand inch.
But enough for a noisy connection.
 
RNS.

[Bad_Driver]

On Saturday I was sitting with my 2nd cup of coffee after breakfast and once again I was scrolling through offerings at EBay for my
FY "improvement" project. Last week I finished my PSU-upgrade and as expected the performance with high Vpp is a little bit better but
that was expected since I increased the lines up to 14.5V +/-. Noise floor is lowered by some dB but that can also be result of my new
PSU filtering stage.

With some items in my basket @ EBay I rethought the situation and all the little weaknesses of the Feeltec device:
- non linear amplitude-frequency response behind 1 MHz, I see a decline of 2.3 dB starting with 1 Mhz up to 60 MHz
- the non existing trigger function with sweeps
- the strange behavior of logarithmic sweeps - I expect a sweep of one decade per time interval as usual for Bode plots
- the strange attenuator stage behind the OPs (thanks to JBG to point this out)
- the  non existing firmware upgrade policy of Feeltec
- the cheap 10 MHz XO of the FY
- ...

I paid last year 89 € for the FY6900  and I learned a lot during the last month about PSU filtering (thanks to JBG), about the FFT functionality of my new SDG 2104X+ (improved to 2154  ) and about my needs on a FG.

I came to the conclusion that I would be able to improve the weak XO with a much better one OCXO (that was the only reason for my PSU change - to have the additional power available) but all the other weaknesses will stay a pain in my a... and will not be fixable by myself.
If the FY FG would be my eternal improvement project as hobbyist I would go further but it isn't.

The FG is as it is, not bad for the money I paid for and I will keep it. But I will stop my efforts to improve it. Maybe I rebuild the PSU to the old one
since I avoid any mechanical changes in the device (I used a selfmade alloy frame for the new industrial grade PSU that fits perfect to the
existing holes). Attached a photo of the FY with new PSU.

On saturday I ordered a Siglent SDG2042X @ Welectron here in Germany (also with the ability for SW-based-improvements - see the EEVBLOG).

[Labrat101]

Bad_Driver    Looks Good . Those caps look interesting If it works its good 
 
There is a trigger bnc input on the back . It did not work well on my FY6800 and I don't use it .
So that is now my 10Mhz reference out out to my HP 5385 frequency counter.
Which I now have 3 . Got a job lot of 3 non working unit . I was hoping to get at least one good one .
Fixed all 3     . Got two spares . One I will sell .
They are nice as they have reference in/out  & 11 digit display and being old school room for addons 
& good from 10Hz to 1ghz

All the best have   
RNS
 

[Bad_Driver]

Hi Labrat,

what price do you expect for your spare counter?
btw: the caps are NOS east german tantalium Cs. I have a bunch of them bye the hand and they seem still ok.

[Labrat101]

Hi Labrat,

what price do you expect for your spare counter?
btw: the caps are NOS east german tantalium Cs. I have a bunch of them bye the hand and they seem still ok.

I will be putting it on ebay for around $250 as I had to do quite a lot of work on it .
 It still need a little work as the previous owner miss used it .
 I have tested it and it is all working as per the service manual  and it reads good to my 10Mhz OXCO which is 0.02 PPB which I reckon is good enough for most of us .
  I have no Idea what the shipping cost would be like and How long it would take .

It the one in the post a few up ..#336 on: June 25, 2020, 04:59:03 pm
 I used it to do the tests after calibration 

[Bad_Driver]

There was a bunch of HP5385 on EBay from Israel, each in working condition für around 100€ during the last weeks.
I hesitated because of the high shipping costs and the unclear Tax/import situation into EU.

I wish you good luck @ EBay. As we all know HP stuff sells for good money these days.
Let me know what money you get.

[Labrat101]

There was a bunch of HP5385 on EBay from Israel, each in working condition für around 100€ during the last weeks.
I hesitated because of the high shipping costs and the unclear Tax/import situation into EU.

I wish you good luck @ EBay. As we all know HP stuff sells for good money these days.
Let me know what money you get.

Thanks I live in Israel there was a lot of X Army stuff the seller is good I get a lot of stuff from him most of it is just loose wires broken or missing knobs  . Its all Military grade so it has some better components in them . there are a few very good suppliers  (tsvi.parts)    Shipping sucks every were .

[Johnny B Good]

Hi Labrat,

what price do you expect for your spare counter?
btw: the caps are NOS east german tantalium Cs. I have a bunch of them bye the hand and they seem still ok.

I will be putting it on ebay for around $250 as I had to do quite a lot of work on it .
 It still need a little work as the previous owner miss used it .
 I have tested it and it is all working as per the service manual  and it reads good to my 10Mhz OXCO which is 0.02 PPB which I reckon is good enough for most of us .
  I have no Idea what the shipping cost would be like and How long it would take .

It the one in the post a few up ..#336 on: June 25, 2020, 04:59:03 pm
 I used it to do the tests after calibration 

 Hi Labrat101,

 I was just reviewing that post a couple of hours ago and puzzling over that photo you'd attached (I'd been a little too preoccupied the first time I'd seen it) so I downloaded it in order to rotate it 90 deg CW in GIMP to save craning my neck trying to work out what was what   I've saved the result which I'll attach to this reply to help make that setup you mentioned (which explains that opened up frequency counter  ) a little clearer to anyone interested in the counter.

 I was rather surprised to see that chimera of a DSO with a CRT display HP scope, along with a 100MHz 200MSa/s logo, no doubt the product of a marketing droid's fevered imagination (70MHz 200MSa/s perhaps with a 200MSa/s ADC for each channel - 100MHz 200MSa/s no way Jose! Not even with one ADC per channel  ).

 That's quite possibly the only example of HP making the mistake of allowing a marketing droid have the final say in the styling of their product line. I doubt that makes it a rarity numbers-wise since there's probably a shed load of these things circulating around in the second hand marketing channel.

 This reminds me of ntcnico's tag line "There are small lies, big lies and then there is what is on the screen of your oscilloscope" only in this case, it's not the screen so much as the front panel markings. If you regard it as a 70MHz BW 'scope (assuming one ADC per channel), it should serve its purpose very nicely (it is an HP after all  ).

JBG

[Bad_Driver]

I got yesterday as mentioned before my new Siglent SDG 2042 which improved itself with a little service
of 5 minutes and the help of an USB-stick into a SDG 2122.
A small step for mankind but a giant step for me! What a great device in relation to the FY6900! 
I will run some tests later with both but this weekend I start family holidays for 2 week and go to the coast.

[Labrat101]

 Hi Labrat101,

 I was just reviewing that post a couple of hours ago and puzzling over that photo you'd attached (I'd been a little too preoccupied the first time I'd seen it) so I downloaded it in order to rotate it 90 deg CW in GIMP to save craning my neck trying to work out what was what   I've saved the result which I'll attach to this reply to help make that setup you mentioned (which explains that opened up frequency counter  ) a little clearer to anyone interested in the counter.

 I was rather surprised to see that chimera of a DSO with a CRT display HP scope, along with a 100MHz 200MSa/s logo, no doubt the product of a marketing droid's fevered imagination (70MHz 200MSa/s perhaps with a 200MSa/s ADC for each channel - 100MHz 200MSa/s no way Jose! Not even with one ADC per channel  ).

 That's quite possibly the only example of HP making the mistake of allowing a marketing droid have the final say in the styling of their product line. I doubt that makes it a rarity numbers-wise since there's probably a shed load of these things circulating around in the second hand marketing channel.

 This reminds me of ntcnico's tag line "There are small lies, big lies and then there is what is on the screen of your oscilloscope" only in this case, it's not the screen so much as the front panel markings. If you regard it as a 70MHz BW 'scope (assuming one ADC per channel), it should serve its purpose very nicely (it is an HP after all  ).

JBG

       Some things are kept out of the eyes . Because their brain fills in the Blanks .. 99% wrong .

Hi JBG ..
This scope is marked up as 100Mhz @ 200mSa/s
Its also an X military and has been modded not by me  . It  will run and see signal of over 5Ghz .
It has some weird panels inside that are not in the manual .
So sorry to disappoint you the label is just a label . The front panel was replaced from another scope.

The other PK scope was modded it is 150Mhz 1gs with a firmware from another section here.
And is just a cheap digi scope with bugs .
.

I was rather surprised to see that chimera

The funny square wave was my error as I wrote above. I forgot to put the loading resistor on the 10Mhz
Reference Divider circuit. The cranking of your neck must have knocked your class off 
..
Why all the sudden are you picking on my scope labels  its is Not 70 mhz per channel .
 This scope is really good maybe old but it has some very nice addon panels in side all with Mil Marks.
& no your not going to have a Peek .

PS   Did I knock your Scope 

[timkoers]

Would somebody be kind enough to provide a list of fixes that you could easily do to the FY6900?

I started at the first page of this topic, but I quickly lost it due to various discussions.

The main thing that I am interested in is not to have AC (by any mains, pun intented) on my BNC's. Could an 1/2W 10k resistor be used in series between the mains connector's earthing cable and the power supply?

[Johnny B Good]

Would somebody be kind enough to provide a list of fixes that you could easily do to the FY6900?

I started at the first page of this topic, but I quickly lost it due to various discussions.

The main thing that I am interested in is not to have AC (by any mains, pun intented) on my BNC's. Could an 1/2W 10k resistor be used in series between the mains connector's earthing cable and the power supply?

 That's almost exactly the same solution I used with my FY6600 (three one quarter watt 33K resistors in parallel to make up an 11K "Static Drain" resistance to ground). In the case of the FY6600, this modification also required an upgrade from the C8 mains socket to a three pin one (C6 as I chose or C14 as almost everyone else did - probably picked out from the "Scrapped desktop PC parts Box" ).

 The psu boards remained the same class II type, with no safety earth requirement. All Feeltech had done was to upgrade the mains input sockets on the 6800 and 6900 models and "angrily bodge" an earth connection between the socket's PE tag and the nearest convenient point of contact with the zero volt rail with the absolute minimum amount of reworking of the wiring, hence that reprehensible hijacking of one of the ground returns in that 6 way PSU to Mainboard ribbon cable connector that we see in the 6600's immediate successor, the 6800 and, unbelievably, in the 6800's successor too, this nice shiny new 6900 model when one might well have expected them to review this bodge, hang their heads in shame and work out a proper solution.

 Apparently, they have absolutely no shame whatsoever which should come as no surprise to anyone who has ever had to remedy the glaring safety issues often seen with other cheap Chinese made electrical kit (eg, the metal cased version of the otherwise excellent KSGER T12 Soldering Station control unit for one  ).

 Getting rid of this unwanted earth wiring loop without reintroducing the ESD hazard of the 6600 model is as simple as (preferably) undoing the damage done to the psu to main board ribbon cable and wiring a 3.3 to 10K half watt resistor between the PE tag on the C14 socket and any convenient ground point connection to the main board.

 This is a modification well worth doing to the 6800 and 6900 models since, not only does it keep the half live touch voltage down to less than half a volt ac, it also eliminates the random DC offsets introduced by the low impedance connection to the noisy mains earth wiring which had fooled some users into thinking their brand new 6800 and 6900 AWGs had not had their DC offsets properly zeroed out at the factory, leading them on a wild goose chase in their misguided efforts to trim it out on the preset pots.

 Long story short: Yeah, just put a 4K7 resistor in series with the wire going to the C14's earth tag.

JBG