Re: FeelElec New Arrival FY-6900 Signal Generator

[Dezmond2]

I just picked up a FY6900 60Mhz. V1.3.1

Does anyone know if it's possible to calibrate the timer on the sweep function?
I'm seeing a timing issue on my box, for example if I try a 2 second sweep I see a 1.8 second sweep on my keysight DSOX1102A scope.
The timing issue seems to multiple as I do longer sweeps, so for example a 10 second sweep will output a 11.8 second sweep and so on.

All modes seem to have the issue

I have same issue with v1.4.1, board version 2.11.

[Quarlo Klobrigney]

Now the FY6900 has memory recall but no sine in the LCD. Just like evava's  photo. https://www.eevblog.com/forum/testgear/re-feelelec-new-arrival-fy-6900-signal-generator/msg3016778/#msg3016778:
I am not impressed. BTW what happened to Feeltech that used to be here in the forum? MIA?

[Quarlo Klobrigney]

So I took a picture of what my screen looks like.
As you can see everything should work but doesn't.
You can also see that the sine drawing is nothing but a flat line.
Yes I did a "repair sine" to no avail.

Anyone have any ideas?

[Labrat101]

This was discussed about a year ago.
There is a really good chance that your cyclone is nothing but a fake chip . If the printed label is faint.  And it's missing the Intel no.  It's not a cyclone.. The Name ALTERA should stand out strong and be readable without a magnifying glass

The Cyclone® IV FPGA family extends the Intel® Cyclone® FPGA series

[Quarlo Klobrigney]

My FY6900 with the bad sine is now fixed!
I had an intermittent power switch so I went in.
The switch was good but the 2 pin connector going to the SMPS was intermittent.
With the lights off, I could see arcing from one connector pin to the SMPS pin.
I couldn't do anything to fix the pin, only remove the connector and solder the wires directly.

While in there I did a look around at all the connectors.
I found the smaller of the two cables to the front panel to be half-on, diagonally.
I pushed it in to the socket and quickly tried the generator.
Voila! it had a sine wave again. 

While there, I also installed a fan in the case since it stays on the bench 24/7 in the main AC switch in on position.

I'd show a photo but you know what a normal display looks like.

[Bad_Driver]

My FY6800 failed the stress test after only 3 hours when the power supply sent out smoke signals  S O S .
(smoked out Smoke) . I asked for a replacement . They gave me $20 refund and said get it fixed   
 Why would you trust their Warranty     Its Void ed

Hope you are doing well during this strange times!

If you do a deep search @ Aliexpress you find the identical PSU for about 1€. It's really an expensive part!

After buying the SDG2042 (now 2122) I reworked the FY a little bit. I added a switch in the back to change the earth connection resistor between 5k and 100k. So I can use the FY as needed
as pseudo "floating" FG for LF work. This helps sometimes.

Since I got a bunch of 18650s I think about making the FY batterie driven. Has anyone done this before?

[Labrat101]

My FY6800 failed the stress test after only 3 hours when the power supply sent out smoke signals  S O S .
(smoked out Smoke) . I asked for a replacement . They gave me $20 refund and said get it fixed   
 Why would you trust their Warranty     Its Void ed

Hope you are doing well during this strange times!
If you do a deep search @ Aliexpress you find the identical PSU for about 1€. It's really an expensive part!

After buying the SDG2042 (now 2122) I reworked the FY a little bit. I added a switch in the back to change the earth connection resistor between 5k and 100k. So I can use the FY as needed
as pseudo "floating" FG for LF work. This helps sometimes.

Since I got a bunch of 18650s I think about making the FY batterie driven. Has anyone done this before?

Welcome back to the replay of the old party 🥳
.  Yes everything here is just  deja vu .
The FY on battery.  It's really power hungry  and it won't really give any better results.
And the guys putting fans . Should only cool the power supply.  And arrange a baffle to keep the air flow to a minimum over the main board.
As was discussed way back . The main board due to its pore design.  Should be alowed to reach a static temperature to reduce drift of low quality components.  .
.

[Johnny B Good]

 I'm afraid to have to disagree here

 In this instance, relying on passive ventilation in such a small a plastic enclosure with some 7 to 9 watts of waste energy to be dissipated with such small ventilation slots and so little height difference between the side intakes and the rear exhaust slots  creates an alarmingly high temperature environment, putting components under a heavy heat stress that's almost certain to result in early failure, particularly true in the case of electrolytic capacitors.

 If you're in the habit of tilting it up on its stand, this will cancel out what little thermosiphonic airflow it had to begin with, making the situation even worse. Plastic is not a particularly good conductor of heat meaning that a very high thermal gradient will be needed to dissipate the heat via the case panels, meaning very high temperatures will build up inside before the heat flow reaches a balance between heat energy input and heat energy output via the case panels alone.

 Avoiding the perceived detrimental effect of a cooling flow of air on 'cheap' components won't offer any more thermal stability since the internal temperatures will still track the ambient as before, the difference being that this will be with a very high temperature offset. Far better imo, to reduce this offset in temperature and have components following room temperature at just ten or fifteen degrees higher than have them track the room temperature some 30 to 50 degrees higher.

 And that is before the detrimental effects of thermally induced fatigue from extreme temperature cycling are taken into consideration. Adding a small fan cooler to actively shift the heat out of the case, even if it doesn't address the temperature stability issue, will at least minimise such temperature cycling effects on reliability as well a allow critical components to run much cooler to extend their service life times. And of course, the most temperature sensitive component inside any such signal generator, the XO frequency/timing reference, will suffer less warm up drift.

 Adding a small cooling fan to one of these cheap AWGs is the single most effective improvement you can make both in regard to their service life and performance... In my opinion, that is!

[Adrian_Arg.]

I put a cooler from an old 486DX CPU, and an aluminum heatsink on top of the FPGA. I prefer to prevent surprises.



yo coloque un cooler de un CPU 486DX antiguo, y un disipador de aluminio sobre el FPGA. Prefiero prevenir sorpresas.

[Johnny B Good]

 That'll help but not as much as you might expect. You'll still have the heat from the mainboard itself reducing the temperature difference between it and the underside of the FPGA to contend with (a major heatsinking source for a plastic packaged IC lacking the heat spreader of the later CPU designs or the more thermally conductive ceramic packaging of the older CPU designs such as the 80486 and early Pentiums).

 It's true that the recirculating air will help keep the opamp heatsink a little cooler and spread the hot air around into the cooler regions as well as increase convective heat transfer to the case walls but allowing a fan to force cool air in and the warmed up air out will offer the most effective cooling benefit all round - everything feels the benefit of this cooling airflow (and without sticking an extra heatsink onto the FPGA - the mainboard temperature will be lowered as well).

 It doesn't need a powerful and noisy fan to dramatically reduce the interior temperatures by some 20 to 30 degrees in the hotter regions. In my FY6600, I used a 40mm square 12v fan running off the 5 volt supply rail for this task, and very effective it was too (it only drew 28mA from the 5v rail - exactly the same as just one of the 5v relays fitted to the mainboard).

 The very obvious hotspots on top of the case over the psu board and the opamp heatsink (as well as the underside by the opamps) all completely disappeared as described here (please forgive my pomposity, I was a lot more "full of myself" back then  ):

https://www.eevblog.com/forum/testgear/feeltech-fy6600-60mhz-2-ch-vco-function-arbitrary-waveform-signal-generator/msg2001344/#msg2001344

[Bad_Driver]

My FY6800 failed the stress test after only 3 hours when the power supply sent out smoke signals  S O S .
(smoked out Smoke) . I asked for a replacement . They gave me $20 refund and said get it fixed   
 Why would you trust their Warranty     Its Void ed

Hope you are doing well during this strange times!
If you do a deep search @ Aliexpress you find the identical PSU for about 1€. It's really an expensive part!

After buying the SDG2042 (now 2122) I reworked the FY a little bit. I added a switch in the back to change the earth connection resistor between 5k and 100k. So I can use the FY as needed
as pseudo "floating" FG for LF work. This helps sometimes.

Since I got a bunch of 18650s I think about making the FY batterie driven. Has anyone done this before?

Welcome back to the replay of the old party 🥳
.  Yes everything here is just  deja vu .
The FY on battery.  It's really power hungry  and it won't really give any better results.
And the guys putting fans . Should only cool the power supply.  And arrange a baffle to keep the air flow to a minimum over the main board.
As was discussed way back . The main board due to its pore design.  Should be alowed to reach a static temperature to reduce drift of low quality components.  .
.

It's just for fun. Nowadays I play a little bit with Low Noise Amplifiers and built one batterie powered. So I'm looking for another
useful or useless project to learn a little bit more about charging/protecting/decharging of 18650.
When it works - ok. When not - again something learned 

[Labrat101]

It's just for fun. Nowadays I play a little bit with Low Noise Amplifiers and built one batterie powered. So I'm looking for another
useful or useless project to learn a little bit more about charging/protecting/decharging of 18650.
When it works - ok. When not - again something learned 

The 18650 batteries are pretty reliable and as long as they are charged to 4.2v  CC then CV the last 10% they are good
 I treated my self to an Agilent 6612C I got from a local surplus deal . Bankrupt stuff .
 It takes all the messing out of charging batteries 
Nice to here you are still enjoying life . Yes the last 2 years have been complicated . But I now have help @ hand for my
wife so I am nearly a happy camper 
 
 The  ""Quote from: Labrat101 on August 31, 2021, 07:58:22 pm""   Was actually a a re quote form mid 2020 . that's why
I said  Deja vu      As they say . Down the old Rabbit hole Again only to find another   older dead rabbit .

@ Johnny B Good .
 As you may have forgot my FY I totally rebuilt into an Alpha Beta . And did not use a SMPS but all Linear . I only cooled
the Power section . and the main board is kept at a constant temp by laminar flow . and the OCXO is sealed in it own
temperature controlled casing . The Piggyback board for the Alpha Beta driver sits on top of the main board .
 So far over the last 9 months the drift has not varied and is staying within 3^-8 as per my Leo Bodnar
 OK this is now a cost maybe close to $400+ but its doing its Job and has been a great help.

 But as for the average person playing with these FY . Horizontal air flow is far better . Use something like an old laptop fan
 mounted on the back top grill blowing outwards and open the bottom side slots on ether side then the angle will not effect the cooling when raised up on the stand . Also fan noise is less . as the inside of the case will always be at a lower pressure .

Wish every one well and good health 

[Johnny B Good]

@Bad_Driver

 I've developed a sudden interest in LiPo powered mini DC UPSes over the past fortnight and searching EEVBlog topics on the subject the latest one of which is this from as recent as last August which you might find useful:

https://www.eevblog.com/forum/projects/suggested-boost-converter-chip-for-3-v-to-5v-500ma-or-so/msg3625624/#msg3625624

 I've been investigating a way to provide built in battery backup for my GPSDO to remove the irritation of the OCXO's mcu oven temperature controller resetting with even just milliseconds of interruption to its supply which yoyos the oven temperature over the next 10 to 15 minutes, leaving the PLL trying to keep up with the resulting frequency/phase excursions.

 It looks like the control algorithm literally assumes a cold start condition when doing such a cold start from reset, applying maximum heating current to heat the crystal up to temperature as fast as possible with the least start up delay rather than testing the actual temperature beforehand.

 This is all fine and dandy when it's installed in a Symmetricom rack mounted GPSDO time and frequency reference generator, powered from a protected supply but a damned bloody nuisance when it comes to swapping DC power sources to a homebrewed GPSDO.

 I discovered this sensitivity to even the briefest of interruptions when trying out a Home Bargains 10,000mAH powerbank that could act as a 5 volt in / 5 volt out mini-UPS due to it being able to continue charging/powering devices plugged into its USB A sockets whilst simultaneously being charged from a 5v wallwart supply via its micro USB charging port.

 Unbeknown to Big Clive who had recently torn down and reviewed a 4000mAH version sold by Poundland, the only fly in the ointment was a 10 to 20ms interruption during the transition from wallwart power to internal LiPo cell power. That little factoid was a surprise he'd left for his audience to discover all by themselves.

 I was already aware of the "reset from cold" issue with the ocxo controller when trying to swap power sources out as fast as possible (a 2 to 5 seconds job at best). I hadn't realised just how short such an interruption it takes to trigger the same response until I started testing with these powerbanks.

In this case, I had to use a 5 to 12 volt boost converter module between the power bank and the GPSDO since it needs a minimum of 6v (maximum of 24 volts) to power it. I had an Atorch USB power monitor plugged into the power bank's 2.1A rated socket to monitor the GPSDO's power demand which on a stone cold start, maxes out just below the 5W mark.

 The power monitor blanked out and restarted every time I disconnected the power bank from the wallwart (thankfully, without clearing its logging data) giving me a clear indication as to why the ocxo (and the M8T) went bonkers every time I dropped wallwart power (transitioning back to wallwart power was no problem). I saw the same behaviour with both the Poundland and Home Bargains power banks.

 As it was obviously a very short lived interruption, I soldered a 4700uF 16v cap directly across the boost converter's output in an attempt to smooth out this interruption. Unfortunately, this bogged the converter down, causing it to generate a very unsteady 2.8v which eventually corrupted the M8T's bbram, leaving me thinking I may have bricked it (I hadn't - when I tested it the next day, after swapping it out for the original one I'd used in the MK II it was back to normal).

 Realising what the issue had been with my quick 'n' dirty capacitor fix, I put a 100 ohm resistor in series with the cap with a reverse biased SB diode across the resistor and tested again with complete success (even when I reduced the boost converter output voltage down to just 6.5 volts to get some idea of how much margin I had against these transitory interruptions).

 That led me to adding the same resistor diode cap modification the ocxo's 12v boost converter output which did mitigate the wild oven temperature excursions but did nothing to stop the M8T from restarting, taking some 3 seconds to lock back onto the satellites. I fixed this issue with a similar circuit but I was still seeing a minor disturbance from its effect on the PLL's LPF.

 In the end, I landed up using a pair of 4700uf 6.3v caps to protect the 5.3v supply which powered everything else (5 to 12v boost converter included) which did allow me to use the 5 to 12 volt converter between battery bank and GPSDO DC jack without the capacitor/resistor/diode bodge attached.

 Only then, did it dawn on me that this GPSDO modification was only good for just one thing, overcoming this very specific drop out issue peculiar to these power banks, causing me to have another think about the 'real problem' of its rather twitchy ocxo, hence my new found interest in how to 'float charge' an internally installed LiPo cell.

 In my case I'm looking to use a 1500mAH protected pouch cell I'd recovered from a Mustool G600 microscope that had arrived with a broken screen for which I received a refund and told to dispose of it myself rather than trouble the seller with return shipping charges (ditto for the 1st one from Banggood). It had taken a third attempt to get an intact one from another ebay seller.

 Anyway the attraction of these protected pouch cells is that they can easily be slipped into place inside the GPSDO between the roof of the case and the top of the sponge rubber pressure pad that holds the ocxo's sponge rubber jacket in place, effectively taking up no further room within the enclosure.

 That just leaves me with the task of rearranging the existing buck and boost converters so the 12v boost converter can be powered directly from the LiPo cell with the 5v buck powered from the boost converter, turning the current arrangement on its head.

 All that's left to do is use another buck converter to keep the LiPo cell charged to 3.85v (a 55% SoC according to this chart https://hobbygraderc.com/lipo-voltage-chart/) from a 5 to 24 volt DC supply and add a cell protection circuit to disconnect the LiPo at an endpoint discharge voltage of 3.2v (supposedly just below the 0% SoC point - the built in cell protection doesn't disconnect until the cell voltage has dropped to 2.6v, way too low a setting made worse by the cut in voltage of 2.9v which the cell immediately bounces back up to).

 I plan on not reconnecting the cell until the return of supply takes the voltage over the 3.7v mark to avoid the built in protection killing the cell by "A thousand cuts". I'm still working out the finer details of this modification.

 In this case, I don't need to charge the cell to full capacity which allows me to keep the charging circuit relatively simple. I expect to get an hour's worth of run time but even just a couple of minutes will serve my need to avoid the ensuing ten or fifteen minutes of disruption when swapping DC power sources. One additional benefit from adding a third buck converter into the 'power train' is that it will extend the input voltage range down to allow me to use 5 volt power sources such as USB wallwart chargers and power banks.

[Bad_Driver]

Oh John,

you are a worth child of Shakespeares nation! Always demanding for a non native speaking guy to follow your poetry!

And you still on the search for the lost picosecond 

I finalized some month again my own GPSDO based on a SAMSUNG UCCM board
https://www.eevblog.com/forum/projects/a-look-at-my-symmetricom-gpsdo-(ocxo-furuno-receiver)/msg3548229/#msg3548229

Before I had played with a SYMMETRICOM board but the Furuno receiver is very unsensitive.

The Samsung UCCM runs without any hick ups since than and I feed all my devices with it. You can see that I use a old video distribution amplifier for it.
I got a bunch of 5 (!!!) for 20 € @ EvilBay and this is the best of it with BW of 120 MHz.

Since we suffer from exploding energy costs in Germany this is the end of my GPSDO efforts. The device takes 12 Watts and this are in Germany about 37 € p.a. nowadays!
That why I stopped all Rb-device searching. All my other devices on the bench are now hard switched to the power supply.

Back to my little FY plastic box. My idea was to use 6x 18650, always 2 parallel and than 3 in series with one of this chinesium BMS devices for charging and protection.
So I get 12.6 volts. The newer FY devices use only a 5 volts PSU which you can get from a step-down-converter but any idea how to get the +/- 13 volts for the board?

Stay healthy and enjoy the announced snow (for the Berlin area) this weekend 

B.D.

[Johnny B Good]

@Bad_Driver

 No, I'm still on the search for the lost nanosecond.

 My own little diy gpsdo takes just 1.4W (DC jack input - the wallwart consumes 2.1W from the mains after the ocxo has warmed up). My rubidium frequency standard project only draws 11 watt (LPRO 101  with an Arduino nano controlled fan cooler fed with 19v from a laptop charging brick).

 One of the attractions for going the DIY route over using older ready made kit is that you get to have some say in how much energy it'll consume. In this case, my main concern wasn't about the energy costs so much as thermal consideration of the component temperatures inside of the rather diminutive and economically priced 100 by 100 by 50mm extruded aluminium enclosure I'd chosen to build it into.

 When you're already burning some 50 watts (NAS box and its UPS) and another 100 watts on a desktop PC running 24/7 on top of the additional 32W consumed by an ancient APC SmartUPS2000 supplying protected mains outlets around the house, another vampire load or three four (2W gpsdo, 11W Rubidium frequency reference, 13W SDM3065X and another 22W on an SDS1202X-E, tends to be 'lost in the noise' (not so true for the 54W consumed by the SDS21504X+ though).

 You touched lucky with those video distribution amps. Every time I've scoured ebay, what few examples there were to be had were all grossly overpriced. I could well land up repurposing the THS3002i I removed from my FY6600 (to get back on topic) to lash up my own 4 channel 10MHz distribution amp if this absence of cheap video distribution amps persists.

 Assuming you're going to be using decent capacity (2AH) 18650s that's the same as a laptop battery pack's 43WH's worth of energy storage which could give you a run time of around 7 or 8 hours. Fitting half a dozen 18650s into the case seems a bit of a squeeze to me though. A set of three 4AH pouch cells might be a better option (less weight and volume).

 As for the question regarding a +/-13v supply, I've looked into this in the past and couldn't find a satisfactory solution (they weren't cheap or suffered too much ripple noise or both). It's worth keeping in mind that as a general rule, buck converters offer the best efficiency and ripple noise performance, followed by non-inverting boost converters and then inverting buck converters with inverting boost converters taking last place.

 IME, trawling through ebay's offerings for cheap inverting buck or boost converters is rather like going on a Unicorn hunt There's a good reason why those who go the analogue mains transformer / rectifier / smoothing cap route to eliminate smpsu switching noise and ripple issues, tend to go for separate secondaries so they can use positive polarity regulators all round whether they be buck converters or those mini-space heaters labelled 78xx or 1117-xx.

 Obviously, you don't want to be using a split battery supply to use the same solution so somewhere between battery and your bi-polar voltage rails, you're going to have to use an inverting converter (buck or boost). Dual rail converters do exist but you're probably going to have to submit a RFQ just to find out the price since I've never seen any on ebay (at least not at a sensible price, if at all).

 We have a saying you're probably already aware of that goes "If you have to ask (the price), you probably can't afford it." I've been mindful of this whenever I've come up against such barriers to getting a price on anything so special that it's seemingly only available brand new direct from the manufacturer or their official distributor so never bothered to submit a RFQ, preferring to carry on looking elsewhere until I either get lucky or start losing the will to live and give the whole idea a well earned rest.

 You can try searching on ebay, Amazon, Banggood and AliExpress if you don't want to pay full price with official distributors like Element 14. You seem to have had better luck than me in snagging bargains (off ebay at least), so you may be able to track down those elusive (to me) Unicorn parts. Keep in mind that with a 3S2P Lithium cell battery you need to assume a discharge endpoint voltage of 10v in your calculations when choosing DC-DC converter modules.

 The latest versions of these FY6900 models have incorporated the +/- 13v converter modules into the main board as you noted, considerably simplifying any diy psu and battery power upgrade projects to just a single 2A 5v psu board or a 2 to 5s Lipo battery pack with one of those mini360 (3A max/2.1A continuous) 5 volt out rated for 24v to 6.5v input converter modules sold by Banggood (and no doubt elsewhere) but that's no help to you (or me!).

 Regarding that weather forecast of snow you mentioned, I have a news flash!

The winds here have been been steadily increasing over the evening. A few hours earlier, I ventured to our attic bedroom floor to check on my gps antenna. It was swaying a little wildly but not alarmingly so. I've just taken another look about half an hour ago (2200GMT) and the picture is very different. The 20cms diameter biscuit tin lid I have mounted onto the top of a 3 metre aluminium pole clamped into the corner of a metal drawer ballasted with four expired car batteries sat on the flat bay window roof is now bending a good 45 degrees southward during the heavier wind gusts and, by looking out of the attic bedroom window onto the bay window roof, rather more alarmingly, I can see one corner of this drawer lifting an inch or so during these gusts, despite the weight of a 36AH car battery holding it down!

 Only this one corner so far but if the wind gusts get much stronger I can see either the pole torn from its ballasted drawer or else the whole lot tipping over and landing in our front garden by the early morning. That gale proof antenna mounting arrangement is not quite so proof against gale force winds as I'd hoped. 

 At the very least I'm going to have to rethink my antenna mounting options when the multiband GNSS timing antenna replacement https://www.aliexpress.com/item/32815205556.html (plus a 20 metre TNC to SMA RG58 patch cable) arrives, hopefully by the promised delivery date in a fortnight's time. I may have to wait until late January though for the one remaining component critical to this upgrade project, namely a 5/8 to 1/4 inch UNC adaptor, to turn up anyway so plenty of time to revamp my antenna mounting arrangement.

 In the meantime, the wind strength seems to have abated, confirmed by my local airport's METAR data and a local amateur weather station. I'd observed two or three briefest of mains drop outs (too brief for the UPS to notice anyway) earlier in the evening before venturing out across our road to get a better Idea of how the GPS antenna was responding to the strongest gusts (it didn't look good) when I observed yet another drop out.

 After returning to my workroom, there were another two dropouts, this time just long enough for the Backups500 to take notice of and issue a warning beep. There haven't been any further dropouts during the past hour or so, all of which suggests we're now through the worst of the weather.

 However, I have noticed another three briefest of barely perceptible dropouts during the past ten minutes, so I guess some part of the high voltage transmission lines must still be taking a beating somewhere in the UK's national grid. Ah well, time enough tomorrow for me to examine the damage and take stock of the situation. There's little I care to do until the weather calms down anyway.

 If it's still there after the weather calms down, at least I can reach out of the window and haul the whole contraption back into the room to repair and strengthen it up to face the next gale force / storm strength winds we can expect to see this winter. At least with the current mag mount patch antenna which I'd also glued to the biscuit tin lid which I'll be taking a pair of tin snips to to trim it down to a lighter smaller and more aerodynamic form.

 As best as I can figure, the GNSS antenna I've ordered must weigh at least 3 to 4 times more. Its more aerodynamic shape will help mitigate the additional weight to some extent but nowhere near enough to carry on using the existing one inch diameter aluminium pole currently clamped to a, no doubt by now, rather over stressed corner of its metal drawer base. I've still got plenty of ballast for the next improved antenna support contraption and a benchmark to guide its design.

 Enjoy your promised snowy weather. 

[Bad_Driver]

Hi John,

no snow last weekend, not cold enough, some snow flakes in the pouring rain Saturday night but it's around 1 degrees Celsius, colder than in an average December.
I'm pretty sure that we will see some white days before Christmas.

I also bought an outdoor GPS antenna and I was lucky enough to get a SYMMETRICOM GPS distribution amplifier from the Bay but the project suffers from a weather proofed idea how to mount the antenna
on the roof (my lab is under the roof). Next spring time project....  So the little puck style antenna has to serve another winter.

I went through my shelves an boxes and found two dual rail converters, one TPS5430 based but only step down
and the following (see picture) and surprisingly this one is a dual rail boost converter. I have no idea where I bought it, it can come from Aliexpress.

I fed it with 12 volts and it gives 2x 13.5 volts with 0.5 amps with an efficiency of 75...80% depending of the input voltage.
Ripple about 40 mV. Now I'm awaiting the Battery Management Module from Aliexpress and than I give it a try.

Maybe this little plastic box will start a second life as portable and floating FG?? We'll see and I keep you informed.

B.D:

[bomarek]

Hello.
This is my first post on the forum related to the generator FY6900H-100M KKMoon, motherboard ver.2.1. The generator works properly, powered only from the USB cable. The problem is that I can't control it from my computer. I have installed the USB driver and software for the FY6900 model. I cannot update this software to version 6.5. The software detects the generator as FY6900H-100M. I run in administrator mode and get the error: Run-Time Error 380, Invalid Property Value. The strange thing is that the software from the FY8300S generator is partially working, I can define my own waveforms and send them to the generator. Someone knows the solution to my problem. Sorry for my English but it's translator's fault

[Bad_Driver]

First test with my batterie powered FY plastic box.
3x2 18650 Li-Ion batteries, 1 dual rail boost-converter for 2x 13.5 volts, 1 buck converter for 5 volts.
The FG needs about 450 mA as you can see. With my used 18650 it run 5+ hours (without any load),
one active channel takes about 20 mA (measured between batterie and converters).

Since I‘m awaiting the batterie management module (for cell protection) I stoped the test at 11 volts (3.7 volts per cell).
Hope to get the module till Xmas. Seems to be on a slow boat from China.

I removed my OCXO (to avoid another 0.4 A at 5 volts) and soldered an used 3.3 volts TCXO in the old place since the cheap original TCXO was killed during desoldering.

I keep you informed about any progress.

B.D.

[Johnny B Good]

@ Bad_Driver

 As test bench lash ups go, that's a very tidy layout. Mine tend to be considerably less tidy affairs (see attached image ).

 I've been tackling a similar problem with my gpsdo where I'm merely trying to avoid upsetting the oven controller in the ocxo I'm using, every time there's even the slightest interruption to the DC power feed (a 10ms break is all it takes!).

 The ocxo is quite obviously using a microcontroller to use a PID control algorithm over the oven temperature. Unfortunately (for me at least), the slightest interruption of power results in a reset which causes the control algorithm to assume a start from stone cold which invokes full heater current to get the process of bringing the oven temperature up with the minimum of delay before it actually starts monitoring the oven temperature.

 This is a sensible approach for an initial startup from a stone cold state but a damned nuisance when it's the result of a momentary interruption of the DC supply voltage as I discovered when testing power banks that could act as 5 volt UPSes by virtue of being able to continue powering or charging attached devices when they themselves were plugged into a USB wallwart (or even another power bank for that matter).

 Transferring to battery power during a loss of external power from a USB wallwart (or exhausted power bank) was no problem but resumption of external power generated a brief interruption, I estimated to be around ten milliseconds or so (ICBA to actually measure it with a 'scope).

 At that time, I had to use a 5 to 12v boost converter (the gpsdo is currently limited to the voltage supply range, 6.5 to 24 volts), so I chucked a 4,700uF across its output which bogged it down to an unsteady 2.8 volts resulting in an M8T with a corrupted BBRAM. My second attempt to overcome this 10ms glitch involved putting a SB charge blocking diode (with a 100 ohm bypass resistor in parallel) in series with the 4700uF cap which eliminated the bogging down issue and provided more than ample energy reserve to ride out the 10ms interruption.

 Since I was well pissed off at this time by the ocxo's response to any interruption of supply (swapping out DC sources or rerouting the jack plug lead), I decided the best option to solve all these issues, rather than just the one specific to the 10ms break generated by these "5 volt UPSes", was to add a 1500mAH LiPo cell with another buck converter to 'float' the cell at 3.85v (50% SoC and recommended storage voltage value) and power the 2.6 to 5.5 v input / 12v output boost converter I was already using to power the ocxo from the current 5.3v buck converter running off the 6.5 to 24 volt DC input jack.

Everything in the MK I had run off a 1.3A 5.2v buck converter, the MK II included a 5 to 12v boost converter for the ocxo powered from a 3A  5.3v buck converter. This LiPo cell add-on to the MK II will add yet a third dc-dc converter to the mix (another 3A buck converter configured for 3.85v output) and reverse the current arrangement of the existing dc-dc converter sequence. The 5.3v buck will be powered from the 12v boost which will now be powered off the 3.85v rail from the additional buck converter or the LiPo cell whenever the external DC supply is lost or withdrawn.

 The additional buck converter will increase the power input by about 100mW (from 1.4 to 1,5 watts) but the payback (apart from the obvious battery backup feature) will be the ability to now also power it directly from a power bank or USB wallwart.

 Since I'm only aiming to keep the ocxo from resetting itself during the shortest of breaks in supply (10ms of a UPS capable power bank or the half minute or less required to swap out power sources) rather than squeeze the maximum run time out of 1500 or 2500mAH LiPo cell (a 3 or 5 hour run time), a simple float charging arrangement, without the complexity of a CC/CV with charge termination set at 0.1C rate, to a safe 50% SoC voltage will more than suffice my requirements (5 or 10 minutes tops).

 I may have simplified the charging circuit in this case but that still leaves me with the problem of isolating the LiPo cell once it has discharged down to a cut off point of 3.2v, along with the means to delay its reconnection by 15 minutes after the DC input power has returned in order to save overloading the input buck converter during the ocxo's initial warm up phase, during which a simple SB diode and 2 ohm series resistor circuit will allow the LiPo cell to receive a 100mA charging current to reduce the voltage difference sufficiently by the time the cell is fully reconnected to the 3.85v float voltage supply to hopefully keep the resulting charge current surge within safe bounds.

 I did contemplate avoiding this complexity by relying on the built in cell protection circuit board (over current and short cct discharge and over charge and discharge voltage protection) but the over-discharge cut off and cut in voltage settings were way too low (2.6 and 2.9 volts) imho, which, coupled with the relatively high discharge rate of half an amp, meant the cell voltage would rebound above the 2.9v limit within milliseconds of being disconnected at the 2.6 volt limit, causing the cell to cycle its connection until it eventually became so completely exhausted as to finally not be able to rebound to the cut in voltage.

 Even if I had decided to rely on the built in protection, I'd have had to use some means to limit the initial charging current since testing with a bench supply had shown charging current values in excess of 3A (but this was, afaicr, when using a CV of 4.2v - at 3.85v it might be a different story altogether - time for more testing perhaps?).

 Actually, after rambling on about my particular backup battery problem, maybe I should do some more testing. Considering that I only really need a reserve of 10 minutes at most, there's plenty of scope for further capacity compromise which may allow me to rely on the built in cell protection for the more pragmatic solution of simply whacking it straight across the 3.85v float charging supply (or 3.6 to 3.7 volts - 5.5 to 15 percent of cell capacity). Allowing the cell to go so flat that the built in protection kicks in should only be a very rare once in a blue moon "Whoops! event anyway.

 I'll post a follow up after I've done some more testing. Whilst this short term battery backup is a little off-topic here, it might still be of interest to anyone thinking of adding a rechargeable battery pack to their FY6900 signal generator. Those who own the later 5v only models could simply use one of those power banks with offer simultaneous recharging whilst powering or recharging connected devices.

 The short 10ms or so interruption on transferring back to wallwart power, that I'd experienced with a couple of 4000mAH Poundland examples and one 10000mAH Home Bargains example, may or may not be a problem. Even if it does cause a reboot, that's only a 3 second delay and if you're in the habit of pressing and holding the OK button for 1 1/2 seconds (or wait for the confirmation beep) to save the current settings prior to switching it off, a restart will be even less of a problem other than as a consequence to any test run you may have in progress.

[Bad_Driver]

Hi John, hi you other FeelElec-frinds/users/haters 

sorry I´m to busy with business and family short before Xmas. No real news.
My main approach with my changes in this little plastic box is always to keep it in a state where I always can go back to the original. So
I try to use the available holes etc.
I removed now the PSU and the mains connector and cut a plate for all the battery stuff which uses the holes of the PSU.
Since I'm awaiting the Battery management module I can't finish it now.

John one idea I got in my mind while reading your post, have you considered a simple Pi-network (C-R-C or C-L-C) for the U-Blox supply to overcome the 10 ms hickup?

regards B.D.

[Johnny B Good]

@Bad_Driver

 I understand your philosophy over making any mods 'reversible'. I have a similar attitude but will modify in a non-reversible way when there's either no alternative  or there's no point in preserving an expired (or Chinese - same thing with Feeltech) warranty.

 As it happens, your post arrived just as I was boxing up my now LiPo cell protected GPSDO and running a final check on its power consumption which is dependant on the LiPo's state of charge. The LiPo cell was not quite at the chosen 3.85v SoC point (around the 60% mark, judging by the earlier 2 1/4 hour run time tests with what now seems to have been a 110% loading factor).

 I set my lab bench supply to deliver 5.0v to the dc jack socket earlier this afternoon with the current draw on the bench supply reading 0.34A initially before dropping to 0.32A a couple of hours later. It is now reading just 0.29A after a further 5 hours. The bench supply's ammeter reads about 10mA low so it looks like the LiPo cell is close to being 'fully charged' to the 3.85v level (taking just a few milliamps). Allowing for the small error in the amperage reading, I'm estimating an energy consumption at 5 volts of just 1.5W which is pretty close to what I had been expecting.

 I've drawn up a 'tidy' version of the new LiPo cell protected supply circuit and attached a picture of the diagram. After doing some more careful testing of the LiPo cell's built in protection, it turns out to be not quite as bad as I'd initially thought so I've ditched the complexity of an overdischarge protection add-on circuit and taken a more pragmatic approach.

 However, I did spend just over a week trying to come up with a 'simple' float charge to only 60% circuit. The main problem being the danger of overloading the first buck converter when cold starting with a depleted LiPo cell. Sticking a 0.33 ohm resistor in series with the cell looked promising at first but the cold start with a flat cell still looked a little too marginal for my comfort and using a 1 ohm resistor introduced an unacceptable volt drop, hence, as you will note, the use of an SK32 diode in parallel to limit the drop to a more acceptable 0.3v which surprisingly, doesn't impact the run time as much as I thought it might (I saw a 2 1/4 endurance with this test setup).

 I'm going to leave it running overnight to make sure the cell has become fully charged up to the 3.85v mark before running a full discharge run time test. From previous test runs, I'm expecting to exceed the previous best 137 minute test run and quite possibly top the 2 1/2 hour mark. This is way beyond my original modest 10 to 15 minutes requirement which could still easily be met with a more compact 500mAH cell.

 Regarding your suggestion to use a LPF to stop the 10ms transient passing through onto the DC supply, you'd still need a very hefty capacitor and a large value inductor. A C-R-C just isn't going to 'cut it' at the very low impedance we're dealing with in this case.

 My own solution of a diode in parallel with a current limiting resistor in series with a shunt capacitor was the most optimal way to deal with such infrequent 10ms transients from these power banks but even this only deals with this one issue unique to the use of such power banks and does nothing to solve the issue during seconds to minutes long breaks in supply due to the need to swap out power supplies without resetting the OCXO's oven temperature controller. Hence my designing a 'simple' self contained LiPo cell backup power add-on circuit to avoid the ten to fifteen minutes of disruption every time I need to swap out psus or reroute the dc jack plug lead.

 Despite my attempts to keep it simple by leaving the discharge protection in the hands of the built in cell protection and simply float charge it to the 60% SoC voltage point, there was still plenty of complexity to deal with, namely the issue of proving that I did need to prevent backfeeding into the first buck converter, followed by working out how to overcome the volt drop issue when a diode alone is used to prevent this. It took a few design attempts before I finally hit on the circuit as shown in the attached photograph below.

 It might seem an inordinate amount of time and trouble to have gone to but, aside from the benefit of built in battery backup, this has now extended the range of wallwart options to include USB chargers and of course, random power banks. Prior to this modification, the dc input voltage range had been limited to a maximum of 24 down to a rock bottom limit of 6 volts, The range now extends down to a minimum of 4.5vdc (7W peak for a stone cold start). Once warmed up, the power consumption remains a constant 1.5W (or even a touch less) over this entire voltage range.

 The only remaining finishing touch is to add an "On Battery Power" indicator. I'm thinking I can simply cause the existing power indicator LED to wink out briefly once every half second or so but I'll save that modification for later on since I still need to complete my Rubidium oscillator project (at least to the boxed up stage when I can fine tune the temperature control).

 With the recent cold weather, I did at least get the chance to test its temperature control down to just below the 16 deg low end I've been aiming for. It looks like, once it's all boxed up, it might be good down to 12 or 13 degrees, possibly even right down to 10 degrees ambient (although retaining control at 30 deg or a little above may become a little more problematical than I was bargaining for )

 I've also attached the photo of the MK II's circuit diagram to lend the LiPo backup circuit some context.

[Labrat101]

@ Johnny B Good
Silly question why are you charging a LiPo to only 3.85v ? They are fully charged @ 4.2 volt . When  charge and check LiPo's 4.2 volt is 100%
3.8 volt is there nominal working voltage . ( new or Good ) . I normally now only use my HP6612A for checking LiPo's . @ CV.  3.8v is about 90 % then the remaining 10% should be charged @ CC . 
Set points V = 4.2 v    I = Cx  (2500Ma/h  500ma ) 5 hrs apx charge,  trickle charge is 1/10^th 25 ma

[Johnny B Good]

@Labrat101

 Not really a 'silly question' seeing as how it proved so difficult to track down cell voltage versus state of charge for LiPo that weren't based on the very high discharge rates (10 to 30 C) used by the RC fraternity.  An example of which can be found here:  https://hobbygraderc.com/lipo-voltage-chart/

 Ideally, I wanted to track down charts based on 0.5 to 1C discharge rates (I'm using a (n estimated) 1500mAH cell) but such charts are few and far between (hint:- the image search option proved to be the most productive way to track down LiPo cell charge/discharge graphs). From what you've just said about a cell voltage of 3.8v representing a 90% SoC, that seems to be more in line with the results I got during testing.

 I did quite a bit of research into the subject of 'float charging' LiPo cells which informed me of the lithium plating out effect when they're floated at a sustained 4.2v "100% SoC" level and the reason for a CC until the cell reaches that 4.2v point, followed by holding at this CV until the charging rate drops to 0.1C before the charge cycle is terminated to prevent plating  the lithium out.

 Since for my purpose, I didn't need the full 100% capacity anyway, choosing a float charging voltage of 3.85v seemed a more than optimal choice between simplicity and long cell life (no plating out risk with a voltage close to the optimum long term storage value, neatly eliminating the additional complexity of a CC/CV with cut off when the charging rate drops down to 0.1C charging regime).

 That voltage versus SoC chart I linked to suggests I should see around 50 to 55 percent of capacity but, after allowing for a discharge rate starting at 3.55v (allowing for the 0.3v drop of the SK32) varying from390mA, increasing to 470mA at the 3.0v endpoint mark, it looks like I'm seeing an extra 20%, presumably on account I'm only discharging 0.3C or less rather than the more typical 10 to 30 C discharge rates used by the RC crowd to whom this table is aimed at.

 Right now, I'm just over 100 minutes into a LiPo cell autonomy test run which I'm expecting to exceed 140 minutes going by previous tests with a dummy load that seems to have been 10% higher than the actual load of the GPSDO itself.

 Unfortunately, with the GPSDO now boxed up, I can no longer monitor the effective cell voltage as I'd been able to do in the previous test runs so I'll just have to keep a close eye on it until it finally dies some time in the next 25 minutes or so. The next 20 minutes are going to be a little intense as I wait out the final outcome of this benchmark test.

[Johnny B Good]

 The autonomy test run managed a total time of 131.5 minutes. Not quite as long as I'd anticipated but way longer my rather modest minimal requirement of ten minutes.

 The indicator lamps (power and lock LEDs) have been blinking on every few seconds ever since as the cell protection cuts out at the 2.7v mark and back in at the 3.0v mark as the cell voltage rebounds between those cut out and cut in points. The intervals have been getting longer each time and this blinking action will eventually come to a complete halt once the cell has become fully depleted to the point when it can no longer rebound to the cut in level.

 These brief battery power restoration events have not resulted in any output to the 'scope which is now only showing the RO reference with the GPSDO trace flat lined. The blink rate has now dropped to once per 25 seconds after some 15 minutes of cell exhaustion. It'll probably take another hour day or two for this final 'Swan Song' to completely fade away.

 This might seem to be an irritating feature of the battery power termination event but since this doesn't seem to present a settings corruption risk, it's actually a handy indication of how recently it ran out of backup power. The blinking has now downgraded its intensity to the briefest of flashes every minute or so after yet another ten or fifteen minutes and I'm going to leave it be for another hour before reconnecting the bench supply (set at 5v) to check how well it copes with a stone cold start and depleted LiPo cell (worst case loading scenario for the primary buck converter - Buck 1 in the diagram).

 I'll post a report on this in the next hour or so. The cell should be as near to totally flat as makes no difference by then.

P.S. I noticed this morning that the 0.29A reading of the bench supply had finally dipped down to late last night, had risen back up to its earlier 0.30A reading. Obviously a temperature effect on the OCXO's heating demand.

 I'm currently using a foam rubber insulation jacket over the OCXO which knocks about a 100mW off the naked 900mW requirement at the internal 27 or so deg temperature of the GPSDO enclosure. I'm planning on upgrading this to a polystyrene foam insulating jacket which will further reduce power consumption of the OCXO, helping to both extend the LiPo cell autonomy and knock a degree or two off the internal temperature both effects, small as they are, being a benefit to the LiPo cell's autonomy and service life.

 I fully appreciate that such insulation if taken to the extreme, can result in compromising the temperature control but I doubt this will be a problem with the limited space available to add more than half an inch thick layer at the most (even a 1cm thick layer will likely be too much of a squeeze).

 I know the vampire loading of these AE CQE OCXOs when the oven current cuts off during the initial warm up phase due to undershoot is 14mA, a 170mW's worth of heating power on a 12v supply (they can function off as little as 5 volts, just!) hence the need to avoid taking insulation to a ridiculous extreme lest this heat source holds the oven above its set point without any assistance from the heater transistors.

 I think I can safely get away with knocking the heating demand down by another two or three hundred mW before the controller starts struggling with the reduced rate of heat loss. Obviously, this hypothesis needs to be tested to confirm this since an important variable regarding the amount of insulation that can safely be used is rather dependant on the ambient temperature immediately surrounding the OCXO within the enclosure.

 In this case (literally!), my best guesstimate is double the delta T ambient of the case which typical shows a 5.5 deg C above ambient IR thermometer reading. Allowing another degree on top to account for the reduced thermosiphonic airflow within the case itself, I think a reasonable estimate is to assume a 12 degree higher than room temperature figure with the LiPo cell running some 4 or 5 degrees cooler (it's pressed up against the 'roof' of the extruded aluminium enclosure).

[Johnny B Good]

 Good news everyone! 

 I restarted the GPSDO using a 5v supply (bench PSU) just over half an hour ago and the gpsdo trace has returned to pretty much where it had been, compared to the RO reference before the LiPo reached the end of its discharge cycle. The bench supply showed a peak current reading of 1.7A (1.71A allowing for the 10mA under-reading discrepancy error) swiftly dropping to 1.6A over the next ten seconds and continuing down to its current steady state (OCXO oven warmed up) of 0.40A now forty minutes later (implying a LiPo cell charging current down from a peak of 600mA to just 130mA).

 After doing some calculations, it looks like the primary buck converter current output at 3.85 volts, assuming a 93% efficiency (synchronous rectifier type) peaked at just under its specified 2.1A continuous without heatsink rating (3A for 20 seconds unless well heatsinked to allow unlimited operation at this 3A maximum rating). It was this limitation that had led me to the additional complexity of my 'simple' float charging circuit.

 I'm glad to see that my LiPo cell backup power modification managed to pass this first worst case scenario real life test with flying colours. It's rather unlikely that there'll ever be any further such extreme stress tests in the foreseeable future but it's reassuring to know that it'll be able to cope should such an extremely unlikely event ever arise again.

 Anyway, that's the essential guts of the LiPo backup power reserve mod completed (the 'battery power' indicator mod can wait for now), allowing me, at long last (damn you Big Clive for reviewing those Poundland 4000mAH UPS capable power banks!), to resume my custom case build for the RO project I've been working on over the past 12 months or more.

 The GPSDO now has more than enough charge to let me transfer it back to its pride of place on top of the FY6600-60M which resides on the shelf above the SDM3065X parked on top of the SDG2042X which I've upgraded from FY6600 to SDM3065X support plinth status.