Author Topic: Low cost <1 ppb 10 MHz OXCOs on ebay  (Read 23994 times)

0 Members and 1 Guest are viewing this topic.

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 7245
  • Country: ca
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #50 on: June 27, 2024, 07:10:01 pm »
I'd noticed a 25kHz sine amplitude-modulation (envelope) at the 10MHz output- but thought it's just my scope aliasing.  It's on my list to see if this is a scope artefact due to lower sweep speed, or if the module has a parasitic oscillation.

Those pictured modules,
The one on the left has newer 200k trimpot, red LED, possible better NS LDO.
The one on the right 100k trimpot, blue LED, crap quality LDO.

I've got a 200k trimpot, blue LED and crappy LDO. They'll use any parts in these as some are not critical.
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #51 on: June 27, 2024, 11:14:47 pm »
Thanks to the comments in this thread I thought to re-look the power supply setting for the Amazon OCXO.  Previously, I had the PS set at 12V and had tried various trimpot settings but could never get the counter to show better than about 10,000,000.7 Hz. 

After reading this thread I decided to try around 5V while also twiddling the trimpot.  The BG7TBL FA-2 counter is taking it's ref from a BG7TBL GPSDO (same as it was when I was operating the OCXO at 12V).  It's hard to know what exactly is causing what, or more specifically, it's hard to know what's accurate and what just looks good, but by dropping the voltage on the power supply to 5V or somewhat lower the Amazon OCXO was able to improve it's performance, at least according to my particular testing configuration.

Main point is that the voltage feeding the OCXO can have an order of magnitude or more impact on the OCXO's performance, I think.  The spec printed on the board says DC7-13V@1A; at 4.64V and 292mA it seems happy.
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 505
  • Country: ca
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #52 on: June 28, 2024, 02:26:26 am »
ah ...   you have a board ....   I assume it was just that metal can only.
about that thingi  i can not comment , honestly ...   as it unclear how it constructed ..

i build a while ago , like 4 years ago ,F ref , on the same OCXO can,  it was significantly thermally isolated, consumed 25 ma
after a while, year or so in ON state , drift was ... hm ... in 4 month approx. , from  precise 10 ...  to around 9.999 999 998 5
sooo  basically 5 10-8  per year  around  ... ( i'm right?)

received today OSC5A2B02, was curious about jitter .... seems nothing ...
I may consider , super duper high jitter was incorrect measurement :
i'm wait 60 seconds , till snapshot , that 7th rise from the trigger


forgot to say .... that high P-P,  due to GPSDO, reference , as the circuit truing to constantly push-pull OCXO adjustment , if you run fa-2 for at least a week , internal oscillator would be reasonable steady, and hook gps to input, you may see as 2 last digit constantly jumping up or dowm,   merely depend on gpsdo algorithms adjustment  ...


« Last Edit: June 28, 2024, 02:35:26 am by GigaJoe »
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #53 on: June 28, 2024, 03:08:55 am »
forgot to say .... that high P-P,  due to GPSDO, reference , as the circuit truing to constantly push-pull OCXO adjustment , if you run fa-2 for at least a week , internal oscillator would be reasonable steady, and hook gps to input, you may see as 2 last digit constantly jumping up or dowm,   merely depend on gpsdo algorithms adjustment  ...

I think the GPSDO does constantly adjust up and down but I'm pretty sure what you saw above for Pk-Pk with the Amazon OCXO is primarily from the Amazon OCXO/board.

Below is the same FA-2 counter (getting it's ref from the GPSDO) taking a 10 MHz input from an Agilent generator (also ref'd to the GPSDO).  Not so much Pk-Pk.  I think this indicates that the GPSDO can be reasonably steady.  If need be I can probably find similar results from measurements made over longer durations.

This stuff can seem like an endless rabbit hole, especially if you don't have a rock solid reference point but after lots of rabbit hole experimenting, I think the BG7TBL GPSDO and counter make for some pretty decent measuring.  Having said that I'm open to figuring out how to measure more accurately and how to be more certain of the accuracy.

Slightly off topic, but one of my motivations (reasons for making and posting the measurements) is to persuade any observers from Siglent that as great as their oscilloscopes are they could be even greater if they offered more models with a 10 MHz Ref in.  When a $14 retail OCXO board can provide 1-2 orders of magnitude better than what comes with a $1k-$3k oscilloscope why not provide a Ref in?  And with a decent GPSDO the performance could be even better.


- no doubt the last 3 digits (as in the 012 7 Avg) will jump around but at this point I think maybe it's good enough; if I could persuade myself that these frequency numbers are accurate at this level of precision I'd be ready to move on to trying for better measuring of amplitude :), and of course phase :)
« Last Edit: June 28, 2024, 03:20:54 am by Electro Fan »
 

Offline Andrew_Debbie

  • Frequent Contributor
  • **
  • Posts: 614
  • Country: gb
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #54 on: June 28, 2024, 12:44:49 pm »
My LBE-1420 GPSDO arrived today.   I'll be checking my "blue light special"  soon.    It does not appear to be overly sensitive to input voltage.   Maybe this one has somewhat better parts.

 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 505
  • Country: ca
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #55 on: June 28, 2024, 02:47:07 pm »
i fry one one of my counter ..   and looking around got BG7TBL FA-2 with 6 buttons.

FA-2 design: in use ATmega328P microcontroller. 
this is 8-bit microcontroller with 2K RAM,  no float operations in hardware
to calculate float value, you need a library that can simulate float on 8bit integer ...  float 4 bytes
for a precision of 7 decimal point to show on display ;  you need double float  in calculation ;  double float 8 bytes

connection OUT to IN - i do observe flickering in LSD  - I may assume it errors in calculation. as the rest are constant. and error due to 8bit INT to FLOAT library

in averaging mode we see 7 decimal point, but even with OUT to IN mode,  everything after 5th digit are random ...  basically noise

so  looks like , there no double float precision, limited memory means no room for precise calculations.
consider that , a may assume that precision for single measurement as calculated value would be 5 decimal point ±2 LSD. 

averaiging: my guess averaging window somewhere 10 numbers,  so running N 500 - no value ( and no space in memory) , as it shows last 10 averaging.
with 5 decimal points,   the rest basically noise  ....

on opposite side , if we have reference 10Mhz , to measure with precision .00000  we need to detect phase shift  better then 100nS / 100000  = 0.01nS , 10pS , I have no any experience with intel altera MAX II EPM570 CPLD  that in use for this counter ....

I may wrong in my conclusions ....  but atmega ...  really?
« Last Edit: June 28, 2024, 03:07:44 pm by GigaJoe »
 
The following users thanked this post: Electro Fan

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #56 on: June 28, 2024, 04:30:27 pm »
i fry one one of my counter ..   and looking around got BG7TBL FA-2 with 6 buttons.

FA-2 design: in use ATmega328P microcontroller. 
this is 8-bit microcontroller with 2K RAM,  no float operations in hardware
to calculate float value, you need a library that can simulate float on 8bit integer ...  float 4 bytes
for a precision of 7 decimal point to show on display ;  you need double float  in calculation ;  double float 8 bytes

connection OUT to IN - i do observe flickering in LSD  - I may assume it errors in calculation. as the rest are constant. and error due to 8bit INT to FLOAT library

in averaging mode we see 7 decimal point, but even with OUT to IN mode,  everything after 5th digit are random ...  basically noise

so  looks like , there no double float precision, limited memory means no room for precise calculations.
consider that , a may assume that precision for single measurement as calculated value would be 5 decimal point ±2 LSD. 

averaiging: my guess averaging window somewhere 10 numbers,  so running N 500 - no value ( and no space in memory) , as it shows last 10 averaging.
with 5 decimal points,   the rest basically noise  ....

on opposite side , if we have reference 10Mhz , to measure with precision .00000  we need to detect phase shift  better then 100nS / 100000  = 0.01nS , 10pS , I have no any experience with intel altera MAX II EPM570 CPLD  that in use for this counter ....

I may wrong in my conclusions ....  but atmega ...  really?

Thanks for the info based on your examination of the counter and GPSDO designs.  Sounds like there are extra digits that are random.

I don't have a good way (that I know of yet) for measuring or comparing against a baseline that is known to be accurate.  So I'm on the lookout for something that could help determine what's what.

I happen to have a Leo Bodnar miniGPS that I use to provide a 49,152,200 Hz clock signal to a radio, or at least I that's the designated frequency.  I fed that signal to the FA-2 (which is taking it's 10 MHz ref from the BG7TBL GPSDO), and got the results shown below.

If I understand you correctly, maybe anything over 10 samples is really just the average of the most recent 10 samples?  And you think 5 places to the right of the decimal might be relevant?  Thanks for any help in clarifying this. 

Attached are the LB specs.

PS, slightly off topic, but the more I look at the designs of all sorts of stuff (from clocks to AI) the more I appreciate the power of floating point and the need to use it appropriately
« Last Edit: June 28, 2024, 05:58:50 pm by Electro Fan »
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 505
  • Country: ca
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #57 on: June 28, 2024, 06:31:12 pm »
the second though that  atmega role not as calculator , but display and buttons manager only,   then all measurement and calculations fall to  intel altera MAX II EPM570 CPLD chip. so my conclusion is incorrect ...

in this case how precise it ....  i cant tell ..
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #58 on: June 28, 2024, 06:57:58 pm »
the second though that  atmega role not as calculator , but display and buttons manager only,   then all measurement and calculations fall to  intel altera MAX II EPM570 CPLD chip. so my conclusion is incorrect ...

in this case how precise it ....  i cant tell ..

I appreciate and admire the ability to have second thoughts and to keep exploring, not to mention your newest idea is more encouraging regarding the likely efficacy of the architecture :)

- further into the rabbit hole we go  :-+
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #59 on: June 28, 2024, 08:36:05 pm »
Nothing like OCD in the rabbit hole.

I recorded 29 consecutive measurements (with 10 second gates); #350 within the series is shown below.  I entered the 29 consecutive current reads ("Fin") and the average reads as shown in the attached spreadsheet.  I can't tell how many reads go into the average calc but I think it's more than the most recent 10.  Maybe someone with a FA-2 can figure it out.
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #60 on: June 28, 2024, 11:12:09 pm »
The previous post used several short BNC cables (with inline connectors) plus a couple BNC to SMA adapters.

This one connects the LB GPSDO to the FA-2 with a single 15' SMA to SMA cable.  Much less Pk-Pk variation.

Might have been some reflections at the several adapters in the prior configuration?

Another variable is the ambient temperature as the AC cycles on and off.

All that said, I think this shows that the BG7TBL FA-2 with a ref in from a BG7TBL GPSDO and the Leo Bodnar MiniGPS as the signal input are pretty happily working together to produce what I think are reasonable results.  I realize "happily" and "reasonable" are ("substantially") more qualitative than quantitative - but if anyone wants to suggest some other tests, please feel free to do so.  Thx
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 505
  • Country: ca
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #61 on: June 28, 2024, 11:18:54 pm »
doesnt work on random ..

so it def. simple averaging, my calculation match with device , N=10, N=20 , then i give-up

and my has gate 10mS
now only last 2 digits changing ...


with N 1000 or more , it def. not atmega
fun thingi :

Gate 0.01S  N=32000  .994 486 3
Gate 10S     N=10       .996 288 8

if gave possible do variable , it would be really awesome
and seeems yes, averaging not equal precision ..
« Last Edit: June 28, 2024, 11:30:28 pm by GigaJoe »
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #62 on: June 30, 2024, 01:05:58 am »
Update:

I tried a couple more Amazon 10 MHz OCXOs.  One happened to have a blue Led, the other a red Led like the first one I tried (see previous posts).

If you try these, I think it would be handy to have a counter that gives you some reasonable confidence and that has more digits than less to the right of the decimal point (even though we know resolution does not = accuracy; presumably your confidence inspiring counter is reasonably in the ballpark and the extra digits will help you see the rate at which things are going up or down - kind of like having an analog DMM but here we don't have a needle on a dial so we're just watching the digits go up or down and the rate at which they are moving).  For this process I find that a 1s gate gives you a feel for what's happening at what rate. 

When you first set up the OCXO it will take some time to warm up and settle down (maybe a half hour or more).  Once it seems to have settled on your counter, experiment with the trimpot to determine which direction adds resistance vs removes resistance - just give it a turn or two and see which way the counter moves.  After you are sure you know which direction adds resistance/decreases resistance and which way the counter moves accordingly as you do that, quit playing with the trimpot.  If you turn the trimpot too far you will run out of scope.  So after your direction confirming testing you want to start the dialing-in process from near the midpoint of the trimpot.

The next step requires a decent power supply where you can control voltage, and current to at least a milliamp.

At this point, you can experiment with some safe but clearly sufficient constant current (1 amp per the board, or less) while changing voltage from about 4.5V to 12V (the board says 7-13V).  Changing the voltage is unlikely to have much if any impact on the performance you see on the counter, at least that I observed but my tests were all pretty short term. 

Now that you have the the voltage set (maybe at 5V or even less) it's time to adjust the max current.  If the counter shows less than 10 MHz, add a milliamp or two and the OCXO will run faster, as you should see on the counter.  If the counter shows more than 10 MHz reduce the current by a milliamp or two and the OCXO will run slower.  You have to be patient with this process or you will overshoot the mark.  You might/probably won't be able to get the counter within much more than a 0.1 Hz of the target 10 MHz by just adjusting the power supply.  I found that 1 mA moves the frequency by more than 0.1 Hz.  So once you get the counter reading as close as you can with the PS adjustments then you will have to start adjusting the trimpot.  (EDIT:  I think the mA to 0.1 Hz relationship might change the longer the OCXO has settled in.)  In any event, the trimpot adjusting process is not the most confidence inspiring thing but it's pretty much all that's left other than the PS.

My main learning (or at least what I think I learned) is that regulating the current is pretty important to being able to control the OCXO's frequency.  Another thing that I'm sure adds a variable is the ambient temperature in your room.  I know from previous experiments that a degree C or F can be enough to change resistance in a resistor by a a few milliamps.  (I learned this by experimenting with a short piece of copper wire and doing some 4-wire Kelvin clip readings).  The point here is that when trying to dial-in a milliamp you could be chasing more than the milliamp of frequency-varying current if the resistance of your circuit is changing due to the ambient temperature changing.  Every time you touch the board to steady it while adjusting the trimpot you will probably heat up the board, which can add some confusion to the trimpot adjusting process.

Feel free to ask questions or point out any thing I got wrong - it's possible I missed on some of this - it's all a learning journey.  Happy metrologying.

EDIT:  After spending more than a couple hours trying to dial-in 10 MHz by adjusting the current coarsly with the PS and more finely but without much luck on the trimpot, I came up with two ideas:
1. Buy a new PS that can set current to at least a tenth of mA (within 100 microamps) - but a quick look-see indicates that would be expensive, so I moved on to:
2. Turn on the air conditioning and within two minutes I can lower the temperature and the frequency sufficient to drop 0.2Hz.  I think idea #2 while not easy to control overshoot is going to be the more cost-effective approach.
« Last Edit: June 30, 2024, 09:52:50 am by Electro Fan »
 

Offline KedasProbe

  • Frequent Contributor
  • **
  • Posts: 653
  • Country: be
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #63 on: June 30, 2024, 02:36:20 pm »
I compared my two OCXOs to each other.
It wasn't very repeatable so I added two curves B and D
the timelab files are attached.

Conclusion is what we already knew, below 10 sec the cheap OCXO is more stable than the leo bodnar GPSDO.

I checked power supply influence but it's below 2mHz per volt.
That is basically how much change there was in 1 hour. (keep in mind I compared them to each other)
Not everything that counts can be measured. Not everything that can be measured counts.
[W. Bruce Cameron]
 
The following users thanked this post: ch_scr

Offline Andrew_Debbie

  • Frequent Contributor
  • **
  • Posts: 614
  • Country: gb
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #64 on: July 01, 2024, 10:22:15 am »
I just recieved the sz-plaza dot com rev.0.0.2 OXCO module.   At first glance it is better than the OSC101.  I paid $12.59+VAT (uk tax) for the complete module.


The LDO regulator + LT1009 reference makes the board far less senstive to input voltage changes and alows a wider operating range.   

Could you set the pot to get it to oscillate above & below 10MHz?


Yes, I can get it to oscillate on either side of 10MHz.       I connected my Racal-Dana 1998 counter to a Leo Bodnar LBE-1420 and let it run over the weekend.   After a day my counter settled at 10MHz +.04Hz.

I am not going to adjust my counter.  I might make it worse.

Now that I have verifed the counter, I'll start doing some check tests on my "Blue Light Special".     

Input voltages between 6V and 7V don't make any short-term change to output frequency within the resolution of my counter.      This one must have a halfway decent LT1009 voltage reference.     

I'm going to let it run for a few days.   
« Last Edit: July 01, 2024, 10:24:59 am by Andrew_Debbie »
 
The following users thanked this post: Electro Fan

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #65 on: July 01, 2024, 05:19:53 pm »
I'm probably far down the knowledge and experience curve from most or all of the posters here but fwiw, I find that adjusting the Power Supply voltage between 4.25V and 12V is only somewhat interesting and impactful; more interesting and impactful is directly controlling (by limiting) the PS current available to the eBay OCXO board.  More current causes the OCXO to run faster, less current slower.  Also, the current seems to impact the OCXO board temperature, and the ambient temperature further impacts the OCXO board.  As temperature increases (due to either increased current or increased ambient temperature) the counter shows the frequency increases, as temperature decreases the counter shows the frequency decreases.  And of course there are time lags everywhere, especially with temperature changes but also with current changes.  No doubt the current and temperature changes are impacting resistance which together with the voltage impacts the current.  It is a lot of around and around and back and forth.

EDIT:  a challenge with the temperature measurement is that while the sensor sits close to (is touching) the OCXO board it is hard to know if the temperature is due to changes in the current drawn by the board (which is limited by how much current the PS makes available), or if the temperature is due to changes in the ambient temperature (if the Air Conditioning is on cooling occurs, if the AC if off heating occurs).  Likely the sensor is showing some cumulative effect of the current and ambient temperature.
« Last Edit: July 01, 2024, 06:00:55 pm by Electro Fan »
 

Offline Andrew_Debbie

  • Frequent Contributor
  • **
  • Posts: 614
  • Country: gb
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #66 on: July 03, 2024, 08:34:42 am »
I'm going to let it run for a few days.

48 hours later. . .

My counter is reading 10.00000001 MHz.    That is .03Hz lower than on Monday.    I don't know what the spec. is for the OC5SC25.    Perhaps the observed drift is due to variations on the adjust pin. 

A real LT1009 from Analog/LT is spec'd 20ppm/kHr @ 25C  typical.  There is no min or max.   Thermal drift over the temperature range in my office is perhaps a few mV.

2300297-0

It shouldn't be hard to swap in an LT1019A which is 3ppm/℃ typical.    One from Digikey would cost more than I paid for the entire board.   Maybe worth it for boards that came with "bad" LT1009 clones?

 
I'm using the internal Option 04A Ovened Oscillator as a reference.      The spec is 3x10-9/day after 3 months continous operation.   My over the weekend check with the LBE-1420 found it to drift less than 1x10-9/day.   When possible I will check the internal timebase drift over several weeks.  weeks.  I might / should change the 35 year-old power supply capacitors first.

I can't use my GPSDO on my desk.  Not yet.  I'm looking into a way to get the antenna a good view of the sky.

« Last Edit: July 03, 2024, 12:25:29 pm by Andrew_Debbie »
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3486
  • Country: us
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #67 on: July 05, 2024, 06:09:55 pm »
It’s hard to say which I find more amazing, the performance or seeing them for $13 as a complete module.

Reading people’s recent comments I find the claims by some that they are junk laughable.  I wish I could find more such “junk”.

Have Fun!
Reg
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #68 on: July 05, 2024, 08:50:58 pm »
It’s hard to say which I find more amazing, the performance or seeing them for $13 as a complete module.

Reading people’s recent comments I find the claims by some that they are junk laughable.  I wish I could find more such “junk”.

Have Fun!
Reg

+1

I have tried several of these and while they vary, if you get a good one it can be impressive for the price.  This is from a $13 unit with no adjustments to the trimpot.

 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #69 on: July 05, 2024, 09:02:04 pm »
fwiw, the Min-Max variation might be somewhat due to the temperature variation.  The sensor is sitting near/touching the OCXO module but it is no doubt measuring not only the module temp but also the ambient temperature which fluctuates with the air conditioning going on and off.  I'm guessing the temperature variation is also due to some power supply current fluctuation.  Point is that for not much $ it's possible to get pretty good accuracy and stability even without much if any attention given to packaging.

(One of the GPSDO's is providing a 10 MHz ref to the counter.)
 

Offline GigaJoe

  • Frequent Contributor
  • **
  • Posts: 505
  • Country: ca
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #70 on: July 05, 2024, 09:43:04 pm »
in average, the oscillator minimizing drift after 2-3 weeks.   
 

Offline rhbTopic starter

  • Super Contributor
  • ***
  • Posts: 3486
  • Country: us
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #71 on: July 06, 2024, 12:40:41 am »
Since when is  0.37 ppb “pretty good”?  Stick it in a box with some polyester batting and I expect it will do near the 0.1 ppb of my first two.  I have checked them a few times since and they stabilize to the measurement limits of my 5386A referenced to one of Leo’s dual channel GPSDOs within 20 - 30 minutes.  It’s quite a bit better than the OXCO in my 5386A.

Of the next 3, one had a cap busted off in shipping and the other two wouldn’t adjust to 10 MHz.  Stability was good, just the correct tuning voltage was out of range of the trimmer. I’ll fix that eventually.

I have 10x too many projects which means only a few get completed. But I have fun and that’s all I care about at 71.  Life is getting short. After learning I would need to drive 75 miles 5 days a week for 8 weeks to get radiation treatment for a recurrence of prostate cancer, I ordered a Toyota GR86.  If I’m going to drive that much I decided I should finally buy a proper sports car.  The 2nd new vehicle in my life.  I still have the first, a 1993 base model Toyota pickup.

Have Fun!
Reg
 
The following users thanked this post: Electro Fan, ch_scr, Andrew_Debbie

Offline Andrew_Debbie

  • Frequent Contributor
  • **
  • Posts: 614
  • Country: gb
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #72 on: July 06, 2024, 08:30:51 am »
It’s hard to say which I find more amazing, the performance or seeing them for $13 as a complete module.

Reading people’s recent comments I find the claims by some that they are junk laughable.  I wish I could find more such “junk”.

Have Fun!
Reg

The OSC101 variant I have isn't nearly as good as the sz-plaza rev 0.2 module.     

The OSC101 powers the adjust pin from a generic 78M05.     It drifts with input voltage.   It drifts with ambient temperature.     Connected to an HP linear bench power supply, observed drift in one hour is about 400x  the sz-plaza drift per day.



« Last Edit: July 06, 2024, 08:34:05 am by Andrew_Debbie »
 
The following users thanked this post: Electro Fan

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #73 on: July 06, 2024, 07:57:47 pm »
Since when is  0.37 ppb “pretty good”?  Stick it in a box with some polyester batting and I expect it will do near the 0.1 ppb of my first two.  I have checked them a few times since and they stabilize to the measurement limits of my 5386A referenced to one of Leo’s dual channel GPSDOs within 20 - 30 minutes.  It’s quite a bit better than the OXCO in my 5386A.

Of the next 3, one had a cap busted off in shipping and the other two wouldn’t adjust to 10 MHz.  Stability was good, just the correct tuning voltage was out of range of the trimmer. I’ll fix that eventually.

I have 10x too many projects which means only a few get completed. But I have fun and that’s all I care about at 71.  Life is getting short. After learning I would need to drive 75 miles 5 days a week for 8 weeks to get radiation treatment for a recurrence of prostate cancer, I ordered a Toyota GR86.  If I’m going to drive that much I decided I should finally buy a proper sports car.  The 2nd new vehicle in my life.  I still have the first, a 1993 base model Toyota pickup.

Have Fun!
Reg

Reg, Thanks for all the insightful info in your posts.  Hope you are enjoying the GR86 and most importantly that the treatments are going as well as possible.  EF
 

Offline Electro Fan

  • Super Contributor
  • ***
  • Posts: 3280
Re: Low cost <1 ppb 10 MHz OXCOs on ebay
« Reply #74 on: July 09, 2024, 02:29:38 am »
When stuck in the rabbit hole, keep measuring :)

Here is the most recent addition to the $13 OCXO module collection.  Unlike the others I tested the trimpot behaves in a consistently predicable manner.  (The trimpots on the other units barely seemed connected to their circuit - adjustments made very little impact and sometimes the impact was unpredictable.)  With this unit I was able to dial in something close to 10 MHz.  According to my GPS disciplined counter it is reasonably accurate and stable even though it is not shielded in any way.  I have some cases so next I'm going to see if I can protect the OCXO module from the ambient temperature swings.  But even without any improvement my main conclusion is that any piece of test equipment over several hundred $ should have at least comparable performance, or a 10 MHz ref in.
 
The following users thanked this post: Andrew_Debbie


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf