EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: erikka on November 11, 2025, 03:17:09 pm
-
This week the sales will start of a new product in the tiny family.
A GNSS disciplined oscillator and two input universal Timer and gap-less reciprocal frequency counter, in short: tinyGTC
The tinyGTC is an extremely accurate (12 digits/s) frequency counter that can measure time, frequency, count and phase.
Time resolution is 40 ps. Input frequencies for both inputs are between 0.1 Hz (unlimited time measurements) and 350 MHz. Input A has an optional GHz pre-scaler extending the frequency input range to 7 or 12 GHz
It has a built-in GPSDO using a good quality TCXO and an external GPS antenna (included) and it can output maximum 2 flexible frequencies.
Is is designed to provide optimal input to tools like Timelab for true gap-less phase based ADEV measurements.
The set of software and communication features is comparable to modern high end counters.
Due to the small size and the low cost the tinyGTC has, compared to modern high end counters, less input protection, less data memory and lower maximum measurement speed.
More information can be found in the tinydevices wiki: https://www.tinydevices.org/
-
Thats an interesting looking device!
Does anyone know how the system operates internally, in order to reach such timing resolution?
-
It uses TDC"s like the TDC7200
-
Offered here
https://www.eleshop.nl/catalogsearch/result/?q=tinygtc (https://www.eleshop.nl/catalogsearch/result/?q=tinygtc)
-
Given the spec mentions picoseconds but the refclk is 10 MHz I expected something like a TDC, but more technical details would be welcome.
I have no idea if anyone has one already, and if any internal pictures are available anywhere.
A schematic would be ideal but I'm not expecting that much :)
-
See https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.Internals (https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.Internals)
-
Additional tinyGTC development info containing some photos of hardware at different stages. From a recent seminar given by Erik in the Netherlands.
https://www.rfseminar.nl/cms/wp-content/uploads/2025/10/RF_Seminar_Counter_design_Erik.pdf (https://www.rfseminar.nl/cms/wp-content/uploads/2025/10/RF_Seminar_Counter_design_Erik.pdf)
Herb
-
Ordered the 7Ghz version....because I'm curious (I also have the tinyPFA)
-
I love the idea to put the abilities of a Pendulum counter into something affordable.
But there are reasons why all and every counter has low-speed high voltage and high-speed low-voltage inputs. Even the design makes it very practical for field use, the missing robustness of the inputs prohibits that.
And finally the form factor just kills it for me. All those flat tablet like things crowding my desk, needing more space than the DUT itself. Also all the cables coming from the top dangling in front of everything. And putting them on the shelf behind the DUT just a little pull on the test lead makes them toppling over.
Build it into a case that that can be put between a bench top multimeter and power supply, put the test lead connectors below or aside the LCD and make the batteries optional. Even that may cost 200USD more, I would buy it.
-
And maybe engage someone with UX/UI experience to do the GUI. That thing is eye cancer.
Less is more, and more screens with less info crammed into each would be a significant improvement.
-
The UI is fully configurable. Please help me to understand what is wrong
-
To be fair, that does seem better. However, the impression it still gives is that quite a lot of the text is rendered rather small, on a relatively small screen. Some of us older folks don't find that so comfortable to use.
Since I've not seen one in reality, and can't find any physical dimensions for it, I could be wrong. And I'll bracket it all by saying that it's not my intention to deliberately offend, nor to just throw eggs at your achievements.
Nonetheless, I think it could be improved if there were scope to hide a lot of things that I may not be using/interested in at any one point in time (e.g. display only one channel's counter and status, and lose the timeclock, battery, output status etc.), and make better use of the resulting extra screen space for larger text and controls. I'm not a fan of button text which runs horizontal in some cases but vertical in others. I'm also not sure why the colours used for the channel config/status panels (green and pink in the initial screenshot) wouldn't then be the same colours used for their respective frequency counter displays below.
As I said, I've not seen one in real life yet, so I'm prepared to concede that my initial impressions might be unfounded.
-
The display is 4" so for sure not big.
For those that have problems with this size I have create a remote desktop utility (a very small PC program) that can mirror and enlarge the display on a PC screen and you can control the device using a mouse from the PC.
-
You can also get rid of the plot but nothing more
Unfortunately the amount of digits is to big to show much larger numbers
-
I assume I can easily check if a PPS source is correct since it should have zero phase difference with your device GPS signal?
I also will try to measure the phase over time of 3 independent 10MHz sources.
Not sure how the software settings go, but it should be possible from what I understood.
P.S. thanks for the monitor screen up-scale, I will need that.
-
Below the screen capture of the setup for measuring 3 clocks simultaneously
And after sending them together to Timelab you can do the 3-cornered hat ADEV
Wat is shown below if the noise floor of the tinyGTC, can't measure sources with stability better then shown
-
I'm also not sure why the colours used for the channel config/status panels (green and pink in the initial screenshot) wouldn't then be the same colours used for their respective frequency counter displays below.
The inputs and the counters are not hard coupled when in Advanced menu mode ( a kind of expert mode). You can measure the Pos Width of input A using counter 1 and counter 2. One triggers on the up edge and the other on the down edge. The time difference between the timestamps is the Pos Width
Of you have two counters with different gate time monitoring the same input. One fast to check for jitter, the other slow checking the frequency drift using a longer gate period
But this is all in advanced menu mode.
It may indeed be a good suggestion to make the colors the same when in basic menu mode when there is a hard link between an input and a counter
-
Also all the cables coming from the top dangling in front of everything. And putting them on the shelf behind the DUT just a little pull on the test lead makes them toppling over.
The UI is user configurable, and the display has a menu option so it can be flipped so that the cables now exit from the bottom.
-
@erikka
About the 1Hz small BW 'random' chosen freq with less measurement accuracy,:
Do you know if expensive counters/timers have a solution for that or are they just not mentioning it?
Maybe they duplicate hardware and choose after the measurement which of the two to use hence covering this small spot. (obviously making it more expensive for little return)
I think on the BG7TBL FA-5 I saw such effect when the count for a chosen gatetime is 10M, not sure if that is the same effect since it looks much wider than 1Hz, but because I know I can work around it with an different gate time.
BTW amazing the amount of settings/features you pack in it for such low price. (assuming it works :) )
-
@erikka
About the 1Hz small BW 'random' chosen freq with less measurement accuracy,:
Do you know if expensive counters/timers have a solution for that or are they just not mentioning it?
For sure HP did have this "accuracy gap" as they mentioned it in their specification (using other words)
Not sure about modern counters. It takes a long time to detect this frequency with any certainty
-
I received mine. :)
I will have some time this weekend to test it.
I just turned it on and I have a few things I noticed:
- HW version on the device is 0.2.8 but in the software it says HW 1.1 (it probably makes sense but is confusing)
- For the tinyRemote.exe the scaling was not sharp, I checked 'Overwrite high DPI scaling behavior' for the exe that fixed it.
(it's a case of windows thinking it is smarter)
- In the tinyRemote.exe GUI you need to select Auto Refresh, default it's off.
(otherwise you are stuck on the same screen, the mouse still works on the device though)
- it adds 3 COM ports each with their own function, the tinyRemote has no problem finding it but no auto reconnect.
- scaling 200% is not saved for next time you start it.
- Auto Refresh checkmark isn't always correct. (on when I start but is off, after pressing on screen with the mouse it's updated)
- brightness and viewing angles are good.
- Weight including GPS antenna is about 250g. It has weight, if you drop it, likely something will break. (normal, it's not a high end smartphone)
-There is some short screen flicker on the device when the menu bar is removed.
-Correct UTC time is displayed
- SW version tinyMonitor.exe: has no version number?
- SW version tinyGTC: V0.1-510
- Build date Oct 23 2025
- USB disconnect during monitor makes it very slow CPU 99% (restart required)
First start up screen attached, later a real test.
Edit, I saw I'm like 10 software versions behind, so I will update first.
-
You can also get rid of the plot but nothing more
Unfortunately the amount of digits is to big to show much larger numbers
Again a lot of wasted space on the screen. Why don't you move the MHz above the number and then use the full width of the screen to display the number?
And yellow is a warning color. Red is an alarm. Use black grey and blue. You can use yellow numbers on black background to have less of an intensity bleeding, but I'd prefer grey or bright gray. This isn't a desk light it is a display showing data.
It is a Chinese thing to be the brightest star on the shelf. But in reality it is stressful to have that bright spot light in sight all the day. And it is very uncomfortable for someone recording or streaming as you always have to switch that thing off when you don't use it. While other instruments can stay on 24/7 without trying to bring their display to attention all the time.
-
Leaving either input connected to an external source for an extended period of time with the tinyGTC switched off will damage your device.
from https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.Safeoperation (https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.Safeoperation)
Is this expected ESD or Heat or not being able to handle the voltage when it's off?
-
Is this expected ESD or Heat or not being able to handle the voltage when it's off?
We had one beta tester reporting this issue. Further testing did not show an acute problem but it is not completely ruled out. The warning is just to be sure.
-
Short video: https://www.youtube.com/watch?v=yqtRUUvnSgQ (https://www.youtube.com/watch?v=yqtRUUvnSgQ)
-
A lot of functionality at a very affordable price.
Don’t need one right now, but have bookmarked the website for a future purchase.
Thanks.
Related question, do you ship to the US?
-
R&L will have stock soon
-
Hi Erik,
great to see another product developed by you! I'm very interested, but I would like to know the following:
Can the tinyGTC be used to measure the time difference between two signals (less than 20MHz in repetition rate, but fast rising, nanosecond width) — for example, between the REF output and one of the inputs (A/B) — with a resolution in the tens of picoseconds? Additionally, is it possible to display this as a histogram (counts vs. ΔT)?
Your dedication is an absolute inspiration.
cheers
-
This is a 16 ns wide pulse with a 5 MHz repetition rate coming from an external source linked to a reference clock send into the A input.
The tinyGTC is using the same reference clock as REF input and it timestamps every 10 ms at up edge and the timestamp is wrapped at 100 ns to get the "time versus the REF"
The measured time is displayed in a histogram with 80 buckets each 3 ps wide over the last 21874 measurements (maximum measurements that can be accumulated into a histogram)
You can see the quantization of the time in the histogram
-
The screen capture was missing.
-
This one is a bit more challenging.
Two 1 MHz, 24 ns wide pulses with fixed time between the two pulses, send in to inputs A and B.
Measurement is set to measure the time difference between A and B
This is more difficult because you get twice the measurement noise due to two inputs being used.
-
New video showing frequency, phase and time stamping measurement and various ways to show the results of these measurements.
https://www.youtube.com/watch?v=bd6Li-4luQk (https://www.youtube.com/watch?v=bd6Li-4luQk)
-
Dammm ....
30 days lead time on Eleshop :scared:
I was on the looks for an early xmas present
-
New video, how to setup the tinyGTC in combination with TimeLab
https://youtu.be/R9YNsT9EPF4 (https://youtu.be/R9YNsT9EPF4)
-
I updated the firmware to version 1.0-12 and did some basic tests.
two remarks
1) is there a reason you start with \r (I use readline() )
\r10000000.0000068303198 CHA\n
Maybe you have the sequence reversed and you end with \n\r instead of \r\n ?
2) The calculation of STDEV is too low are you using an other calculation? I verified with USB log data.
(see screenshot, 30.4µHz is correct)
Data of the TinyGTC looks better than the GB7TBL FA-5 (measuring its own reference)
Clearly no gap since I get 1000ms interval, with FA-5 I get unstable something like 1023ms
-
Dammm ....
30 days lead time on Eleshop :scared:
I was on the looks for an early xmas present
You can get it from China if it's so important to get it before Christmas.
Zeenko/Hugen is the OEM that Erik is using, so it's the same backland that European sources like Eleshop are sourcing from.
Their official store.
https://aliexpress.com/item/1005010361182255.html?
The biggest issue is that you seem to be based in Denmark, and the Aliexpress system seems kind of messed up lately, as 80% of my wishlist was suddenly overnight greyed out with a blank white slide "not available to your region"/Denmark.
Seems to be centered around Denmark for some strange reason, as if I change my region to other EU countries, incl. neighboring countries, most of these items are again purchasable.
The TinyGTC was, in fact, purchasable from Denmark just a few days ago from Zeenko's Aliexpress store, but now its "region verboten" and locked away behind a white regional slide, while it's still purchasable from these other EU countries like Germany, Spain, France etc. https://imgur.com/a/ro7I4yw
Not sure why Aliexpress lately has gone so ballistic in censoring "listings" away from Danish consumers, quite annoying. Labgear, mobilephones, DJI drones & many other science & tech segments just gone if you have a Danish address... instead it's primarily low-cost Temu/Wish-like chinese garbage that is left.
One can always use parcel forwarding and rely on a flip address in all these other EU countries, and simply bounce a parcel through there, but it adds time, alongside significant expenses.
And in regard to TinyGTC, there is close to no financial gain getting it directly from China, as the 407/412 prices look to be around the same as EU vendors (at least for the moment) the primary benefit would be time-related from lack of EU stock, and get it before Christmas.
-
Time resolution is 40 ps.
The measured time is displayed in a histogram with 80 buckets each 3 ps wide over the last 21874 measurements
So is the resolution 40ps or 3ps? 3ps is really impressive!
-
New video, how to setup the tinyGTC in combination with TimeLab
https://youtu.be/R9YNsT9EPF4 (https://youtu.be/R9YNsT9EPF4)
Hey, Erik -- can you send an ASCII text dump to john (at) miles.io with some frequency readings that demonstrate the erroneous phase slope that you observed in the video? This isn't something I've run into before, so I'd like to make sure there's nothing unexpected going on. I don't usually import frequency data myself, so maybe there's some room for improvement that I've overlooked.
-
John,
Will do
When I change the settings to eliminate the linear regression the phase derived from the frequency does not show this drift so I think this is an artefact from the linear regression or this is due to some rounding somewhere in the tinyGTC
-
So is the resolution 40ps or 3ps? 3ps is really impressive!
The Time resolution (single event) is 40 ps
The time data is derived from partial higher resolution data so I use 10 ps internal calculation resolution
The 3 ps mentioned was the bucket size in the specific histogram and as the time is rounded to 10 ps only one in every 3 buckets was used.
-
two remarks
1) is there a reason you start with \r (I use readline() )
\r10000000.0000068303198 CHA\n
Maybe you have the sequence reversed and you end with \n\r instead of \r\n ?
Each line ends with \n\r. At one line per second it is easy to see that the \r is sent right after the \n.
2) The calculation of STDEV is too low are you using an other calculation? I verified with USB log data.
(see screenshot, 30.4µHz is correct)
I can see that too. The factor between the two is a bit smaller here but almost constant 1.2 to 1.24 tested at some frequencies between 10kHz and 100MHz
-
Each line ends with \n\r. At one line per second it is easy to see that the \r is sent right after the \n.
The factor between the two is a bit smaller here but almost constant 1.2 to 1.24 tested at some frequencies between 10kHz and 100MHz
The line ending will be corrected in the next release.
And I will check the STDEV calculation
-
This is a fascinating instrument. I used it too check the outputs of a LBE-1421 2-port GPSDO locked clock source set to 1.23456789Hz and 1.23456789GHz simultaneously.
I used these frequencies because they are close to the lower and upper end of the range of the LBE-1421 and it is easy to count the digits.
Both GPS antennas were indoors.
-
It looks many people (including me) who ordered ZC412 from official Zeenko Store on aliexpress finally gets ZC407. Discussion with Zeenko is tough so watch out. Finally he blamed me this was my fault (!!!) but actually there was multivariant offer which was later changed and it looks they did not managed it properly.
Anyway, is it possible to convert ZC407 to ZC412? There is ADF4107 on the A input which acts as a prescaler. What chip is used in the 12GHz version and what about other updates (SW) to get 12GHz?
-
See this message from hugen regarding receiving a ZC407 after ordering a ZC412
https://groups.io/g/tinygtc/message/1623 (https://groups.io/g/tinygtc/message/1623)
-
Actually that thread triggered me to write here as discussion with Zeenko customer support lead to nowhere. It looks they even trying to block possibility of item return on aliexpress (I was asked to provide official parcel weight confirmation from post office). Zeenko customer support only offer was 15$ coupon (not even refund) to buy another ZC412. Really?! :D
Anyway meter you have designed is really nice. Please consider providing details on converting it to 12GHz.
-
To all that received a 407 after ordering a 412
You will be contacted by zeenko..
PM me next week if that did not happen
-
I confirm that in version 1.0-16 USB- log line ending and STDEV are fixed.
(if only big companies fix bugs so fast :) )
-
The line ending will be corrected in the next release.
And I will check the STDEV calculation
Thanks for these correcetions and the other improvements in V1.016.
I think I ran in two other issues, one of them again concerning line endings. When sending SCPI commands the answer ist terminated with CR CR LF in a lot of cases and CR LF in some others.
The other issue is more serious. With the GHz prescaler activated my ZC407 shows wrong Count and Total Count values. The displayed values are off by a factor of 100/24.
-
The other issue is more serious. With the GHz prescaler activated my ZC407 shows wrong Count and Total Count values. The displayed values are off by a factor of 100/24.
Will be solved in release due next week
-
Both solved in release V1.018
-
Thank you.
I think I just ran into another minor issue: When using an external reference and that stops or is disconnected the device crashes with a fatal error and a stackdump and then restarts.
-
This is a known issue and impossible to solve at the moment as when on external reference it is literally the clock for the CPU and any disruption of that clock will cause problems/crash. A solution would increase the cost of the device.
-
I connected the tinyGTC to my Leo Bodnar GPS disciplined oscillator and measured the output at 10 MHz.
I leave the Bodnar unit running 24/7 as my lab frequency reference. It's fast output edges are not liked by some of test so I run it through a 24 MHz 6 pole LP filter and into a splitter to connect to my test equipment that prefers sine waves.
To get the readings shown I turned on the tinyGTC, connected the GPS antenna, connected the Bodnar output to input A, and let it run for 2+ hours while I watched (American) football.
So long as I do not move either unit for a few minutes before I take a reading, they agree to within ±2mHZ or less.
[attachimg=1]
[attachimg=2]
[attachimg=3]
P.S. given the post above I need to update the firmware.
-
This is a great tool!
Are there any measurements of the internal oscillator stability when not supplying GPS or an external reference signal?
Sometimes I need the device to be portable and dont want to have a GPS antenna around - I am worried about how much the accuracy will suffer.
I am happy to measure this myself once I get the hardware, too.
-
So long as I do not move either unit for a few minutes before I take a reading, they agree to within ±2mHZ or less.
P.S. given the post above I need to update the firmware.
Get the latest version of the tinyRemote.exe too to control the device without touching it and to take screenshots.
-
Are there any measurements of the internal oscillator stability when not supplying GPS or an external reference signal?
See below
-
For simply looking at frequency drift below plot makes more sense.
-
Short video demonstrating some of the clock output capabilities and limitations of the tinyGTC
For more information visit the tinyGTC wiki at https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.Homepage (https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.Homepage)
https://www.youtube.com/watch?v=NpAAQwEr4y4 (https://www.youtube.com/watch?v=NpAAQwEr4y4)
-
About this tutorial video:
When you sync the 1PPS the reference is not set to INT GPS (but only INT)
Does that mean it just uses the gps anyway like INT GPS, or should I understand that differently?
-
When you sync the 1PPS the reference is not set to INT GPS (but only INT)
Does that mean it just uses the gps anyway like INT GPS, or should I understand that differently?
There is always an internal time base.
When on INT it is defined by the frequency of the internal oscillator. When on INT GPS it is linked to the GPS PPS.
In the demo I did not want to start the GPSDO as I plan to do a video on the GPSDO
The purpose of the ALIGNED PPS is to have an output pulse that is as close as possible to top of second of the currently used timebase.
Of course when synced with the GPS PPS you can use the ALAGNED PPS as an actual clock tick.
-
I understand it's just that I assumed that the INT GPS setting is a requirement for PPS to work (because it needs GPS)
-
I understand it's just that I assumed that the INT GPS setting is a requirement for PPS to work (because it needs GPS)
The PPS output is linked to the internal notion of time which is like an internal clock. If it is running on random time the PPS is synced with the top of second of the internal random time, but when you have synced the internal notion of time with GPS time the clock will stay "on time" (to some extent) even when the internal clock is no longer actively kept in sync with gps time.
-
Are there any measurements of the internal oscillator stability when not supplying GPS or an external reference signal?
[attachimg=1]
The input is my Leo Bodnar GPS at 10MHz. This ran for several days in my garage where the temperature varied from 12.2C to 29.4C.
-
The range of 7 kHz is suspicious as the maximum tuning range of the TCXO is 200 Hz.
Can I suggest the following:
- Update to the latest FW. This will show the AEDEV at tau = 1 s instead of the RANGE
- Repeat the measurement with CHART/STYLE/FOREVER AVERAGE selected for TRACE 1. This can show a multi day plot by averaging the compressed data, great for seeing a trend. Or you can use FOREVER MIN_MAX if you keep all the noise
- Check if the first measurements are correct. Best is to always hit RESTART for a row doing statistics when the first few measurements do not make sense. E.g. the first ADEV should be below 5e-10
-
Show video demonstrating how to enable the tinyGTC GPSDO function and how to check the locking using either a separate GPS module or another tinyGTC with GPSDO enabled
https://www.youtube.com/watch?v=gVTbCdV-NWM (https://www.youtube.com/watch?v=gVTbCdV-NWM)
-
A tinyGTC set to 10 MHz output at 19 degrees C cooled down to -15 degrees C shifted less than 2 Hz
7 kHz range is impossible.
-
A tinyGTC set to 10 MHz output at 19 degrees C cooled down to -15 degrees C shifted less than 2 Hz
7 kHz range is impossible.
It is probably due to EMI / external interference. The stddev seems to be 22Hz so 7kHz is definitely an outlier. The problem with min/max measurements is that they'll catch any outlier and keep that value. What would be interesting is to capture min/max as pairs of results based on 1, 2 and 3 sigma so min/max also have some statistical relevance due to getting rid of false measurements. But this is getting close to a histogram.
-
The tinyGTC does support histogram
But better would be style forever min-max as this shows the min and max per time bucket covering an unlimited time
-
Frequency drift of tinyGTC with 10 MHz output starting at -18 degrees Celsius and ending after 4 hours at 19 degrees Celsius.
Total range is 2.287 Hz
-
Doing a similar measurement I think I found another small error: the time in the lower left corner of the chart was about twice the real value. When I tried to reproduce it and take screenshots everything was correct. It took me some time to reproduce what I had done initially. I had started a chart in LAST style and then had switched to FOREVER MIN_MAX before reaching the 400s limit. At 400s the time in the chart switched to 800s and from there continued to count too fast.
-
Exactly a factor 2 too fast?
-
Almost exactly, sometimes it lags a bit behind as in my last picture with a displayed value of 1.79 kiloseconds after n=900 frequency measurements at a gate time of 1s.
When starting in LAST style with a limit of 60 things seem to get even more weird when switching to FOREVER MIN_MAX. Perhaps it would be better to restart the trace when the style is switched.
-
I could not get my ZC407 to normal counting in the chart yesterday.
I tried it again today with differeent settings after reflashing the firmware and a clear config:
Up to 400s the dispayed time increments every second and then jumps to 800s. After that point the displayed time increments by 8 every 4 seconds. When the displayed time reaches 1ks the numer format changes and it is no longer visible by what amount it increments in each single step.
I somehow missed to see the increment by 8 yesterday as it was already getting too late for me.
-
New release: Version 1.024 should solve this problem
-
I noticed that the Time displayed on the device is not updated on time (anymore).
I added a LED to the PPS OUT (20µs via a function generator to 50ms)
In this video you can see it is missing seconds and goes out of sync.
https://www.youtube.com/shorts/8CKgj5GSjNo (https://www.youtube.com/shorts/8CKgj5GSjNo)
-
Sometimes the seconds count on the display skips a update and increments 2 seconds after skipping. Will see if this can be improved
-
Hi, interesting device. I have a 5370 and other counting equipment. I plan to purchase one when I can find it in the US.
Can you explain what you mean in the following from your site? Does that mean that the GPSDO does not process sawtooth information?
"It consists of a simple GPS module that can receive GPS and BDS constellations only and delivers a 1 PPS as timing output without sawtooth correction info being available."
Thank you.
Jerry
-
Does that mean that the GPSDO does not process sawtooth information?
Correct. To fit the total cost target sawtooth correction was not possible.
Adding a true timing enabled GPS module would double the price.
-
randl.com
-
when do you think randl.com will have them? not listed that I see.
thanks
-
https://www2.randl.com/index.php?main_page=advanced_search_result&search_in_description=1&keyword=Tinygtc (https://www2.randl.com/index.php?main_page=advanced_search_result&search_in_description=1&keyword=Tinygtc)
-
https://www2.randl.com/index.php?main_page=advanced_search_result&search_in_description=1&keyword=Tinygtc (https://www2.randl.com/index.php?main_page=advanced_search_result&search_in_description=1&keyword=Tinygtc)
Thanks, the title above the unit said TinySA, that's what threw me off. I backordered one of the 12G units.
In the write-up, you mentioned about switching the TDC used. I've used the 7200s and wondering what you ended-up selecting. The 7200 had 35ps stdev and a resolution of 55ps listed in the datasheet. I assume the chip you used exceeded that? I've not looked for a replacement in a while.
thanks
-
Phase data calculated from single shot timestamp every 100 ms using 10 MHz input
-
This is a bit more challenging measurement 10.000000001 MHz so a full phase rotation every 1000 s
The histogram below shows the quantization and the stdev.
The phase plot shows both the timestamp based measurement and the phase based measurement. The later uses multiple timestamps and linear regression to enhance accuracy
You can clearly the the about 20 ps phase pulling repeating every 1000 s as I did not use ground loop isolation transformers.
-
Can you show a picture of the setup for the measurement above? Also wondering what you are using for a reference oscillator.
-
No picture needed.
Put a SMA Tee on REF and connect one side of the Tee to A and the other side to your source.
The quality of the source is not very important as long as the phase noise is not too bad as it is use for both the REF and A inputs, I'm using the REF output of a signal generator so you get the 10 MHz needed for the REF input
-
Sorry, the 10000000.001 Hz variant uses the output of a SG into A and REF out of the SG into REF
-
Hi, do you have a residual noise floor test histogram with an ADEV as well? Just ref split to the input? Or just upload the .tim file.
I'd like to compare it a 5370B that I have. I ordered the 12g model, waiting for it to get here as it is backordered at randl.
Lots of testing has been done with the 5370s, I almost hate to get mine out as that usually kicks off months of testing. I changed the processor, upgraded the voltage reference, etc., in my 5370b.
https://www.febo.com/pages/hp5370b/ (https://www.febo.com/pages/hp5370b/)
Thanks
Jerry
-
Attached the .tim files of both the phase and timestamp based noise floor measurements at a 1 mHz frequency offset and a phase measurement base .tim file with REF = A input
Let me know if you some something else
-
Just need my unit now... please send a 12g to randl with my name on it. :-DMM
Thanks for the data.
Jerry
-
I'm waiting for my 12G device to get here to do some testing. I plotted the .tim file against the HP5370B data I have and that is known. I also have the DMTD reference line on the plot, though achieving that line plotted by Bill Riley escaped me. I had problems with my DMTD board and cross-talk from 10Mhz signals across the room caused a sine wave type error down in the -13 to -14 range, drove me crazy. I'm amazed that the Tiny can get so close to the 5370B.
-
I like this item. :)
It's already ordered.
-
Erikka,
The data in the .tim files you posted, is that raw data coming off of the timer(s), right out of the GTC into timelab, or is it averaged/processed by the device?
-
The timestamp data is raw data coming from the HW
In below phase plot you can see in blue a the timestamps of 10 MHz with 10 Hz, 0.5 degrees phase modulation. +/- 130 ps
and in pink the same 10 MHz without phase modulation
Gate time was 10 ms
-
Impressive. Now you just need to create a phase noise test set. I am dying to test it myself against my 5370B. Your .tim files are almost identical to the 5370 which even today, is still a great performer.
Two more questions:
1) Is the performance of the 7G and 12G versions the same under 5Ghz? Most of my testing would be using ultra low noise oscillators in the 26Mhz range but I'm curious if the noise baselines are the same for the two versions of the unit, and if they differ, at what frequency does it begin?
2) Given R and L still doesn't have the 12G version, if I purchase the 7g version, can it be upgraded?
Thanks again.
Jerry
-
Impressive. Now you just need to create a phase noise test set.
I did, but without correlation yet and the very high performance OCXO required as baseline would be rather expensive when bought new so it will not become a product.
With correlation you need the memory and processing power of a PC
1) Is the performance of the 7G and 12G versions the same under 5Ghz? Most of my testing would be using ultra low noise oscillators in the 26Mhz range but I'm curious if the noise baselines are the same for the two versions of the unit, and if they differ, at what frequency does it begin?
Below 5 GHz performance is identical
2) Given R and L still doesn't have the 12G version, if I purchase the 7g version, can it be upgraded?
No
-
Phase noise is expensive even with a DIY test setup like I have that was designed by Andrew Holme http://www.aholme.co.uk/PhaseNoise/Main.htm (http://www.aholme.co.uk/PhaseNoise/Main.htm) It uses an FPGA, external ADC, correlation in the attached PC and has an amazing noise floor. Dual, low noise reference oscillators cost a bundle even used. Then you need filters, power splitters, good cables and connectors, and special power supplies.
Thanks for the additional input.
Jerry
-
Will get mine tomorrow :popcorn: :horse:
-
I'm having trouble with the firmware update!
No problem with TinySA Ultra. Just installed the latest update.
But not with the TinyGTC.
In normal operation, the device shows 3 COM ports!?
None are available in dial-up mode, except for the AT32....
Maybe someone knows more?
Strange.
-
I had the same problem, It is a known problem. Use a simple USB 1.1 or 2.0 hub to make it work.
-
I had the same problem, It is a known problem. Use a simple USB 1.1 or 2.0 hub to make it work.
I've already tried it. Unfortunately, the same thing happens.
Should I try it on a different PC and USB 2.0?
-
None are available in dial-up mode, except for the AT32....
This is normal behavior
In DFU mode the device runs a completely different FW showing only the DFU port.
-
None are available in dial-up mode, except for the AT32....
This is normal behavior
In DFU mode the device runs a completely different FW showing only the DFU port.
The program ( STM32C...)isn't working on TinyGTC!?
It can't find any connections...
No problem with the TinySA Ultra
(http://)
-
Follow guidelines published here: https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.FirmwareUpdate (https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.FirmwareUpdate)
-
Follow guidelines published here: https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.FirmwareUpdate (https://www.tinydevices.org/wiki/pmwiki.php?n=TinyGTC.FirmwareUpdate)
So, it worked!
With Buddy 0.2...
Seek and ye shall find.
I thought it was another Python...
-
Hi,
My GTC arrived, impressive device, will be testing it this week now that my Cs beam and phase noise analyzer are back online and running great thanks to 5065Guru.
I had a hard time updating the firmware using DFUbuddy. By the way, you need to read the readme at the start of the github to find the windows and linux releases. Anyway, my computer, a dual CPU behemoth with a ton of memory, LTO5 tape, SAS drives etc, would not power the device fully during the DFUBuddy update. I needed to connect a powered USB hub to get it to erase the unit, the first step of programming the firmware. Success followed.
Thanks for another cool device.
Jerry
-
Yes it would be better to add the actual download link instead of the project page
https://github.com/sourcebox/dfu-buddy/releases/
I see you are still updating the firmware regularly, any plans to add change notes? (even if they don't give much details)
-
what is the best mode for the input on the GTC and the best way to measure it before applying it to the device?
I haven't fully explored all the options, but what is the default input impedance? And in that mode, what is the highest voltage level, 3.5V as below? I have power meters including HP sensors, but they are all 50ohm, I believe. I would like to check the voltage input because for instance, my 5061b cesium PPS is 10V, if I remember correctly. So if the GTC can be run in 50ohm, is that the preferred safest mode and if so, then the numbers below for 50ohm would be applied. Or would it be better to measure it with a scope in high impedance mode? I can't remember how the HP power meters read PPS, if it is accurate or not. I know some of the sensors measure heating but probably not those I have.
Absolute maximum short term DC input level: between -10 and +10 Volt
Recommended maximum DC input level: between -3.5 and +3.5 Volt
Absolute maximum short term AC input level: 10 Vpp
Recommended maximum AC input level: 3.5 Vpp <--- this one and the one above seem to contradict each other.
Absolute maximum short term power into 50 ohm: +20 dBm
Recommended maximum power into 50 ohm: +10 dBm
Trigger level,: between 0 and 3.5 Volt
Input frequency A input: between 0.1 Hz and 7 GHz (for ZC407) or 12 GHz (for ZC412)
Input frequency B input: between 0.1 Hz and 350 MHz (max 50 MHz without prescaler)
Input trigger maximum time interval: No limit when measuring timestamp.
Input impedance: 1 MOhm/2 pF or 50 ohm (above 100 kHz)
Input hysteresis: 5 mV
Minimum input level in 50 Ohm: -20 dBm
Power meter range in 50 ohm input mode: -20 dBm to +20 dBm between 100 kHz and 1 GHz (see frequency dependent power level deviation chart below)
Thanks,
Jerry
-
With high frequencies like 10 MHz its best to use 50 ohm to avoid reflection that could become unstable and depending on cable flexing.
Keep the input level between 0 and +10 dBm
-
Hi,
i'm having a problem measuring the delta between two 1PPS signals. I carefully measured the 1PPS coming out of my cesium beam that started at ten volts peak until it was down below 2V peak. I have the tinygtc in 50ohm mode on that one port. The tiny gtc, though, reports it as -27dBm and it should be closer to +7dbm. The other port I can't put it in 50ohm mode because the signal isn't that strong. I don't know if this low -27dbm signal is my problem or not but i'm getting glitches that weren't there when I was using my HP 5370b. I had a hard time trying to figure out how to get the tinygtc to output the delta between the two signals, but I think I have that correct now. It tracks well for the first several hundred or so points but then it starts glitching, -10e3 or -10e4 additive glitches. I'll get a bunch of glitches and then it'll clear up and then I'll get a bunch more. It's not a phase transition. I'm using the tinygtc's internal GPS as a reference. I'm going to try using the cesium as a reference and 1PPS tomorrow.
The primary issue is with setting levels. It would be great If you made an external box that made setting levels easy. For 1PPS signals, i've used a variable resistive divider in the past but I only have one.
Thanks
Jerry
-
As specified in the wiki the power meter does not function at 1 Hz.
One extra complication is that for 1 Hz you also cannot use show levels .
You do not mention setting the trigger level.
Keep in mind the tinyGTC inputs are DC coupled and require the correct setting of the trigger level.
With a 2 V PPS between 0 and 2 V, suggest to set the trigger level at 0.5 V and no need to enable 50 ohm
To reduce large input signal you can use a 50 ohm attenuator that can handle the power
-
I had set the trigger level to .6v as that was the same level used on my 5370b. I am using attenuators, screw-on sma type, tested with high impedance inputs on my scope. I'd have to measure the high input impedance of the GTC, and the attenuators weren't as effective (mathematically) as they should have been due to the impedance mismatch. What do you mean by show levels? I don't remember seeing that in the menus. I'll have to measure the other channel, the one that is coming from Leo Bodnar's new GPSDO into the high impedance channel A. I'll have time for more testing tomorrow.
-
When using a 20 dB attenuator your get a factor 10 reduction in voltage. Your 20 Vpp source will become 2 Vpp after the 20 dB attenuator,
The tinyGTC input impedance in High-Z mode is 1 MOhm/2 pF, but with 20 dB attenuation the input impedance of the tinyGTC can be ignored.
COUNTER/SHOW LEVELS will only work for input frequencies above 10 kHz.
-
Have a look at this video: https://www.youtube.com/watch?v=05LMwBLIQMQ (https://www.youtube.com/watch?v=05LMwBLIQMQ)
-
Thanks Erikka, i found the problem and it now is plotting the ADEV right on top of the 5370b plot. Issue was the timestamp. I turned off that option on the time difference and it corrected the problem. Maybe Timelab was picking up the timestamp sometimes that leads the channel B digits sent to Timelab.
-
This video shows the tinyGTC doing measurements over its entire frequency range starting at 0.1 Hz and going up to above the specified maximum 12 GHz input frequency. Your tinyGTC is not guaranteed to reach the same maximum frequency but for sure is will surpass 12 GHz.
https://www.youtube.com/watch?v=S25D6l0kah4 (https://www.youtube.com/watch?v=S25D6l0kah4)
-
A short video showing how to measure a mechanical device without an electrical connection
https://www.youtube.com/watch?v=dOOVTutfwNc (https://www.youtube.com/watch?v=dOOVTutfwNc)
-
A short video showing how to measure a mechanical device without an electrical connection
https://www.youtube.com/watch?v=dOOVTutfwNc (https://www.youtube.com/watch?v=dOOVTutfwNc)
May I ask where you got the stand (3 legs) from?
-
Erikka: I received my TinyGTC recently. I seem to be having trouble using or recognizing a square wave from my of my rubidium oscillators. I've attached an image with it's corresponding sine wave. Wondering if you can advise me on what I did wrong or need to set up?
I also Blender modified a stand from one for the NanoVNA-H4 (aka TinyPFA) that I found on Printables from Rik HORKM (NanoVNA-H4 Stand). Since I can't seem to attach it I may have to find a different way to include or place it somewhere if someone is interested in it.
I've attached it as a Zip.. Print 2 of them and slip it over the lip on the bottom it won't slip seems okay for me for now..
-
The square wave seems symmetrical around zero volt so a zero volt trigger level should work
What settings did you use?
-
Don't know about the stand, it was a gift
-
The square wave seems symmetrical around zero volt so a zero volt trigger level should work
What settings did you use?
Source A/B (doesn't amtter), Gate 100ms, Presc: Off, 50 Ohm: checked, edge: Up, Trig Lv: 0, Show: levels, debounce: Off
-
Is there no triggering?
Does the self test succeed?
Can the tinyGTC mease its own OUT at 10 MHz?
-
Is there no triggering?
Does the self test succeed?
Can the tinyGTC mease its own OUT at 10 MHz?
Shows TOUT
Selftest: Pass:
Output: Square Wave - 10M shows it is counting it. Also attached the output from the GTC on the scope below.
Using your latest Jan 22 firmware - 1.0.63
Also cleared the config, and self tested again.
-
I fail to understand the the exact nature of the problem you mentioned. "I seem to be having trouble using or recognizing a square wave from my of my rubidium oscillators."
Please explain the problem
Is there not triggering at all? Did you check the DC levels? Was your scope AC coupled? The tinyGTC has a DC coupled input. Use COUNTER/SHOW LEVELS to see the input levels.
Is triggering bad?
-
I fail to understand the the exact nature of the problem you mentioned. "I seem to be having trouble using or recognizing a square wave from my of my rubidium oscillators."
Please explain the problem
Is there not triggering at all? Did you check the DC levels? Was your scope AC coupled? The tinyGTC has a DC coupled input. Use COUNTER/SHOW LEVELS to see the input levels.
Is triggering bad?
The scope is AC coupled to make it easier to compare and align.
When I put in this particular square wave it is not triggering at all.
Show levels:
Vmin=0.7200 Vmax=3.8672V
But that would explain it, show levels shows Vmin as 0.7200.. I changed the trigger to 2v to be on the safe side and it triggered, that little "ripple" in the square wave threw it all off. Thanks for the help!
-
Just to be safe you may consider a 3 or 6 dB attenuator between your source and the tinyGTC. This will lower the input voltages a bit and provide extra protection.