Jump to content

Chinese Timegraphers


Recommended Posts

On 1/9/2023 at 2:52 PM, Kalanag said:

The Chinese microphones to be used with a timegrapher software or app get better and cheaper! Here is what I got today. I used the free tg timegrapher software (tg-timer 0.5.0)

D62928FC-40EC-42B2-87EC-F1CF09495576.jpeg

Is anybody aware if these get sold within the EU?

Here in Denmark, if the item is shipped from outside the EU, one pays through your nose on "handling fees" and "import duties". "Handling fee" is by definition already 160DKK = €21.50. On top comes 25% VAT over the (item-value+shipping cost).

Link to comment
Share on other sites

2 hours ago, Endeavor said:

Is anybody aware if these get sold within the EU?

My item was shipped to Germany without additional tax or shipping costs.

If you are looking for a seller inside the EU you should search ebay

Edited by Kalanag
  • Thanks 1
Link to comment
Share on other sites

48 minutes ago, Kalanag said:

If you are looking for a seller inside the EU you should search ebay

In most EU countries you are better of. In DK they went bananas (€5 for a postcard stamp!).

So far, eBay.de shows mostly China and a few in the UK ........ I guess I've to wait a bit 😉

Link to comment
Share on other sites

51 minutes ago, VWatchie said:

Shopping on AliExpress my two most recent purchases were delivered from Holland without any additional charges

Perhaps I can ask AliExpress and make sure, else there is a chance that I've to pay the PostNord extortion.

  • Like 1
Link to comment
Share on other sites

Today the Aliex prices use to include your VAT. The seller paid it in advance. 

Frank

19 hours ago, xyzzy said:

I wonder how the timing accuracy is on the CM108? 

xyzzy,
your fine analysis in your post was dead-on!
CM108 come as different variants (e.g. ...B and ...AH). One uses a 12MHz quartz, the other (cheaper) uses an on-chip RC oscillator.

All recent chinese USB soundcards (with case) seem to come without own quartz and they are more or less useless for our goal. I tested several of them.  Better ones show small rate variations, others ridiculous high rate variations. None is appropriate for precision tests.

Those naked boards have the quartz visible, what is good, and still have one until ...

Frank

Link to comment
Share on other sites

13 hours ago, xyzzy said:

There is a difference.  With this device, there are two clocks.  One is in the device, it samples the audio.  For a computer, you are done here, one clock.  But with the USB device, the audio must be sent over USB.  USB is digital, there is a clock here too.  But this clock is in the computer.

Both clocks must agree: every second, 48000 audio samples.  But are two different clocks going to run at exactly the same rate?  No, they are not.  So stuff must be done about this.

And that stuff to deal with these two clocks that must be the same but aren't, it can affect timing things via audio at a rate measurable in s/d.

Fine.  We are measuring tick tocks, not Def Leppard (or Beethoven...choose your medicine).  If you are making the point that analog audio directly into the PC has better SINAD, nobody is going to argue with you.  This chip has horrible SINAD out of the box.

2023-01-11 08_03_26-CM108_DataSheet_v1.6-1.pdf - Foxit PhantomPDF.png

Link to comment
Share on other sites

3 hours ago, Endeavor said:

Perhaps I can ask AliExpress and make sure, else there is a chance that I've to pay the PostNord extortion.

Perhaps the delivery from Holland is all legit or for the present time a loophole in the EU extortion machinery 🤔 Perhaps they (the Chinese) have some method to get the stuff into Holland and then deliver from there to make it look legit. It's been pretty fast too, three to four weeks.

  • Thanks 1
Link to comment
Share on other sites

16 hours ago, xyzzy said:

Both clocks must agree: every second, 48000 audio samples.  But are two different clocks going to run at exactly the same rate?  No, they are not.  So stuff must be done about this.

And that stuff to deal with these two clocks that must be the same but aren't, it can affect timing things via audio at a rate measurable in s/d.

USB is a digital protocol, and the audio sampled by the sound card is chunked into packets suitable for USB transfer. The speed at which the transfers happens should be irrelevant to actual playback on the host system, because the sound card tells the host its sample rate (among other things) before enumeration. This Silabs application note outlines how the USB audio class works. As long as the USB connection is fast enough for audio, everything should be fine.

I would expect that the software on your PC uses information about the sample rate to infer timing information from the audio stream, not the "USB packet time of arrival" if we can call it that. In other words, the accuracy of your timing setup is entirely dependent upon the poor 12 MHz crystal in the frontend. I had a quick look at Farnell and 10 ppm seems to be the highest frequency tolerance (at 25°C) you can reasonably get. Using one of those, you'd be 0.86 s/d off in the worst case (10e-6*86400=0.864). You can of course get better oscillators with internal temperature compensation and all the bells and whistles, but I wouldn't expect those to be found in Chinese clones like these. It would be interesting to learn which oscillator they use in these.

Edited by fellerts
  • Like 2
Link to comment
Share on other sites

10 hours ago, LittleWatchShop said:

If you are making the point that analog audio directly into the PC has better SINAD, nobody is going to argue with you. 

No, absolutely not SINAD.  I doubt that makes a lot of difference for a timegrapher unless it's absurdly bad.  I'm just considering issues that directly relate to timing.  I know these will cause real errors because I've written timegrapher software.

Take a look at this NXP app note, https://www.nxp.com/docs/en/application-note/AN13364.pdf §2.7 Table 5.  The way the chip maintains synchronization between the audio clock and USB is to switch the audio sample rate between approx -700 s/d, 0 s/d, and +700 s/d.  

Now in my opinion, the chip in that app note is being (ab)used to do something it wasn't really designed for and is very bad at it.  One would hope a purpose designed chip like the CM108 isn't that bad!  But if it were, you would see errors in the timegrapher of ±700 s/d.  That isn't nitpicking signal quality, that's totally useless.

8 hours ago, fellerts said:

This Silabs application note outlines how the USB audio class works. As long as the USB connection is fast enough for audio, everything should be fine.

That's not quite how it usually works.  Due to the design of the isochronous transfers, the USB device might need to send data at *exactly* the rate the host will control via the USB clock.  In that app note, look at §4.1.12, the endpoint is defined as asynchronous mode, which lets the audio clock be independent (asynchronous to) the USB clock.  They further mention synchronous and adaptive mode. The CM108 mentioned previously doesn't use asynchronous mode, rather it uses adaptive mode, "where the stream’s clock source varies according to the buffer requirements of the stream."  That variation is a big problem for using the stream's clock as the source of the timing information (which is how timegrapher software works).

9 hours ago, fellerts said:

I would expect that the software on your PC uses information about the sample rate to infer timing information from the audio stream, not the "USB packet time of arrival" if we can call it that. In other words, the accuracy of your timing setup is entirely dependent upon the poor 12 MHz crystal in the frontend.

Quite right, if we used asynchronous mode.  However, we can use USB packet arrival time to calibrate that poor 12 MHz crystal, if the PC can timestamp the packet arrival with something accurate, like its own clock that has been adjusted via a GPS or with NTP to better than the crystal's 10 ppm.

But that calibration only works for long term errors in the crystal's frequency, i.e. this crystal is always 7 ppm slow.   It doesn't work for a short term change in frequency due to the crystal's changing temperature.  And the very short term, and possibly very large, changes in frequency due to USB adaptive mode? It absolutely does not work for that.

21 hours ago, JohnR725 said:

So simplistically are saying that using a USB interface for timing is out?

Not exactly.  The CM108, especially one with a poor crystal or an internal RC oscillator, might be out.  But if I were designing a really good, but still a reasonable product, timegrapher microphone, I'd use USB.  The problem with analog is you are dependent on the PC/labtop/etc having good audio.  Some are perfectly ok, but some aren't.  By using USB, you remove the PC's audio from the equation and can be sure it's done right by doing it yourself.  But you have to do it right.  So I'd use asynchronous mode and not adaptive mode.  Such USB audio hardware certainly exists.

I'd also be tempted to try Bluetooth. Yes, batteries, a pain.  But no cables and then have a tablet with the timegrapher display, also no cables, wouldn't that be nice?  You could design the microphone to work as completely functional movement holder.

  • Like 1
Link to comment
Share on other sites

7 hours ago, xyzzy said:

The CM108 mentioned previously doesn't use asynchronous mode, rather it uses adaptive mode, "where the stream’s clock source varies according to the buffer requirements of the stream." 

You're right, I missed that detail in the datasheet. I wonder why they chose to use adaptive mode?

7 hours ago, xyzzy said:

I'd also be tempted to try Bluetooth. Yes, batteries, a pain.  But no cables and then have a tablet with the timegrapher display, also no cables, wouldn't that be nice?  You could design the microphone to work as completely functional movement holder.

I would be inclined to use Bluetooth for this as well. Nordic Semiconductor (whose HQ I could throw a stone at right now!) recently released the nRF5340 which would be an absolute perfect fit for this application. It has the necessary low-jitter hardware for interfacing with a microphone and is tailored for LE audio. LE audio hasn't seen that much adoption yet but I think that will change in the next couple of years. 

This would be a very fun project...

Edited by fellerts
quoted wrong section
Link to comment
Share on other sites

13 hours ago, fellerts said:

I would be inclined to use Bluetooth for this as well. Nordic Semiconductor (whose HQ I could throw a stone at right now!) recently released the nRF5340 which would be an absolute perfect fit for this application. It has the necessary low-jitter hardware for interfacing with a microphone and is tailored for LE audio. LE audio hasn't seen that much adoption yet but I think that will change in the next couple of years. 

I saw that chip too.  Has I2S for interface to an external ADC.  Maybe the internal 12 bit 200 ksps ADC can be used?  But lack of adoption of LE audio is an issue.  I haven't look at the LE audio much yet, for instance how does it do timing?  With A2DP you have device generated RTP timestamps.  But there just isn't much available for Bt classic.  Venerable ESP32 has it and I2S.  Also a 2Msps ADC when Wifi is disabled, but I think it doesn't have DMA and so is probably not useful.

Nice option could be new Bouffalo Lab BL616/BL618.  These have support for Bt Classic/LE and a built in audio ADC.  The Sipeed M0S module using it is supposed to be out soon.  Specs looks great for the problem.  < $2, built in audio ADC, Bt Class, Wifi, USB.  But the Bouffalo is one of those Chinese silicon companies that don't do documentation, so I dread trying to make something with their chips.

Link to comment
Share on other sites

  • 2 weeks later...
On 1/10/2023 at 8:18 AM, Kalanag said:

May I ask why you don‘t just use the oscilloscope function of Tg0.5.0 for free?6740958E-6FE8-4C11-B04D-12346BDAD250.thumb.jpeg.aef764d4198bcffbd9029482a2c59a66.jpeg

Well my microphone from Aliexpress has arrived, and I downloaded Tg 0.5.0 but can't get it to work.

I have occasionally got a plot like the above pic, but then it all goes yellow, displays wrong values, and stops.  I can't calibrate it as it won't detect a quartz watch.

image.png.db69bb4cfc4b6ee56897b69a93b0dd94.png

As there's no instruction manual, I'm at a loss to know what to do. I guess it's the sound card settings?

The device comes up as 'C media USB headphone set'. What should the volumes be?

image.thumb.png.499843620d258d953c8604033318d7b2.png

 

I tried Watch-o-scope, get a plot like this :

image.thumb.png.64e1bcba3f4fd25b4e4053ab84e1a8bf.png

But then it says it can't detect a consistent beat, or I get random dots. 

image.png.ffbf7dfb9f7707130a681a5615ce9440.png

 

 

 

Edited by mikepilk
Link to comment
Share on other sites

If the watch symbol in the upper left corner of the tg 0.5.0 screen is red the signal is weak.

No idea what might be wrong.

Here is my sound device screen. The microphone gets adjusted continuously with a kind of automatic gain control.

4C43762E-714B-4117-87FB-A3689EEFE371.jpeg

Edited by Kalanag
Link to comment
Share on other sites

37 minutes ago, Kalanag said:

If the watch symbol in the upper left corner of the tg 0.5.0 screen is red the signal is weak.

No idea what might be wrong.

Here is my sound device screen. The microphone gets adjusted continuously with a kind of automatic gain control.

 

I'm using an old pocket watch, I can hear the tick across the room, so I don't know what's wrong.

And now the blue LED doesn't light when I plug it in, and it isn't recognised by Windows. I think I might have bought a dud.

My first, (and probably last) purchase from AliExpress !

Edited by mikepilk
Link to comment
Share on other sites

1 hour ago, mikepilk said:

blue LED doesn't light when I plug it in,

As usual there seemed to be variations my LED is red.

 

3 hours ago, mikepilk said:

The device comes up as 'C media USB headphone set'. What should the volumes be?

The adapter they used is capable being a microphone and an audio out. But you only want to use it as a microphone because there is no audio out even though it's in the chip. You want to use your computer's audio if you want to listen to the sounds. Which might be helpful now if you're having a problem you can see how it sounds just be really careful with the volume level. Because the way this thing is designed it's extremely sensitive to everything so if you have audio how do your laptop for instance it will pick that up so it has to be kept really really low is not the best microphone because of that but it's relatively cheap. I like some sound dampening inside the case would help with external audio other than the watch.

Then looking at your oscilloscope image I circled something that shouldn't be there at all? Plus the rest your oscilloscope it looks like your gain is turned up way too high. Which asked and noticed on my microphone it seemed to be the signal level was much higher than what I normally had used in the past for anything that I had.

 

 

 

oscilloscope bad.JPG

Link to comment
Share on other sites

It's stopped working. Took it apart and can't see anything wrong with any of the soldered joints. I'm an aero, not electronics engineer, so don't know how to test anything.

Pity, I was looking forward to using this.

I guess if you pay peanuts, you shouldn't be surprised if you get useless crap. 

  

Link to comment
Share on other sites

Odd the cmedia chip would just die.

Gain being too high looks like a possible problem.  My new version of tg has a signal meter to help with setting the gain. It could audio be background noise. Maybe something mechanical that is vibrating the desk? The spectrogram in my TG version also helps identify noise. 

Link to comment
Share on other sites

7 hours ago, xyzzy said:

Odd the cmedia chip would just die.

Gain being too high looks like a possible problem.  My new version of tg has a signal meter to help with setting the gain. It could audio be background noise. Maybe something mechanical that is vibrating the desk? The spectrogram in my TG version also helps identify noise. 

The output on Watch-0-scope did suggest that the gain was too high.

The connector is OK, power is getting to the board - I put a meter on C11,C12. Beyond that, I've no idea !

1.thumb.jpg.9510ca7b2e937c843bad230f0156636d.jpg

Link to comment
Share on other sites

2 hours ago, mikepilk said:

Beyond that, I've no idea !

One of things I find interesting is the disk wires look to be relatively stiff? Whereas ideally they should be very fine wire with perhaps a little coil in them so that their free to move around. At least all the ones I've ever seen in a timing machine there are always very tiny wires. Oh and then your volume adjustment is in a different position then the previous one we have a picture of.

 

Link to comment
Share on other sites

4 hours ago, mikepilk said:

The connector is OK, power is getting to the board - I put a meter on C11,C12. Beyond that, I've no idea !

Flip the board over and take another picture and post.

That potentiometer at the far right of the board adjusts the gain as shown in this schematic.

image.png.dcdb1ae85dc467e12983e43e10590337.png

CM108 has gain control internally, but it is under software control.  There is an edge-triggered input for volume up and volume down but those pins do not appear to be connected.

Edited by LittleWatchShop
Link to comment
Share on other sites

2 hours ago, LittleWatchShop said:

Flip the board over and take another picture and post.

That potentiometer at the far right of the board adjusts the gain as shown in this schematic.

CM108 has gain control internally, but it is under software control.  There is an edge-triggered input for volume up and volume down but those pins do not appear to be connected.

I guessed that's what the pot does.

Problem is, it's dead, bereft of life, it's kicked the bucket. 

When I plug the USB in, the computer doesn't see it, and the LED doesn't light. 

All that way from China, and it worked (?) for about an hour.

I guess it must be the main chip that's failed - so there's nothing I can do ?

image.thumb.png.3289d939aec0f5993ba62f9580aa1683.png

Edited by mikepilk
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.




×
×
  • Create New...