New 2ch pocket DSO+SG - Zeeweii DSO2512G

[jebem]

Hi Jebem!

despite the average/poor input sensitivity of 10mV/div that in fact points to the probable design choice in use.

10mV/div is just fine for the lowest price range.
There are more expensive devices from e.g. Fnirsi that have 20mV/div or even 50mV/div.
...
I didn't buy it because of any bandwidth or sampling rate specifications, but because two channels can definitely do more than one and it was the cheapest two channel device.
If it can show a nice square wave on both channels at, say 4MHz, I'm happy for that price.
...
It is small and handy, runs on battery and has a rubber cover, so it does not fall down when you put it somewhere in the engine compartment, for example. 

Yap, I also was looking for a cheap scope that had the highest input sensitivity and found this one.
Also it features fair to good bandwidth.
And the price was just right. 


 

[tatel]

I will dismantle my unit after it arrives to have a more detailed look and try some sort of reverse engineering, as I usually do for all my acquired devices, specially when the manufacturer do not release service documentation.

Very good news. I'll follow it.

[Aldo22]

I will probably never understand what exactly is going on here.
Today I got the attenuators for the TinySA and two BNC<->SMA connectors.

You may have read about the ugly signal I saw when connecting the Si5351 directly via coax and load resistor.

With the -10dB attenuator and the coax, I see a pretty stable square wave at almost 10MHz.
It looks even more stable than with the probes.
Hmmm...

Maybe I should buy a real oscilloscope after all, just to see if the problem is with the DSO2512G. 

[jebem]

I will probably never understand what exactly is going on here.
Today I got the attenuators for the TinySA and two BNC<->SMA connectors.

You may have read about the ugly signal I saw when connecting the Si5351 directly via coax and load resistor.

With the -10dB attenuator and the coax, I see a pretty stable square wave at almost 10MHz.
It looks even more stable than with the probes.
Hmmm...

Maybe I should buy a real oscilloscope after all, just to see if the problem is with the DSO2512G. 

Hi Aldo,

At a first glance, it seems this scope behaves better on small signals than on higher levels.
Surely a poor design of the vertical amplifier/attenuator stage can be causing it (or a component failure).

If I am understanding correctly, it looks that there is a behavior pattern for the same square wave frequency signal:

- When using the scope probe (I assume you used in X10), or when using the 10dB attenuator, both connections are giving a better square wave shape (adequate and correct harmonics reaching the scope vertical amplifier);
- When using a direct coax cable and a load resistor (what value, was it lowish like 50R?), it results in distortion of the square wave.

Before assuming this is a scope fault, other possible reason is signal source distortion based on the imposed load at that square wave frequency.
If that is the case, then the scope is just reading what is being supplied at the input of the vertical amplifier.

If I am not mistaken:
- A X10 probe puts a typical 9MR in series in the transmission line and the scope input puts 1MR in parallel;
- an 10dBv attenuator also uses a series resistor (like 800KR?)  with a parallel resistor (like 3KR?) at its output, plus the scope input 1MR parallel resistor.
In both cases the signal source sees a lighter load than the direct line with load termination.

When using a direct coax cable with load termination, I would only use low load values if the signal generator output impedance is guaranteed to present similar low values.

As you say, another (and better) scope would be required to analyse the issue.
Meanwhile, you could try again the direct connection coax but without the load resistor, to compare the results.

P.S..- Here is the scope that I should have bought instead of my OWON 1102:
The RIGOL DS1202Z-E, 2-Channels, 200MHz,  1 GSa/s (single channel), 500 uV/div Vertical max. sensitivity.
https://www.rigol.eu/Public/Uploads/uploadfile/files/20221124/20221124215236_637f77244604d.pdf





 

[wasedadoc]

an 10dBv attenuator also uses a series resistor (like 800KR?)  with a parallel resistor (like 3KR?) at its output, plus the scope input 1MR parallel resistor.
In both cases the signal source sees a lighter load than the direct line with load termination.

No.  Reply 151 has the link to the attenuators Aldo purchased.  They are 50 Ohm input and output impedance.

[Aldo22]

Hi jebem and wasedadoc

Thank you!

Yes, the attenuator is this and the load resistor is this one.

I made another observation.
When I switch the vertical sensitivity from 200mV to 500mV, I hear a relay click.
After that, the displayed signal looks more nervous and distorted.
So it seems that at lower vertical sensitivity something is switched on or off, so that it shows a nicer signal at 10-200mV vertical sensitivity than at 500mV.
Maybe you know what this could be about.

So it could be that the attenuator is just bringing the input voltage into that range.

It's not quite that simple, though.
If I put the load resistor (LR) before the attenuator, the signal also gets a uglier again.

So it is also the LR, which makes the signal "ugly" in this setup (not with the probes).
In the first image (attachment) you can see 2MHz at 3.33 Volt PkPk directly connected from the Si5351 module via coax without LR and attenuator!
In this frequency range still no problem, even with 500mV sensitivity.
The second image shows exactly the same, but with LR connected in between.
Already in this snapshot you can see the problem. On the screen it looks very unstable, even if it has lower voltage than the direct connection.

Which signal is more "honest" I can't say, because I don't have a better oscilloscope.
I only have the TinySA as another instrument, but that can also only be fed with max. 700mV, so I would have to feed it with attenuator, which already provides a quiet picture on the oscilloscope.
So I can't feed the "ugly" input to it at all.

Sometimes I think this oscilloscope is made to measure something in the Arduino range with the probes. Thereby you don't come across these questions at all. 

Thank you!

[tatel]

Maybe I should buy a real oscilloscope after all, just to see if the problem is with the DSO2512G. 

Maybe you should look at other oscilloscopes, just to see how they behave, then draw conclusions about DSO2512G

Don't get me wrong, I'm not saying this device is as bad as radiolistener thinks, but it isn't as good as you seem to think, either.

Please look how this recently purchased Hameg HM303-3 behaves. It's a 30 MHz oscilloscope. Square wave is said to be good up to 3 MHz. This device has nuances of his own, but, hey, it doesn't look as high on amphetamines. I can look at the trace and see things, just because it's a nice stable trace.

https://youtu.be/THvpo1G16mE

[Aldo22]

Hi tatel!

Don't get me wrong, I'm not saying this device is as bad as radiolistener thinks, but it isn't as good as you seem to think, either.

I never said it was "that good."
I'm just showing screenshots.
If someone says it can't measure 120MHz, I show a screenshot of it doing so.

That doesn't mean I don't realize it's limited.

The problem is imho more the other way around.
Some seem to be comparing it to a $500+ device just because it has 120MHz/500M in the spec sheet.
That's ridiculous.
If you pimp a Fiat/Seat 600 to go 250km/h, that doesn't make it a Rolls Royce.
But if it can do that, it's still a fact. Maybe you wouldn't want to sit in it at 250. 

The Zeeweii is nowhere near the quality of, say, your GW Instek, no sane person would want to compare them.

But for $80 is quite usable and it is fun.
I think the guys have made an effort to bring a cheap, easy to use device/toy to the market.
I don't regret buying it.

Please look how this recently purchased Hameg HM303-3 behaves. It's a 30 MHz oscilloscope. Square wave is said to be good up to 3 MHz. This device has nuances of his own, but, hey, it doesn't look as high on amphetamines. I can look at the trace and see things, just because it's a nice stable trace.

It looks nice. Congratulations!
Looking at the CRT may calm the nerves, but to be honest I don't see any nuances above 1MHz there. 

[wasedadoc]

If I put the load resistor (LR) before the attenuator, the signal also gets a uglier again.

That is not a valid configuration.  If you do that everything is mismatched.  Amplitudes will be incorrect and reflections in the cable will distort the waveform.

[Aldo22]

If I put the load resistor (LR) before the attenuator, the signal also gets a uglier again.

That is not a valid configuration.  If you do that everything is mismatched.  Amplitudes will be incorrect and reflections in the cable will distort the waveform.

That's what I suspected, too.
But the valid configuration (I assume coax plus 50Ω LR is valid) shows an even uglier signal (second screenshot above).

Thank you!

[4thDoctorWhoFan]

Thought I would join in the fun. 

I'm using a HP 3325B as my square wave source.
My scope has Firmware: v1.2.9C

1st Image:  Direct cable connection with no load.  Source = 2Mhz @ 1.5Vpp.  I'm using 1.5Vpp because with no load the scope will show a 3Vpp signal.

2nd Image:  Source is same setting but now with a 50 ohm load and the scope measures properly at 1.5Vpp.

3rd Image:  Increased source to 3Vpp so scope will show 3Vpp with 50 ohm load.

[Aldo22]

@4thDoctorWhoFan: Thanks, interesting!
#1 looks almost exactly as mine, #2 looks much better than mine.

Possible differences are:
- Signal Generator: I am only seeing these problems with the Si5351 module, the internal SG looks OK @2MHz (attachment).
- Load Resistor?
- Firmware (v1.2.8C here) or even Hardware.

Attached are the screenshots of the internal SG and coax. The first with LR, the second without.
The one with LR looks better, but I don't understand the voltage.

Edit: Btw. I soldered a small connector for the internal SG (3rd picture).

[4thDoctorWhoFan]

Here are the results of using the internal SG on my pocket scope.  I also checked the waveforms and voltages on my HP 54616C scope for comparison and got about the same waveforms and voltages as shown below.

1st image is with no load
2nd image is with 50 ohm load.

[Aldo22]

@4thDoctorWhoFan: Did you get my PM reply? It's the first time I'm sending a PM here, so I'm not sure if it works. I've had problems with this on other forums.

Regarding the attenuator and the dummy load, I experimented a bit.
I get the best signal from the Si5351 module with the coax cable and the -10dB attenuator directly at the input of the DSO2512G.
This shows a quite stable signal over the entire frequency range.

The signal looks a bit fat at higher frequencies. It's not going to win a beauty contest, but for all practical purposes, it undeniably shows the square wave up to 40MHz.
I mean a $80 toyscope that can display a 40MHz square wave?
Come on, that's amazing. 

[jebem]

I start to believe that this ZEEWEII DSO2512G scope can be a cheaper clone of the more expensive OWON HDS272.

The OWON HDS272 specs are 70MHz bandwidth, 250MSPS, 10mV/div max. sens.
Based on what I have seen published by several owners, the above specs are more realistic, and more aligned with what the DSO2512G is capable to do (despite the alleged 500MSPS using a mysterious unidentified IC ).

OWON uses 500MSPS but on the HDS2102 (100MHz bandwidth), and 1GSPS on the HDS2202 (200MHz bandwidth).
Again, more acceptable specs for the claimed sample  rate.

There are now several hardware revisions of this scope, under several marketing brands (SigPeak, ZEEWEII, etc., probably coming the same manufacturing plant).
The later hardware revision (v9.15 ?) should improve things with the help of a better F/W, otherwise I don't see any point of having the trouble to do so.

[jebem]

I get the best signal from the Si5351 module with the coax cable and the -10dB attenuator directly at the input of the DSO2512G.
This shows a quite stable signal over the entire frequency range.

I wonder if the Si5351 circuit implementation in use really is able to support 50R loads on a TL for such high signal levels seen in the screenshots (100 to 200mV).
Is the circuit using a dedicated output driver, or is it just using the Si5351 direct clock output buffered lines?

The Si5351 datasheets specifies a 50R output impedance *when* running the VDDO supply at 3.3V (this is the Output Buffer Voltage at pin 7 on a 10-MSOP package, not the VDD supply at pin 1).

But the Output Buffer Supply Current per clock (IDDOx) is specified at jut 2.2mA with a very low CL of 5pF.
This current seems too low to be able to cop a large frequency spectrum of a square wave over a 50R load, unless the signal output is decreased drastically to a few milivolt.
Can you programmatically decrease the output signal to, say, 20mVpp or 50mVpp and see if the square wave improves over the 50R load?

The datasheet application examples do not explicitly include loads of 50R. The closer example would be  HCSL compatible outputs, but the termination load is done using a series 511R followed by a parallel 240R.

[Aldo22]

Hi Jebem!

I start to believe that this ZEEWEII DSO2512G scope can be a cheaper clone of the more expensive OWON HDS272.

I don't think so. I don't see any connection between Zeeweii and Owon.
From hearsay I have that it is rather related to Fnirsi or that some people from Fnirsi went to Zeeweii.
But I don't know anything for sure.

From what I see in videos, it does not look technically related.
The HDS272 shows a smoother signal, but the frequency measurement drops out at 85MHz, while the Zeeweii can still measure the 120 MHz in OC mode and 96MHz in normal mode.
Can you show a picture of the HDS272 with a 30 or 40MHz square wave? I would be interested in that.

The device does exist under several marketing brands, but Zeeweii/sigPeak is the manufacturer.

From their website:

05/13/2021 Due to trademark disputes, sigPeak were abandoned and now we are zeeweii.

[Aldo22]

I wonder if the Si5351 circuit implementation in use really is able to support 50R loads on a TL for such high signal levels seen in the screenshots (100 to 200mV).

I've wondered the same thing, but I don't understand enough about it.
It is after all a ready made module. https://alexnld.com/wp-content/uploads/2018/12/de599e1e-1bd6-4817-9801-aec8e95ffdd2.jpg
And no, I don't know of any way to control the amplitude of the Si5351 programmatically.

Thank you!

[wasedadoc]

I get the best signal from the Si5351 module with the coax cable and the -10dB attenuator directly at the input of the DSO2512G.
This shows a quite stable signal over the entire frequency range.

I wonder if the Si5351 circuit implementation in use really is able to support 50R loads on a TL for such high signal levels seen in the screenshots (100 to 200mV).
Is the circuit using a dedicated output driver, or is it just using the Si5351 direct clock output buffered lines?

The Si5351 datasheets specifies a 50R output impedance *when* running the VDDO supply at 3.3V (this is the Output Buffer Voltage at pin 7 on a 10-MSOP package, not the VDD supply at pin 1).

But the Output Buffer Supply Current per clock (IDDOx) is specified at jut 2.2mA with a very low CL of 5pF.
This current seems too low to be able to cop a large frequency spectrum of a square wave over a 50R load, unless the signal output is decreased drastically to a few milivolt.
Can you programmatically decrease the output signal to, say, 20mVpp or 50mVpp and see if the square wave improves over the 50R load?

The datasheet application examples do not explicitly include loads of 50R. The closer example would be  HCSL compatible outputs, but the termination load is done using a series 511R followed by a parallel 240R.

Taking at face value the amplitudes of 2.51 unloaded and 0.717 into 50 Ohms gives the output impedance of the 5351 module as 125 Ohms.

[Aldo22]

Hi wasedadoc

Taking at face value the amplitudes of 2.51 unloaded and 0.717 into 50 Ohms gives the output impedance of the 5351 module as 125 Ohms.

You refer to 4thDoctorWhoFan's measurements, but he didn't have the problems!
I measure 3.33V PkPk without LR and 2.51V PkPk with LR, but heavily distorted.
https://www.eevblog.com/forum/testgear/new-2ch-pocket-dsosg-sigpeak-dso2512g/msg4859288/#msg4859288

Thank you!

[jebem]

I wonder if the Si5351 circuit implementation in use really is able to support 50R loads on a TL for such high signal levels seen in the screenshots (100 to 200mV).

It is after all a ready made module. https://alexnld.com/wp-content/uploads/2018/12/de599e1e-1bd6-4817-9801-aec8e95ffdd2.jpg
And no, I don't know of any way to control the amplitude of the Si5351 programmatically.

I had a closer look to the Si5351 specs and the AN619 Application Note doc.
I could not find any register setting that could change the output signal level either.

However there is one parameter that could be changed to increase the output current ability as long as it is only enabled in one of the outputs at a time to not overload the IC.
This envolves the edition of the control software program that you are using, to do a minor change of one configuration setting.

If the clock output is taken from Clock0 line (for the other two lines pls check the AN619):

Register 16. CLK0 Control, bits D0 and D1 (CLK0_IDRV[1:0])
Function: CLK0 Output Rise and Fall time / Drive Strength Control
Possible values for D1, D0:
00: 2 mA
01: 4 mA
10: 6 mA
11: 8 mA

I suspect it is currently set to the default "00" value.
I would try to test it with a value of "11".






 

[Aldo22]

Hi jebem!

I suspect it is currently set to the default "00" value.
I would try to test it with a value of "11".

Thank you! I will test this later, probably not tonight.

And answering my own question (half way).

Can you show a picture of the HDS272 with a 30 or 40MHz square wave? I would be interested in that.

The more expensive Owon does not show a nicer square wave, rather the opposite, see the example for 10MHz in attachment.
For comparison: https://www.eevblog.com/forum/testgear/new-2ch-pocket-dsosg-sigpeak-dso2512g/msg4857962/#msg4857962

https://you tu.be/6v2UhBz3xZU?t=645
Btw. Does anyone know how to reference a youtube video in this forum without embedding the video?

[Aldo22]

Hi Jebem

I suspect it is currently set to the default "00" value.
I would try to test it with a value of "11".

I had a quick look at the source code.
I think 8mA is already set there, right?

Code: [Select]

  /* Configure the clk control and enable the output */
  /* TODO: Check if the clk control byte needs to be updated. */
  uint8_t clkControlReg = 0x0F; /* 8mA drive strength, MS0 as CLK0 source, Clock
                                   not inverted, powered up */
  if (pllSource == SI5351_PLL_B)
    clkControlReg |= (1 << 5); /* Uses PLLB */
  if (num == 0)
    clkControlReg |= (1 << 6); /* Integer mode */
  switch (output) {
  case 0:
    ASSERT_STATUS(write8(SI5351_REGISTER_16_CLK0_CONTROL, clkControlReg));
    break;
  case 1:
    ASSERT_STATUS(write8(SI5351_REGISTER_17_CLK1_CONTROL, clkControlReg));
    break;
  case 2:
    ASSERT_STATUS(write8(SI5351_REGISTER_18_CLK2_CONTROL, clkControlReg));
    break;
  }


[jebem]

Hi Jebem

I suspect it is currently set to the default "00" value.
I would try to test it with a value of "11".

I had a quick look at the source code.
I think 8mA is already set there, right?

Code: [Select]

  /* Configure the clk control and enable the output */
  /* TODO: Check if the clk control byte needs to be updated. */
  uint8_t clkControlReg = 0x0F; /* 8mA drive strength, MS0 as CLK0 source, Clock
                                   not inverted, powered up */
  if (pllSource == SI5351_PLL_B)
    clkControlReg |= (1 << 5); /* Uses PLLB */
  if (num == 0)
    clkControlReg |= (1 << 6); /* Integer mode */
  switch (output) {
  case 0:
    ASSERT_STATUS(write8(SI5351_REGISTER_16_CLK0_CONTROL, clkControlReg));
    break;
  case 1:
    ASSERT_STATUS(write8(SI5351_REGISTER_17_CLK1_CONTROL, clkControlReg));
    break;
  case 2:
    ASSERT_STATUS(write8(SI5351_REGISTER_18_CLK2_CONTROL, clkControlReg));
    break;
  }


Hi Aldo,

Yap, I think so.
clkControlReg = 0x0F  0b00001111
The last two binary digits D1 and D0 are set to 1, so it is already at maximum current.

One other optimization would be to run the VDDOx pin 7 at 3.3V, but the kit you are using most probable uses that voltage already from the LDO regulator. The 3.3V are important because the IC specs states that is a condition to have the alleged Zout of 50R.
I could not read the LDO regulator reference code from your pictures to learn the output regulated voltage, but the easier way would be to use a multimeter to check the voltage at pin 1 (VDD) and pin 7 (VDDOx).

Other additional optimization would envolve the addition of two bypass capacitors between pin 7 and GND, soldered as close as possible to the pin 7 of the Si5351 IC.
One of them should be a small capacitance, like a 100nF, with wv of 50V minimum, ceramic capacitor, to clear fast RF transients;
The other one should be a larger, electrolytic type capacitor, like a 10uF or maybe a 22uF, 105°C grade, with a wv of 25V or higher to have a low ESR, able to clear lower frequency fluctuations.


Sorry, I missed your request on the OWEN HDS272 on waveforms quality.
I would need to check the reviews to find out.
But as you found out, the OWON shapes are not exactly perfect either and even look worse on that review than what our scope here can do.


It seems that the only big issue with this scope is the bandwidth figure changing with the vertical amplitude of the signal.
This is really annoying and a big red flag on this scope if it is confirmed to be happening due to a design flaw.

I already commented initially about this PGA (programmable gain amplifier) section in scopes.
Not all are designed in the same way, but all of them need to present an adequate signal to the ADC, not too small and not too excessively large.
This implies that the input signal needs to be amplified if it is too small, and needs to be attenuated if it is too large.

At least on these low/medium cost scopes, we see the use of signal relays (mechanical or SS) to insert/remove RC networks to change the attenuation, depending on the selected vertical sensitivity, hence the clicks the mechanical relays do when changing selectivity. 
My 200EUR OWON SDS1102 100MHz 1GSPS does the same.

So this scope decreases the usable bandwidth when the vertical sensitivity is decreased.
This seems to be related to the insertion of a RC network in series with the input circuit in order to attenuate the input signal.

Either it is a design flaw or else a production batch using wrong/defective component values.

In both cases I hope that the manufacturer have detected and corrected the issues.
I expect to get the latest H/W revision (v9.15)  and hope to get a good free of red flag bugs unit.


I am afraid I will not be able to test my scope when it arrives, because due to personal reasons, I am away from my home where I have lab equipment, and I will not return before December.
Here at my mother's house I have just a LCR, a couple of multimeters, a small lab power supply and the usual tools including an soldering iron. I would need a proper RF signal generator and another scope.
 
Meanwhile, I will use my new scope to help on the repair of a few vintage PLL double conversion heterodyne radios in my collection, starting with a pair of Grundig Satellit 500.



   

[Aldo22]

Hi jebem.

I could not read the LDO regulator reference code from your pictures to learn the output regulated voltage, but the easier way would be to use a multimeter to check the voltage at pin 1 (VDD) and pin 7 (VDDOx).

Everything runs on 3.3 volts.
As controller I use an NodeMCU (ESP8266), which supplies the voltage for the module. I control the whole thing via a web interface, so I don't need any display module, buttons and switches for the test setup..

You can see in the first picture here that 3.33 volts are actually measured.
The picture shows a direct coax connection to the Si5351 module without dummy load or attenuator.
This image also shows that it doesn't have to look that ugly at 500mV sensitivity. Only when the dummy load comes into play, it gets bad (2nd image).
I mean, it ist not as good as with the attenuator at higher frequencies but it is definitely getting worse with the LR.

Other additional optimization would envolve the addition of two bypass capacitors between pin 7 and GND, soldered as close as possible to the pin 7 of the Si5351 IC.
One of them should be a small capacitance, like a 100nF, with wv of 50V minimum, ceramic capacitor, to clear fast RF transients;
The other one should be a larger, electrolytic type capacitor, like a 10uF or maybe a 22uF, 105°C grade, with a wv of 25V or higher to have a low ESR, able to clear lower frequency fluctuations.

I can only solder something at the outputs. Around the IC it is too narrow for my equipment (5 pins on 2 mm).
I had once experimented with capacitors at the output between GND and CLK.
For example 15pF cuts off a lot of the overshooting, but makes the signal washy at higher frequencies.
I can try again.

So this scope decreases the usable bandwidth when the vertical sensitivity is decreased.

Probably. I am not sure what is going on.
The signal can be displayed without massive distortion at 3.3 volts (see picture in the linked message above).
In any case something does not fit with the dummy load.

I would like to test with higher voltages to see how this develops, but I don't have an SG that goes above 3.3 volts and no amplifier for these frequency ranges.

But well, as I said before: If you work with the probes in the "Arduino range", you don't see anything of this problem. At low frequencies it doesn't matter either.
Since I am aware of the problem now, I can regulate down with the attenuators if necessary.
I think I can live with it.
It is not a $1000 tool.

Thanks and have a good time!