EEVblog #359 - QDSO Pocket Oscilloscope Review

[T4P]

So dodgy diyertool took down the item immediately
VERY SORRY FOR THIS STUPID PRODUCT! FORTUNATELY, WE INVITED DAVE AND MIKE TO POST THEIR PROFESSIONAL REVIEW. FACTS HAVE PROVED, THIS IS A FAILED PRODUCT.
HERE WE PROMISE: WE WILL* NEVER SELL THIS PRODUCT AND PROMOTE PRODUCTS FROM THIS MANUFACTURER!

Yeah, some of my notes :
The STM8S is actually one of the fastest 8bit architectures around (or so ST claims) at least faster than Atmel/Microchip
And i wasn't partially surprised the UI is a steaming heap of S#@!
The power supply part was funny though, "Spare no expense" 
It is such a terrible product that the UT81C although not exactly "pocket" sized does much better per dollar (the UT81C i suspect uses the same AD9288 clocked at 80MSPS)

[SeanB]

Caps on input are for probe compensation, but of course you cannot adjust them with the case closed.

OHL at his finest, wonder who will email the designer and ask why???

[bullet308]

Seems this product is a few dozen man-hours and about $20 away from being useful.  :-/

I think they *almost* have something there. Well, maybe not *almost*...

[AndyC_772]

So, will it blend? 

[ModernRonin]

This product might have a snowball's chance in hell of being, well, *passable* in the very low end of the "starving college student" market, if only:

1) They QUIT INTENTIONALLY LYING about the specs. Even with a 100 MS/s ADC, we shouldn't expect it to be able to usefully display a signal faster than 10 MHz. Maybe, if everything inside was absolute duck's guts (which it clearly ain't) then *maybe* 15 MHz. But any product brief that includes the figure "40 MHz" is being intentionally deceptive, and the idiot marketing-droid who wrote the copy needs to be shot. In the head. Twice (to be certain).

2) They stop using the word "oscilloscope" to describe it. That word carries connotations of something that's a piece of scientific grade Test and Measurement equipment. This wasn't designed to be that, it didn't turn out to be that, and it never will be. Again, they need to QUIT LYING. This thing is a "waveform viewer", and not a particularly great one either. That's fine, as long as they stop calling it an oscilloscope. Call it what it is. Quit fucking lying.

[kkp]

I was expecting Dave to bend out the common-mode suppression capacitor, and find it to be a 63V type (I found this to be the case in a china switcher). Y-rated capacitors are expensive. The factory will find a much cheaper substitute :-(.

[nitro2k01]

18:00-18:30 (The 20 MHz signal). The issue seems to be differing frequency response on the attenuator stages. Same-ish reason why you get overshoot on the square on some ranges. As Dave and Mike setablished, what a pile of

[lewis]

I was expecting Dave to bend out the common-mode suppression capacitor, and find it to be a 63V type (I found this to be the case in a china switcher). Y-rated capacitors are expensive. The factory will find a much cheaper substitute :-(.

I noticed that too, absolutely lethal. It looks to me like a Guangzhou-Ning-Ping-Shin-Happy-Special-Condensor-Co-Ltd ceramic crossing the isolation boundary in that power supply.

Dreadful scope too. Absolute shite.

[hans]

If I look at the DSO, it seems an approach where they can do the digital design (RAM, LCD controller, FPGA) but didn't spent any effort on the analog side of things. YEs the split ground plane, but I also learnt that the ground plane should be joined at the ADC in most cases. So go figure .
And then they do have a working digital side, and put CRAP firmware on it. I mean, the Rigol 1052 has a 320x240 screen, and it can put 2 waveforms on it decently with good menu's.

The analog bandwidth ofcourse is complete bullshit. 200MS/s is 20MHz at most. But without a sin(x)/x interpolation, it doesn't look pretty at all.
The micro seems like only handling buttons. If you follow the board you see it moving straight across to the button side and just reading the damn inputs. Seems like they were short on I/O or something. Shift registers?

The A/D seems legit in that it's actually a dual 100MSPS ADC. Well, if it's not counterfeit. Mike pointed out the laser markings were not aligned straight...

Seems the input is pretty simple. Voltage follower (with diode clamps on the input, look at the size of those - how long would the survive an overload?), crappy voltage gain (with selector feedback resistor, great!) and into the A/D. Great, maybe they should buy a VGA/PGA next time.

[Neilm]

If I look at the DSO, it seems an approach where they can do the digital design (RAM, LCD controller, FPGA) but didn't spent any effort on the analog side of things. YEs the split ground plane, but I also learnt that the ground plane should be joined at the ADC in most cases. So go figure

There are some EMC experts who are against splitting the ground plane especially if there are tracks crossing the split as there pretty much all over it. I assume that the board was a dual layer not 4 layer.

It would be interesting to know if the noise that was seen had coupled through onto the input - those thick tracks going by the input caps looked like they might be carrying power so noise from the power supply could be coupled onto the input jack.

Neil

[T4P]

The analog bandwidth ofcourse is complete bullshit. 200MS/s is 20MHz at most. But without a sin(x)/x interpolation, it doesn't look pretty at all.
The micro seems like only handling buttons. If you follow the board you see it moving straight across to the button side and just reading the damn inputs. Seems like they were short on I/O or something. Shift registers?

I don't actually think you know about the johnson-nyquist sampling thereom. Explain why the Highest DSOX3000 is 1GHz at 5GS?

[nitro2k01]

I don't actually think you know about the johnson-nyquist sampling thereom. Explain why the Highest DSOX3000 is 1GHz at 5GS?

That's assuming that
1) you've got a perfect brickwall low pass filter with no roll-off
2) you're fine with having your f/2 Hz sine wave represented as a 2 samples per cycle square wave on the screen.
In practice you need a little headroom if you care about being able to see phase information in the recorded data. Whether that means an x5 or x10 sample rate depends on the rest of the design. In the case of QDSO the problem was obviously a crappy input stage. Or at least the first problem.
And also, I believe you mean the Nyquist-Shannon theorem.

[T4P]

I don't actually think you know about the johnson-nyquist sampling thereom. Explain why the Highest DSOX3000 is 1GHz at 5GS?

That's assuming that
1) you've got a perfect brickwall low pass filter with no roll-off
2) you're fine with having your f/2 Hz sine wave represented as a 2 samples per cycle square wave on the screen.
In practice you need a little headroom if you care about being able to see phase information in the recorded data. Whether that means an x5 or x10 sample rate depends on the rest of the design. In the case of QDSO the problem was obviously a crappy input stage. Or at least the first problem.
And also, I believe you mean the Nyquist-Shannon theorem.

Oh,damn. Brain not working in the morning.  Well yeah, it was a SERIOUSLY crappy input stage.
Have you seen the UT81 series? The B version goes to 8MHz at x5 sample rate ... and it doesn't seem to be typical chinese crap

[hans]

The analog bandwidth ofcourse is complete bullshit. 200MS/s is 20MHz at most. But without a sin(x)/x interpolation, it doesn't look pretty at all.
The micro seems like only handling buttons. If you follow the board you see it moving straight across to the button side and just reading the damn inputs. Seems like they were short on I/O or something. Shift registers?

I don't actually think you know about the johnson-nyquist sampling thereom. Explain why the Highest DSOX3000 is 1GHz at 5GS?

Lol, funny.
Of course I do. You need to sample  at 2x sps to capture a sine wave of frequency x Hz. But that's purely mathematical. I was talking about practical situations where a rule of thumb 10 samples per period is good. This scope doesn't obey. 4-5x sounds okay, but then at least have the scope designed for it.
I mean: anti-aliasing filters and decent drawing methods like sin(x)/x interpolation. Because at 5 samples per period, linear interpolation, it would probably look something like this:

I don't know about you, but I call that nasty. Of course this view suggests a pure sine wave at a frequency pushing the boundaries of the scope.
We should actually care about the -3dB bandwidth of the scope, which should be situated at 40MHz and roll-of nicely.
But c'mon look at the input stage. Where is the bandwidth filter? Dave experienced HORRIBLE ringing at 10MHz square waves,.. that tells me the input stage doesn't have a very 'flat' frequency response.

Q(rap)DSO

[Psi]

I think what you want is something that folds in half.
One half is the screen and the other has thumb-wheel encoders for all your typical controls plus tack buttons for everything else.

So when you fold it out it really is pretty much like a normal scope, just smaller.

[george graves]

This review was almost silly(and yet - I enjoyed it!) - but it was just to review something bad for the sake of making fun of it - Ha!  Well....

Someone soon is going to nail this form factor with performance as-good-as-or/better then an entry level rigol - 640x480, 2 channel and a touch screen.  It's only a matter of time.

And it will be $150-$200 USD at the most.  I think even Dave will admit - the race to the bottom of test gear cost is on!  They'll be some duds, and some outstanding price/performance stuff along the way....now if we only had someone to test all this upcoming gear! 

Great video BTW - and don't forget to click the "like" button!

[Psi]

Threw together a mockup of a folding pocket scope with proper controls.

However now i'm thinking it would be better to folder on the bottom rather than the right.
So the controls were below.

[mikeselectricstuff]

Threw together a mockup of a folding pocket scope with proper controls

Horizontal format probably not good for handheld use - you want to be able to hold in one hand and reach most of the controls with a thumb.

Don't really need buttons or LEDs for AC/DC - these don't get changed often enough to justify the space.
Knobs don't need to be that wide - make them narrower and have more of them - at least x position and timebase, and y position/scale.
Maybe also push functions for time zero, autoscale, zero and ac/dc select respectively.

[Psi]

Yeah, there are lots of layout options.
Folding on the bottom would definitely be better.

I probably did go a bit overboard on the buttons.

Those thumbwheels were just what i found Google image searching.
Any control would work so long as it's analog and easy to use.

[AndyC_772]

I wouldn't go with a pocket format at all, it's uncessary and results in too many design compromises that do limit its usefulness. The smallest that a scope needs to be is one which is small enough to hold one handed or to find room for on a cluttered desk.

I'd go with a deeper design, with proper BNC probe connections on the sides for its 2 channels, and a battery that's big enough to power more capable electronics and still maintain a decent lifetime. A scope with a flat battery is useless.

However, in case of a flat battery I'd have the option of dc power from an external brick, with an isolated internal power supply that allows it to make floating measurements even when it's not running on its internal batteries. USB power might be a possibility.

I hate touch screen scopes, so I'd have a few proper controls on the front panel - maybe a single rotary encoder and a bunch of hard buttons which control what it does. From the front it might look a bit like a Philips Pronto TSU9600 (google it).

2 channels means it's possible to do serial decoding on I2C, CAN and UART interfaces, or one direction of an SPI port at a time. The serial decode feature would come as standard, meaning it might actually be cheaper in some cases to buy a handy miniature scope than it would be to add serial decode options to a conventional bench scope. Costly serial decode options are a complete rip-off, IMHO.

The hardware would be able to do an intensity-graded display, to give a much better impression of irregular waveforms. (The extra memory and FPGA resources would be the main driving force behind needing the deeper form factor - it's not trivial to do this by any means).

My miniature scope wouldn't be particularly cheap, simply because it can't be both cheap and right. Existing toy pocket scopes have proved this already. A scope that doesn't give a true, useful indication of a waveform is useless, and therefore a waste of money at any price.

[KedasProbe]

Nice video (a bit too long), building up to the moment we know we were all waiting for: Throwing the thing away.
Maybe hitting it with a hammer would have been nice to end the video 

[TheWelly888]

It is unbelievable that someone decided to sell a device with an on/off button that doesn't work as such!! Such counterintuitive products are very very annoying if you don't use them everyday.

A shame that such a facepalmy product actually went to market! But I am glad that the main dealer withdrew the device before it caused any more grief.

Well done Dave and Mike!

[Psi]

The "pocket" format is what makes it useful.
If it's too big to fit in my pocket then i might as well use a proper scope.

The idea is to have a scope you can take anywhere, something that goes with one of those tiny pocket multimeters.

[mikeselectricstuff]

I wouldn't go with a pocket format at all, it's uncessary and results in too many design compromises that do limit its usefulness. The smallest that a scope needs to be is one which is small enough to hold one handed or to find room for on a cluttered desk.

I think there are probably two useful product points for the ultimate portable scope:

"Pocket" - for quick testing where a DMM isn't enough. Maybe 60MHz, very small & light, basic functionality, maybe 1 and 2 channel options. Chargeable from USB. dedicated probes with small connectors acceptable. UI probably touchscreen and/or side mounted edge-wheel encoders. Must be fully operable one-handed. >2 hour battery life.

"Portable" - cose to full features (comparable to, say. rigol 2000 & Agilent MSOX2000), intensity graded display, at least 200MHz with comprehensive triggering options and protocol decode.
Option for probes isolated from each other (extra cost option as many people won't need it). Small and light enough to use handheld for short periods, with basic controls operable handheld- about the size of the OWON - the Agilent portables are too big and heavy.

[jpelczar]

If they opensourced the scope's firmware, I'm sure people would improve it (both in FPGA and uC)