OWON VDS1022I Quick Teardown (versus the Hantek 6022be)

[Gyro]

As this just came up in another post, and I can't find any other pictures on the web (Plenty of the Hantek 6022be) I though it was about time to do a teardown of my Owon VDS1022I. Naturally there will be quite a few comparisons with the similarly targeted Hantek.

A teardown of the Hantek can be found in this thread:
https://www.eevblog.com/forum/testgear/hantek-6022be-20mhz-usb-dso/

I realize that this won't be of much interest to the Rigol DS1054Z crowd, but when a bunch of these came up cheap a couple of years ago on ebay uk I thought it was worth a go. I already have a Picoscope ADC216 16bit for audio use, a Tek475A 250MHz Analogue for high frequency stuff and a decent logic analyzer for long trace digital, so I couldn't justify yet another bench-top, I just needed a 'fill in' for the middle. I also reasoned that the first things to fail on a Chinese bench scope would be SMPS, front panel controls and fan so getting something with non of these would be a bonus!

Anyway, on with the teardown. The VDS1022I is a dual channel 100Msps (per channel) 25MHz. This seems a bit more realistic sampling speed than the 20MHz 48Msps Hantek. The 'I' suffix indicates USB Isolation - you can float the scope ( the front panel BNCs are 'touchproof' style too). There's also a non-isolated version. Other features include more versatile triggering (Edge, variable Slope and Pulse, TV and with variable hold-off - the Hantek only has edge) Alternate triggering is also included (the ability to trigger on two unrelated signals at the same time) and Variable persistence too. There is also a dedicated Trigger connector which can be used as Trigger In, Trigger Out, and Pass/Fail out, using internally set rules.

The one real downside, spec-wise of the Owon is it's 5k samples per channel memory depth, versus a claimed 1M for the Hantek, not an issue for me, but might be for others. The Owon also uses (only needs) USB 1.1 vs the Hantek's USB2. Both of these are side effects of the significant internal architectural difference.

It is in the architecture that I think the Owon wins. Where the Hantek just uses an EzUSB micro for interfacing the ADC to USB, the Owon includes a Silicon Labs SiM3U156 32 bit ARM Cortex-M3 CPU together with a Xilinx Spartan-3 FPGA. There's clearly a lot more happening on-board than in the Hantek. Presumably the triggering is all handled on board (especially as there is a Trigger-out), together with data compression needed to pipe the data via USB1.1 at decent refresh rate. It also results in very low PC overhead with minimal impact on running other applications even on a low-end PC. The 5k memory depth is clearly a less desirable side effect. I'm not sure where the Hantek's triggering is handled, as it's edge only it could either be on-board or in the PC software.

The USB 1.1 interface also allows the USB isolation, which is performed by an ADUM3160. USB power is isolated by a TI DCP020505 2W isolated DC-DC converter. This has the lower isolation voltage of the two devices at 1kV. PCB clearances are very good. There are also un-populated footprints for a non isolated RS-232 interface. Some of the higher end models have LAN too, but there are no footprints on the board for this.

On the Analogue side, the A/D is an AD9288 dual in its 100Msps variant. ADC drive is via a pair of LMH6551 370MHz Differential op-amps and at least part of the gain switching is accomplished by a pair of AD603 90MHz pin selectable variable gain amplifiers, according to the datasheet these are capable of a 42dB gain range at 30MHz.  The only hand mods visible are beefier supply caps on the LMH6551s. The front end, which presumably includes the higher voltage attenuators, AC/DC selection and input amps is heavily screened in cans top and bottom side which I have no intention of trying to remove (another difference from the Hantek). Video triggering is implemented using a dedicated Rohm BA7046 sync separator rather than all in s/w. There are several other packages that I haven't managed to identify in the analogue area, and there is a scattering of supply regulators around the board.

All in all then a fairly large component cost compared to the opposition. It's clearly a more comprehensive design which presumably shares a lot of its acquisition circuitry with the PDS5022 desktop model, which shares the same  bandwidth, sample rate and memory depth. The supplied probes were 60MHz bandwidth X1/X10 which seem pretty good.

In terms of PC software, this is very subjective from user to user. I like the Owon for its lack of pseudo knobs (!) general screen layout and number of useful shortcuts (once you find them) and of course the added functionality. FFT and Maths traces can be selected to come up on a separate grid to the main traces. Persistence can be set from zero to infinity. Note that interpolation is fixed as sinX/X and trace colors, which match the probes and input connectors cannot be changed. The original s/w had a few bugs around single trace but the current s/w seems fine. I did flag one bug in the memory system at the end of last year which their engineers fixed (sending me a couple of test versions to confirm) before releasing, which I think could be described as 'responsive'.

Full disclosure: I have no connection with either company, other than having bought the Owon.

Pictures are attached and questions / comments are welcome. Thank you if you've made it this far! 

[rf-loop]

Thank you about this.

Is it possible you make one very tiny test,

If you have 50ohm BNC terminator use these, if not then input connectors open. (not probe, just open input BNC)
Both channels on, coupling DC.
Set vertical sensitivity to lowest voltage where is not BW rejection.
Set channels vertical positions so that other is 2div over center and other 2 div under center.

Set horizontal speed to 1ms/div,  if you can set persistence 1s or more

Scope display image.

Set horizontal to 1us/div, same persistence

Scope display image.

So we can see scope front end internal noise level.
(keep fingers and all far away from inputs for avoid external signal noise, also depending computer look that its (possible) EMI noise do not affect scope open inputs)

If you have 50 ohm terminators then repeat and take pictures with terminators connected
(in some cases if there internal noise levels this test may change displayed noise level (and how it change may also tell something, specially if it rise)

----

There is higher models also, and lets hope some day someone show VDS 2000 or 3000 series some teardown and tests. (10M memory, 500M and 1G samplerate, up to 4 channels etc)

[Gyro]

Hi rf-loop,

Yes, happy to help. I do have 50 ohm terminators so these are fitted.

I've attached the two saved images for 1ms/div (sampling at 250kS/s) and 1us/div (Sampling at 100MS/s) as you asked. The input stages are pretty quiet!

Settings were:

- 5mV/div (most sensitive)
- Persistence set to infinite (ran for about 2 mins each time)
- Sampling, Peak detect

I also turned on the measurements for Vp-p and Vrms.

I don't think pc induced noise is an issue due to the USB isolation.

Yes I'd like to see the higher models too! It's a shame the USB isolation only extends up to the 250MSps model.

I hope this gives you what you need. Please ask if you want other measurements.

[Gyro]

Sorry, I just re-read your post and realized that you also asked for screenshots with terminators removed (empty sockets, no scope probes). Please find these attached too. All other settings the same, still nice and quiet. 

[rf-loop]

Thank you. 
 
Result is not bad at all (some amount better what I expect). Of course it need note that analog BW is 25MHz so also analog front end natural noise is less what makes it quite easy.  (Higher analog BW makes it much more complicated and  there also circuit design and component selection is critical and in high role).  Also it is good that terminated inputs result is same or less that unterminated.

[Gyro]

Glad it was helpful, as you say the 25MHz bandwidth helps minimize the noise. I'll try an analogue sweep tomorrow and see what bandwidth it actually achieves, I've never thought to do that before.

P.S. I've just been browsing the Hantek 6022be, thread and seen that the claimed 100M samples is actually its memory buffer in the PC software, it only has the 2k fifo in the EzUSB micro as on-board memory, and can't continuously stream 48Msps to the PC. While the Owon's 5k samples per channel is still small in current terms, it is at least real, which is nice.

[Gyro]

@ rf-loop,

Just to follow up, I ran a frequency scan today as I indicated. I used an Adret 730A synthesized sig gen with 50 ohm through-terminator directly at the scope input. At 1Vrms signal level. Averaging was turned on:

Channel 1 was -3dB at 34.7MHz
Channel 2 was -3dB at 31.2MHz

I guess this makes the 25MHz spec a little conservative, but easily enough to cope with worst case units.

[Mark_O]

Gyro,

thanks a lot for the teardown, and posting this!  Very timely.  (I had been thinking of doing a comparative review vs. the 6022BE, and just ordered the Owon recently for that purpose.  So you saved me quite a bit of time.)

The one real downside, spec-wise of the Owon is it's 5k samples per channel memory depth, versus a claimed 1M for the Hantek, not an issue for me, but might be for others.

Well, I'd consider the Hantek claim to be fraudulent, since the 1M isn't internal to the acquisition module.  They could as easily have claimed 10M/channel, or 100M.  IMO, if you're going to pull that kind of crap in your advertising, you have a responsibility to make it clear that you're leveraging the memory in the PC.  Which Hantek "accidentally" still fails to do (after having only years to correct it).  The 6022BE has 1k/channel internally, and they should be honest, and admit that.  Much of what we know about the Hantek device was only discovered after many folks here obtained them, and dug into their internals to expose the truth.  Hantek played no role in that.

The Owon also uses (only needs) USB 1.1 vs the Hantek's USB2.

That's interesting.  I say that because I was told specifically, and unequivocally by an Owon tech support person at Owon HQ that the USB1.1 was a leftover documentation error, and the actual interface was USB2.0.  However, I see from the ADC spec-sheet that only USB-FS is possible through the isolation device.  Which makes me suspect that the VDS-1022 (non I) has USB2.0, while the isolated version can do only USB1.1.  Unless their Tech was FOS.  There's certainly no reason, beyond the ADUM3160, that the Owon couldn't do USB2.0 w/o difficulty.  (But also, not much reason to do so.  Unlike on their higher-level VDS models, which have 5M or 10M/channel.)

It is in the architecture that I think the Owon wins. Where the Hantek just uses an EzUSB micro for interfacing the ADC to USB, the Owon includes a Silicon Labs SiM3U156 32 bit ARM Cortex-M3 CPU together with a Xilinx Spartan-3 FPGA. There's clearly a lot more happening on-board than in the Hantek.

No kidding!  The 6022BE cut every corner possible, and eliminated every component they could.  The device is essentially nothing more than a low-speed sample streamer, with extremely limited front-end gain scaling (prone to truncating traces, on many gain settings).  That's actually a very good chip-complement on the Owon.

...together with data compression needed to pipe the data via USB1.1 at decent refresh rate. It also results in very low PC overhead with minimal impact on running other applications even on a low-end PC.

I'm not sure any compression is required.  With just 5K/channel, even sending it all over (which isn't necessary in real-time modes), that's at least 100 screen refreshes/sec.

The 5k memory depth is clearly a less desirable side effect.

While that's certainly very low, in comparison with many current DSOs, it actually better than what most DSO's had available for many, many years.  At one point, 2K or 2.5K/chan was a "big deal".  And with decent triggering (which the Owon has), 5K can provide very good results, along with the ability to zoom in a bit (perhaps up to 10x).  While that's not going to challenge DSOs with 1000x, or 10,000x zooms, it's not bad for such an inexpensive device.

I'm not sure where the Hantek's triggering is handled, as it's edge only it could either be on-board or in the PC software.

There is no on-board triggering of any kind, and it's ALL handled in the PC software.     With pretty poor results at some sampling rates (highly unstable trace rendering).

The front end, which presumably includes the higher voltage attenuators, AC/DC selection and input amps is heavily screened in cans top and bottom side which I have no intention of trying to remove (another difference from the Hantek).

Yes, that's a big difference, and the residual noise levels are quite impressive on the Owon... FAR below those on the Hantek, even after extra shielding had been provided by owner-mods.

Video triggering is implemented using a dedicated Rohm BA7046 sync separator rather than all in s/w. There are several other packages that I haven't managed to identify in the analogue area, and there is a scattering of supply regulators around the board.

And again, it's impressive that they'd go to the trouble to incorporate a special chip, just to be able to provide Video triggering!

In terms of PC software, this is very subjective from user to user. I like the Owon for its lack of pseudo knobs (!) general screen layout and number of useful shortcuts (once you find them) and of course the added functionality.

Not having had an opportunity yet to operate one yet, I can't comment on usability.  But in terms of look and feel, and cleanness of the UI layout, I'd say it's head and shoulders beyond the software provided for the 6022BE.  As in, "generations apart".

Considering that the 6022BE generally goes for about $70, and the Owon VDS-1022 for about $50 more, I'd say that the Owon wins hands down.  Unless that $50 is so important to you that you're willing to put up with endless aggravation with the 6022BE, in general-purpose scope use.  OR you can make good use of the streaming capabilities of the 6022BE (which the Owon can't do at all).  Streaming on the Hantek is quite powerful, so if you need to capture a large amount of data (at either a high OR low speed), then the Hantek has that use-case nailed.  Just because the scope program that Hantek supplies limits you to 1M samples, doesn't mean that a simple App using their SDK can't capture far, far more than that.  Capturing a gig of samples shouldn't be difficult.


Thanks again for your efforts, Gyro, and taking the time to share.   

[Mark_O]

Quick question...

The 'I' suffix indicates USB Isolation - you can float the scope ( the front panel BNCs are 'touchproof' style too). There's also a non-isolated version.

I see that the Model ID on the case of your unit says 1022, not 1022I.  Is there some more detailed info, perhaps on a bottom sticker?  I ask because what I ordered was a non-I model, but may wind up getting an I variant anyway.  I'm wondering if I'll have to take it apart to tell the difference?  (It's used, and I don't think the original package is included.)

[Gyro]

Hi Mark,

Thanks for your comments and interest, there's nothing sadder than watching your treasured teardown sinking without trace for lack of interest  . Grateful for rf-loop's questions of course and glad I've saved you some time too.

When I bought the Owon I hadn't really discovered EEVblog so hadn't seen the 6022be thread, it was really 50:50 whether I went for the Hantek or the Owon I did it purely on higher sample rate vs the extra cost, a lucky shot.

Well, I'd consider the Hantek claim to be fraudulent, since the 1M isn't internal to the acquisition module.  They could as easily have claimed 10M/channel, or 100M.  IMO, if you're going to pull that kind of crap in your advertising, you have a responsibility to make it clear that you're leveraging the memory in the PC.  Which Hantek "accidentally" still fails to do (after having only years to correct it).  The 6022BE has 1k/channel internally, and they should be honest, and admit that.  Much of what we know about the Hantek device was only discovered after many folks here obtained them, and dug into their internals to expose the truth.  Hantek played no role in that.

Me too, I didn't actually realize until I dug deeper into the 6022be thread. I'm not completely sure where the Owon has it's buffer, presumably in the ARM chip, or fifos implemented in the Xilinx  FPGA or both. Curiously the desktop version spec says 6k per channel, it's the only difference in spec from the 1022I. I guess the problem here is that it's going to be pretty much impossible to reverse engineer the Owon unlike the Hantek (unless you're a lot cleverer than me anyway!)

That's interesting.  I say that because I was told specifically, and unequivocally by an Owon tech support person at Owon HQ that the USB1.1 was a leftover documentation error, and the actual interface was USB2.0.  However, I see from the ADC spec-sheet that only USB-FS is possible through the isolation device.  Which makes me suspect that the VDS-1022 (non I) has USB2.0, while the isolated version can do only USB1.1.

Curious, maybe they think saying that something isn't actually USB2 isn't as impressive.    As you say, there's no was that USB2 is going to squeeze through the isolator. I've looked on the PCB and there's no way that the 'I' could be depopulated to a 'non I' so they must be different PCBs. Maybe the 'non-I' really is USB2.

No kidding!  The 6022BE cut every corner possible, and eliminated every component they could.  The device is essentially nothing more than a low-speed sample streamer, with extremely limited front-end gain scaling (prone to truncating traces, on many gain settings).  That's actually a very good chip-complement on the Owon.

Yes, certainly no sight on cost cutting, the BOM cost must be way higher for the price. Actually, another difference I've noticed between the VDS1022 'non-I' and 'I' specs... The 'non-I' states maximum input of 40Vp-p whereas the 'I' states 400vp-p. Now whether they are confusing input with input-to-ground with the USB isolation I don't know. There are certainly multiple relays under the shield, you can hear them click on some range changes. Of course as we've established that they are two different PCBs  40V is enough for 400V with a x10 probe anyway.

By the way, I don't know how the 6022be does its AC/DC input switching, I don't see any relays. Can it?

I'm not sure any compression is required.  With just 5K/channel, even sending it all over (which isn't necessary in real-time modes), that's at least 100 screen refreshes/sec.

You're probably right, less for the PC s/w to do too. Of course it's impossible to say how many screen refreshes it's doing, there's no spec as there would be on a desktop, but it's certainly fast enough. The PC overhead really is very low, you can't tell it's running on my humble T42 laptop under XP. DISCLAIMER, I haven't been able to try it the s/w on Win7 or later.

Yes, the 5k doesn't seem to be much of a limitation in practice (certainly coming from an analogue). It doesn't have protocol decoders so you're unlikely to be decoding long data streams (an LA is better for that anyway). The Zoom works fine and there are zoom assist cursors too. You can also set the trigger holdoff and position on screen which helps.

There is no on-board triggering of any kind, and it's ALL handled in the PC software.     With pretty poor results at some sampling rates (highly unstable trace rendering).

Again, hadn't realized that until I read the 6022be thread, that's horrible! Having decent triggering certainly seems to be a VDS1022 strength, it seems very stable. Having a proper ext trig is helpful - I haven't tried the trig-out functionality yet but hopefully its latency is good enough to be useful, it could certainly be used to post-trigger an logic analyser.

Not having had an opportunity yet to operate one yet, I can't comment on usability.  But in terms of look and feel, and cleanness of the UI layout, I'd say it's head and shoulders beyond the software provided for the 6022BE.  As in, "generations apart".

It certainly feels like it was designed by an engineer! Worth downloading a copy to play with the menus (and try it on Win7+) even if you can't get traces. I did so a comment somewhere complaining that menu driven was 'so last year' but it's very intuitive. I'm not sure how closely the manual writer was to the s/w engineer, there are lots of undocumented shortcuts - eg. hover over the v/div or t/div values and turn the mouse scroll wheel and they change without needing to bring up a slider. Click on a channel icon and it immediately brings up the channel menu, AC/DC/Off on the fly, same with trigger slope, little arrow next to the on-screen measurements that brings up the measurement menu etc. Non of them documented (I'm still finding them).

I see that the Model ID on the case of your unit says 1022, not 1022I.  Is there some more detailed info, perhaps on a bottom sticker?  I ask because what I ordered was a non-I model, but may wind up getting an I variant anyway.  I'm wondering if I'll have to take it apart to tell the difference?  (It's used, and I don't think the original package is included.)

Haha, yes, the only indication is the small barcode label serial number label on the bottom. It reads VDS1022Ixxxxxxx, a very skinny 'I' too, easy to miss.

I sense that I've rambled on again  Glad it helped.

Chris

[Mark_O]

By the way, I don't know how the 6022be does its AC/DC input switching, I don't see any relays. Can it?

Nope.  DC-coupled only.  If you want AC, you add your own external cap, I guess.    And no Gnd/Off option either, of course.  Relays would have totally blown the budget on the Hanteks.  (It's really rather ingenious how far they managed to strip it down.  And I'd still recommend them for certain purposes.  Just not as a scope, except for those on the tightest of budgets.)

Thanks for the additional details, Chris.  It seems a shame that the software has so many undocumented goodies/shortcuts/usability enhancements.  Perhaps you could help fill that void.  It sounds like you've uncovered quite a few.

[Mark_O]

I'm not completely sure where the Owon has it's buffer, presumably in the ARM chip, or fifos implemented in the Xilinx  FPGA or both.

Unless one of the smaller chips you couldn't ID was a RAM chip, the dual 5K buffers must be located on the FPGA itself.  All the data has to flow through there anyway, for triggering to be handled. 

But it can't be passed on directly to the ARM chip at capture rates, because there is no external data channel capable of handling a 200MB/sec transfer rate.  At least not on any of the Cortex-M3's I have ever worked with!  It probably gets dumped from FPGA to ARM over a SPI-bus channel (DMA-driven).  That could potentially run as fast as ~25 MB/sec (Quad-SSP).  And then the ARM sends it out over USB, even slower yet.

[Gyro]

Clearly the man for a bit of reverse engineering then 

I've taken some time to get the IPA swabs out and get some better pictures and identify a few more bits...

- The SOT23-5 opamps U1 and U4 between the shield and the AD603 VGAs are AD8061 300MHz rail to rail.

- There are a couple of SGM 8292YS8 High voltage dual rail to rail op-amps by the power converted (one top side, one bottom) with lots of passives around them. I'm wondering if they're being used as charge pumped supplies.

- A pair of SOT23-5s marked YJAA (SGM2019 250mA LDO 30uVrms), obviously low noise supplies for the front end.

- Various other SOT23-5s marked T081(?), S40B (? Switcher), 2 x SC0DA (? Switchers, logic supplies). All look supply related anyway.

There's a single 100MHz clock U76 feeding the ADC and FPGA directly.

Definitely no standalone RAM chip so data must be fifo'd in the FPGA as you said.

There seem to be quite a lot of parallel data signals between the FPGA and ARM so maybe not necessarily SPI. I've tried to get a decent close-up of the data from ADC to FPGA to ARM data paths anyway, nothing significant happening on the back.

During cleaning, a few more small hand mods became apparent. A couple of cuts in the horizontal tracks bottom left of the FPGA, just above U35, maybe related to the not fitted and jumpered-out U29 in the analogue section. There's also a perfectly executed USB signal crossover using diagonal series 24R terminating resistors R99 and R100, just off the top right corner of the ARM - happens to the best of us I guess    . Apart from that just a few resistor mods and the beefed up caps on the LMH6551s.

One other thing I can't think of is why include an option for fitting a non isolated RS232 port (there's a footprint for a MAX232 on the bottom side). Remote control? data logging? Console?

I've captured a closer view of the analogue section too (U12 is the video sync separator).

Hopefully clearer than my previous efforts anyway.

[Gyro]

Nope.  DC-coupled only.  If you want AC, you add your own external cap, I guess.    And no Gnd/Off option either, of course.  Relays would have totally blown the budget on the Hanteks.  (It's really rather ingenious how far they managed to strip it down.  And I'd still recommend them for certain purposes.  Just not as a scope, except for those on the tightest of budgets.)

Thanks for the additional details, Chris.  It seems a shame that the software has so many undocumented goodies/shortcuts/usability enhancements.  Perhaps you could help fill that void.  It sounds like you've uncovered quite a few.

Oops, didn't see your first email.

True, the Hantek  does still have its place - it is still one up from the LCD handheld scope type products and USB AX (analogue channel) and for less money. Still might be be a case of cutting the cloth a bit small for what it might have been for $10 more, Looking at the 6022be thread i sense a feeling of "it's ok but a bit disapointing". I do wonder how the manufacturing+BOM costs vs price ratio of the Hantek and Owon compare. I'm still surprised that the Owon hasn't achieved more traction, all I get in a google search are sales sites, virtually no forum type hits. Just looking back, I paid GBP84 for it last year from a UK ebay seller (there were loads). Only 1 ebay UK hit for it now at GBP212 I think you can get it much cheaper (nearer the $100 mark?) through the right China channels.

Yes, sure, I'll sit down and have a serious play and see how many undocumented mouse shortcuts I can find and post here. Maybe a marked up screenshot if I can draw well enough.

Just one other thought, as the ADC is direct clocked by the oscillator, it is always sampling flat out at 100MSps. I wonder what it does with all the samples that won't fit at lower speeds - does it have enough intelligence to be able to average them / oversample at lower timebase settings???

[Mark_O]

True, the Hantek  does still have its place - it is still one up from the LCD handheld scope type products and USB AX (analogue channel) and for less money. Still might be be a case of cutting the cloth a bit small for what it might have been for $10 more, Looking at the 6022be thread i sense a feeling of "it's ok but a bit disapointing".

I do think it's good value for money, but it's just not very good as a scope.  And yes, one of these is much cheaper than a USBee AX, though you then lose the time-correlated analog+digital data stream.  The 6022be would be great to use in a risky environment, where it could get hosed.  Smallest $ loss.

I'm still surprised that the Owon hasn't achieved more traction, all I get in a google search are sales sites, virtually no forum type hits.

The thing is that USB DSO modules have a bad rep to start with, and Owon isn't doing a lot of marketing.  Also, while the 1022 is priced affordably, as you move up the VDS series, the 2000/3000 are expensive enough that folks (rightfully) start to think that for a few $$ more, they could just get a DS-1054z.

But yes, I agree that the VDS1022 is definitely overlooked, which is why I ordered one myself, to shed some light on it.  Which you have already achieved.  There are boatloads of craptastic "USB DSO's", but I think the VDS1022 has some unique value at it's price point.

Just looking back, I paid GBP84 for it last year from a UK ebay seller (there were loads). Only 1 ebay UK hit for it now at GBP212 I think you can get it much cheaper (nearer the $100 mark?) through the right China channels.

They're readily available new for $119, on *bay for example.

Just one other thought, as the ADC is direct clocked by the oscillator, it is always sampling flat out at 100MSps. I wonder what it does with all the samples that won't fit at lower speeds - does it have enough intelligence to be able to average them / oversample at lower timebase settings???

That's actually a good thing, and justifies the full-time sin(x)/x curve-fitting (which could lead to serious aliasing problems if the sample rate dipped too low).

I'm sure the FPGA handles the extra samples properly, either decimating, averaging, or aggregating.  They perform those tasks really well.  The later two, of course, would be preferred.  With aggregation providing excellent peak-detect.

There seem to be quite a lot of parallel data signals between the FPGA and ARM so maybe not necessarily SPI.

In that case, you're right.  SPI isn't necessary.  The FPGA sample-memory block(s) could be mapped into an address space on the ARM chip, and byte-streamed across directly via a DMA channel.  That would be even faster than SPI, but probably still not enough to manage the 200MB/sec required for direct streaming.  I'd have to check my ARM manuals to make sure what max rates are possible in that scenario.  In the designs I've built, the SPI channel gave me all the speed I ever needed.

Thanks for the additional chip insights, and the higher quality pix!

[Gyro]

Owon isn't doing a lot of marketing.  Also, while the 1022 is priced affordably, as you move up the VDS series, the 2000/3000 are expensive enough that folks (rightfully) start to think that for a few $$ more, they could just get a DS-1054z.

True, they do just seem to be relying on their retailers, they're even hard to find on the Owon site without a specific part number search  Yes I noticed that, certainly on ebay prices anyway, the higher speed ones are straight into Rigol territory, you'd have to have a pretty specific space reason for NOT wanting the screen!

The FPGA sample-memory block(s) could be mapped into an address space on the ARM chip, and byte-streamed across directly via a DMA channel.  That would be even faster than SPI, but probably still not enough to manage the 200MB/sec required for direct streaming.  I'd have to check my ARM manuals to make sure what max rates are possible in that scenario.  In the designs I've built, the SPI channel gave me all the speed I ever needed.

Knew you were the man for the job 

I was having another look at the board, there seem to be a few traces between the ARM and that dual high voltage opamp. It might be tied in with the on-chip ADCs/DACs for some purpose... the self-cal function maybe. Too slow for any sort of triggering, the SGM 8292YS8's GBP is only 1.5MHz. I wish we had a schematic!

I'm looking forward to yours turning up now! 

[Mark_O]

Yes I noticed that, certainly on ebay prices anyway, the higher speed ones are straight into Rigol territory, you'd have to have a pretty specific space reason for NOT wanting the screen!

Surprisingly, the value proposition just  isn't there, at the higher-end.  A 4-channel unit, 5M/chan, max 1 GSa/s,  and 100 MHz BW, is priced at $400.  Considering how much more a Rigol offers, it's hard to see why they'd price a VDS3104 that high.  Unless you've always got a laptop handy, and want to add just an acquisition module for travel, a standalone 1054z would almost always make more sense.

The FPGA sample-memory block(s) could be mapped into an address space on the ARM chip, and byte-streamed across directly via a DMA channel.  That would be even faster than SPI, but probably still not enough to manage the 200MB/sec required for direct streaming.  I'd have to check my ARM manuals to make sure what max rates are possible in that scenario.  In the designs I've built, the SPI channel gave me all the speed I ever needed.

Knew you were the man for the job 

I took a quick look at the Ref Manual for the SiLabs ARM chip, and with it clocked at the max-rated 80 MHz, DMA transfers thru the EBI would run 20-40 MB/sec, depending on # of clock cycles required per transfer (looked like ~3), and if it was an 8-lane or 16-lane data bus.

[Gyro]

Hmm, I guess it all depends whats happening in the FPGA then. I might try tracing out what pins are in use on the ARM (no point at the the FPGA end ).

Just wondering what would happen if it re-made those two cut traces, what magic function........ No not worth the risk  It will be interesting to compare with your board though!

[Gyro]

As requested, the definitive (ish) list of Owon VDS1022 Software shortcuts....

I haven't included keyboard shortcuts (those are documented in the manual). All of the menu's referred to can be accessed by the large 'home' icon on the right hand side, but some of the shortcuts bring up the relevant one directly.

The numbers tie up to the attached Screenshot:

(1) Clicking on the small arrow directly brings up the measurement cursors menu for time and / or voltage.

(2) Clicking on the small '+' icon (only appears when the mouse pointer is in the measurements area) immediately brings up the measurements selection menu.

(3) Clicking on either of the channel numbers immediately brings up the channel settings menu.

(4) Volts/div selection, Clicking brings up a normal popup+slider However if you just hover the mouse over it instead you can use the scroll wheel to increment/decrement directly, much quicker.

(5) Clicking here toggles between DC / AC / GND for that channel.

Additional, hovering in either of the channel boxes also brings up a small 'x' icon to turn off the channel. Clicking in a greyed out channel box will turn it on again. (couldn't show this and (2) at the same time).

(6) Clicking in the 'T' box brings up a slider for trigger position. It's normally easier just to drag the red trigger position flag at the top of the screen, but the 'T' box click also brings up a useful 'Reset' button to restore to the middle.

(7) Time/div... Similar operation to the V/div. Clicking brings up the pop-up/slider but mouse-over and scroll wheel changes the value directly.

(8 ) Clicking toggles the trigger source between Chan1, Chan2 and Ext (trigger marker on right hand edge of screen changes color to match).

(9) Clicking toggles trigger polarity, works on edge, slope and pulse (icon changes to match trigger type). In Video trigger mode it toggles Odd, Even, Line etc.

(10) Click to alter trigger level - Actually it's far easier and more accurate to drag the trigger marker on the right hand side if the screen, but clicking here also brings up two useful buttons to 'Reset' the trigger level and 'Set to 50%'.

Obviously clicking the Trigger button takes you straight to the Trigger menu.

Some, but not all of the above shortcuts may be documented somewhere, but certainly not all (eg. the scroll wheel ones). Avoiding having to navigate the menu system for most things apart from initial setup speeds thing up quite a lot.

Hope it helps...

[Gyro]

Hey Mark_O has your 1022 turned up yet??? (in the spirit of you show me yours and I'll show you mine ).

It would be interesting to see the layout differences between the -I and non-I PCBs.

In the analogue area I'm wondering if the U29 footprint (or equivalent) on yours is configured the same way as mine. It's a shared SSOP10 device shared between both channels (the only shared IC in the analogue section). The 2 cut tracks from the Cortex CPU that I mentioned originally went to U29 pins 5 and 6 via RC filters (configured to be outputs from the CPU, inputs to U29), possibly PWM I suppose.

On mine, pins 7 and 9 are linked (Channel 1 side) but the equivalent pins (1 and 3) for Channel 2 aren't. When I added a link the channel 2  voltage noise readings changed. It settled after and Autocal but the noise readings were still higher than before. I tried it a couple of times and autocal was needed after each add/remove. I don't want to try again (or remove the Channel 1 jumper) for fear of losing pads. Pins 10 and 4 are +ve and -ve supplies. Inputs on 1 and 9, outputs on 3 and 7, f/b network to 2 and 8?

A not-fitted part always makes me wonder about additional functionality but not sure what it would have been, a switch? Mux?, another VGA?, Filter? Something controlled by the CPU anyway. Its position is between the gain stages and ADC driver amps. It may have been removed because it degraded cross-talk performance. I haven't turned up any likely candidates based on pinout.

As I say, I'm curious to see if yours is configured the same way.

[Mark_O]

Hey Mark_O has your 1022 turned up yet???

Yes, it arrived some time ago.

It would be interesting to see the layout differences between the -I and non-I PCBs.

I agree.  However, I've been unable to open mine.    The jewelers screwdrivers I have are the only one's small enough to fit, but I can't get enough torque to loosen the screws.

[Gyro]

I hope it's living up to expectations.

I agree.  However, I've been unable to open mine.    The jewelers screwdrivers I have are the only one's small enough to fit, but I can't get enough torque to loosen the screws.

Ah, that's frustrating. Self-tap screws do tend to stick in anodized aluminium. A pair of mole grips clamped to the jewelers screwdriver handle often yields results of one sort or another  (or the local application of heat    )

[Mark_O]

Ah, that's frustrating.

No kidding.  They didn't make it any easier by recessing the screws so deeply in a narrow shaft.  I can't get any larger screwdrivers in there.   

What I really need is something with a smaller shank, but a nice hefty handle.  Hmm... something like this:

[Mark_O]

OK.  That did the trick.  Now what did you want to see?

U37 and U9 are gone... replaced by a handful of 0-ohm resistors.  So no USB-isolator on this non-I board, or USB1 slowdown.

[Mark_O]

You were asking about U29.  I don't remember where it was on your board, but on mine it's in a spot I think was empty on your's.

(BTW, my board rev is V1.4, dated 2014.08.04.  About a year newer than your V1.2 board.  Not surprising, considering how long you've had yours.)