19-07-2015, 11:31 PM
A few weeks back, totally on impulse, I picked up a JYE DSO138 'scope kit for £15.80 delivered, from Banggood.com: http://www.banggood.com/DIY-Digital-Osci...84002.html
(I note that now, they appear to be selling the latest version. The version I have is the older version. The difference between them is 1 resistor, and accordingly different firmware. In theory, the newest version should have less noise. But, with this version, JYE aren't making the firmware available (to protect against cloning), so if you wish to experiment with different firmware that has been developed by the community, you have to change the resistor to the original value - and accept that if you re-programme the processor, you can't go back to the latest specification.)
It arrived on Friday, and I have just assembled it.
I chose the version that has all the surface-mount components already mounted. It is possible to save a pound or so if you are willing to install them yourself (minus the CPU), but you'd be mad to consider it! The quality of the soldering is first-rate.
These kits have been widely cloned, so an obvious question is this: are Banggood selling clones? Well, as far as I can tell from the data available, these seem pretty genuine. Certainly, no problem with the quality of either the PCB or the supplied components. The printed instructions are very good, on high quality glossy paper. Everything feels fine...
Assembly starts with the resistors. These are 1/8 watt metal film types, and have the 5 band colour coding convention. Perhaps in the past I'd have no problem seeing these, but today I was using a x4 loupe to read them, and was double-checking them with a DMM. The trouble with 5 band 1% resistors is there are combinations that could be read in either direction, so it's best to verify them. Incidentally, all were comfortably within spec. As were the no-name electrolytic caps...
The instructions provide you with spaces to tick off the components as you go. They suggest the order of assembly, starting with the smallest and working up in size, as you might expect. The text and pictures are clear and logical, and providing your fingers and eyesight are OK with the small components used, assembling this should present few problems. There is a warning about the input BNC needing a fair amount of heat, but my Weller TCP had no problems at all.
Once finished, the test procedure has you measure the output of the voltage regulators, and if good, a soldered link is made to connect the 3.3V rail to the ARM Cortex CPU. The display is attached, and power is applied. Boringly, it worked first time!
I took my time assembling this, and reckon I probably spent between 3 to 4 hours on it. Yes, you might be able to do it in half that time or less, but it depends on how neat you want the final assembly to be. I solder in each component one at a time, giving due regard to how it sits on the PCB, and how each solder joint works. But that takes time...
So, how does it work?
Again, bear in mind that I bought this on impulse. If it worked at all, then it was worth the exercise. If it proved to be even mildly useful, then that's a bonus. Remember, we're talking about £15 delivered here!
JYE developed this kit as an educational product. Students can have some experience with soldering and assembly, and they have something to show for it at the end. In the past, we have paid this price to buy those really nasty multimeters (that are available for less than £5 fully assembled!) in kit form for the same goals. In the future, I'd consider these kits - if only our purchasing systems would let us! Should RS or Farnell start to stock them, then no problem!
At power-up, a couple of splash-screens are presented, and the unit is up and running in between 5 and 10 seconds. It appears to remember all the settings from the last time it was used, which is convenient.
The controls are a mixture of slide switches and push-buttons. The three slide switches are controlling the analogue signals:
Switch 1: Coupling - GND, AC, DC
Switch 2: Sensitivity 1 - 1V, 0.1V, 10mV
Switch 3: Sensitivity 2 - X5, X2, X1
The two sensitivity switches let you cover a range of 10mV/Div to 5V/Div. OK, sliding the two switches to and fro is nowhere near as user-friendly as rotating a knob, but rotary switches are rather expensive. Pragmatism...
But, each switch has two poles, and while 1 pole deals with the analogue signal itself, the other pole is used to report back to the CPU. So the display always includes the selected sensitivity and coupling settings - this makes the sliding switches a lot better than you might expect. Points for effort
Moving on to the buttons, one is a reset switch. Not something you'd need every day.
That leaves the following: SEL, +, -, OK.
The idea is that you move a highlight around the screen with the SEL button, and then you modify the parameter with the + and - buttons. At power-up, the timebase speed is highlighted. This can be varied between 500 seconds per division and 10us/div.
Pressing SEL moves the highlight to the timebase mode setting. The options here are AUTO, NORM and SING. If you are familiar with analogue 'scopes, you'll understand what these do. Press OK in SING mode to make it sweep.
Press SEL again to move the highlight to the trigger slope selection. It defaults to rising edge, but can be changed to falling edge if required.
Another press of SEL takes you to the trigger level setting. A pointer on the right hand side of the graticule turns cyan, and can be moved up and down by the +/- buttons. The threshold voltage is shown on the top-right of the display in magenta.
The next press of SEL moves the highlight to the pan control (top of the screen). The +/- buttons allow you to move along the data memory, allowing you to see about a screenful of the trace either side of the centre that it defaults to.
Finally, pressing SEL again takes you to the Y position control - a pointer on the left of the graticule indicates the 0V reference for the trace. This needs an occasional resetting (press OK for two seconds while the input is set to GND).
Pressing and holding OK when in timebase speed mode will toggle a display of measurements (which was a pleasant surprise!). The measurements displayed include frequency and amplitude (including volts RMS). Very useful, and reasonably accurate too...
Generally, briefly pressing OK stops and starts the timebase.
So, a bit clunky compared to "real" controls, but again, being pragmatic about the aim of the product, I think it's fine. I suppose that if you were really keen, you could connect a rotary encoder to the unit and modify the firmware accordingly (prior to the latest release, it's open-source). Or ask someone else to do it for you. Including a rotary encoder would probably add a couple of pounds to the final price, so I can see why they didn't.
So how does it perform?
The specification suggests an analogue bandwidth of 200kHz. I'm not sure I'm getting that. More exhaustive testing required - I have only just built the thing!
The sample rate of 1MS/s realtime ain't bad for a 200kHz BW.
Memory depth of 1024 samples isn't bad either.
The analogue signal is processed by a TL084. One section is a follower, following on from the input attenuator (x1, x0.1 and x0.01). Following that is a passive attentuator that does the x1, x2 and x5 selection. Then that's followed by another section that provides a gain of two ahead of the ADC in the microcontroller (the gain is increased to x8 in the latest version). Finally, the trigger signal is derived by a third section of the TL084, which is used as a comparator - the threshold comes from the microcontroller. The 4th section is unused.
The trace is a little noisy. The firmware does not appear to use averaging or any other technique to reduce trace noise. First impressions are that the analogue section might be picking up noise, rather than generating it itself, but I've yet to confirm that. It seems to be unrelated to the switching action of the -ve PSU.
The second - and related - problem is trigger jitter. Looking at the signal at the comparator, it's very jittery. The trace hops left to right on screen at higher frequencies. It's not unusable, but if you're expecting the rock-solid trace of an analogue 'scope, you won't be getting that here.
Of course, this 'scope will alias at slower timebase speeds. To be expected, and easy to spot if you're experienced in using 'scopes. But a trap for the unwary. Better digital 'scopes (with more sample memory) avoid this problem by running the ADC as fast as they can, consummate with the timebase speed and the available memory, so hopefully the sample rate is kept high enough at all times.
Overall, better than expected, and I can see this being a useful project for someone who fancies the challenge of putting it together. It won't replace a decent analogue 'scope - as few digital 'scopes can - but it's definitely worth it for the measurement functions.
It runs from 9 to 12V - the screen brightness is affected by supply voltage. Current consumption is just short of 120mA at 10V. A series diode is provided to protect against incorrect polarity. The micro-USB connector and 0.1" headers either side of it aren't used for this application, but they are provided for developers (the unit doubles as a dev environment for the ARM Cortex microcontroller).
I can see this mounted inside one of Yorkie's oak boxes with a clear perspex lid and rechargeable batteries hidden beneath
(I note that now, they appear to be selling the latest version. The version I have is the older version. The difference between them is 1 resistor, and accordingly different firmware. In theory, the newest version should have less noise. But, with this version, JYE aren't making the firmware available (to protect against cloning), so if you wish to experiment with different firmware that has been developed by the community, you have to change the resistor to the original value - and accept that if you re-programme the processor, you can't go back to the latest specification.)
It arrived on Friday, and I have just assembled it.
I chose the version that has all the surface-mount components already mounted. It is possible to save a pound or so if you are willing to install them yourself (minus the CPU), but you'd be mad to consider it! The quality of the soldering is first-rate.
These kits have been widely cloned, so an obvious question is this: are Banggood selling clones? Well, as far as I can tell from the data available, these seem pretty genuine. Certainly, no problem with the quality of either the PCB or the supplied components. The printed instructions are very good, on high quality glossy paper. Everything feels fine...
Assembly starts with the resistors. These are 1/8 watt metal film types, and have the 5 band colour coding convention. Perhaps in the past I'd have no problem seeing these, but today I was using a x4 loupe to read them, and was double-checking them with a DMM. The trouble with 5 band 1% resistors is there are combinations that could be read in either direction, so it's best to verify them. Incidentally, all were comfortably within spec. As were the no-name electrolytic caps...
The instructions provide you with spaces to tick off the components as you go. They suggest the order of assembly, starting with the smallest and working up in size, as you might expect. The text and pictures are clear and logical, and providing your fingers and eyesight are OK with the small components used, assembling this should present few problems. There is a warning about the input BNC needing a fair amount of heat, but my Weller TCP had no problems at all.
Once finished, the test procedure has you measure the output of the voltage regulators, and if good, a soldered link is made to connect the 3.3V rail to the ARM Cortex CPU. The display is attached, and power is applied. Boringly, it worked first time!
I took my time assembling this, and reckon I probably spent between 3 to 4 hours on it. Yes, you might be able to do it in half that time or less, but it depends on how neat you want the final assembly to be. I solder in each component one at a time, giving due regard to how it sits on the PCB, and how each solder joint works. But that takes time...
So, how does it work?
Again, bear in mind that I bought this on impulse. If it worked at all, then it was worth the exercise. If it proved to be even mildly useful, then that's a bonus. Remember, we're talking about £15 delivered here!
JYE developed this kit as an educational product. Students can have some experience with soldering and assembly, and they have something to show for it at the end. In the past, we have paid this price to buy those really nasty multimeters (that are available for less than £5 fully assembled!) in kit form for the same goals. In the future, I'd consider these kits - if only our purchasing systems would let us! Should RS or Farnell start to stock them, then no problem!
At power-up, a couple of splash-screens are presented, and the unit is up and running in between 5 and 10 seconds. It appears to remember all the settings from the last time it was used, which is convenient.
The controls are a mixture of slide switches and push-buttons. The three slide switches are controlling the analogue signals:
Switch 1: Coupling - GND, AC, DC
Switch 2: Sensitivity 1 - 1V, 0.1V, 10mV
Switch 3: Sensitivity 2 - X5, X2, X1
The two sensitivity switches let you cover a range of 10mV/Div to 5V/Div. OK, sliding the two switches to and fro is nowhere near as user-friendly as rotating a knob, but rotary switches are rather expensive. Pragmatism...
But, each switch has two poles, and while 1 pole deals with the analogue signal itself, the other pole is used to report back to the CPU. So the display always includes the selected sensitivity and coupling settings - this makes the sliding switches a lot better than you might expect. Points for effort
Moving on to the buttons, one is a reset switch. Not something you'd need every day.
That leaves the following: SEL, +, -, OK.
The idea is that you move a highlight around the screen with the SEL button, and then you modify the parameter with the + and - buttons. At power-up, the timebase speed is highlighted. This can be varied between 500 seconds per division and 10us/div.
Pressing SEL moves the highlight to the timebase mode setting. The options here are AUTO, NORM and SING. If you are familiar with analogue 'scopes, you'll understand what these do. Press OK in SING mode to make it sweep.
Press SEL again to move the highlight to the trigger slope selection. It defaults to rising edge, but can be changed to falling edge if required.
Another press of SEL takes you to the trigger level setting. A pointer on the right hand side of the graticule turns cyan, and can be moved up and down by the +/- buttons. The threshold voltage is shown on the top-right of the display in magenta.
The next press of SEL moves the highlight to the pan control (top of the screen). The +/- buttons allow you to move along the data memory, allowing you to see about a screenful of the trace either side of the centre that it defaults to.
Finally, pressing SEL again takes you to the Y position control - a pointer on the left of the graticule indicates the 0V reference for the trace. This needs an occasional resetting (press OK for two seconds while the input is set to GND).
Pressing and holding OK when in timebase speed mode will toggle a display of measurements (which was a pleasant surprise!). The measurements displayed include frequency and amplitude (including volts RMS). Very useful, and reasonably accurate too...
Generally, briefly pressing OK stops and starts the timebase.
So, a bit clunky compared to "real" controls, but again, being pragmatic about the aim of the product, I think it's fine. I suppose that if you were really keen, you could connect a rotary encoder to the unit and modify the firmware accordingly (prior to the latest release, it's open-source). Or ask someone else to do it for you. Including a rotary encoder would probably add a couple of pounds to the final price, so I can see why they didn't.
So how does it perform?
The specification suggests an analogue bandwidth of 200kHz. I'm not sure I'm getting that. More exhaustive testing required - I have only just built the thing!
The sample rate of 1MS/s realtime ain't bad for a 200kHz BW.
Memory depth of 1024 samples isn't bad either.
The analogue signal is processed by a TL084. One section is a follower, following on from the input attenuator (x1, x0.1 and x0.01). Following that is a passive attentuator that does the x1, x2 and x5 selection. Then that's followed by another section that provides a gain of two ahead of the ADC in the microcontroller (the gain is increased to x8 in the latest version). Finally, the trigger signal is derived by a third section of the TL084, which is used as a comparator - the threshold comes from the microcontroller. The 4th section is unused.
The trace is a little noisy. The firmware does not appear to use averaging or any other technique to reduce trace noise. First impressions are that the analogue section might be picking up noise, rather than generating it itself, but I've yet to confirm that. It seems to be unrelated to the switching action of the -ve PSU.
The second - and related - problem is trigger jitter. Looking at the signal at the comparator, it's very jittery. The trace hops left to right on screen at higher frequencies. It's not unusable, but if you're expecting the rock-solid trace of an analogue 'scope, you won't be getting that here.
Of course, this 'scope will alias at slower timebase speeds. To be expected, and easy to spot if you're experienced in using 'scopes. But a trap for the unwary. Better digital 'scopes (with more sample memory) avoid this problem by running the ADC as fast as they can, consummate with the timebase speed and the available memory, so hopefully the sample rate is kept high enough at all times.
Overall, better than expected, and I can see this being a useful project for someone who fancies the challenge of putting it together. It won't replace a decent analogue 'scope - as few digital 'scopes can - but it's definitely worth it for the measurement functions.
It runs from 9 to 12V - the screen brightness is affected by supply voltage. Current consumption is just short of 120mA at 10V. A series diode is provided to protect against incorrect polarity. The micro-USB connector and 0.1" headers either side of it aren't used for this application, but they are provided for developers (the unit doubles as a dev environment for the ARM Cortex microcontroller).
I can see this mounted inside one of Yorkie's oak boxes with a clear perspex lid and rechargeable batteries hidden beneath