While it might be nice to use a $4,000 oscilloscope in a lab at a university or well-funded corporate environment, a good portion of us won’t have access to that kind of equipment in our own home shops. There are a few ways of getting a working oscilloscope without breaking the bank, though. One option is to find old CRT-based unit for maybe $50 on craigslist which might still have 60% of its original 1970s-era equipment still operational. A more reliable, and similarly-priced, way of getting an oscilloscope is to just convert a device you already have.
The EspoTek Labrador is an open-source way of converting a Raspberry Pi, Android device, or even a regular run-of-the-mill computer into a working oscilloscope. It’s a small USB device with about a two square inch PCB footprint that includes some other features as well like a signal generator and logic analyzer. It’s based on an ATxmega which is your standard Arduino-style AVR microcontroller but geared for low power usage. It looks as though it is pretty simple to use as well, and the only requirements are that you can install the software needed for the device on whatever computing platform you decide to use.
While the Labrador is available for sale at their website, it is definitely a bonus when companies offer products like this but also release the hardware and software as open source. That’s certainly a good way to get our attention, at least. You can build your own if you’d like, but if you’d rather save the time you have pre-built options. And it doesn’t hurt that most of the reviews of this product seem to be very favorable (although we haven’t tried one out ourselves). If you’d prefer an option without a company backing it, though, we have you covered there too.
Comparing this to a fully-fledged oscilloscope is silly, let alone a $4k one, even if it’s a CRT. 100kHz bandwidth is barely usable for anything in the real world other than audio.
I’ll keep my “bulky CRT DSO”, thank you very much.
exactly. And it probably doesn’t support voltages higher than whatever its supply voltage is, which is probably 5V. Anything higher and you’ll have to construct your own voltage divider. Cumbersome.
Please read the datasheet!
It has an input range of -20V to +20V, and the frontend load resistor can easily be modified to increase that range if need be (at the cost of precision).
I used a USB-based instrument with a desktop computer to probe a system that was also plugged into AC. Bumped a battery + terminal with the ground lead. So much for that motherboard…
This.
You’re continuing a time-honored tradition of letting the magic smoke out. One’s not a real maker till they’ve broken something in their life. :-j
Experience gained is directly proportional to the amount of equipment destroyed.
Chris, I think this article did your product a huge disservice. Your product is what it is, and I’m sure people can find uses for it, but I’m also sure that use won’t be as an oscilloscope replacement in the vast majority of cases. My criticism is towards the article and not your product, which seems actually pretty decent.
Let me pull out my Agilent catalog and see if I can find something better for $29.
I really don’t understand your complaint. It seems like you’re complaining about the article as if you only read the (admittedly) clickbait-y title:
“While it might be nice to use a $4,000 oscilloscope in a lab at a university or well-funded corporate environment, a good portion of us wonβt have access to that kind of equipment in our own home shops. There are a few ways of getting a working oscilloscope without breaking the bank, though. One option is to find old CRT-based unit for maybe $50 on craigslist which might still have 60% of its original 1970s-era equipment still operational.”
What exactly about that seems like a head-to-head comparison to you?
I purchased 4 of these oscilloscope’s
In working condition from a college in
NJ
I was wondering what price I should attempt to sell them for and is there a place that would purchase all four from me
Your best bet is a Ham Radio Flea Market.
Don’t expect to get rich.
keep ’em all for parts
When I was a kid the electronics rags always had the ad in the back for the plans to turn an old B&W to into a scope in that price range. How times don’t change. Back then I think the scope was more audio bandwidth, so things have gotten a wee bit better, but still. Another piece of click bait on HAD.
Yeah the CRT scope comparison is not exactly correct
getting a 10mhz CRT scope is quite easy and i think takes away from the actual usefulness of the device for lower frequency projects and data acquisition
MHz != mHz
I’d like to hear about someone who came across a useful reason to measure something in millihertz. I’m sure they’re out there, somewhere.
Subsonic waves are being recorded on several places all over the world to check if somewhere nuclear experiments are being performed.
You can also use that range to detect earth quakes
Yes, earthquake L-waves can be lower in frequency than 1 mHz.
Biomedical signals. Respiration’s active band is usually 0.1Hz but I could see setting your high pass down to 0.01Hz.
Baseline wander for ECGs is similar, and you may want to capture that signal as well.
Millihertz matters.
If doing anything with digital, you want a scope that’s at least 5X the bandwidth of the highest frequency signal you will be probing. 10X is better. E.g. a 1MHz square wave has 1, 3, 5, 7, 9, … MHz frequency components. A 1MHz scope will show you a distorted sine wave. A 100kHz scope will show you an attenuated or aliased sine wave, at best.
100kHz is probably useful for audio. Questionable if you have a switch-mode audio amp. Keep in mind that you need to see any oscillations, which may be high frequency.
More likely your 1MHz square wave on a 1MHz scope will show a very very slow square wave as you alias the slight difference between sampling and the signal. The 100kHz will show a mess, depending on you mean bandwidth or sample rate.
The 3dB bandwidth is much higher than 100kHz, but I advertise it as such because at 750kSa/s you can’t usefully measure things much higher than that frequency.
Also worth noting is that there’s a digital mode that samples at 3MSa/s – obviously only one bit per sample though, so can’t see rise or fall times.
Comedicles is correct in saying that it appears as an aliased square wave. You’ll definitely want more than one sample per clock cycle!
You’ll spend more time debugging the makeshift oscilloscope, than debugging your actual circuit.
Not worth the money and the time. A used $50 crt oscilloscope from Ebay will save you time and money in the long term.
Not at all, actually.
Just download the software, plug in the board and a live trace will appear on your screen. Run the calibration routine it and said trace will be accurate to within tens of mV.
I spent two and a half years actively developing this, and a huge chunk of that was making sure that users didn’t have to deal with show-stopping bugs. There’s a full 80+ hours of work in there that uses the XMEGA’s timer counter to eliminate clock skew between the USB host and the device’s internal clock.
This is INDEED an excellent device and well done for making it, providing schematics and code thats all open source, really does inspire.
there are DIY solder together oscilloscopes on aliexpress too but your design helps MAKE USE OF old tech we got lying around not being used anymore like that old android tablet/s that are pilling up in everyone’s rooms LOL
Very nice to be able to re-use tech like that – stopping/slowing down the garbage dumps & land fills….
For a tiny device that can do al that – hats off to ya matey !
Any of you already-haters out there – YOU TRY making something useful before overly criticizing some fellow who’s got off his backside and one something POSITIVE… Instead of nagging and winge’ing… lol
What have you made today ? !!!
and Done something POSITIVE
When I said ‘debugging’ I also meant not having proper probes, impedance mismatch, EMC problems, and ground current from the PC’s power supply, and the hundred other problems you will run into, that the designers of the oscilloscope have already solved for you.
I have a LHT00SU1, that someone mentioned here, and we have PC based oscilloscopes at work, but nothing of that is really worth the trouble.
Those problems you mentioned by name are problems whether you’re probing with a $1 scope or a $1million scope.
Absolutely. I designed the thing, and if you’re doing serious work where EMC and noise from ground loops is an issue, I’d recommend spending the money on something more serious. But for hobby-level stuff, a (stable) USB scope is very useful.
Would I simply use alligator clips with soldiered on wires to use this to probe circuits?
Depends on how much fun you are having working on the scope. If you want to do one as a hobby project, enjoy!
OTOH, if you want to get a different project going, this will slow you down.
An oscilloscope is selling a few thousand, it has technology and accuracy built within. $50? What is the result are you going to get during your circuit troubleshooting?
Jack of all trades master of none. My good old CRC 569A might have some problems regarding horizontal sweep, but I’d trust it more than this gadget.
Geesh I have lots of old crt and DSO cros – though this one might be a useful training example for a few techs still brewing into from electrical cognitive overload not yet exposed to electronics, they need a zap to induce more capacity to get them out of their comfort zones but, sadly face resistance, what are they thinking ;-)
Exactly. for the newbs just wanting to get into electronics and electrical engineering – something like this could suit newbie’s wallets and help learn about how to do things and in the meantime contemplate about how to get a PROPER Oscilloscope…
This aint exactly been serious to be comparing it to a Β£4K Oscilloscope – that one really hit ya’ll hard eh ?
This is a serious waste of effort. You can buy inexpensive USB scopes already for not much more (there are several) that will far outperform this thing. A $4k lab scope is a cheap one BTW. If you need a good quality, capable, affordable scope, save for a few weeks and get a Rigol. I guess this is a fun tinker toy but for $30 there are far more useful tinker toys.
I’m curious about your list of “far more useful tinker toys” for $30. I already have a (much better) scope.
The hantek USB hardware scopes aren’t terrible, but they are let down by the software. Any suggestions for a FOSS equivalent of sigrok?
> Any suggestions for a FOSS equivalent of sigrok?
Well, Sigrok?
Well duh! I thought it was only for digital signal analysis. In fact it supports my hantek out of the box!
try a dscope much better than this hantek crap.
can confirm, its the jam
I guess this thing will find some good use for someone.
After all you’ll get what you pay for.
What makes me anxious is, who is going to buy devices like this one.
Is it pepople who want to start a new hobby at a low budget?
Or is it people who know well about capabilities and limitations of this low budget device?
I.e. it is obvious to me that I cannot use the Labrador for HF Debugging.
What I really like about this project is the user interface, as mentioned somewhere in the github/wiki a lot of design time was spent for software.
Just from the label I would say, software is where this project really shines.
for HF up to 30MHz, you are best served with an ADC of at least 120Msps. Which, in turn, will choke a USB 2.0 connection to your PC. That means adding DSP filtering to the ADC, or moving to USB 3.0 interface.
USB 3.0 is the better option, but it excludes a lot of cheaper or older PC hardware.
Or buffer… not like it has to be realtime, memory is cheap.
Actually you won’t get what you pay for with the crappy 8-bit AVR for anything to do with ADC. There are much better chips from most vendors. A low end STM32F030 below $1 (qty 1) has12-bit ADC @1M samples/sec with DMA, hardware compare and 4K RAM.
The F030 doesn’t have USB. But the F070 does.
And they’re only a smidge over $1 in volume.
Thank you. This chip looks like it could get the same performance as the current XMEGA design for half the price. The 5th DMA channel would be handy too.
Or for less money you can buy an LHT00SU1, that does all that the Labrador does, but with more bandwidth and channels. Seriously, who thought that an ATMEGA series chip was a good choice for this kind of application?
> Seriously, who thought that an ATMEGA series chip was a good choice
Yes, really wired. Specially since such devices based on AVRs are available since many years and they have all the same problem, very limited bandwidth. This will not change, just because a AVR does not have the power.
> Seriously, who thought that an ATMEGA series chip was a good choice
That would be me! Although it is important to point out that it’s an AVR ATXMEGA, not an ATMEGA.
It still has an 8-bit CPU, but there’s a much, much greater density of peripherals (especially the analog peripherals).
http://ww1.microchip.com/downloads/en/devicedoc/atmel-8387-8-and16-bit-avr-microcontroller-xmega-a4u_datasheet.pdf
As for why I chose that particular chip, the scope was designed so that every sample read by the microcontroller is transmitted to the PC and buffered in main memory. This provides both unlimited sample depth, as well as the ability to zoom in on past data without losing detail. It’s actually the thing I’m most proud of about Labrador – I haven’t seen anyone else do it that way and the result is much, much nicer to use than USB scopes with on-device buffers.
However, this means that the amount of samples per second that can be processed is the limited by the bandwidth of the USB stack. In the case of USB-FS, this means 1023 kB/s for the primary stream. At 8 bytes per sample, this is basically 1Msa/s. As a side note, the chip’s ADC has a sample rate of 2MSa/s, which is actually quite high for chips around this price point, however anything above this 1MSa/s would have avoided performance bottlenecks.
As for how these samples get from the ADC to the USB PHY, it’s almost entirely DMA. Every 64 cycles, an ADC sample complete event is fired, which triggers a DMA channel to move that sample from the ADC register to a buffer in SRAM. On the USB SoF (every 1ms), a USB-DMA transfer is set up to move this buffer from the SRAM to a FIFO in the USB controller and when requested the USB device transmits the packet to the PC.
In short, the CPU doesn’t do much. It is only used to set up the peripherals and respond to interrupts. You can actually look at the source code, the main loop is just a series of nops.
Yes, even when I began development in 2015 there were 32-bit ARM devices that had 10x the CPU performance of the XMEGA for the same price, but the CPU is not in any way a performance bottleneck. What mattered was that the IC was peripheral-rich and that there was an effective way of connecting these peripherals, and ATXMEGA was the lowest-cost chip I could find that met every performance requirement. The more powerful ARM devices at XMEGA’s price point either had a slow ADC, no DAC, or lacked an on-chip USB PHY (if PSoC4 had that, I would have definitely selected it instead).
Still to this day I can’t find a chip better enough to justify a redesign, although that being said I’d be happy to take suggestions. Would be especially interested in something with USB-HS and a >20MB/s ADC, for a higher end version of Labrador.
I hope that makes some sense. It’s quite late in Melbourne.
Haters gonna hate, but your design has merit. I do resonate with those that say the input bandwidth of the scope function is too limited to be useful for long. That said, you mentioned USB limitations that restrict the bandwidth. Is there a way to get the output of your device to a format other than USB? Or is that a design limit of the Atmel chip?
USB 2.0 can do about 40 MB/s if you have a proper controller pushing the data through. This is a limit of the AVR.
Luke, USB 2.0 has multiple speeds.
FS has a bandwidth of 12MHz (approx 1MB/s), HS has a bandwidth of 480MHz (Approx 40MB/s, as you said).
Most low-cost microcontrollers with in-built USB 2.0 PHYs support only the slower 12MHz mode.
Walter, unfortunately not. Other Microcontrollers might have faster SPIs, but I’d then need a second IC to translate that to USB (and it’d need to be using Isochronous transfers for this application). Those ICs usually cost more than the micro itself, to the point where it would make more sense to get one with an in-built 480MHz USB PHY.
I started in electronics in 1961. In a dirt floor grarage. Went into the Army to work on microwave communcations. I had to build a lot of my own test equipment. I understand designing things to do a job with limited resources and constrances. It looks like a find tool to do the job you set out to do. I’m interested to see what you come up with next. Keep up the good work.
Rich Little
Illuminating – thx for glimpse into developer’s mindset.
Last time I’ve tried http://openhantek.org/ was way better than sigrok.
Awesome, thank you!
I got the bitscope, and it just about does the job for me, however is more cumbersome than I thought. You can get a rather thin 100MHz osc from siglent with color screen, probes and all for around 300$
You can buy pocket DSO of eBay for same price with bandwidth of few MHz and has color LCD and includes probes and open sources software.
I have found MEGA2650 (SUMP firmware) to work nicely with OLS (OpenLogicAnalyzer) as 4MHz logic analyzer (8 ch, 7K samples) at no extra cost.
https://github.com/gillham/logic_analyzer
Or just grab a cheap FX2 board. (I would like to see it updated to support FX3 – something like 32 channels at 100MHz!)
You have 100mhz
While a lot of effort has been put into this project and the designer deserves some credit for the work they have done, hyping it as a replacement for a proper oscilloscope is a joke and has caused all the negative remarks. If the article had touted this as a pretty good effort as a replacement for a USB scope then I don’t think you would have got so many negative comments.
This is a neat little project for as an open replacement for a basic USB scope and would be seen in a much more positive light if presented as such.
Thanks for that, Darren.
Your explanation makes a lot of sense.
To be honest, I was a little bit surprised at the negative feedback. The device has been out for almost two years now and the feedback I’ve received in that time has been overwhelmingly positive.
There’s very few products at this price point that have had as much design effort put into them as Labrador and all of the customers I’ve spoken to understood the bandwidth limitations and how this affects what the scope can and can’t do.
Of course the device is not a direct alternative to a DS1054Z, but a lot of people have found it very useful and not all of them can justify the money or bench space for a fully-specced scope.
Nevermind the naysayers. Most people probably feel the need to justify their $$$k purchases, or to brag they got a really good one for $50.
But Ebay/craigslist doesnΒ΄t work like that for all the world.
And for applications that doesnΒ΄t need a high bandwidth, this is good. And those uses where one can easily fry the oscilloscope. Beter fry a cheaper part that you can have a spare stashed away, than something thag cost $4k and is hard to replace/justify the fixing cost.
Along with that , have you considered using some of the Cypress ( or others ) usb3 interface chips ? That could help alleviating transfer speed to the computer …
I have considered using a faster USB PHY, but then the ADC would limit the bandwidth. And if the ADC was upgraded, then the frontend would limit bandwidth. And if I upgraded the frontend, then the probes would would be the limiting factor. It’s unfortunately not an easy problem to solve without a full redesign!
Not all that worried by the naysayers. I’d probably be a bit crushed if this was the feedback I got on launch, but everything else has been pretty positive and I think people are just being a bit reactionary.
Besides, I’ve seen 2 months’ worth of orders in the few hours since the article went up. I have no right to complain!
“Of course the device is not a direct alternative to a DS1054Z”
With the title of the article suggesting it is an alternative to a real scope, you shouldn’t have been surprised at the negative feedback.
While you may have put a lot of work, and come up with a good design given the constraints, I think you made some bad choices for the initial constraints. It seems rather obvious to me (and likely many others here) that you should’ve gone with a USB 2.0 (480mbps) interface. Off the top of my head, the Cypress FX2LP is under $10. Other people may be able point out even better options.
However there’s already PC scopes out there with decent bandwidth for under $100 (i.e. Hantek), so it may be a tough market to compete in.
I think the value in the work you’ve done is in the software, and if you were to port it to work on the STM32F103, then everyone with a $2 blue pill could use it. Something like that would be great in a classroom where it is often hard to find the money for a real scope for every student.
One shouldn’t. This is after all a tech forum, not Forbes. Things can get a little rough.
Very well said, these folk CLEARLY cant take a joke lol
The main downsides of this scope is though its lack of bandwidth.
750 KSPS doesn’t get far to begin with, especially when it is shared between the two channels.
A resulting 100KHz bandwidth is a bit low for most thing. The 12 bit ADC could be useful for some things.
But at the end of the day, a Rigol DS1054z costing about 400$ having 4 channels and a 50MHz bandwidth out of the box is arguably far more performance for the money. And 400$ isn’t all too expensive for an oscilloscope. (Though, there are cheaper scopes around with similar 50MHz bandwidths on more then one channel.)
But if one’s budget is really constrained, then 30$ might be reasonable for 100KHz.
Though, a DS212 costs about 100$ and has 1MHz of bandwidth (on both channels at the same time), though it might not be as polished on the software side of things, but it is open source. (in the end it has a 10x increase in bandwidth for about 3-4x increase in price.)
Then one can always search around on Ebay for second hand scopes. Though this is can be a lottery…
I coach a team of school robotics students, and this looks really interesting. At $29 a board, I can afford 3 or 4 of them. If one gets toasted, as sometimes happens, we wouldnβt break the bank. Perfect to check if both channels of an encoder have output, and if the phase is correct. Perfect to check if the I2C, SPI or CAN bus is talking. We donβt need to decode and debug protocol. We just need to see if itβs there. Perfect to log voltage over time.
Iβm not impressed by all the βit will never workβ and βjust buy this on eBay or Amazonβ posts. I read HAD for whatβs new and interesting. Those comments are neither.
I’ll second the recommendation for a Rigol. 400 bucks gets you a solid, lightweight but not cumbersome dual channel 50mhz scope. It’s very easy to take screen captures which cam be retrieved via USB for your write-up.
Picked up one of these thinking it would just be a good cheap signal generator to play around with, and it greatly exceeded my expectations.
The trigger mode works like a charm and the logic analyser was a tool I never knew I’d need until I tried it out.
Obviously for anyone serious about their electronics, paying extra for a heavy duty oscilloscope is a worthwhile investment.
But I’d still recommend the labrador to anyone just wanting an easy way to get started. And it’s come in handy quite a few times as an all-in-one tool when I’m away from the lab working on group assignments.
p.s. thanks espotek for sharing the code
Would this be helpful to someone like me, a spare time tinkerer? I have been trying to find the uart settings for a doorbell I took apart and havenβt been able to get the aerial connection working despite trying most every combination. (Screen garbage.)
Perfect match I’d say
A real scope is much more than just an ADC on a microcontroller. BTW Atmel s one f the worst choice for its slow ADC which can only be triggered from sotware. On ARM chips, ADC can be directly triggered from timer, with higher 12-bit or even 16 bit resolution, faster sampling rates easily into mega samples/sec, multiple ADC, hardware compare of ADC and much more RAM.
A real scope has decent analog front end, repeatable triggering circuit, decent timebase, signal processing etc. These are also important for their oversampling feature allowing them to piece together a higher time resolution from repetitive waveform.
I guess if you call this a scope, you haven’t really learn to appreciate one.
I think this is great for teaching basic electronics. I am a MakerSpace organizer; and resources are so very limited. For years I have used the Gabotronics Breadboard scopes; my only negative with that is they would never fix the GUI to scale to the PC’s display so critical controls were below screen where the mouse could not reach them. I am also a professional with 30 years of experience with lasers, fiber optics, radio, avionics, embedded computing and am astonished how opinionated some of these comments are! Be more positive for god’s sake.
It is quite strange that some semiconductor manufacturer does not have any low cost analog-oriented DAQ SOC in their product line. Some quad core OpenRISC with 1GSPS ADC coupled with 64MB sample RAM and a medium density programmable logic matrix for triggering etc. And of course a monolithic front-end with switches, attenuator/PGA, S&H and all the requisite stuff. Something like Zynq, but more integrated and for $15 a pop. I think that it would be beneficial for a lot of applications like SDR, automotive, industrial etc.
I read the title with apprehension…
The (older) 120 MHz Digital Sampling O-scope (CRT) I recently purchased (sub $100) was a waste of money?
Whew! I’m glad it is not so!
The software is what makes it attractive. I’m ordering one now. Is there a schematic without using KiCAD to look at it? I need to see the front end.
Shoot me your email address and I’ll send you a PDF.
The ADALM2000 is a nice USB Analogue+Digital input device
https://www.analog.com/en/design-center/evaluation-hardware-and-software/evaluation-boards-kits/ADALM2000.html#eb-overview
I’m ordering one, just because I like Chris Esposito’s attitude.
So much inappropriate hatred here, which I hope that Chris Esposito won’t take seriously. I have audio-frequency applications where it would useful to have a simple, inexpensive waveform monitor and signal generator permanently attached in various locations, and the Labrador would be π±π¦π³π§π¦π€π΅ for that. I intend to order a handful or two of them.
this kida hatred in the electronics world is rife and STILL shocks the hell outta me at times, where there absolutely no need for it even being or drawing breath in positive places like hackaday or any other place where we love coming for advice, how to’s or to showcase our latest collaboration or ouw own concoction..
cant really rule out all the haters are from just one place too, they are all over the place lol, one could be looking over your shoulder right now, LOL
Unfounded nonsense in the highest order, usually by people who feel threatened by the people they are hating on, thus far, thats how iv seen the game unfold anyways..
Leave that crap at the door or dont bother coming in lol
If I didn’t already have a nice Fluke 97 I would be all over this like white on rice. Neat project again on Hackaday…..
I’ve had an idea for a while to make an “electronics tricorder” basically just a handheld device that would have basic multimeter, logic, oscilloscope, and signal generator abilities. Doesn’t have to do one thing really well, just all the things adequately. This would work quite well for such a project.
PERFECT !!! Now this is the way forward !!!
Downloaded the ‘Labrador’ S/W from https://github.com/EspoTek/Labrador/releases/tag/continuous-win32-build188 & Avast AntiVirus blocks with: “Infected with IDP.Generic” – ??
Please submit a sample to Avast if they give you the option.
When the software first came out, McAfee used to give a similar warning. Got a lot of complaints about that for a couple of weeks until they updated their definitions.
Note that IDP.Generic doesn’t mean that the software is malicious, but that their machine learning algorithm has decided that, for whatever reason, it looks like it might be malicious.
I believe that they pick it up because it bundles libusbK (which has a kernel component) as part of the installation.
I’ve scanned it with Virus Total just in case this is an issue with other AVs and it looks to be clean:
https://www.virustotal.com/gui/url/c6fa7820aea9e8f1986887e9eea0c9bdecb495c672c47cf557a060d1ad7fbbc8/detection
You’re welcome to do the same and post your results here.
Thanks for reply – yes, maybe Avast giving false positive,,