05:20 offset & noise amplitudes may wander a bit during the first few minutes after power-on. I've done part-per-billion high-precision frequency measurements [flicker-free at 28 bits in a 32-bit resolution counter] where I need to wait ~ 20 minutes for the reference oscillator to settle down. Patience is a virtue - set up the test, power it on, and head for the coffee machine. Thick wall shielding box wrapped in a bath towel is another virtue, not only for EMI rejection but to add thermal mass. Gotta love these high-resolution measurements - everything is a thermometer if you peer deeply enough into the swamp.
@@walterdove8727 ...Agreed! Audio transducers are best evaluated either in an expensive anechoic chamber or outside in the great outdoors, if you can find a sufficiently quiet location [without rainfall]. Experiment design requires thoughtful attention to context.
From the high resolution measuring guys(toolmaker, etc), "everything is rubber" + so easy to influence the measurements just by people moving on the same part of the slab they have the testing stuff setup on
Well that is the best case scenario. In practice people will use the test instruments soon after power up, so you better measure the worst case scenario.
This is more proof that TI still makes the best parts in the industry, analog or digital. And when TI gives a spec for a product it is a real spec. Not an idealized specification like most other manufacturers give (whenever possible) for their parts. PS: The in depth looks at this type of thing are great! Thanks for the look at how you do this type of testing.
They write the DS to match the product after running through production testing. This gives you the bounds for voltage, temp and process skews. If the spec is good enough, they release, If not then it needs some silicon tweaks. ALL reputable manufactures do the same or similar, Maxim, ADI, etc. What you really want to look for is Max vs Typical specs. That shows those that have tested widely and have a very good product, vs those that don’t....
They might make good parts, but their user experience, when working with more complex parts is horrible. When I buy a current product from them and need to use some ancient version of Code Composer Studio based on ecplipse, which is horrible enough, something is very wrong. On top of that, they are pretty monopolistic in many regards which is never good
Thank you so much. I don't need a lower noise floor uCurrent, but that doesn't keep me from wanting one badly. As soon it's out, I'll upgrade. If they werent so expensive I'ld update one myself just for fun.
I don't think you can; the battery holder on my Rev 5 uCurrent Gold has a battery connector that would touch the +ve terminal of both batteries. If you were on the moon and had some insulating tape you could cover the +ve terminal of one battery to make that work.
@@martinda7446 Back then, I could only find the PLASTIC ones.. even tried "coating" a plastic one in aluminum paint.. never saw an all metal cast one- (I ended up using a metal box of sorts- not cast, but just shaped and spot-welded that worked- but a cast one (6" x 6" x 3") square configuration) I could never find) in-short, if Radio Shack didn't have it, it didn't exist. )
I used the OPA189 in a project recently, it also handles saturation very very well with some input resistors (never tested it without) at least in my testing of 3 for a few minutes
Hi Dave, regarding the high frequencies spikes for the OPA189 opamp. If you vary the analyser sampling frequency you could rule out an aliasing effect.
Iceteavanill, ah didn’t realise the 121GW wasn’t CAT IV rated. I didn’t hear any announcements or news either, but I wouldn’t notice if it was posted on twitter or the EEVblog forums for example.
3:45 Are these not supposed to be the same? Maybe some of the feedback components have to be changed to make the new one accurate? Maybe they are close enough for noise tests.
I'm curious what people are measuring with the uCurrent that such low noise becomes significant. I seem to remember that the uCurrent was originally developed as a solution to deal with DMM burden voltage, so wouldn't noise voltage on this order of magnitude be a "nothing burger"? I'm also curious how you intend to power the OPA189 uCurrent, though it occurs to me that the case it comes with could easily accommodate a 3 AAA battery holder.
Good point. I could have needed a μCurrent when I needed to measure sleep current of a data logger in the μA range (with the microcontroller occasionally waking up and consuming 10mA or so). Wouldn’t have needed much accuracy there.
It seems that the gains aren't really worth the rework and versioning trouble. Good to see that such an old design is still holding its own with the newer competition!
EEVblog Agreed. Still, I wonder how many people are using the uCurrent at or near the edge of its operating envelope... Regardless, the comparison series is super interesting and I think it’s great that you continued working to make the device the best it can be (in a price bracket that makes it obtainable for the masses).
Great video. Interesting and informative. If you’re asking should you bother filming and uploading this, then YEP! 👍 You shouldn’t have to ask in my opinion 😝
Awesome series. Maybe you could do a 'fundamentals friday' and give a simple overview of your bench tool choices here? I know the move to the old lab has probably had an impact, but I'm curious why, other than the sig scope (and maybe that keysight DVM), most of the tools used were out of pocket acquisitions. I know these tools are top shelf stuff. I would like a better understanding of the thresholds dictating their use. As a hobbyist, I haven't progressed to the point where I need to dive deep into the specs and math, and probably never will... If (future +~5 years) Sagan were to ask 'why these tools vs newer featured stuff like those R&S PSU's' how would you explain it to him? Then, (assuming he has chosen to follow in dad's footsteps), how would you explain alternative techniques to make similar measurements with his more basic Uni-dorm-room limited bench kit to get an additional edge up on his studies?
IS THAT A NEW 121GW!? Looks like it lost a few pounds 👍 And watts? Milliwatts? Hell yes. Finally some accurate power draw readings for all these weird ass micros. P.S. 4 wire measurements pls? I know it's unnecessary but it would be so hot 😁
I wonder why you not use a i-u op amp converter instead shunt with amp? This will allow us to use only one precision resistor instead of a precision 4 resistors in the amplifier and shunt.
The video content was good. Thank you! Too bad the videos are still being made in an improvised bench corner. We like to see the bench full, with all the equipment and space. In a way the channel is more amateurish than it was at the beginning in the garage. I hope that one day the bench will be complete and tidy.
Why are there more noise at low frequency for opa189? Looks like 1/f noise, which is not suppose to be there with chopper opamp. Also if you don’t terminate the BNC in the box, then aren’t you just measuring the Johnson noise of the input impedance of the DSA which is 1Mohm which might contribute to high noise floor.
That DSA has the usual 1/f noise all of its own. There’s a reason they only have 400 lines of frequency resolution: make no sense to have more since it’s very meh around DC from what I could see.
Kuba Ober the 400 line limit is probably because of the processing power not much about the 1/f corner. Pretty much all DSA have 1/f just how low you can push the corner. In the case of SRS they have a
When Dave asked, 'Can we find better after X years?' I knew the answer. Not really! (if he did a decent search before). Since the late 70s I don't think cutting edge op amps have changed that much. Especially in audio people have been poking about for best noise performance for 40 years and the shortlist just stays the same.
@@a1nelson Yes, I think more basic physics. The designers choose the areas they want their IC to perform well in - which usually comes down to bandwidth and noise and cost.. often they make a great job of sucking the last ounce out of silicon. It is circuit design too, which certainly has not advanced in 50 years. We use the same building blocks now as we did then. (This is my guess).
I think the price difference has been mentioned between the two parts. The other parameter for the part that hasn't been mention is the age of them. When did each part first hit the market and what is the expected life of the part? Don't want to switch to the OPI189 only to find out it will soon become unavailable while the original continues to be available. On the other hand, if the original part turns in to unobtanium the OPI189 can take its place.
Nice piece of old HP equipment, but why not just using a decent audio device an measure it in software? SNR numbers of 115-120dB are very doable nowadays, even (much) more when you average over time. In worst case you can even amplify the signal. Not that there is anything wrong with this setup, but just an idea for people who can't find or effort such equipment. Good video Dave!
Once your kids are too old to suffocate themselves chewing on plastic bags, we let them back into the house along with polyester fabrics and adopt a less overgeneralized vocabulary for them a as other than “paper”, “plastic”, and “gross”. But if you still babysit, ‘Meh’ should do fine.
That’s what we call a spa in hotels. Jacuzzi at home, whereas a billiards hall which in a pub is also a pool hall, but is not a pool room. So if you’ll not be going swimming at a hotel, then accuracy matters.
05:20 offset & noise amplitudes may wander a bit during the first few minutes after power-on. I've done part-per-billion high-precision frequency measurements [flicker-free at 28 bits in a 32-bit resolution counter] where I need to wait ~ 20 minutes for the reference oscillator to settle down. Patience is a virtue - set up the test, power it on, and head for the coffee machine. Thick wall shielding box wrapped in a bath towel is another virtue, not only for EMI rejection but to add thermal mass. Gotta love these high-resolution measurements - everything is a thermometer if you peer deeply enough into the swamp.
In this case, good enough for Australia. But yeah, for really precision stuff settling if very important.
Thanks for that colour jrb. Super interesting insight, especially thermometer observation.
@@walterdove8727 ...Agreed! Audio transducers are best evaluated either in an expensive anechoic chamber or outside in the great outdoors, if you can find a sufficiently quiet location [without rainfall]. Experiment design requires thoughtful attention to context.
From the high resolution measuring guys(toolmaker, etc), "everything is rubber" + so easy to influence the measurements just by people moving on the same part of the slab they have the testing stuff setup on
Well that is the best case scenario. In practice people will use the test instruments soon after power up, so you better measure the worst case scenario.
Three 20 minute eposiodes means it's the microamp hour!
Definitely enjoying the "series" and look forward to more.
This is more proof that TI still makes the best parts in the industry, analog or digital. And when TI gives a spec for a product it is a real spec. Not an idealized specification like most other manufacturers give (whenever possible) for their parts.
PS: The in depth looks at this type of thing are great! Thanks for the look at how you do this type of testing.
They write the DS to match the product after running through production testing. This gives you the bounds for voltage, temp and process skews. If the spec is good enough, they release, If not then it needs some silicon tweaks. ALL reputable manufactures do the same or similar, Maxim, ADI, etc. What you really want to look for is Max vs Typical specs. That shows those that have tested widely and have a very good product, vs those that don’t....
ti making the best parts is kind of an overstatement.
They might make good parts, but their user experience, when working with more complex parts is horrible. When I buy a current product from them and need to use some ancient version of Code Composer Studio based on ecplipse, which is horrible enough, something is very wrong. On top of that, they are pretty monopolistic in many regards which is never good
Yes please continue with this series.
Sorry that i don't comment all your Videos. I learned so much with your channel 😀. Greetings from Austria 🇦🇹
Using the uCurrent to measure the power supply current of the uCurrent is a very pleasing recursion, lol
Thank you so much. I don't need a lower noise floor uCurrent, but that doesn't keep me from wanting one badly. As soon it's out, I'll upgrade. If they werent so expensive I'ld update one myself just for fun.
Enjoying this series very much thx
You can put stack of two CR2016 instead of one CR2032 and get 6V without changing battery slot.
I don't think you can; the battery holder on my Rev 5 uCurrent Gold has a battery connector that would touch the +ve terminal of both batteries. If you were on the moon and had some insulating tape you could cover the +ve terminal of one battery to make that work.
LOOOOOVE the box! (thats EXACTLY what I NEEDED for back in my pirate radio days.. and now I see one 30 years too late! oh, well... )
What? The bog standard cast box that has been that way since..er 1960??? At least..
@@martinda7446 Back then, I could only find the PLASTIC ones.. even tried "coating" a plastic one in aluminum paint.. never saw an all metal cast one- (I ended up using a metal box of sorts- not cast, but just shaped and spot-welded that worked- but a cast one (6" x 6" x 3") square configuration) I could never find) in-short, if Radio Shack didn't have it, it didn't exist. )
@@WarpRadio We had our Maplin catalogues back in the late 70s.
I used the OPA189 in a project recently, it also handles saturation very very well with some input resistors (never tested it without)
at least in my testing of 3 for a few minutes
What happened to the silkscreen at the bottom right @2:07, Mr. "I vi I L. J n-s" ;)?
Absolutely no idea!
I thought it might be intended as an amusing way to detect blatant knockoffs. Pretty sure everyone buying a uCurrent knows just who designed it.
After seeing it here in the comments I picked up on what it says. "Mr. Evil Genius". :)
Loved this series.
I like this edited videos. Shorter and complete. 👍
these tutorials are great !
Hi Dave, regarding the high frequencies spikes for the OPA189 opamp. If you vary the analyser sampling frequency you could rule out an aliasing effect.
Excellent "series" - thanks Dave!!
3:28 -> Is that a updated version of the 121GW
Hmm 🤔 I wondered what multimeter that was...
Electrical tape on the model number?
@@davadoff It also go NCV and CAT IV certification so definitely a new model. Have I missed an announcement or is is not yet announced?
Iceteavanill, ah didn’t realise the 121GW wasn’t CAT IV rated.
I didn’t hear any announcements or news either, but I wouldn’t notice if it was posted on twitter or the EEVblog forums for example.
It's good to enjoy a better performance mod, satisfying.
Opa is Dutch for grandpa. Seriously love your work Dave big thums up!!!!!!!
*Bam* just ordered one, thanks Dave!
Any plans to compare with the ADA4528 that was mentioned in the first video? Would be interesting to see how it stacks up.
3:45 Are these not supposed to be the same? Maybe some of the feedback components have to be changed to make the new one accurate? Maybe they are close enough for noise tests.
Love it. More please.
I'm curious what people are measuring with the uCurrent that such low noise becomes significant. I seem to remember that the uCurrent was originally developed as a solution to deal with DMM burden voltage, so wouldn't noise voltage on this order of magnitude be a "nothing burger"?
I'm also curious how you intend to power the OPA189 uCurrent, though it occurs to me that the case it comes with could easily accommodate a 3 AAA battery holder.
Good point. I could have needed a μCurrent when I needed to measure sleep current of a data logger in the μA range (with the microcontroller occasionally waking up and consuming 10mA or so). Wouldn’t have needed much accuracy there.
So Dave, when do you want floating measurements instead of it being ground?
It seems that the gains aren't really worth the rework and versioning trouble. Good to see that such an old design is still holding its own with the newer competition!
Lower noise, higher bandwidth, and greater dynamic range supply are pretty useful benefits.
EEVblog Agreed. Still, I wonder how many people are using the uCurrent at or near the edge of its operating envelope... Regardless, the comparison series is super interesting and I think it’s great that you continued working to make the device the best it can be (in a price bracket that makes it obtainable for the masses).
Great video. Interesting and informative. If you’re asking should you bother filming and uploading this, then YEP! 👍 You shouldn’t have to ask in my opinion 😝
How is the work going on the improved HP 3566x frontend?
Would be really interesting to see if it's possible to improve that instrument.
Wasn't easy. Used a dual matched fet, and changing this would potentially have screwed up all sorts of stuff.
Dave, I think you need to get yourself a uCurrent so you can measure the current of the uCurrents
It's turtles all the way down.
I would have tried to screw the enclosure cover. It might have significantly increased the shielding.
Awesome series. Maybe you could do a 'fundamentals friday' and give a simple overview of your bench tool choices here? I know the move to the old lab has probably had an impact, but I'm curious why, other than the sig scope (and maybe that keysight DVM), most of the tools used were out of pocket acquisitions. I know these tools are top shelf stuff. I would like a better understanding of the thresholds dictating their use. As a hobbyist, I haven't progressed to the point where I need to dive deep into the specs and math, and probably never will...
If (future +~5 years) Sagan were to ask 'why these tools vs newer featured stuff like those R&S PSU's' how would you explain it to him? Then, (assuming he has chosen to follow in dad's footsteps), how would you explain alternative techniques to make similar measurements with his more basic Uni-dorm-room limited bench kit to get an additional edge up on his studies?
Is uCurrent used to sense High side current or just Low side?
How come a piece of tape on the meter please ?
Love it!
At low frequencies, a steel box will provide much better shielding...
Do it with LM324! Just to compare regular OP-AMP like that to better ones
if anything a lm358 because you dont need the 4 amps in the 324
2:12 That's a triple triple A battery holder.
So will there be a µCurrent Platinum with the new opamps?
It is 'usual' and Silver, I think. :)
IS THAT A NEW 121GW!?
Looks like it lost a few pounds 👍
And watts? Milliwatts? Hell yes. Finally some accurate power draw readings for all these weird ass micros.
P.S. 4 wire measurements pls?
I know it's unnecessary but it would be so hot 😁
I wonder why you not use a i-u op amp converter instead shunt with amp? This will allow us to use only one precision resistor instead of a precision 4 resistors in the amplifier and shunt.
where to buy
I wonder what is practical application of such device?
Care to comment on the 1/f noise of the 189? The 4239 didn't appear to exhibit that.
189 can only be lower.
DC-10Hz is supposed to be lower, need to re-run for say DC-100Hz bandwidth.
@@EEVblog Was the measuring equipment FFT based? DC offset will bleed into the low frequency range.
In my mind’s eye I can see a ferrous project box arriving at a lab in southern Australia’s lab .
@@HighestRank That's what biscuit tins and Altoids boxes were designed for.
You can use OPA333 to replace LMV321
The video content was good. Thank you! Too bad the videos are still being made in an improvised bench corner. We like to see the bench full, with all the equipment and space. In a way the channel is more amateurish than it was at the beginning in the garage. I hope that one day the bench will be complete and tidy.
The OPA189 switching frequency is tpically around 300 KHz
I put an ~50 KHz output filter on me visible light oscilloscope probe to dump the chopping noise.
I don't understand what the fuck this is but **I'M LOVEING IT!!!!!**
Next, lets see how they both handle transient response :)
Why are there more noise at low frequency for opa189? Looks like 1/f noise, which is not suppose to be there with chopper opamp. Also if you don’t terminate the BNC in the box, then aren’t you just measuring the Johnson noise of the input impedance of the DSA which is 1Mohm which might contribute to high noise floor.
“This is not trivial”, so expect falling right into the new-player’s trap, though I’m not sure why it’s tailing up there.
That DSA has the usual 1/f noise all of its own. There’s a reason they only have 400 lines of frequency resolution: make no sense to have more since it’s very meh around DC from what I could see.
Kuba Ober the 400 line limit is probably because of the processing power not much about the 1/f corner. Pretty much all DSA have 1/f just how low you can push the corner. In the case of SRS they have a
If Maxim still have their supply chain problems it'll be good to escape those too :-)
When Dave asked, 'Can we find better after X years?' I knew the answer. Not really! (if he did a decent search before). Since the late 70s I don't think cutting edge op amps have changed that much. Especially in audio people have been poking about for best noise performance for 40 years and the shortlist just stays the same.
Martin D A Curious: what’s the reason for the lack of progress? Basic physics, limited demand or something else?
@@a1nelson Yes, I think more basic physics. The designers choose the areas they want their IC to perform well in - which usually comes down to bandwidth and noise and cost.. often they make a great job of sucking the last ounce out of silicon. It is circuit design too, which certainly has not advanced in 50 years. We use the same building blocks now as we did then. (This is my guess).
‘best noise’- that’s audio in a nutshell.
why don't you use AD8628
Would that fix the compatibility problem with the power supply rail?
@@HighestRank yes, and AD8628 have lower noise
The EEV Blog Multimeter looks interesting ....... Is that still a secret?
I think the price difference has been mentioned between the two parts. The other parameter for the part that hasn't been mention is the age of them. When did each part first hit the market and what is the expected life of the part? Don't want to switch to the OPI189 only to find out it will soon become unavailable while the original continues to be available. On the other hand, if the original part turns in to unobtanium the OPI189 can take its place.
The OPA189 is only a few years old IIRC
So, a new micro current ?
Really worth the ic change? Maybe the Maxim its ok!!! Just a comment...
Kind of curious if the OPA would've puttered into life with the coin cell...
Nope, amplified like a drunken sailor.
When is the uSupply thing going to be released?
I dont even know what these micro current devices are used for?
www.eevblog.com/projects/ucurrent/
Measuring very small currents accurately.
Measuring current with lower burden voltage (100 times lower).
Testing a shunt resistor’s suitability for use as an antenna. “Microcurrent applications”.
I guess the new one should be the "uCurrent Platinum"
"this video is long enough" nahhh it isn't!
Nice piece of old HP equipment, but why not just using a decent audio device an measure it in software?
SNR numbers of 115-120dB are very doable nowadays, even (much) more when you average over time.
In worst case you can even amplify the signal.
Not that there is anything wrong with this setup, but just an idea for people who can't find or effort such equipment.
Good video Dave!
Dave’s luck with software setup has already been well-documented.
suspicious multimeter ... 😎
Me likey!
the pink packaging stuff in the box that is so gross. dont do that. just put nylon standoffs on the board and put it in the box directly...
Meh.
Once your kids are too old to suffocate themselves chewing on plastic bags, we let them back into the house along with polyester fabrics and adopt a less overgeneralized vocabulary for them a as other than “paper”, “plastic”, and “gross”. But if you still babysit, ‘Meh’ should do fine.
Why is your voice so high?
close enough. going straight to the pool room.
That’s what we call a spa in hotels. Jacuzzi at home, whereas a billiards hall which in a pub is also a pool hall, but is not a pool room. So if you’ll not be going swimming at a hotel, then accuracy matters.
it is a movie phrase use in an iconic Australian film called 'The Castle'. sorry for not clarification on this one
So many neins!
Whackers