Author Topic: µCurrent Gold - HW Design question  (Read 8503 times)

0 Members and 1 Guest are viewing this topic.

Offline black-earth

  • Newbie
  • Posts: 3
  • Country: de
µCurrent Gold - HW Design question
« on: February 22, 2014, 07:50:51 pm »
Hello world,

I am very interested in measuring current consumption of ultra low power wireless sensor node devices for overall current optimization via firmware.
The µCurrent Gold seems to be a realy nice tool to measure dynamic current consumption with the help of an connected oscilloscope.

Therefore I want to undersand the hardwaredesign of the µCurrent Gold designed by Dave L. Jones.
I came across a design detail, which I don't understand right now.

What is the bottleneck about located between the negative current input and the Resistor R1 (R010) (signal line name "-IN") ?
Isn't that the backpath were in "mA" settings up to 1,25 A could run through ?
 

Offline Legit-Design

  • Frequent Contributor
  • **
  • Posts: 562
Re: µCurrent Gold - HW Design question
« Reply #1 on: February 22, 2014, 08:07:15 pm »
Fusible track Dave talks about it in several places, on his Channel and on TheAmpHour. I think he mentions Saturn pcb (he demonstrated saturn pcb on one his blogs) has calculator for it?
Apparently using fusible track is not exact science, but more of a try and guess.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 15088
  • Country: za
Re: µCurrent Gold - HW Design question
« Reply #2 on: February 22, 2014, 08:38:27 pm »
Fusible track is more of a defined failure point at massive overload. Will not be exact ( like any other fuse) but will fail first before any other PCB traces, so you will have less to repair and not have all the tracks burnt off the board.
 

Offline black-earth

  • Newbie
  • Posts: 3
  • Country: de
Re: µCurrent Gold - HW Design question
« Reply #3 on: February 22, 2014, 10:16:47 pm »
Ah, that sounds interisting.

First I thought it would add an additional resistance to the 10mR shunt, but after calculation the trace resistance should only add ~305µR :-).
(But thats additional 3,05% of 10mR (which has tol. of +/-0,1%) isn't it ? Maybe my striplinecalculator is joking, otherwise it could have influence on the measurement result.) My fault, VGND is refered to a point between that bottleneck and the neg. terminal of R1, very nice!

But what about the contact resistances of the range switch ?
Datasheed only says:

Quote from: Datasheet
CONTACT RESISTANCE: Below 10 mR typ. initial @ 2-4 V DC, 100 mA, for both silver and gold plated contacts.
Thankfully this will only add up on the burdon voltage and will not influence the current measurement.

--------

Another question: The recommended operationg voltage of the OpAmps is 2V7...5V5.
Would it make sense to set Vcc e.g. up to 5V to enlarge the measurement range ? Or would it have other negative sideeffects ?

Higher currents are available by increasing the battery voltage, to a maximum of 5.5V, which would give +/-2250mA/2250µA/2250nA maximum)
Lol I skipped that text and run directly into the scematic and pcb design - sorry for that.
Seens, that i could answer many of my questions myself.

At least a 3V6 lithium battery would improve system lifetime by nice factor ... (a coin cell only holds abut 260mAh)
« Last Edit: February 23, 2014, 11:58:09 am by black-earth »
 

Offline MLXXXp

  • Regular Contributor
  • *
  • Posts: 221
  • Country: ca
Re: µCurrent Gold - HW Design question
« Reply #4 on: February 23, 2014, 11:11:26 am »
Another question: The recommended operationg voltage of the OpAmps is 2V7...5V5.
Would it make sense to set Vcc e.g. up to 5V to enlarge the measurement range ? Or would it have other negative sideeffects ?

At least a 3V6 lithium battery would improve system lifetime too ... (a coin cell only holds abut 260mAh)
The only real problem with doing this is the "battery ok" circuit. At 5V the LED will run at approx. 10mA, which is beyond the 5mA absolute maximum output current rating of the TPS3809 voltage supervisor IC (although I'm not sure why Figure 3 of the datasheet graphs much higher currents?). Even at 3.6V the current will be close to 5mA.

You could increase the value of R4 to lower the LED current. However, if you want a greater measurement range that you can trust, you would want the LED to go out at a voltage above the desired range. You would have to put in a version of the TPS3809 with a higher threshold voltage.
 

Offline black-earth

  • Newbie
  • Posts: 3
  • Country: de
Re: µCurrent Gold - HW Design question
« Reply #5 on: February 23, 2014, 12:00:38 pm »
Damn, I was editing my post while you answered :-).

I would change the voltage supervisor to meet the higher Vcc for sure. Additionaly i would use a 1mA LED with new calculated resistor.

--------

Another idea in my mind is to modify the pcb design to interface the pcb through a closed HF-shielding-case instead of a plastic case with PCB as top case-cover.
That would result in higher system-robustness and better noise performance i think.



Just a few things to do: Change the golden contacts to isolated case-installable-terminals and change switches to jumperable (ultra-low-resistance <<1mR) isolated terminal.
If using a HF-case with a snap-on lid, battery changing is not a problem (i would go for 3V6 Lithium (size AA or 1/2AA) anyway to improve battery life. Maybe you can omit the On/Off switch if system current is low enough.
The voltage supervisor IC can be changed to check on an other BAT level for 3V6 baterys.

Omit/replace the range switch will end up in even lower burden voltage for mA range (~10,3µV/mA burden voltage instead of ~20µV/mA).
« Last Edit: February 23, 2014, 12:12:18 pm by black-earth »
 

Offline bronson

  • Regular Contributor
  • *
  • Posts: 172
  • Country: us
Re: µCurrent Gold - HW Design question
« Reply #6 on: March 12, 2014, 07:13:36 pm »
Has Dave (or anybody) actually tried out the fusible trace?

Any photos of the result?
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29658
  • Country: au
    • EEVblog
Re: µCurrent Gold - HW Design question
« Reply #7 on: March 12, 2014, 09:48:30 pm »
Fusible track Dave talks about it in several places, on his Channel and on TheAmpHour. I think he mentions Saturn pcb (he demonstrated saturn pcb on one his blogs) has calculator for it?
Apparently using fusible track is not exact science, but more of a try and guess.

Correct in every respect.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29658
  • Country: au
    • EEVblog
Re: µCurrent Gold - HW Design question
« Reply #8 on: March 12, 2014, 09:50:12 pm »
Has Dave (or anybody) actually tried out the fusible trace?
Any photos of the result?

Hmm, I'm thinking a FF video coming on...
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29658
  • Country: au
    • EEVblog
Re: µCurrent Gold - HW Design question
« Reply #9 on: March 12, 2014, 09:55:49 pm »
But what about the contact resistances of the range switch ?
Datasheed only says:
Quote from: Datasheet
CONTACT RESISTANCE: Below 10 mR typ. initial @ 2-4 V DC, 100 mA, for both silver and gold plated contacts.
Thankfully this will only add up on the burdon voltage and will not influence the current measurement.

Correct.
The fusible trace is also there in an attempt to protect the switch, hopefully that will break before the switch is damaged. Never tried it though.

Quote
Another question: The recommended operationg voltage of the OpAmps is 2V7...5V5.
Would it make sense to set Vcc e.g. up to 5V to enlarge the measurement range ? Or would it have other negative sideeffects ?

Correct.
No side effects with the chips used. I have done a video using other Analog devices chips that change their offset voltage with supply voltage. I was going to use those until I discovered that issue.

Quote
At least a 3V6 lithium battery would improve system lifetime by nice factor ... (a coin cell only holds abut 260mAh)

Yes, plenty of battery option.
I included a 3 x AAA battery holder in some early uCurrent units.
 

Offline MLXXXp

  • Regular Contributor
  • *
  • Posts: 221
  • Country: ca
Re: µCurrent Gold - HW Design question
« Reply #10 on: March 13, 2014, 12:14:41 am »
At least a 3V6 lithium battery would improve system lifetime by nice factor ... (a coin cell only holds abut 260mAh)
Yes, plenty of battery option.
I included a 3 x AAA battery holder in some early uCurrent units.
As I said in a previous post in this thread, I calculated that running the uCurrent at 4.5V would cause the Batt OK LED to draw more current than is safe for the TPS3809 voltage supervisor IC. Dave has indicated that it's OK to run the uCurrent at 4.5V, so I must be mistaken and I'd like to know why.

Would someone please look at the datasheet for the TPS3809:
http://www.ti.com/lit/ds/symlink/tps3809l30.pdf
and tell me what, on page 2, "Maximum low output current, IOL ... 5 mA" and "Maximum high output current, IOH ... -5 mA" mean and what would cause these values to be exceeded?

Also, on page 5, why do the curves in figures 1, 3 and 4 show currents far exceeding these 5mA and -5mA maximum values?
« Last Edit: March 13, 2014, 01:12:21 pm by MLXXXp »
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29658
  • Country: au
    • EEVblog
Re: µCurrent Gold - HW Design question
« Reply #11 on: March 13, 2014, 01:22:11 am »
As I said in a previous post in this thread, I calculated that running the uCurrent at 4.5V would cause the Batt OK LED to draw more current than is safe for the TPS3809 voltage supervisor IC. Dave has indicated that it's OK to run the uCurrent at 4.5V, so I must be mistaken and I'd like to know why.

I did not consider the max led current in that chip, that was an oversight, I think I just looked at the curves way back. Yes, you'll need to resize the resistor for 4.5V operation if you are concerned. Might be a bit dim near the cutoff voltage though. Ultimately the chip will likely survive just fine, I've tried it no problems.
« Last Edit: March 13, 2014, 01:36:44 am by EEVblog »
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29658
  • Country: au
    • EEVblog
Re: µCurrent Gold - HW Design question
« Reply #12 on: March 13, 2014, 01:36:10 am »
Also, on page 5, why do the curves in figures 1, 3 and 4 show currents far exceeding these 5mA and -5mA maximum values?

Well that's interesting. They have characteristic curves outside the absolute max spec figure, so obviously they have test them.
At ambient temp, assuming a 1.9V drop LED @4.5V input, with 270 dropper, the maximum LED current is 9.6mA assuming an zero drop in the output transistor of the chip. Of course it's not going to be zero, it will match the curve, so here is where it will "self balance" itself for want of a better term. At 9.6mA it would drop 2.5V in the pass transistor, but of course that changes the LED current, so its will be less than that. How much less? you'd have to guess, or measure it.
I just measured it: at 4.5V the LED current is 7.2mA
0.61V drop on the output transistor, a fair bit off the expected curve value.
 

Offline MLXXXp

  • Regular Contributor
  • *
  • Posts: 221
  • Country: ca
Re: µCurrent Gold - HW Design question
« Reply #13 on: March 13, 2014, 01:00:43 pm »
I just measured it: at 4.5V the LED current is 7.2mA
0.61V drop on the output transistor, a fair bit off the expected curve value.
But closer to what you expect by extrapolating the VOH values given in the table on page 3 of the datasheet.
 

Offline EEVblog

  • Administrator
  • *****
  • Posts: 29658
  • Country: au
    • EEVblog
Re: µCurrent Gold - HW Design question
« Reply #14 on: March 13, 2014, 01:03:29 pm »
But closer to what you expect by extrapolating the VOH values given in the table on page 3 of the datasheet.

That's because they rudely vary both VDD and output current in that table, both will have an effect, so you don't know what and by how much.
 

Offline MLXXXp

  • Regular Contributor
  • *
  • Posts: 221
  • Country: ca
Re: µCurrent Gold - HW Design question
« Reply #15 on: March 13, 2014, 01:27:50 pm »
Yes, you'll need to resize the resistor for 4.5V operation if you are concerned. Might be a bit dim near the cutoff voltage though.
Well, if I were going to use 3 x 1.5V cells to run the uCurrent at 4.5V, I'd also change the voltage monitor to a TPS3809K33, for a 2.93V threshold. Then, the guaranteed "Batt OK" measurement range would increase to about +/-1400xA and the LED wouldn't dim as much at the low end.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf