Uni-T UT61E backlight mod with touch sensor [tutorial]

[Nisei]

Several people have asked me what parts I've used and how I've wired up the UT61E backlight mod with touch sensor I posted in a different topic so I thought I might as well dedicate a thread to the subject so it's easier to find for people searching for it.
If you want to use any of the pictures in this post then feel free to do so but please mention who and where you got them from. It has taken me quite an amount of time to come up with this so please have the courtesy to mention me and link to this thread.

https://youtu.be/EHStaKB0BwI

While looking for a tutorial to install a backlight I came across a post on diyaudio by user Tolik.
I loved the touch sensor but he built everything from scratch and there are quite a few components involved.
I don't really like the fact that when you let the UT61E's ES51922A chip control the backlight, it turns off again after 60 seconds. Besides that, his circuit uses around 35mA which I think is quite a lot (compared to what the meter itself uses).
So, I started looking around if I couldn't build something myself with less components and not connected to the ES51922A chip.

The result is a very simple circuit with only a few components.
It draws only a few micro amps when the backlight is off and a few milli amps when the backlight is on.
The output port sink current on the touch module is 8mA but I've tested with way higher values and it doesn't seem to damage anything.
The resistor used depends on the specs of the LEDs so the one in the tutorial is not correct.

This tutorial may look for dummies but that's because I'm a dummy myself when it comes to electronics. I only have very basic knowledge of it and build things by logical reasoning and learn a little as I go along.
Thanks to helpful member mariush in this thread and a member on Hackaday (who covered this mod on their blog), I have revised this mod which has drastically improved efficiency.

Parts list (click on link to be taken to item on eBay):
2x 3528 LED red
1x TTP223 touch sensor module
1x 3.3V low quiescent current voltage regulator MCP1703T-3302E/DB
1x resistor (value depends on LEDs used)

First off, modify the touch sensor module. We don't need the touch area because we're going to make our own. Bridging the W1 terminals (pins 5 & 6 of the TTP223 chip) turns the module from a momentary switch into a toggle switch:



Your module should now look like this:



Since we've removed the LED (D1) there's room to put the voltage regulator on the board to keep things compact.
The middle leg of the voltage regulator can be removed since we're using the tab on the top.
The touch module has a capacitor on the input side so we can solder the MCP1703 regulator straight to the board.



Left of the battery compartment is plenty of room to install the circuit.
Watch out for wires getting trapped between the battery compartment and board when you put the back of the meter back on.
The 9V we're using is the battery power which gets switched by the rotary selector so the circuit turns on and off when you turn the meter on and off.
I've used green color to draw the +9V signal because red on red isn't really clear to see
Again, note the resistor is NOT the correct value in this picture.



To find the right value for the resistor it's handy to try out a few values or just use a variable resistor so you can set the brightness to your liking.
Keep the brightness as low as possible. The touch module isn't really made to supply more than 8mA of current although in real life it can handle at least twice that amount.

Solder a wire (I've used 30 awg wrapping wire for all wiring) to C3 on the edge of the board.
This wire runs to the top of the meter and connects to a piece of aluminum foil tape. If you're having difficulty soldering to the foil tape, just strip off a fair amount of insulation and use another piece if tape to attach the wire to the foil tape.
The larger the piece of foil tape, the more sensitive the sensor. If too large, the backlight will turn on even before your finger touches the meter. Experiment to find the correct size.



You can use the 2 holes seen in the picture to lead your LED wires to the other side of the board.
DO NOT SOLDER THEM ONTO THESE TERMINALS AND MAKE SURE THEY'RE PROPERLY INSULATED!
30 awg wrapping wire is perfect.



We're wiring the LEDs in parallel.



Using a very tiny amount of hot glue, stick the LEDs onto the foil tape that's on the plastic plate behind the LCD display.
The cutouts are there, because Uni-T also uses these plates for the other 61x models that do have backlight already installed.
The LEDs should be emitting light into the plate (to the top of the picture).



And here's a view from the other side.



To make sure nothing's interfering with the meter I've wrapped the added circuit board in aluminum tape (without taking the backing off) and connected it to GND (VB-) so it's completely shielded.

When re-installing the board and cover, make sure no wires get trapped or damaged.

If you ruin your meter by installing this mod, don't blame me.
I can't be held responsible if anything gets damaged in the process.

Any comments, tips and annoyances are welcome

[mariush]

Awesome mod.

I would only work on improving the efficiency, as 9v batteries aren't that great when it comes to higher current draws ... ut61e only uses 3-6mA in various modes and your light draws 3-4x times that, and at 20-25mA a 9v battery will drain much faster (and the voltage will drop prematurely due to the battery's internal resistance) ... it's enough to see the difference between 2mA and 10mA in this datasheet to see how much it matters: http://ww2.duracell.com/media/en-US/pdf/gtcl/Product_Data_Sheet/NA_DATASHEETS/MN1604_6LR61_US_CT.pdf

UT61E works fine with as little as 5.6v but AMS1117 has a voltage drop of at least 0.7v so in theory it would be fine... but the suck part is the high quiescent current consumption, typically about 5mA according to datasheet, which is more than what the whole meter uses in some modes.

Seems like the TTP223 sensor works with 2.0 - 5v, so I would probably change the voltage to something like 2.8v or 3.3v, which in turn would allow you to use a much smaller current limiting resistor for the LED (you'd want to limit current to 5-10mA anyway).

Ideally, I'd use a switching regulator for best results (there are some that only require an external inductor and maybe a couple of capacitors to work) but for this application you could go with some nice ldo regulators like

lp2980 : http://uk.farnell.com/texas-instruments/lp2980aim5-3-3/v-reg-ldo-3-3v-smd-2980-sot235/dp/1212201
lp2985 : http://uk.farnell.com/texas-instruments/lp2985im5-3-3/v-reg-ldo-3-3v-smd-2985-sot235/dp/9778284

TLV70133 : http://uk.farnell.com/texas-instruments/tlv70133dbvt/ldo-3-3v-150ma-sot23-5-full-reel/dp/2437506

mic5213 (3v) : http://uk.farnell.com/micrel-semiconductor/mic5213-3-0yc5-tr/ldo-volt-reg-0-08a-3v-sc-70-5/dp/2431764

You just have to be careful to get something that can handle 10-12v input voltage, because fresh 9v batteries can often have voltage higher than 9v. These above will work, but there's lots of tiny ldos that can only work up to about 5.5-6v, so again, must be careful about it.

With LDOs the difference between input and output voltage is wasted as heat, so for example with 10mA , you have (9v - 3.3v )v x 0.01a = 0.057w .. i guess it's reasonable, the lower the battery voltage the lower the waste.

[analogix]

This tutorial may look for dummies but that's because I'm a dummy myself when it comes to electronics. I only have very basic knowledge of it and build things by logical reasoning and learn a little as I go along.

Don't sell yourself short because that's one of the better and more detailed tutorials I've come across -excellent work! And you seem to have more than basic knowledge if you ask me
As a simpler alternative to this modification -I read about someone just connecting two LEDs (through a resistor) to the battery terminal and through the mode switch so that the backlight is always on as long as the meter is powered on. Does anyone know how much more power the meter would consume with this and if it would cause any issues and drain the battery noticeably faster than without the LEDs?

[commie]

Hi

There is something confusing me here, does the UT61E come with a back light as standard?, The reason I asked, I have been reading some threads going back 4 years regarding the UT61E and it would seem it doesn't have a back light? However, if I read the up to date spec. on the UT61E it claims it comes with a back light. So does the UT61E come with a backlight as standard or what?

[mariush]

UT61E has no backlight. The manuals and tech specs are for all models in UT61 series, some models have backlight, some don't.  There's usually a  YES / NO  or a checkmark / x  icon , or it says (model x only) when it comes to each feature.

Also, the english uni-t version of the site is sometimes updated less often, so it could be the tech specs on that version of their website are not current (but I doubt it)

This mod adds backlight by adding surface mount led on the side of the lcd display which is controlled by that small pcb.

There's another mod out there which uses a tiny microcontroller to hook into the functionality of a button (i think the yellow one but I'm not sure) and when you kept the button pressed for a particular amount of time it would turn on a led.

[commie]

UT61E has no backlight. The manuals and tech specs are for all models in UT61 series, some models have backlight, some don't.  There's usually a  YES / NO  or a checkmark / x  icon , or it says (model x only) when it comes to each feature.

Okay thanks for the reply, so what does this mean: http://www.rapidonline.com/pdf/85-4068.pdf

[mariush]

It means that in September 2010 (date is embedded in the pdf file), some support desk/geek from Rapid Online went to Uni-T's site and hit "print to pdf" to create that document.

Since 2010, Uni-T has corrected the tech specs page, you can see it here: http://www.uni-trend.com/en/product/2014_0729_802.html  (hit the Parameter or Contrast buttons).

[commie]

It means that in September 2010 (date is embedded in the pdf file), some support desk/geek from Rapid Online went to Uni-T's site and hit "print to pdf" to create that document

Hmmm...yeah I see what you mean, thats a shame because it is ideal for 4..20mA calibration work.Thanks anyway.

[Nisei]

Awesome mod.
I would only work on improving the efficiency, as 9v batteries aren't that great when it comes to higher current draws...

Thank you so much for all the information.
Like I already mentioned I have very little experience with electronics. Perhaps you get an idea of my inexperience when I tell you that 2 weeks ago I didn't even know what a voltage regulator looked like
I'm already glad I could build something that only uses 3mA quiescent current (that's what the AMS1117 draws) and 15mA when in use compared to the original 30-35mA the original posted circuit was using.
But your comment stimulates me to read further (I have no idea what some of the things you mention actually are, yet) and understand what can be improved and most importantly; why.
Nevertheless, I'm happy with it and 9V batteries can be bought dirt cheap nowadays

[Nisei]

Don't sell yourself short because that's one of the better and more detailed tutorials I've come across -excellent work! And you seem to have more than basic knowledge if you ask me

Thanks for the kind words but really, I hardly know anything about electronics.
When I'm interested in something I read, read and read and apply that little knowledge to come up with something myself. Ask me what a transistor is and does and I have to answer that I don't have the slightest idea
About the always-on mod; no idea. But I can only imagine its current draw would be 3mA less than my mod when the LED's are lit and being the same brightness.

[Maxlor]

Really neat! Your solution inspired me to investigate those touch sensors, and I think I found a way to do it with a single chip, thanks to Microchip AN1298. Let me sketch it out:

You need an MCU with a 10bit ADC, and your touch pad connected to, say, channel ADC0 (I used a jumper wire attached to a piece of aluminium foil, just like Nisey). Now repeat these tree steps:

1. Sample VDD with the ADC to charge the internal sample&hold cap. Since the Atmega32u4 can't do it internally, I connected Pin ADC1 to VDD and sampled that.
2. Set the ADC0 pin as an output and drive it low to drain any charge from the touch pad. wait a few microseconds, then configure ADC0 as an ADC input again.
3. Sample ADC0. The stored charge in the sample&hold cap will start charing the touch pad, until they reach equilibrium. The voltage is lower as the capacitance of the touch pad increases.

I got typical value of around 630 (out of 1023) with the foil at rest, and around 605 when my finger was hovering above it. Thus I programmed a simple algorithm that samples every 50ms and stores the last 20 values. It checks whether the current value is at least 16 units lower than the maximum seen in those last 20 values, if so, it assumes a button press. I added some hysteresis and a release delay, and got pretty reliable "press" detections.

What I plan to do next is:
replicate the same with a PIC12F1822 in PDIP form factor (because I have those here.) Use BAT- (negative battery terminal) and GND (many places, e.g. one of the screws holding the display) as power source. Note: in the UT61E, GND sits 3V above BAT-. Connect the PIC to the BKLIT pin of the ES51922 chip in the UT61E, and connect the LEDs to the BKOUT pin (so yeah, I'll be using the 60s timeout, which works for me.) Since I only need the single chip, I can glue it dead bug style onto the PCB and solder some wire directly to the pins.

What's not yet clear to me is whether the BKOUT pin can source enough current for two LEDs, and whether it really swings by 6V as the datasheet claims, or whether that's a typo and it's only 3V.

The only thing I lack at the moment are small enough LEDs. I guess I'll wait a bit longer until the next parts order.

[Penguin36]

Very nice mod!

Q: Whats the max. current on TTP223 output?

[Nisei]

Very nice mod!
Q: Whats the max. current on TTP223 output?

Thanks.
Hadn't thought about that one yet. It's nowhere in the specsheet. Strange...

[Nisei]

Really neat! Your solution inspired me to investigate those touch sensors, and I think I found a way to do it with a single chip, thanks to Microchip AN1298. Let me sketch it out:

You need an MCU with a 10bit ADC, and your touch pad connected to, say, channel ADC0 (I used a jumper wire attached to a piece of aluminium foil, just like Nisey). Now repeat these tree steps:

1. Sample VDD with the ADC to charge the internal sample&hold cap. Since the Atmega32u4 can't do it internally, I connected Pin ADC1 to VDD and sampled that.
2. Set the ADC0 pin as an output and drive it low to drain any charge from the touch pad. wait a few microseconds, then configure ADC0 as an ADC input again.
3. Sample ADC0. The stored charge in the sample&hold cap will start charing the touch pad, until they reach equilibrium. The voltage is lower as the capacitance of the touch pad increases.

I got typical value of around 630 (out of 1023) with the foil at rest, and around 605 when my finger was hovering above it. Thus I programmed a simple algorithm that samples every 50ms and stores the last 20 values. It checks whether the current value is at least 16 units lower than the maximum seen in those last 20 values, if so, it assumes a button press. I added some hysteresis and a release delay, and got pretty reliable "press" detections.

What I plan to do next is:
replicate the same with a PIC12F1822 in PDIP form factor (because I have those here.) Use BAT- (negative battery terminal) and GND (many places, e.g. one of the screws holding the display) as power source. Note: in the UT61E, GND sits 3V above BAT-. Connect the PIC to the BKLIT pin of the ES51922 chip in the UT61E, and connect the LEDs to the BKOUT pin (so yeah, I'll be using the 60s timeout, which works for me.) Since I only need the single chip, I can glue it dead bug style onto the PCB and solder some wire directly to the pins.

What's not yet clear to me is whether the BKOUT pin can source enough current for two LEDs, and whether it really swings by 6V as the datasheet claims, or whether that's a typo and it's only 3V.

The only thing I lack at the moment are small enough LEDs. I guess I'll wait a bit longer until the next parts order.

Very interesting!
Although I don't understand most of what you're describing I think it's really cool you can program a chip like that yourself.
But just wondering; are you going to do it for fun or to save money? I've bought 10 of these TTP223 modules for only $7.50

[Maxlor]

Very interesting!
Although I don't understand most of what you're describing I think it's really cool you can program a chip like that yourself.
But just wondering; are you going to do it for fun or to save money? I've bought 10 of these TTP223 modules for only $7.50

Just for fun And to apply a newly learned trick.

[Nisei]

Just for fun And to apply a newly learned trick.

Ah, great! I'm like that as well. Can you keep me posted how things are going please?

[Nisei]

With LDOs the difference between input and output voltage is wasted as heat, so for example with 10mA , you have (9v - 3.3v )v x 0.01a = 0.057w .. i guess it's reasonable, the lower the battery voltage the lower the waste.

I'm still not sure why you prefer bringing the voltage down to 3.3V.
Like you said, the bigger the difference between input and output voltage, the more heat and waste. The efficiency I'm looking for is most important for regular use, which is when the backlight is switched off. So isn't it logical to drop the voltage as little as possible? Efficiency with backlight switched on may be less because of the higher value resistor used but that's only in the odd occasion when you need it (probably less than 5% of the time for most people).
I'm thinking about using the MCP1702T-5002E/CB.
The pinout is different from the AMS1117 but I've already finished the drawings and can change the tutorial as soon as I've got it confirmed working.

[mariush]

You're right it won't be much of a difference when the led is lit but it will make a difference while the whole thing is idle, waiting for you to tap the meter to turn the backlight on.

The whole point of your circuit is to light up a LED, let's say one with a rated forward voltage of 3v and 10mA.

Your circuit consists of that AMS1117, that sensor chip, the led and the resistor.  The AMS1117 you say it uses 3mA when idle (quiescent current), let's say the sensor chip uses 1mA , the led and resistor will use 10mA

Without doing anything, your circuit will use 3-4mA of current from your battery regardless of voltage, so if your battery is 9v you'll have power loss of  (9v-5v)x0.04a = 0.16w
A better regulator will use maybe 0.1mA when idle and the sensor chip itself will probably use less current at lower voltage but let's say it's still 1mA , so now your power loss is (9v - 3.3v ) x 0.011a = 0.06 so about 3 times lower energy wasted when idle.

If you power led from 5v, you'll need a resistor to limit the current at 10mA :  5v - 3v = R x 0.01  , 2 = R x 0.01  => R  = 200 ohm   The power dissipated in the resistor will be P = IxIxR = 200 x 0.01 x 0.01 = 0.02w
If you power led from 3.3v, the resistor can be much smaller, less waste 3.3v - 3v = R x 0.01 , 0.3 = R x 0.01 = 30 ohm and you'll only waste 0.003w in the resistor.

I just think it's better to concentrate the heat into one location (the regulator) instead of producing heat in several places (resistor and regulator and maybe sensor chip also)

But the difference between consuming 11mA (10mA for led and 1mA for the rest) vs 15mA (10mA for led, 4mA for others) is the  important part because of using 9v batteries.

The UT61E will show a low battery voltage at around 5.6v but it will happily run with as little as 3v without any errors in measurements.

Let's imagine you have a 9v battery that's close to being discharged, let's say it's measuring 6v when taken out of the multimeter.  By default, with the multimeter's normal consumption of 2-6mA (depending on mode), this is low enough that the battery will be capable of operating properly.

If you add the 3-4mA of idle current from your circuit, you're going into the area where such a depleted battery may have problems keeping the voltage and delivering 8-10mA at the same time, so the battery's voltage will sag down from 6v to maybe 5.8v or even less.  If you turn on the led, you add an additional 10mA to the load and IF the battery will be still powerful enough to do 16-20mA, the voltage will definitely sag a lot, maybe even lower than 5.6v and you'll see the low voltage indicator prematurely.
Of course, you could argue that you don't care, you'll just replace battery when you see low voltage because 9v batteries are cheap. I can understand that.

I personally would be pissed to have a circuit that uses more power than the whole multimeter.. like I said, the meter uses maybe 2mA when measuring voltage, with small pulses twice a second when data in transmitted. 3-4mA quiescent current for your circuit is a lot in comparison.

MCP1702 seems like a very regulator for this.

[onlooker]

From the datasheet: The touch sense chip TTP223 comsumes 1.5uA to 6.5uA when no load. It can deliver 8mA to the load. The working voltage is from 2V to 5.5V.

Maybe it is better not to use any regulator, instead, just power it from an existing lower voltage rail,similar to what Maxlor is trying, but not to involving anything else on the main processor IC.

[Nisei]

Thanks for the extensive reply but I think you may have misread my post.
I'm already convinced about using an LDO instead of the AMS1117 because of its high quiescent current. The AMS1117 is getting replaced with an LDO.
I was just curious about your choice to bring down the voltage to 3.3V
I look at it like this:
If your car could be made 50% more efficient and they gave you the following options, which one would you choose?
A: 50% efficiency gain when driving forward, 0% efficiency gain when in reverse
B: 25% efficiency gain when driving forward, 25% efficiency gain when in reverse
To me, the times you're using the backlight are so rare that I'd rather concentrate on making the circuit as efficient possible in circuit idle mode.

[Nisei]

The touch sense chip TTP223 comsumes 1.5uA to 6.5uA when no load. It can deliver 8mA to the load.

Ooooo.K. So I'd missed this: 8mA sink current, I didn't even know what that meant...
Amazing it held up 'til now even while I've been overloading it...
Sink current is what it's able to pass through right? When I'm replacing the voltage regulator with the LDO I'll see if 8mA is enough to get decent brightness from the LED's.

Maybe it is better not to use any regulator, instead, just power it from an existing lower voltage rail,similar to what Maxlor is trying, but not to involving anything else on the main processor IC.

Well, that was my first idea and I posted a question where on the circuit board I could safely tap a 5V or 3.3V signal from. But no-one responded so I had to come up with something else.

[Carko]

I'm already convinced about using an LDO instead of the AMS1117 because of its high quiescent current. The AMS1117 is getting replaced with an LDO.

By the way, AMS1117 is a LDO 

You probably meant a buck converter (step down voltage regulator).

[Nisei]

By the way, AMS1117 is a LDO 
You probably meant a buck converter (step down voltage regulator).

Doh! Goes to show how little I know
No, I have replaced the AMS1117 with a much more efficient LDO now (MCP1703) which only uses 2uA quiescent current instead of 3mA.
I've revised the tutorial.

[Maxlor]

I've finally gotten around do finishing this little mod.

I've used a PIC12F1822 (because I some spares), glued onto the meter's board with double sided tape. It's powered using the multimeter's -3V and GND rails from two nearby caps. As a sensor, I use bit of copper tape connected to one of the PINs of the PIC. For sensing, I use the PICs capacitive sensing module, since it gave me more robust results in testing than the charge divider method sketched in an earlier posting. Microchip has actually deprecated that method because of noise susceptibility, but I'm getting really good results from it, even without any filtering or double-sampling. It triggers on touch without fail, and so har has never mis-triggered when putting the meter down.

One thing that interferes is the speaker. If the sense wire runs near it, a beep will trigger the backlight. The issue goes away if the wire keeps its distance.

I run the PIC off its 31kHz internal clock, which is fast enough so for my main loop to run at 8Hz. Power usage is around 20uA, negligible.

Originally I planned to use the Cyrustek's builtin backlight function (it has two pins, an input that you pulse low to switch the backlight, and an output pin for the LEDs. Btw, note that the logic level is -3V for low, and GND for high, but that the backlight out pin goes from -3V to +3V.) but it turns out that the Cyrustek only has minuscule current driving ability in the vincinity of 1.5mA. Since I didn't want to add a mosfet, I decided to drive the LEDs directly from the PIC instead. It's advertized source/sink rate is 25mA per GPIO. Unfortunately, it turns out that my chip here only gives me 5.5mA from a single GPIO pin, maybe because it's running at 3V, not 5. So I paralleled two GPIOs, that gives me 8.5mA. Still less than I would have liked (I was aiming for 20mA for both LEDs), but it's usable.

If anyone cares about the code, let me know and I'll post it.

Things to note: that 0.5mm pin pitch on the cyrustek is a pita to work with. I spent about 2 hours trying to solder a wire to the BKOUT pin, wicking up solder bridges that would inevitably be created, etc. Also, with the limited current sourcing ability of the PIC, I needn't have bothered with the series resistors. Next time, test this first.

[Nisei]

Still less than I would have liked (I was aiming for 20mA for both LEDs), but it's usable.

Nice work Maxlor!
Do you have a pic of it showing the backlight in action?
I'm wondering why you were aiming for 20mA for both LEDs. Or did you use white ones? 8mA is enough for the red LEDs. You don't need to use the meter as a flashlight