Silicon ChipAD584 Precision Voltage References - July 2020 SILICON CHIP
  1. Outer Front Cover
  2. Contents
  3. Subscriptions: PE Subscription
  4. Subscriptions: PicoLog Cloud
  5. Publisher's Letter: Publisher's Statement
  6. Publisher's Letter
  7. Feature: Techno Talk by Mark Nelson
  8. Feature: Net Work by Alan Winstanley
  9. Back Issues: Net Work by Alan Winstanley
  10. Project: Speech Synthesiser with the Raspberry Pi Zero by Tim Blythman
  11. Project: AD584 Precision Voltage References by Jim Rowe
  12. Project: AM/FM/CW Scanning HF/VHF RF Signal Generator by Andrew Woodfield
  13. Project: High-current Solid-state 12V Battery Isolator by Bruce Boardman
  14. Feature: Circuit Surgery by Ian Bell
  15. Feature: AUDIO OUT by Jake Rothman
  16. Feature: Make it with Micromite by Phil Boyce
  17. Feature: PICn’Mix by Mike Hibbett
  18. Feature: Max’s Cool Beans by Max the Magnificent
  19. Feature: Electronic Building Blocks by Julian Edgar
  20. PCB Order Form
  21. Advertising Index

This is only a preview of the July 2020 issue of Practical Electronics.

You can view 0 of the 72 pages in the full issue.

Articles in this series:
  • Techno Talk (July 2020)
  • Techno Talk (July 2020)
  • Techno Talk (August 2020)
  • Techno Talk (August 2020)
  • Techno Talk (September 2020)
  • Techno Talk (September 2020)
  • Techno Talk (October 2020)
  • Techno Talk (October 2020)
  • (November 2020)
  • (November 2020)
  • Techno Talk (December 2020)
  • Techno Talk (December 2020)
  • Techno Talk (January 2021)
  • Techno Talk (January 2021)
  • Techno Talk (February 2021)
  • Techno Talk (February 2021)
  • Techno Talk (March 2021)
  • Techno Talk (March 2021)
  • Techno Talk (April 2021)
  • Techno Talk (April 2021)
  • Techno Talk (May 2021)
  • Techno Talk (May 2021)
  • Techno Talk (June 2021)
  • Techno Talk (June 2021)
  • Techno Talk (July 2021)
  • Techno Talk (July 2021)
  • Techno Talk (August 2021)
  • Techno Talk (August 2021)
  • Techno Talk (September 2021)
  • Techno Talk (September 2021)
  • Techno Talk (October 2021)
  • Techno Talk (October 2021)
  • Techno Talk (November 2021)
  • Techno Talk (November 2021)
  • Techno Talk (December 2021)
  • Techno Talk (December 2021)
  • Communing with nature (January 2022)
  • Communing with nature (January 2022)
  • Should we be worried? (February 2022)
  • Should we be worried? (February 2022)
  • How resilient is your lifeline? (March 2022)
  • How resilient is your lifeline? (March 2022)
  • Go eco, get ethical! (April 2022)
  • Go eco, get ethical! (April 2022)
  • From nano to bio (May 2022)
  • From nano to bio (May 2022)
  • Positivity follows the gloom (June 2022)
  • Positivity follows the gloom (June 2022)
  • Mixed menu (July 2022)
  • Mixed menu (July 2022)
  • Time for a total rethink? (August 2022)
  • Time for a total rethink? (August 2022)
  • What’s in a name? (September 2022)
  • What’s in a name? (September 2022)
  • Forget leaves on the line! (October 2022)
  • Forget leaves on the line! (October 2022)
  • Giant Boost for Batteries (December 2022)
  • Giant Boost for Batteries (December 2022)
  • Raudive Voices Revisited (January 2023)
  • Raudive Voices Revisited (January 2023)
  • A thousand words (February 2023)
  • A thousand words (February 2023)
  • It’s handover time (March 2023)
  • It’s handover time (March 2023)
  • AI, Robots, Horticulture and Agriculture (April 2023)
  • AI, Robots, Horticulture and Agriculture (April 2023)
  • Prophecy can be perplexing (May 2023)
  • Prophecy can be perplexing (May 2023)
  • Technology comes in different shapes and sizes (June 2023)
  • Technology comes in different shapes and sizes (June 2023)
  • AI and robots – what could possibly go wrong? (July 2023)
  • AI and robots – what could possibly go wrong? (July 2023)
  • How long until we’re all out of work? (August 2023)
  • How long until we’re all out of work? (August 2023)
  • We both have truths, are mine the same as yours? (September 2023)
  • We both have truths, are mine the same as yours? (September 2023)
  • Holy Spheres, Batman! (October 2023)
  • Holy Spheres, Batman! (October 2023)
  • Where’s my pneumatic car? (November 2023)
  • Where’s my pneumatic car? (November 2023)
  • Good grief! (December 2023)
  • Good grief! (December 2023)
  • Cheeky chiplets (January 2024)
  • Cheeky chiplets (January 2024)
  • Cheeky chiplets (February 2024)
  • Cheeky chiplets (February 2024)
  • The Wibbly-Wobbly World of Quantum (March 2024)
  • The Wibbly-Wobbly World of Quantum (March 2024)
  • Techno Talk - Wait! What? Really? (April 2024)
  • Techno Talk - Wait! What? Really? (April 2024)
  • Techno Talk - One step closer to a dystopian abyss? (May 2024)
  • Techno Talk - One step closer to a dystopian abyss? (May 2024)
  • Techno Talk - Program that! (June 2024)
  • Techno Talk - Program that! (June 2024)
  • Techno Talk (July 2024)
  • Techno Talk (July 2024)
  • Techno Talk - That makes so much sense! (August 2024)
  • Techno Talk - That makes so much sense! (August 2024)
  • Techno Talk - I don’t want to be a Norbert... (September 2024)
  • Techno Talk - I don’t want to be a Norbert... (September 2024)
  • Techno Talk - Sticking the landing (October 2024)
  • Techno Talk - Sticking the landing (October 2024)
  • Techno Talk (November 2024)
  • Techno Talk (November 2024)
  • Techno Talk (December 2024)
  • Techno Talk (December 2024)
  • Techno Talk (January 2025)
  • Techno Talk (January 2025)
  • Techno Talk (February 2025)
  • Techno Talk (February 2025)
  • Techno Talk (March 2025)
  • Techno Talk (March 2025)
  • Techno Talk (April 2025)
  • Techno Talk (April 2025)
  • Techno Talk (May 2025)
  • Techno Talk (May 2025)
  • Techno Talk (June 2025)
  • Techno Talk (June 2025)
  • Techno Talk (July 2025)
  • Techno Talk (July 2025)
Using Cheap Asian Electronic Modules by Jim Rowe AD584 Precision Voltage References These three low-cost precision voltage reference modules are based on the AD584 IC from Analog Devices, but each uses a different version of it and has a unique designs. Two are ‘naked’ boards, while the third comes in a transparent laser-cut acrylic case. T he ML005-V1.2 is the smallest module, with a PCB measuring 32 × 32mm. You can purchase it from AliExpress for around £4.50 (including delivery): www.aliexpress.com/item//32853943748.html The slightly bigger module has no ID, but its PCB measures 50 × 50mm and it is available from Banggood for around £13.25 (including delivery): https://bit.ly/pe-jul2020-ad584 The largest module, from KKMOON, comes in an acrylic case, measuring 70 × 52 × 35mm overall. It is available from suppliers like Banggood and eBay for around £13.00 (including delivery): https://bit.ly/pe-jul20-kkm Each of the modules are based on different versions of the AD584 precision voltage reference device made by Analog Devices (the datasheet can be found at: https://bit.ly/pe-jul20-ad584data). Let’s start by looking at how this chip works. The AD584 device Analog Devices describe the AD584 as a ‘Pin Programmable Precision Voltage 22 Reference’. It comes in a number of versions, all of which are available in an 8-lead hermetically sealed TO-99 metal package. The two lowest-precision versions are also available in an 8-lead plastic DIP. The metal-package versions have an ‘H’ suffix, while those in the plastic package carry the ‘NZ’ suffix. All versions are made using laser wafer trimming (LWT) to adjust the output voltages and also their temperature coefficients. Originally, five versions were available: the AD584J, AD584K and AD584L, all specified for operation from 0-70°C; and the AD584S and AD584T, which are specified for operation between −55°C and +125°C. However, the AD584LH version was discontinued by Analog Devices in 2012, so presumably, those used in modules like the one described here are either ‘new old stock’ (NOS) or have been ‘recycled’ from used equipment. The basic specifications of the AD584JH, AD584KH and AD584LH are summarised in Table 1; which can be found at the end of the article. The AD584JH version is the least accurate, while the AD584LH is the most accurate. But note that all three versions have identical specifications when it comes to noise output and long-term stability. A simplified version of the AD584’s internal block diagram is shown in Fig.1. At the heart of the device is a high stability band-gap reference diode providing a 1.215V reference. This is followed by an op amp used as a buffer amplifier, with its voltage gain set by the string of divider resistors connected between its output (pin 1) and common (pin 4) terminals. Internal feedback from the lowest tap of the divider string (pin 6, VBG) ensures that the buffer amp maintains VBG at very close to 1.215V, the bandgap voltage. So if a DC voltage between +12-15V is applied to the device between pins 8 and 4, and no external connections are made to pins 2, 3 or 6, it will provide a nominal output voltage of very close to 10V at pin 1. But if pins 1 and 2 are joined externally, the voltage at pin 1 will drop to Practical Electronics | July | 2020 Fig.1 (left): the AD584 voltage reference IC used in all these modules contains a very accurate and stable 1.215V laser-trimmed bandgap reference, plus a precision op amp and resistors to amplify that reference to provide four possible output voltages (2.5V, 5V, 7.5V and 10V) depending on which combination of pins 1, 2 and 3 are tied together. Right: the ML005-V1.2 module shown at nearly twice actual size. Note that searching for ‘ML005’ online will not find this module, so you will need to search for AD584. Inside the AD584 chip very close to 5V, and if pins 1 and 3 are joined, it will be very close to 2.5V. If pins 2 and 3 are joined, it will settle very close to 7.5V. Notice also that pins 1, 2 and 3 can be used to source 10V, 5V or 2.5V independently, although pins 2 and 3 cannot provide significant current without affecting accuracy, and so if used, the voltages should be fed through unitygain buffers. More on that later. Note that you can’t get a buffered 1.25V output from pin 1 by tying pins 1 and 6 together, turning the op amp into a unity gain buffer. This is because the 2.5V tap is used for internal biasing. There are two pins we have not yet explained in Fig.1: pin 7 (CAP); and pin 5 (STROBE). Pin 7 is provided so you can connect a small capacitor (usually 10nF) between this pin and pin 6 (VBG), to lower the bandwidth of the internal op amp and reduce the output noise level. Pin 5 is provided to allow the AD584 to be switched on or off by a logic signal. If no current is drawn from pin 5, the device operates normally, but if the pin is pulled down to common/ ground, it effectively switches off. Now let’s look at how it’s used in the lowest cost module of our three. The ML005 module Fig.2 shows the full circuit of the ML005 module, plus the basic map of its PCB. As you can see, this module is essentially a ‘bare minimum’ design. It contains little more than the AD584 chip plus a few support components and some SIL headers used for input and output connectors, and for programming the desired output voltage. It uses the ‘JH’ version of the AD584 chip, so we shouldn’t expect too much from it in terms of output precision or temperature stability. Diode D1 is presumably to protect the AD584 from damage from reversed supply polarity, while LED1 and its rather high-value series resistor is to provide power-on indication. The 10nF capacitor connected between pins 7 and 6 of the device reduces the output noise level, while SIL header J5 allows setting the module’s output voltage by fitting a jumper shunt to one of the four possible positions. The current drain of the module when operating is less than 1mA, but this will rise if current is drawn from any of the outputs. Before we move on to look at the next module, you might like to know how easy it is to give the ML005 module three fixed and buffered outputs of 10V, 5V and 2.5V. Fig.3 shows all you need to do this: a low-cost dual op amp like the LM358 or the TL072, wired as shown to provide two unity-gain buffers. One is for the 5V output of the module, and the other for the 2.5V output. The 10V output is already buffered by the op amp inside the AD584, so it doesn’t need any further buffering. Note though that this buffer op amp’s ‘input offset voltage’ error term will slightly reduce the accuracy of the output voltages, although typically this figure is no more than a few millivolts. However, it can change with temperature and time. So if you need maximum accuracy, use a precision or chopper stabilised op amp, which will have offset voltages in the microvolt range. So is it possible to trim the outputs of the ML005 module, to set the output voltages closer to nominal? Yes, it is, using the trimming circuit shown in Fig.4. As you can see it’s fairly straightforward; just a 10kΩ multi-turn trimpot connected across the output from ML005 Precision Voltage Reference module Fig.2: the circuit and general layout of the basic ML005 reference board. It’s a minimalist implementation of an AD584based voltage reference, with pin header J5 provided to select the output voltage using a jumper shunt. Practical Electronics | July | 2020 23 J3 (VOUT) to J4 (0V), with a 10kΩ resistor in series and with its wiper connected to the 2.5V pin of J5 via a 3.3MΩ series resistor. This allows the outputs to be adjusted over the range of about ±20mV; more than enough to achieve calibration. The trimpot should be a 25-turn cermet unit, to allow fine adjustment and also provide a low temperature coefficient. The two fixed resistors should also be metal film types. The 3.3MΩ series resistor can be reduced in value for a wider adjustment range, but its value should not be lower than 300kΩ as this would adversely affect the module’s stability. The KKmoon module Now we turn our attention to the module with all the ‘bells and whistles’; the KKMOON (www.kkmoon. com/p-e0555.html). It comes housed in a laser-cut transparent acrylic case. The case can be easily disassembled for servicing, if needed. The designers of this module seem to have gone out of their way to add every feature they could think of. For a start, they’ve built in a 3.7V/500mAh lithium-polymer (LiPo) battery, so the unit can be used away from mains power. Of course, the battery will need to be charged when you are back in your workshop, so they’ve built in a charger as well, with a 5V input (microUSB socket). Since the battery only provides about 4.2V even when fully charged, they’ve also included a DC/DC boost converter to step up the battery voltage to around 13.5V for the AD584. They’ve also added circuitry so that the various voltage ranges of the AD584 can be selected in sequence using a single pushbutton switch and LEDs to indicate which output voltage is currently selected. The circuit (Fig.5) shows the parts they have added to provide all these extra features. The heart of the unit is still the AD584 (IC1). The ‘KH’ version of the AD584 is being used in this module – the one with performance specifications about twice as tight as those of the ‘JH’ version. All of the circuitry at the top and far left in Fig.5 is associated with the unit’s battery power operation. The Li-ion cell is charged via IC2 at upper left, using power from a 5V USB source fed in via CON1. IC2 is a Linear Technology LTC4054 charge controller, with pin 3 connected to the positive pole of the cell. The resistor connected from pin 5 of IC2 (PROG) to ground sets the charging current level, while pin 1 24 Giving the ML005 module three fixed, buffered outputs Fig.3: this circuit shows how to get multiple different reference voltages from the ML005 module simultaneously. While you could use a low-cost dual op amp as suggested here, the voltages would be more accurate and stable if a precision or chopper-stabilised op amp was used. Trimming the ML005 10V/7.5V/5V output levels Fig.4: it’s quite easy to connect a trimpot to the ML005 module, so that you can adjust its output voltages to be close to the nominal values. You need a very accurate voltmeter to do this. This will work with the output voltage set to one of the 10V, 7.5V or 5V options. of the device (CHRG) goes low when charging is taking place. It’s used to indicate when the battery is being charged, via LED1. The circuitry at centre and lower left is intended to protect the Li-ion battery from damage from overcharging or over-discharge. IC4 is a DW01P ‘Li-ion protector’ chip which monitors the battery voltage via its VCC pin (pin 5) and controls battery charging and discharging via pins 3 (CGO) and 1 (DGO), connected to the gates of Q8, an FS8205A dual N-channel power MOSFET. However, oddly, in the modules we’ve seen, the sources and drains of Q8 are shorted together by solder blobs, disabling the protection circuitry by permanently connecting the negative side of the battery directly to ground. Perhaps this has been done because the LTC4054 has its own protection circuitry, which may well be sufficient for this application. IC3 and its associated circuitry at upper right is the boost converter, which steps up the Li-ion battery voltage to around 13.5V, to run IC1. It’s a standard configuration using the MC34063A switchmode converter chip. MOSFET Q1 is used as an on/ off switch for the boost converter, and hence for IC1 as well. Practical Electronics | July | 2020 It’s controlled in turn by IC5, shown at lower centre, which is an unmarked microcontroller unit (MCU) in an 8-pin SOIC package. The MCU is also used to perform the output voltage switching of IC1, as well as the indication of the selected output voltage. This is all in response to presses of switch S1, connected between the ‘SW’ pin of IC5 and ground. Different outputs of IC5 are used to select the various output voltages available from IC1 by switching on one of the transistors Q5, Q6 or Q7, which then in turn switches on one of the P-channel MOSFETs Q4, Q2 or Q3. These latter devices perform the same purpose as the jumper shunt links on the ML005 module (see Fig.2). The LEDs indicating which voltage is selected are powered by the base drive currents for Q5, Q6 or Q7. Because none of the links need to be fitted for IC1 to deliver its 10V output (ie, all those transistors are switched off in this case), the MCU simply activates LED5 via its ‘10V’ output (pin 3) when that output voltage is selected. So the KKMOON module is much more complex than the ML005 we looked at first, which probably explains why it costs about three times as much. But it does offer a number of extra features, like portable operation and control using a single button. It also uses the superior AD584KH. Mind you, using a high-frequency step-up converter to provide the 13.5V supply for IC1 might increase the noise level, while using MOSFETs Q2-Q4 to select the lower output voltages might also turn out to have unexpected consequences. We’ll look at these aspects a little later. The unnamed module The third module is the one on a 50 × 50mm PCB, which carries no ID as such, but is marketed as a ‘high precision’ module. This is perhaps because it features SMA coaxial connectors for the three main outputs, and is also claimed to use the AD584LH chip, which has the tightest specs of all versions. The only aspect of the AD584LH which raises one’s eyebrows is that, as mentioned earlier, it was discontinued by Analog Devices in 2012, suggesting that the makers of this module either bought a large quantity before then and are still using them up, or that they have salvaged some from used equipment. That’s assuming they are genuine AD584LH devices, of course. The circuit for this module is shown in Fig.6. It’s much less complex than the KKMOON module, and only a little more complex than the ML005. It’s designed to run from 15-24V DC, fed in via J1, a standard concentric power jack. S1 is the on/off KKMOON AD584KH-based Precision Voltage Reference Fig.5: the circuit of the KKMOON voltage reference module is substantially more complicated, since it includes a DC/DC converter to boost the Li-ion battery voltage to a suitable level, as well as battery protection, a battery charger and output voltage selection via pushbutton S1. Practical Electronics | July | 2020 25 switch, while regulator REG1 derives a steady +12V to power IC1, the AD584LH. RF choke L1 and its associated capacitors ensure that the supply to IC1 is quite clean. LED1 provides a power-on indication. Apart from the use of SMA sockets for the 10V, 5V and 2.5V outputs from IC1, the rest of the circuit is similar to that of the ML005 module. However, there are two subtle differences, apart from the different AD584 version. One is that if you want a 7.5V output, this can be achieved by fitting a jumper shunt to SIL header P4. Then, SMA socket P1 delivers 7.5V rather than 10V. The other difference is that the three main outputs of IC1 are also brought out to four-pin header P2, together with a ground connection. This may not seem significant, but it does make it easy to connect a voltage-trimming adaptor like that shown in Fig.4 to this module. Trying them out When we received the three modules, we put them through their paces. In each case, we applied power and allowed the module to warm up and stabilise for about one hour. At the same time, we also switched on our very accurate Yokogawa 7562 6-1/2 digit DMM, and allowed it to stabilise as well. We then measured the four different DC voltage levels from each module, along with the noise levels, as shown in Table 2. Overall, the output voltages from each module were within the specifications that were given by Analog Devices for the AD584 version used in that module. In fact, the measured output voltages from all three modules were all within the specs given for the superior AD584LH device, with those for the ML005 and the KKMOON modules actually tighter/better than those for the module using the actual AD584LH. How surprising! The box for the KKMOON module came with a stick-on label listing the actual output voltages for that module as measured at 23°C using an Agilent 34401A DMM. These were shown as 10.00393V, 7.50163V, 5.00292V and 2.50014V. Our measured figures were quite close to these, as you can see. The ML005 module didn’t come with any equivalent figures, but the module using the AD584LH device had a similar stick-on label on the sealed plastic bag it was packed in. The KKMOON module has a LiPo cell mounted on This ‘high-precision’ the underside of the main PCB, which is held inside the acrylic case by two tapped spacers. module did not state the meter that had been Our measurements for the noise levused to make the measurements, but els from each module are somewhat they were shown as 10.004V, 7.503V, higher than the AD584 specs would 5.003V and 2.501V; again within the lead you to expect, although they’re AD584LH specs and also quite close still quite low. to the figures we measured. Fig.6: the ‘high precision’ voltage reference uses the more accurate AD584LH chip. Otherwise, it’s a pretty basic module, with a linear voltage regulator, power indicator LED and four different output sockets (P1-P3 and P5). With the exception of the 10V/7.5V outputs at P1 and P2, the others must be connected to very high impedance loads (eg, the inputs of CMOS or JFET-input op amps) to avoid inaccuracy. AD584LH ‘High-precision’ Voltage Reference module 26 Practical Electronics | July | 2020 This might be due to a shortcoming in the millivoltmeter used to make the measurements as its resolution below 1mV is rather poor. We were interested to see if there was any adverse effect on the output stability or noise levels of the KKMOON module outputs as a result of its use of MOSFETs to control the output voltage and that high-frequency DC-DC boost converter, but we couldn’t find any. The reference outputs of that module seemed to be just as stable and clean as those from the other two. Trimming the AD584LH The output measurements of the AD584LH-based module were a little disappointing, so we decided to try it out with a trimming adjustment adaptor. Fig.7 shows the adaptor circuit connected to the AD584LH module. The components were fitted to a small piece of ‘stripboard’, with the 25-turn trimpot at one end and a 4-pin SIL socket at the other, to mate with pin header P2 on the module. Using this simple adaptor we were able to adjust the 10.00497V output of the module down to 10.00003V at 26.4°C, with no increase in the apparent noise level. It was then left operating undisturbed for four hours, during which the ambient temperature rose to 27°C and the measured output fell to 9.99997V – a drop of only 0.06mV or 60µV. So our impression is that together with the trimming adaptor, the AD584LH module can be used to make a very stable and accurate voltage reference. Which to choose? If you just want a reference for checking 3-½ digit DMMs, analog meters and the like, the ML005 module would be ideal and has the price advantage over the other two modules. But if you want a portable reference for checking instruments ‘in the field’, the KKMOON module would be the one to go for. If you want the highest accuracy and stability, we’d suggest you choose the module based on the AD584LH device, together with the trimming adaptor circuit shown in Fig.7. This gives you a voltage reference comparable to commercial units costing over 10 times its modest cost of around £13.00. You can find a quick overview of the same three modules over at: www.markhennessy.co.uk/ad584_ references/ Trimming the AD584LH module output Fig.7: the voltage reference can also be trimmed with the addition of just four components. As this is the most stable of the references described here, it would make sense to adjust it to be as close to the nominal voltages as possible. It should then remain accurate in the long term. The alternative ‘highprecision’ AD584-based module. It uses an AD584LH as opposed to the AD584JH used in the ML005 module. However, when measured, this module displayed worse accuracy than the other two. Reproduced by arrangement with SILICON CHIP magazine 2020. www.siliconchip.com.au Practical Electronics | July | 2020 27