My Voltage Standard project

[tron9000]

So I recently just finished this project:

This is my 6 selectable output voltage reference based on an AD780 and IN134. Schematic attached. I'd been watching some of Scullcom Hobby electronics videos on Youtube and I wanted a reference powered off AA batteries, with a number of selectable outputs of a fixed value. I was on a next to nothing budget - so all the parts I wanted to use were what I'd collected. I wasn't sure about accuracy to be honest,I just picked the most accurate and stable reference I had in my parts box - the AD780BN - and would determine the accuracy after I'd built it.

I also wanted an auto-off feature and one push on button and for it so stay on for 3-5mins. I also wanted it to be portable and fit nicely on my bench. Battery life: about 4 hours non-continiuosly, so decided on 2S2P configuration for my battery supply, that meant using a DCDC converter that could work between 3.0V and 1.6V, though I'd be happy if worked on a minimum voltage of 2.0V. I also wanted a low battery warning LED.Naturally running directly off the output of the DCDC converter supply would not be ideal, so I would have to filter and regulate it with an LDO.

The AD780 had a selectable output by grounding or floating pin-8 on the AD780 - so either 2.5V or 3.0V by using a toggle switch (SW2). The INA134 had pretty good gain linearity. By reconfiguring the connections to the inputs and output feedback, I could get a gain of 0.5, 1.0 and 2.0. I did this using a rotary switch (SW3).

So this meant I could have the following voltage outputs: 1.25V, 1.5V, 2.5V, 3.0V, 5.0V & 6.0V

I built up the DCDC converter based on an LT1073. I needed a supply rail at least 1.6V greater than my maximum required voltage output. with some inductor selection teething and resistor selection I settled on an output voltage of about 8.7V after filter and regulation. A bonus of using the LT1073: it had a free error amp built in, this became the low battery warning circuit. The regulator is an LT1121 LDO and I used a PI filter to get rid of as much switching noise from the DCDC converter.

I then built a number of iterations of my auto-off circuit. The first of which were based on high side switching. these didn't work (rather couldn't get them to work) so decided on a low side switch. The circuits pretty simple: SW1 bypasses the FET Q2. By doing this  Voltage on the output of the regulator will have come on, the voltage on the cap will be 0V and the transistor Q1 will be conducting thus turning on FET Q2. The Base current through Q1 charges up C2 until the voltage on the cap is high enough that Q1 stops conducting, turning Q2 off. When Q2 turns off the voltage on C2 discharges through R9, D4, R10. It works well enough for my requirements, however just before it switches off, the current drawn from the supply increases and the low bat light comes on, call this a feature: warning the the unit is about to power off.

One of the iterations:

Once I assembled my reference and amplifier circuit I first measured the output at the reference and confirmed I could switch between the 2 voltages on the fly, which I could. Then I measure the voltages at the output of the amplifier and went through the different switch positions. At lower voltages I couldn't get below 1.35V in the lowest voltage setting (1.25V). This was cos the output of the amp was hitting its minimum rail which is about 1.3 to 1.4V above 0V. This meant sticking in an inverter circuit. An ICL7660 IC  was kicking about so hooked that up and connected it up to the amplifier negative supply rail. A retest and I was able to get 1.25V.

So before committing my design to veroboard, I gave it a quick teat to see how the accuracy was. I had access to a HP 34401A and measured the output voltages:


Test setup:


I then stuck it all on veroboard and in a convenient case that already had a battery compartment and aluminium panel, which I drilled some holes for LED's, switches and connectors. Yes connector spacing is 19.05mm or 3/4".

After sticking it all in a box I gave it another test, this time using a Keysight 34972A DAQ at work, fresh a month out of the box! Here are the results:


Test setup:


Seems that 2.5V based outputs are less accurate than 3.0V. Not sure why this is, but either way I'm happy with the performance. I printed out my results and stuck them to the box for reference:

I plan to test my unit against temperature also using the DAQ at a later date.

I'm pretty happy it turned out anyway, thought I'd show and tell it, and see if anyone can offer any insight into anything I could have done better, or I did wrong - feel free to pick at it.

[tron9000]

any feed back would be greatly appreciated. Many thanks.

[bktemp]

Using a 7660 is a bad idea: It is probably the worst power suppy I can imagine. It generates a large current spike at every switching cycle at both the input and output voltage. Since there is almost no capacitance on Vcc there is probably quite some ripple. An inverting buck-boost converter connected to V+ should be less noisy. If you want to keep your existing charge pump, move it far away from the output as possible and add some filtering at both input + output to keep the switching noise away from your voltage reference and the output amplifier.

[tron9000]

well there is some capacitance on Vcc (~1u43F in the form of smoothing and decoupling) but are you suggesting there should be more?

I have actually placed the 7660 as far from the amp and reference as possible, as I did have concerns about this:

on the left had side of the board is the AD780 and the IN134, bottom right is the 7660 circuit.

So in terms of filtering: another two pi filters? One on input and output of 7660 circuit?

thanks for feedback.

[bktemp]

There was a thread some time ago showing how bad a 7660 is and trying to find ways to clean the noise:
https://www.eevblog.com/forum/projects/an-evening-with-the-icl7660/

From my experience you need a lot of capacitance to get rid of the ripple produced by a 7660, because the switching frequency is rather low.

[tron9000]

 Eek! That's shocking! although it appears to get better with the application of some ferrite beads and some choice passives. Will have a look through that thread for a better insight.

That's a really useful post is that! I did look at the VDD supply rail before and I have to admit I wasn't quite aware that the noise was that bad, but put it into the prespective of that post: this could be kicking the performance of my output right out the window.

Never come across a capacitor multiplier before either - useful circuit to know!

[max_torque]

A nice little project.  Can be great learning experience as well as making a useful little tool to use as well!

One thing i have to say is that these days, i've pretty much given up doing analogue circuits for general tasks, like soft on/off buttons etc.  You can buy small 8b microcontrollers for very low costs, and these need just about no external components beyond a simple bypass capactior, and make software functions, like pressing a button and staying on for 5mins or whatever so, so simple.  I've got a box of ATtiny25's for just such uses, and i use them all the time now, rather than spend hrs messing around with caps and transistors etc!

[ZeTeX]

A nice little project.  Can be great learning experience as well as making a useful little tool to use as well!

One thing i have to say is that these days, i've pretty much given up doing analogue circuits for general tasks, like soft on/off buttons etc.  You can buy small 8b microcontrollers for very low costs, and these need just about no external components beyond a simple bypass capactior, and make software functions, like pressing a button and staying on for 5mins or whatever so, so simple.  I've got a box of ATtiny25's for just such uses, and i use them all the time now, rather than spend hrs messing around with caps and transistors etc!

But analogue is fun!
programming a micro controller is boring.

Great project! I would not use any converters, only batteries.

[max_torque]

A nice little project.  Can be great learning experience as well as making a useful little tool to use as well!

One thing i have to say is that these days, i've pretty much given up doing analogue circuits for general tasks, like soft on/off buttons etc.  You can buy small 8b microcontrollers for very low costs, and these need just about no external components beyond a simple bypass capactior, and make software functions, like pressing a button and staying on for 5mins or whatever so, so simple.  I've got a box of ATtiny25's for just such uses, and i use them all the time now, rather than spend hrs messing around with caps and transistors etc!

But analogue is fun!
programming a micro controller is boring.

Great project! I would not use any converters, only batteries.

I guess it depends which bit of the project you want to invest the most time and effort in.  For something like a voltage standar  :-DDd, i'd spend the time getting to a decent standard of standard, and not in making useful, but boring, things like the power button work!

[tron9000]

I'm with ZeTeX on this one, but maybe its just the way people like to do things. Me: I like the analogue route and now I've developed a circuit I know works and keeps my skills sharp.

I did consider using PP3's, but I find their capacity lacking and I can get AA's for cheap <taps nose>. I do have some 18650 lithium cells salvaged from old laptop batteries, but have yet to validate their condition (am currently constructing a roughcut charger).

I thought that if I used enough filtering I'd overcome the supply switching, but it appears I've not used enough. I think if I can get some lithium cells charged and validated, then I think just 4S configuration would be enough for capacity and enough voltage for the INA134 to operate.

However I found some max4198 differential op-amps in my spares bucket, that are identical in construction to the INA134 and are rail to rail, so could do away with the 7660 circuit to widen the amps operating range. This will allow me to power directly off battery voltage. However will that be a good trade off against gain stability Vs. temp and non-linearity, which the INA134 does seem to have better performance. Just trying to work with what I have.

thanks

[CraigHB]

One thing i have to say is that these days, i've pretty much given up doing analogue circuits for general tasks, like soft on/off buttons etc.

I also tend toward that.  Basic 8 bit ucontrollers are inexpensive and I have a bunch laying around.  I typically do projects for some specific purpose and it's usually faster and more simple to program an MCU to do a task.  Plus there's flexibility you don't have with an analog circuit.

Switching regulators can be problematic when you need a really clean output.  It's better when switching frequencies are higher allowing the use of low ESR and ESL chip capacitors.  The high value electrolytics required with a lower switching frequency have a relatively high ESL.  It can make a clean output rather elusive.  You need an array of capacitors to mitigate those reactive effects and then it's not worth the trouble.  Better to just start with a higher switching frequency.  I usually avoid anything under half a mega-hertz, though there are trade-offs there as always with circuit design.

[max_torque]

One thing i have to say is that these days, i've pretty much given up doing analogue circuits for general tasks, like soft on/off buttons etc.

I also tend toward that.  Basic 8 bit ucontrollers are inexpensive and I have a bunch laying around.  I typically do projects for some specific purpose and it's usually faster and more simple to program an MCU to do a task.  Plus there's flexibility you don't have with an analog circuit.

Precisely.  Had too many meetings where someone says "Can you just" and everything has to be totally reddone if it were analogue!  For example, say you decide you want to have a double button press make the unit not autopower down?  You analogue circuit is now useless, whereas with probably 3 additional lines of code, and about 10min, i can add that to my logical power controller!  or maybe you want to flash the power led for 30sec BEFORE it turns off so the operator knows it's about too, or a million other functional improvements!

I love fiddling round with analogue stuff, and you learn a LOT doing it, but there is a reason digital logic has taken over the world 

[CraigHB]

I love fiddling round with analogue stuff, and you learn a LOT doing it, but there is a reason digital logic has taken over the world

I think you'd be hard pressed anymore to find an electronics product that does not have a micro-controller at heart.  Still analog stuff will always have a place and it is more interesting to work with.  Writing code can can get boring and monotonous.

[Cyberdragon]

I love fiddling round with analogue stuff, and you learn a LOT doing it, but there is a reason digital logic has taken over the world

I think you'd be hard pressed anymore to find an electronics product that does not have a micro-controller at heart.  Still analog stuff will always have a place and it is more interesting to work with.  Writing code can can get boring and monotonous.

Voltamort does not like it when all this digital gubbins randomly malfunctions due to shitty chinese programmers.

"If you spit out ONE MORE random letter with an error...!"

G
Error 2938828

*GASP*