# Multichannel constant current box

#### enahs

Jun 15, 2011
10
I am looking for a multichannel box (12 outputs) that can output a very precise constant current ranging from 0.2 uA to ~200 uA with a voltage of up to 20V or so.

Would anybody know where I could get something like this from? Something controllable by a computer would be great.

#### poor mystic

Apr 8, 2011
1,074
I don't know straight off how to meet your requirements, but perhaps if you say a little more there might be something we can think of.
Just how precise a current do you want? Is there money available? Do you want something off the shelf or do you want to build it yourself?

Last edited:

Jun 10, 2011
443
A constant current source is fairly easy to build with an op amp that drives e.g. a MOSFET (see here for an example). You need to specify what you mean by "very precise" though.

You also need to specify how much money you're willing to spend. I could likely easily meet your needs by telling you to buy 12 Keithley 6620 current sources, but unless you're an oil sheik, you'll probably complain about the cost.

I doubt there's a relatively low-cost off the shelf solution. But an EE or experienced electrical hobbyist could probably build you a box that would do what you want and include e.g. a microcontroller that both set the current outputs and provide a serial interface. Depending on the specs, it would probably cost you 1 to 5 kilobucks US or so (typical design engineering rates in the US are in the $50-$150 per hour).

#### (*steve*)

##### ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,505
Firstly we need to know whether you want a highly precise, or a highly accurate current source (or some mix of both).

A highly precise 10uA current source might give you 9.5410 uA, but vary only by 0.001 uA over some load and line variation, temperature and time.

A highly accurate 10uA current source may give you 10.001uA under prescribed conditions of temperature, load, etc., but vary by 0.5uA under other circumstances.

In general, accuracy is improved by allowing an adjustment of the output current (in this case), whereas precision is enhanced by using highly stable components. They are somewhat at odds because the device used to trim the current may not be highly stable and thus may decrease precision.

OK, lets assume you want something realistic, perhaps within half of a percent or better.

To do this you need to ensure that the sum of all your tolerances (well, the ones which matter) are less than 0.5% in both absolute value, and over the temperature range you're contemplating (and also over time)

The simplest circuit uses an op-amp (driving perhaps a mosfet or bipolar transistor for higher currents) to maintain a fixed voltage across a sense resistor. The voltage it is compared against comes from a voltage reference.

The sources of loss of accuracy and precision are:

1) voltage reference accuracy and precision
2) voltage divider (if used) for voltage reference
3) input offset of op-amp
4) sense resistor accuracy and precision
5) op-amp input current (primarily from sense voltage)

Add to this the accuracy and precision of any current measuring device you have if you are going to calibrate the current source.

Let's say we can use a 2.5V 0.3% voltage reference. And lets say we can pick a 0.1% resistor that will drop 2.5V at the required current -- we've used up 0.4% already. We need to ensure that the effects of the input current and offset voltage of the op-amp do not add more than 0.1% more inaccuracy. Basically that means the input current for the op-amp needs to be as low as possible (0.05% of the set current) and the offset voltage less than 0.05% of the sense voltage.

Of course, if you know the characteristics of the op-amp, you may be able to allow for a relatively constant input current, and you may be able to trim away some of the offset voltage, but you still have to consider their variances over temperature (at least) and if you are trimming things, the accuracy and precision of your trimming.

So yes, the question remains: what do you mean by precision?

#### enahs

Jun 15, 2011
10
I need a highly precise constant current source. So if I set it at "10uA" but am only getting 9.768uA that is fine, as long as the fluctuations only vary slightly (0.002uA or so at most).

Ideally I am looking for a box that that has the 12 outputs, and each output is adjustable. I then want to have 7 of those boxes! Or equivalent number of total outputs.

I just can not seem to find any ready made device that can do it, or be easily modified to do it. We have a full electronics shop here and can do it, but they would charge around $6,000 per box and was hoping to find a cheaper solution. Ideally I would like something off the shelf that can do it, or that can be modified easily to do it. If a piece of electrical equipment has to be modified, it has to be done by the electronics shop (stupid rules and regulations here) and so hopefully it would be something that could be done quickly and easily. And ultimately, I would love it if these output settings were able to be controlled by a computer. Last edited: #### daddles Jun 10, 2011 443 Do you have the equipment to verify you're only seeing 0.02% variations in a 9 uA signal? It should have at least 10 times better resolution than the minimum variation you're after. To my knowledge, there isn't such a tool available off the shelf, other than such as the ones like the one I already mentioned. Thus, unless you have the skills and time available yourself, you're going to have to pony up the money to pay for someone to build it for you. Thirty years ago I worked at a well-known company in the test equipment area and their going rate for outside engineering work was$120/hour. That's \$5000 per week and that's not an unreasonable amount of engineering time to build a prototype of one of these current sources, document it, etc. (in fact, it might be a bit too low of an estimate). If you want things on a PC board and a computer interface, it will take more time and money. You could pay to have the first one developed, then have a cheaper outside shop duplicate the units.

Or, you could spent some time to learn a little electronics yourself and fiddle to make a current source (there are lots of circuits out there on the web). That's what I had to do, as I'm a scientist, not an EE -- but I wanted electrical things and didn't have the cash to pay someone to make them for me.

#### enahs

Jun 15, 2011
10
I am a scientists too, but the lab I work out I am not certified to do the electronic work and thus would not be allowed too even if I took the time to make them.

We already have some similar boxes that are 20 years old, but the guy who was making them for us has passed away.

Though, my numbers were overly extreme, I would be happy with 0.05 uA precision.

Was just hoping their was something commercially available.

#### poor mystic

Apr 8, 2011
1,074
Perhaps you'd do well to inquire in Russian and Chinese as well.

Jun 10, 2011
443
ehahs, where do you live? If you're in the US, I might be able to undertake the design and construction work for you. I can't claim to be an expert, but I have a number of EE friends locally and I'd imagine the wizards on this board would be willing to help with advice if/when I get stuck. If you'd like to talk more about this, please PM me.

Sounds like you might be doing some electrochemistry or materials research stuff...

#### poor mystic

Apr 8, 2011
1,074
This is becoming a very interesting thread. Here I was trying to think of a way to do the work and I'd got as far as wanting to keep a box at an elevated constant temperature with DAC inside tied to a bandgap reference.

Jun 10, 2011
443
Again, I'm a scientist, not an EE, so I'd need help from you guys if I do undertake such a task. But I've built a current source using an op amp and a MOSFET; I used a voltage reference IC and a voltage divider to generate a reference voltage for the op amp. Then the other op amp input came from a shunt resistor in series with the current-controlling MOSFET. The basic idea came from Maxim's app note 106, but there are numerous other similar current sources on the web.

The nice thing about this circuit was that I was able to get a constant current for microamps to amps -- it simply depended on the size of the shunt resistor. I'd have to build a prototype and test it to see what kind of stability it would have. The one I built is stable to at least 0.1%, as that's all I needed. I had some gorgeous Daven resistors that don't change their resistance that I used as the shunts (two of the 0.4 ohm resistors in parallel). Crappy sand resistors from Radio Shack suck. So finding a good low tempco shunt resistor is one requirement. Ditto for the voltage divider. I'm sure you experienced EEs have some other tricks up your sleeves for dealing with problems from temperature changes.

Yes, I had thought of the same thing (a closed-loop controlled oven) to maintain the voltage reference at a constant elevated temperature. I believe National has at least one bandgap reference source that contains an internal heater for this purpose, so I'd look at that first (and it has a much smaller time constant than anything I'd build from scratch).

#### poor mystic

Apr 8, 2011
1,074
Then of course enahs' requirement for digital control has to be met. That's not easy to visualise. I thought I was on the track of a way to use a commercial DAC and current mirrors at first but now I think I was dreaming.
What, then? How about digitally-controlled, mercury-wetted relays and extra-nice milspec resistors? Scary to get mercury hot but there is a safe zone and I can't quite believe in dry contacts, which will surely develop changes in resistivity, and ruin repeatability.

Jun 10, 2011
443
If the voltage reference/op amp/MOSFET/shunt method is the way to go, then to get digital control might mean to put a digital pot in to make the voltage divider or a commercial DAC. Or, even better, some pot/DAC with an I2C interface, as then a low end PIC could be used to talk to the 7 current sources in a box and a TTL serial interface to the chip could be used.

But the OP hasn't piped up with enough specs, so it's mostly speculation so far. It would be a fun project (I love current sources), but would probably tax my skills.

#### poor mystic

Apr 8, 2011
1,074
One thing we haven't considered here yet is the method by which current would be measured. I presume it'd be best to use an inamp across a series resistance in the feed?

Next question: How quickly does Enahs want to modulate his current source?

Speculation: If it's a chemical or biological load it won't be a pure resistance - it'll change with supplied current. Some algorithm would have to be decided on for sampling the current and correcting the applied voltage.

... of course you are correct, it's all too speculative until we hear back from Enahs.

#### enahs

Jun 15, 2011
10
Sounds like you might be doing some electrochemistry or materials research stuff...

Exactly. I am in the US; but with this being a government institution, even though I have a decent amount of money to spend, there are so many rules and regulations as to what and how we purchase it is ridiculous.

#### enahs

Jun 15, 2011
10
One thing we haven't considered here yet is the method by which current would be measured. I presume it'd be best to use an inamp across a series resistance in the feed?

Next question: How quickly does Enahs want to modulate his current source?

Speculation: If it's a chemical or biological load it won't be a pure resistance - it'll change with supplied current. Some algorithm would have to be decided on for sampling the current and correcting the applied voltage.

... of course you are correct, it's all too speculative until we hear back from Enahs.

Not very quickly.

Essentially we put this small amount of current over some electrodes that are placed in solutions, and over weeks or months small crystals grow on the electrode. Yes the resistance of the electrode changes over that period, but VERY slowly.

#### poor mystic

Apr 8, 2011
1,074
Perhaps Enahs could give us some information.
I think we need to know
1) how often he wants to change the settings on his machine
2) how quickly the change in effective resistance in the load must be corrected for
3) how does he want to communicate with his machine?

#### poor mystic

Apr 8, 2011
1,074
So.
That sounds like enough information.
Thank you Enahs.

#### enahs

Jun 15, 2011
10
Digital controls would be nice, but are not a deal breaker if their is something that can be purchased ans easily modified.

As I said, we do have a complete electronics shop and have money. But we also want/need A LOT of these sources (12*72), and they would just charge more then we can afford. But if it was modifying an existing device, we can do that.

#### enahs

Jun 15, 2011
10
Perhaps Enahs could give us some information.
I think we need to know
1) how often he wants to change the settings on his machine
2) how quickly the change in effective resistance in the load must be corrected for
3) how does he want to communicate with his machine?

1) At most, once a day.

2) 1 hours would be acceptable.

3) Communication with the box is not a deal breaker, but would be nice.

P
Replies
15
Views
3K
Fred Abse
F
Replies
1
Views
731
M
Replies
3
Views
705
JosephKK
J
W
Replies
1
Views
2K
Martin Griffith
M