Tiny PCBs

[Mr Evil]

The smallest PCBs I have ever made. One constant current source and one voltage reference. Details can be found on this page.

If anyone can think of a way to make the current source variable by only one resistor without compromising the performance, I'm all ears...

[cunningham]

Ahhh surface mount...tiny


This is one way I too have found to make a complicated circuit fit on a small board. Because you don't need lots of current for things like bias current sources, switches and other circuit controling features, SMD's can be used and save lots of space...if you can solder the small parts without going crosseyed!

They are a pain to change out sometimes for when you screw up and use the wrong part.

[Mr Evil]

I first saw it done a few years ago when I disassembled an old receiver and saw it had separated out quite few circuits onto small daughterboards made with SMT components.

Since then I've tended to try to do the same, but every time I need to solder or desolder things I vow never to use SMT again!

[Andreas W]

The best way to make a constant current device in ou r line of work, is to use a JFET, gate to neg terminal and a variable resistor from source to neg terminal, in case of a N fet. drain is positive terminal. There are a bunch, not so big nowadays of JFETs, out there. All of these have their own geometry rendering them different parameters. My favourite is is siliconix J201, pulling dowm 0.5mA approx at Vgs 0V. This is a really low noise maker, sounding very uninteresting to Linn ears, ie extreme low distorsion and broad bandwidth. In differential amps they need to be matched, but that is a walk in the park given you buy at least 25 and match them to Vgs at a certain Id. They really sound nice, soft and clear. And JFets are sturdy, they don't have fragile oxide layers that are punctured by a few dozens of volts. they can take abuse like a BJT. Actually they were invented long before the BJTs. Use them, you will like them. See siliconix, homesite for data and tips, for these underdogs of modern sound amplification. They are also useful for switching analog signals, where their sturdiness comes in handy. Excuse me for linguistic errata, since we had a somewhat wet dinner to night.

Andreas

[jwb]

Where are you going to get a transistor with an Hfe of exactly 315.3? Ditto what the last guy said: a JFET or cascoded JFET will make a spiffy current source, and they come in handy 2-terminal packages.

[Andreas W]

The voltage reference is NOT a reference, it is a thermometer. The cirquit is called Vbe multiplier, and is used in supplying a variating bias to power amps that keeps track of temperature. Temperature is one great big evil when it comes to making DC stable cirquits, and this cirquit is not the answer to that. If you don't believe me invest in a breadboard and measure your reference voltage, then pinch the transistor whith your 37 C body and notice the refreence change, on a dinky DVM. You will notice that this is not what would be called a reference in any sense of the word. Better try with a bandgap ref or a buried zener LM399.
There are dozens of refs out there with temp-dependicies of 10ppm/ K.

Andreas

[Mr Evil]

Originally posted by Andreas W
The best way to make a constant current device in ou r line of work, is to use a JFET...

Certainly seems like a good idea. I tend to avoid JFETs because of very poor availability and higher cost (5-10 times a BJT), but I'll look into it. The simplicity is appealing.

Originally posted by jwb
Where are you going to get a transistor with an Hfe of exactly 315.3?..

The simulator I use just happens to display hfe of the model used; the exact value is not critical to either circuit. Since they are intended to be variable, no transistor parameters are critical.

Originally posted by Andreas W
The voltage reference is NOT a reference, it is a thermometer...

I did do just that test on it, hoping that it would go some way to explaining the difference in output impedance between reality and simulation. Strangely the voltage did not change at all. In fact that's another thing that differs from the simulation, which does show change with temperature. But anyway, yes it could be a lot better.

[Steven]

Originally posted by Mr Evil
The smallest PCBs I have ever made. One constant current source and one voltage reference. Details can be found on this page.

If anyone can think of a way to make the current source variable by only one resistor without compromising the performance, I'm all ears...

Hi Mr Evil,

W.r.t. constant current source, keep in mind that your circuit has two stable current settings. One is the one you wanted, the other is the zero current condition. These kind of circuits may not start up. In integrated circuits you often see a kind of start up current injector forcing a small current into the circuit, after stabilizing the current the injector is blocked by changing DC conditions of the current source circuit (e.g. a diode that passed the injector current is becoming reverse biased after start up).

To make the circuit variable with one resistor, just replace the top or bottom current source with a current mirror. But also then the zero current condition can happen.

Steven

[Mr Evil]

Originally posted by Steven
...These kind of circuits may not start up....

Hmmm... I can see how that could be so, but I would expect leakage currents to be enough to start it off. If it were made of MOS devices it seems that would be more likely to be a problem. Have you seen this happen with BJT based current sources in reality? Startup circuitry would require a larger PCB, negating one of the reasons I made it, so I wouldn't add it unless absolutely necessary.

Originally posted by Steven
...To make the circuit variable with one resistor, just replace the top or bottom current source with a current mirror. But also then the zero current condition can happen....

Thanks. It seems obvious now you've pointed it out, but sometimes the obvious solution is the hardest to see. I've attached a schematic of the improved design. A simple two-transistor current mirror is not accurate enough, resulting in low output impedance, so a Wilson current mirror is used.

[destroyer X]

I can see too much people discussing theories, and very feel people doing really diy things.

I feel happy to see that some can do it themselves, that some have the good and old real spirit.

My deep and sincere congratulations

regards,

Carlos