I've built one for ~ $30 in parts from Radio Shack using an LM317 current limiter. There is an odd bootstrapping problem involved with tDCS devices: you need to have the device connected to comprehend the instructions! Researchers speak of anodal and cathodal stimulation, and the convention is that + is anodal and - is cathodal. So far so good--at least this comports with vacuum tube conventions, which will be familiar to all readers of Hacker News. Researchers speak of anodal stimulation of the left dorsolateral prefrontal cortex. The tDCS device has to be connected (with the positive connection over the left dorsolateral prefrontal cortext) to even begin to comprehend that left means left from the perspective of the subject, but right from the perspective of a viewer viewing the subject's face. The right negative contact can go over the right eyebrow (but there are other protocols). This is one of the clearest: http://www.ncrrn.org/papers/symposium_tdcs/hamilton_tdcs.pdf.

As you can see, there is an inviolable convention among researchers not to unambiguously mark the anode + contact on the left dorsolateral prefrontal cortex, and the cathode - contact over the subject's right eyebrow. Instead, what we see on the third slide is a cartoon drawing of a subject in which two contacts are labeled "active" and "reference". There is an arrow leading from the active contact to the reference contact labeled "anodal" and another arrow from the reference contact labeled "cathodal". This refers to an ancient convention in electronics, in which current was assumed to flow from the anode to the cathode, even though electrons flow from the cathode to the anode. The literature is replete with such helpful diagrams. One would like a completely unambiguous statement: the anode is the positive connection, and it is secured to the left side of the subject's scalp; the cathode is the negative connection, and it secured over the subject's right eyebrow. Or one of the eight combinations possible by arbitrarily switching anode with cathode, positive with negative, and left with right. But I have been unable to locate a statement in the literature that would condescend to so direct and simple a description.

Here is another article with a diagram showing node placement for tDCS. http://www.sciencedirect.com/science/article/pii/S0165017306.... Note that the anode is placed over the left dorsolateral prefrontal cortex, and the cathode is placed on the forehead. That beats the anodal and cathodal arrows of the previous diagram. What we still don't have is the identification of + with anodal and - with cathodal. And so one must continue searching through the literature--which should be done in any case--until one comes to something approximating a consensus on node placement and the meaning of anodal and cathodal versus positive and negative.

Finally, here's a compilation from a less than unimpeachable source: http://www.drmueller-healthpsychology.com/tDCS.html. The author misreads 1mA as one microamp. That's one milliamp. But we do get a straightforward statement about node placement from Fregni, F., Boggio, P., Nitsche, M., et al. (2006). Letters to the Editor: Treatment of major depression with transcranial direct current stimulation. Bipolar Disorders, 8:203-205. This is confirmed in http://www.ncbi.nlm.nih.gov/pubmed/22031874, though you have to know that anode means positive and cathode means negative.

I should say that in my initial reckless period of self-experimentation, I managed to induce phosphenes by accident -- blue white flashes in the entire visual field, blanking out everything else. Both contacts were in the supraorbital region. I ceased my experiments for a while and returned to the literature.