First, if you are a dissertator, that is not a bad time to have kids. The PhD can stretch out and while it might be painful, you don't have to get maternity leave.
Second, if you are going to be at a powerhouse school, get a little established first. But the problem is, fertility does decline with age. It's complicated, (see articles in The Atlantic and Salon), but it does decline.
My best advice would be, do it now, but I'd be very interested to hear what other mentors say.
This is always a challenging question, and always one where I try to remember that God is sovereign. Many couples who are great planners get surprised by an unexpected pregnancy, or a child with special needs, or twins, and others use their best abilities to plan and then discover that it takes much longer to get pregnant than they had anticipated. So my first suggestion would be to pray for your hearts to be submissive to God's timing.
Next, there are several recent resources that you could consult to learn about the impact of having children on training and career for women in the academy. Mary Ann Mason at Berkeley is one of the experts in this area, and her new book, co-written with Marc Goulden and Nicholas H. Wolfinger, Do Babies Matter? Gender and Family in the Ivory Tower might give you some perspectives in addressing your needs with your advisor or colleagues if you do get pregnant. (Mason summarizes research in a Slate article, “In The Ivory Tower, Men, Only.”) I found her earlier book, Mothers on the Fast Track: How a New Generation Can Balance Family and Careers, to be very illuminating. Definitely also read Sheryl Sandberg's Lean In for useful points about not planning so much ahead of time that you opt out of opportunities before you even have a baby.
Finally, I think it's a great sign that you ask about the time that "we" will need to care for an infant. If your husband expects to be an active parent who will help to support your work by doing hands-on caring for the baby, the likelihood is much greater that at any stage, you will be able to continue to work and make progress. If he is planning to be a stay-at-home dad, you obviously would have much more flexibility in terms of timing and work requirements. The pros for having a baby during training include more flexibility and less responsibilities, being younger (which is good for fertility and for lower risk of some birth defects and pregnancy complications), and the tenure clock hasn't started yet. However, it's likely that your financial resources are more limited. The pros for waiting until you have a "real job" include having accomplished the big task of getting your PhD, greater financial resources, and being settled in a new place. However, these days it seems like many academics take a long time to get that "real job."
Ultimately, expect having a baby (or two or three) to take much more time and emotional and mental energy than you could plan for. Do not expect to be able to care for a baby and work at the same time. You will need help, either paid childcare, or your husband or other family member caring for the baby, in order for you to get your work done.
I encourage people to use as much maternity leave as they can in order to spend as much time as possible with their babies, to give their bodies time to recuperate, to compensate for the sleep deprivation that will inevitably follow, to allow for breastfeeding which is easier and more efficient than pumping, and to just allow time to focus on the baby and not have to try and switch gears back into work mode.
If you're in thesis-writing stage, work is likely to stop, but in this stage you might be more able to take the time you want with your baby, and then, in time, shift back into work mode with child care. Of course at any point where there are classes or timetables — whether you are a student or a professor — you lose some of the flexibility to spend more time at home, and flexibility is like gold for moms with professional careers.
I wish you and your husband the best!
Biologically speaking, we know earliest is best. In the sciences, how easy the whole process will be totally depends on the advisor and your relationship with the advisor. Most people will have a very good idea of how the advisor will react to a student having a baby. Or maybe I should say that you will likely know if it is not a good idea to get pregnant in your current lab. In some cases it will be best to wait until the next stage in your career. If you are not sure, then I strongly encourage you to have the conversation with your advisor.
Assuming that you have chosen an advisor who would allow some time off and some schedule flexibility for having children, then I would try to be as strategic as possible. It is very difficult to predict how long a project and publication will take to complete (it is potentially even more difficult to time a pregnancy), but to the best of one's ability, it is best to aim for having a child near a natural transition time such as a paper submission.
In terms of the time an infant requires, I found the challenge was not so much the time spent caring for the child (infants can sleep a lot), but the mental energy required to remember all sorts of details like bottles, pumping, diapering schedule, keeping the diaper bag stocked, etc. Prior to having children, I could devote much more of that "organizational" energy to designing and carrying out experiments. The lack of uninterrupted sleep can affect ones' ability to concentrate too, making organizational energy all that much more valuable.
I found that trying to work experiments around the baby's feeding schedule or a pumping schedule took quite a bit of planning, but can be done. It requires creative planning, and it helps to have a strong support system. You might be surprised — having to keep a tight schedule might make you more productive and efficient.
Your work schedule will likely change. Unless you can find non-conventional, flexible childcare, your childcare hours will be approximately 7:30 to 5:30. It is very difficult to fit some types of lab work into that schedule. I have chosen research projects based in part on the hours that I knew I could spend in the lab.
One other note: there are reports that the number of DNA changes in the child's genome relative to the parents' genomes is largely dependent on the age of the father. This is attributed to the male making sperm throughout his lifetime (as opposed to the female being born with all the eggs she'll produce) and so the older the dad, the more risk that additional mutations have cropped up that will be incorporated into the sperm. It seems that dad's clocks are ticking too.