Need much more information - this is a very specific question. I have spreadsheet built to compute this exact figure for my wife and I, and for our current situation, it comes out that I need to send in 22.6% (I round up to 23%) of my check, however you could have a very different situation than us:
1. We live and work in a state with no state income tax - if you are in a state with a state income tax, odds are you need to hold back more for taxes than we do.
2. I put 18.5% + 17,500 into a solo 401K - if we didn't do this, we'd need to send in significantly higher estimated taxes
3. My wife is maxing a 457 plan through her job - if she didn't do this we'd need to send in more estimated taxes, and her withholding would be higher
4. If my wife had filed her W-4 differently, that can change how much I need to send in
My spreadsheet assumes that we take the standard deduction - if you have high enough itemized deductions to itemize, you should probably use that figure instead.
For us, Items 2 and 3 there are extremely valuable - without those, the amount I would need to send from my check would be something like 33% - that is a big difference.
So project your income for the calendar year, factor in any deductions, apply applicable taxes (remember to include self-employment taxes), then divide the total by the income. If you are worried about the accuracy of the projection, use a high estimate for your income and a low one for any deductions. I've noticed that in our case, I think largely due to the 18.5% I'm putting into the Solo 401K, my percentage comes out to between 22 and 23% over a large range of incomes - your situation might be more or less sensitive to inputs than mine though.
Or you could just hold back 40-50% and almost certainly be safe, however this is probably way too much, unless you are in a high-income tax state and have a high income. Remember too that you have to send in your estimated taxes quarterly (at least for federal) - don't just save the money for taxes, remember to send it in on time as well.