What does it take to derive the ideal gas law in themodynamics? 
How can the ideal gas law be derived from the following assumptions/observations/postulates, and these only ?
  
  
*
  
*I'm able to measure pressure $P$ and volume $V$ for gases.
  
*I notices that if two systems of gases come into thermodynamical equilibrium, that the quantities $PV$ for gas 1 and gas 2 coincide.
  
*I assume there is an energy function $U$.
  
*The first and the second law of thermodynamics hold by axiom, together with all the consequences, which can abstractly derived from the two.
  
*For a free expansion of my gas $(\Delta U=\Delta Q=\Delta W=P\Delta V=0)$ I find that the expression $PV$ for my gas is the same before and after the experiment.
  

If I'd postulate $PV=nRT$, then I'd easily see that $U(T,V)$ is only a function of $T$ and I can derive everything else, pretty much even without the second law.
Here I know of the second law, which gives me a bunch of expressions and I wonder if that suffices to identify the temperature.

If it's not possible, are there other experiments I can do in my position which would do the job?


Here is a weaker version of the question:

Can the ideal gas law be derived if I additionally know that $U$ is really only a function of $T$, or even that $U(T,V)=C_V T\ $?



And what role does the zeroth law of thermodynamics play here?

The essence of the question is not necessarily gas physics. It's rather about how temperature is defined by the laws of Thermodynamics, if there is merely the assumption that the system can be described by internal energy $U$ and is experimentally accessible only by its parameters like $P,V,M,H,\dots$. And there is no connection to a theory lying above. 
 A: According to the second law, thermal equilibrium between two systems means that they both have the same temperature $T$. The fact 2 that $PV$ coincide whenever two gases are at thermal equilibrium (and, I assume, for the same $n$) means that $PV$ is only a function of $T$. In othor words, there is a function $g()$ such that
$$\begin{align}PV&=g(T)&   P=\frac{g(T)}{V}\end{align}$$
The goal is now to show that $g()$ is linear, i.e. that $g(T)=Tg'(T)$. In order to show that, one can use a Maxwell relation, more specifically, the one which is linked to Helmholtz free energy $A=U-TS$ which can be defined because of condition 3.
We have then (including condition 4 for the computation of the derivatives)
$$\left(\frac{\partial S}{\partial V}\right)_T = \left(\frac{\partial P}{\partial T}\right)_V = -\frac{\partial^2 A }{\partial T \partial V}$$
From the first equation, we have 
$$\left(\frac{\partial P}{\partial T}\right)_V = \frac{g'(T)}{V}.$$
Except if $PV=g(T)$ is constant over a range of temperatures (which can be checked experimentally), the condition 5 implies that $U(T,V)=U(T)$. If one make as mall isotherm transformation of our gas, one has
$$\begin{align} 0=dU&=\delta Q - PdV& \delta Q=PdV\end{align}.$$ 
The second principle tells us then that
$$dS=\frac{\delta Q}{T}=\frac{P dV}T=\frac{g(T)}{TV}dV.$$
By definition, in this case $dS=\left(\frac{\partial S}{\partial V}\right)_TdV$, so we get this partial derivative from the last equation..
Equating the two partial derivatives according to the Maxwell equation then gives us $g(T)=T g'(T)$, which implies $g'(T)=nR$, where $nR$ is an "arbitrary" constant.
Hence,
$$PV=nRT.$$
A: Perhaps the following comments might be of interest.  A gibbsian thermodynamical system is specified by the equations of state $T=f(p,V)$ and $S=g(p,V)$ for two functions $f$ and $g$ which are constricted to satisfy the Maxwell relations.  The relevant thermodynamical quantities can then be expressed in terms of $p$, $V$, $f$, $g$ and their partials.  For the present discussion, we require $\left (\dfrac{\partial U}{\partial T}\right )_V$ and $\left (\dfrac{\partial U}{\partial V}\right )_T$.  These are can be easily computed to be $\dfrac{f g_1}{f_1}$ and $-p+\dfrac f{f_1}$ respectively (the subscript denotes partial differentiation with respect to the first variable $p$). (This is a simple computation involving  the chain rule and inverse function theorem---for a systematic treatment, see the arXiv article 1102.1540). From this we can draw several conclusions.
Firstly, $U$ depends only on $T$ if and only $\dfrac f{f_1}=p$ and this means that $f$ has the form $p\phi(V)$ for some function $\phi$ of one variable.
Secondly, $\left (\dfrac{\partial U}{\partial T}\right )_V$ is a constant if and only if $\dfrac{fg_1}{f_1}=c$, i.e. $g$ has the form $c \ln f+ \psi(V)$ for some function $\psi$ of one variable.
Thirdly, both of the above conditions hold if and only if $f=\phi(V)$ and $g=\ln p +  \ln \phi(V)$ for some function $\phi$.
If we combine these with the condition that the isotherms are as in Boyle's law i.e. are the hyperbolae $pV = c$, then we can deduce that the equations of state are those of the ideal gas.
A: Assumption 1 is standard.
Assumption 2 is not experimentally true.  It holds only in the
limit of very low pressure, but in fact holds for no real gas.
Assumption 3 is standard as is Assumption 4.
Assumption 5 is crucial.  It also is not experimentally true,
but what it says is that the energy is not a function of
volume.  Or, more exactly, it says that $(\partial U/\partial V)_{T,n}$
is zero.  This means that the energy is independent of the volume of
the gas and so that the energy is independent of the distance between
molecules.  If that is true, the molecules cannot interact with
each other, i.e. no attractive or repulsive forces.  And so such
a gas must be ideal.
Going further thermodynamically we end up with
$$\left( \frac{\partial U}{\partial V}\right)_{T,n} = T\left(\frac{\partial p}{\partial T}\right)_{V,n} - p $$
This expression is, of course zero for an ideal gas.
which can be applied to any equation of state to find out what
the change the energy of the gas with volume actually is.
