Gurobi solving SOCP problem

Hi,

I'm using Gurobi to solve an SOCP problem and have tried two different formulations.

In the first approach, I define the constraint directly as follows:

m.addConstrs(v[k,q,i,h]**2- (4*(1 + 1) * w0[q][0] 

                                       + 4*gp.quicksum(1 * w[q][i_prime][h_prime]*(x[i_prime,h_prime]**2) for i_prime in var_i for h_prime in var_h)
                                      + ((w0[q][0]+gp.quicksum(w[q][i_prime][h_prime]*x[i_prime,h_prime] for i_prime in var_i for h_prime in var_h))-(2-pi[k,q,i,h]))**2) >= 0 for k in var_k for q in var_q for i in var_i for h in var_h)

In the second approach, I introduce an auxiliary variable z3 to represent the third long term:

z3= m.addVars(K,Q,N,H, vtype=GRB.CONTINUOUS,lb=-gp.GRB.INFINITY, name="z3")

m.addConstrs((w0[q][0]+gp.quicksum(w[q][i_prime][h_prime]*x[i_prime,h_prime] for i_prime in var_i for h_prime in var_h))-2+pi[k,q,i,h]
             == z3[k,q,i,h] for k in var_k for q in var_q for i in var_i for h in var_h)


#SOCP

m.addConstrs(v[k,q,i,h]**2- (4*(1 + 1) * w0[q][0] 
                                      + 4*gp.quicksum(1 * w[q][i_prime][h_prime]*(x[i_prime,h_prime]**2) for i_prime in var_i for h_prime in var_h)
                                     + z3[k,q,i,h]**2) >= 0 for k in var_k for q in var_q for i in var_i for h in var_h)

The first formulation produces an optimal solution, but the second one results in an infeasible or unbounded model. I suspect there might be an issue with how I introduced z3.

Could you please help me check whether I made any mistakes in defining the new variable and modifying the constraint? Any guidance on how to fix this issue would be greatly appreciated. Thank you!